Dataset Viewer
title
stringlengths 1
827
⌀ | uuid
stringlengths 36
36
| pmc_id
stringlengths 5
8
| search_term
stringclasses 44
values | text
stringlengths 0
2.08M
|
|---|---|---|---|---|
Clinical outcomes following total hip arthroplasty for bony ankylosed hips: a propensity score-matched analysis | c28db74b-3f6e-470f-8fca-e42d3b278c6d | 11849327 | Surgical Procedures, Operative[mh] | Patients with spontaneous ankylosis due to hip arthritis or those who have undergone arthrodesis in adolescence for hip disease often achieve pain relief but experience difficulty with activities of daily living and disability of adjacent joints, such as the lumbar spine and knee . Therefore, pain due to adjacent joint disorders caused by ankylosed hips is an indication for total hip arthroplasty (THA) for ankylosed hips . THA is known to be one of the most useful treatment procedures for hip arthritis . However, the postoperative outcomes of THA in patients with bony ankylosed hips are nonconclusive. To our knowledge, no studies have compared clinical outcomes following THA in patients with bony ankylosed and non-ankylosed hips using propensity scores. In this study, we examined the postoperative results of THA for bony ankylosed hips using propensity score matching.
This retrospective cohort study used data obtained from the THA database of our institution. Data obtained included the Japanese Orthopaedic Association (JOA) hip score, laboratory data, postoperative complications, and computed tomography (CT) images. The study protocol was in accordance withSaga University Hospital (Reference number: 2020–06-R07). All patients provided written informed consent prior to participation. Patients Propensity score matching Surgical procedure Surgical indication Types of implants Outcomes Statistical analyses Statistical analyses were performed using JMP Pro software (version 15.2.0, SAS Institute Inc., Cary, NC, USA). Categorical variables were expressed as absolute and percentage values, and continuous variables were expressed as mean ± standard deviation. The Shapiro–Wilk test was conducted to evaluate the normality of the distribution of the continuous variables. A two-tailed F-test was used to evaluate variance. The Wilcoxon signed-rank test was used to compare the postoperative JOA hip score at the last follow-up and post-propensity score matching with the preoperative and pre-matching values, respectively, in the same group. Comparisons between the bony-ankylosed and control groups were performed as follows: a two-tailed Student’s t-test was used to compare the post-propensity score-matched height, weight, preoperative total JOA score, haemoglobin level, and alkaline phosphatase on the day prior to surgery and postoperative day 1. A two-tailed Welch’s t-test was used to compare the pre-propensity score-matched height, body mass index, postoperative blood loss, and creatine phosphokinase level on postoperative day 1. The Mann–Whitney U test was used to compare the preoperative JOA hip score, except for the total score, postoperative JOA hip score at the last follow-up, preoperative and postoperative use of walking aids, length of hospital stay, follow-up period, operating time, intraoperative blood loss, allogeneic blood transfusion, laboratory data except for the haemoglobin level, alkaline phosphatase on preoperative day and postoperative day 1, and creatine phosphokinase level on postoperative day 1, and postoperative CSA of the gluteus medius and CT values at that site. Post-hoc analyses were performed for the postoperative total JOA hip score at the last follow-up (effect size d = 1.13, two-sided alpha = 0.05, sample size = 40 and 40), CSA of the gluteus medius (effect size d = 1.88, two-sided alpha = 0.05, sample size = 11 and 8), and the CT value of the gluteus medius (effect size d = 1.39, two-sided alpha = 0.05, sample size = 11 and 8), resulting in calculated power values of 0.99, 0.96, and 0.79, respectively. Regarding CT imaging studies, intra-rater reliability was assessed by two measurements recorded by one orthopaedic surgeon 1 week apart, and inter-rater reliability was assessed by measurements recorded by two orthopaedic surgeons. The intra-rater intraclass correlation coefficient (ICC) and inter-rater ICC values were 0.9448 and 0.8133 for the postoperative CSA of gluteus medius and 0.9446 and 0.8512 for the postoperative CT values of gluteus medius at CSA, respectively, all of which were sufficiently reproducible to be classified in the first class by Wheeler’s Evaluating the Measurement Process method . Following Cochran’s rule, Fisher’s test was used for the number of hips that underwent revision surgery and time of postoperative CT imaging, where more than 20% of the squares had an expected number less than five . The chi-square test was used for to analyse sex, preoperative and postoperative use of walking aids, number of hips with allogeneic blood transfusions, number of hips with postoperative complications, where no squares had an expected number less than five . Survival analysis was performed using Kaplan–Meier and log-rank tests for the acetabular side, femoral side, and both sides of the prosthesis, with total hip arthroplasty revision as the endpoint. The relationship between intraoperative blood loss and other variables, including the type of ankylosis, was evaluated using Spearman’s correlation coefficient in the bony-ankylosed group. For all analyses, statistical significance was set at a p -value < 0.05.
Data were extracted for 3338 hips (2863 patients) that underwent primary unilateral THA at our institution between January 1999 and December 2011 (Fig. ). Cases with follow-up periods of less than 10 years and those with missing data were excluded. The hips included in the study were then temporarily divided into two groups: a bony-ankylosed group (40 hips, 38 patients) and a non-ankylosed (control) group (829 hips, 729 patients). Bony ankylosed hips were defined as cases with joint ossification on simple radiography and no range of motion (Fig. A). The remaining cases were defined as non-ankylosed hips.
To minimise confounding factors, propensity score matching was used to match bony ankylosed to non-ankylosed hips. Using logistic regression, propensity scores were calculated using the five variables of age, sex, height, weight, and body mass index, which were selected based on previous studies . Propensity score matching was then performed using nearest neighbour matching without replacement, with each bony ankylosed hip matched to a control hip . We used a calipre width of 0.2, which is the standard deviation of the logit of the propensity score . To check the balance of the matches, a standardised mean difference threshold of 0.1 was set to determine the residual imbalances .
THA was performed using the posterolateral approach under spinal anaesthesia with identical cementless implants. In the bony-ankylosed group, 33 of the 40 hips underwent an additional approach from the anterior edge of the gluteus medius.
The main surgical indication for the bony-ankylosed group was pain due to adjacent joint disorders caused by the ankylosed hips and for the control group it was a decline in activities of daily living due to pain and limited range of motion in the hip joints. Walking training within the allowable pain range was initiated without weight-bearing limitations after drain removal on day-2 postoperatively.
In both groups, proximal hydroxyapatite-coated cementless femoral components with a proximal porous coating consisting of pure titanium (PerFix-HA femoral component; Kyocera, Kyoto, Japan) and hydroxyapatite-coated cementless hemispherical acetabular shells with a porous coating consisting of pure titanium (AMS-hydroxyapatite acetabular shell; Kyocera, Kyoto, Japan) were implanted. An alumina or zirconia ball and ABS (alumina ceramic inlay mechanically fixed to a polyethylene liner) or AMS (polyethylene liner) liner were used for the bearing surface (Fig. B).
The outcomes of this study were the JOA hip score and use of walking aids, preoperatively and at the last postoperative follow-up; length of hospital stay; follow-up period; operating time; intraoperative blood loss; postoperative blood loss calculated from a drain tube; allogeneic blood transfusion; number of hips with allogeneic blood transfusions; number of hips with postoperative complications; number of hips that underwent revision surgery; laboratory data obtained preoperatively and at postoperative days 1 and 7. Postoperative haemoglobin included data for patients who received allogeneic blood transfusions. The JOA hip score consists of four subcategories and is calculated on a 100-point scale: pain, 40 points; range of motion, 20 points; gait ability, 20 points; and activities of daily living, 20 points. Laboratory data, including platelet count and creatine phosphokinase levels, were evaluated using routine blood tests. Additionally, data were missing after propensity score matching; however, postoperative CT images were used to evaluate the gluteus medius on the surgical side in the transverse section, and the image information unification system ShadeQuest/ViewR (version 1.26.10; Yokogawa, Tokyo, Japan) was used to take measurements. A complete-case analysis was used for the evaluation. The midpoint of the line connecting the superior anterior iliac spine and the apex of the greater trochanter was defined in the frontal simple radiograph image of the pelvis for positioning when the CT image was captured. The cross-sectional area (CSA) and CT value of the gluteus medius were measured at the midpoint level (Fig. A and ) .
Propensity score matching Clinical results Laboratory results Radiographic results Survival of acetabular and femoral components The cumulative survival rates of the cup, stem, and overall, with revision as the endpoint, were not significantly different between the bony-ankylosed and control groups ( p = 0.5620, p = 0.1753, and p = 0.3378, respectively; Fig. A–C).
The propensity score-matched population consisted of 40 bony ankylosed (38 patients) and 40 non-ankylosed (40 patients) hips as matched controls. The area under the receiver operating characteristic curve for fitting the propensity score was 0.68. This value is an appropriate value for calculating the propensity score as it ranges from 0.6–0.9 and is not extremely close to 0.5 or 1.0, which are considered inappropriate . The standardised mean differences in baseline demographics for the matched study population are shown in Table and Fig. : all variables achieved an appropriate balance (standardised mean difference < 0.1). In the bony-ankylosed group, the mean duration of ankylosis, as assessed by preoperative interviews with patients, was 36.1 ± 19.0 years, with 22 cases of spontaneous ankylosis and 18 of arthrodesis.
Intragroup comparison Comparison between the bony-ankylosed and control groups Preoperatively, the JOA hip scores for “total” and “pain” in the bony-ankylosed group were significantly higher than those in the control group ( p < 0.0001 and p < 0.0001, respectively), whereas the scores for “range of motion” and “percentage of use of walking aids” were significantly lower ( p < 0.0001 and p = 0.0368, respectively). No significant differences were observed between the two groups with regards the JOA hip scores for “gait ability” and “activities of daily living” ( p = 0.5654 and p = 0.0563, respectively). Regarding the postoperative values, the JOA hip scores at the last follow-upwere significantly lower in the bony-ankylosed group than in the control group ( p < 0.0001, p = 0.0382, p < 0.0001, p < 0.0001, and p = 0.0083, respectively). The percentage of patients who used walking aids postoperatively, length of hospital stay, operating time, intraoperative and postoperative blood loss, allogeneic blood transfusion, number of hips with allogeneic blood transfusions, and number of hips with postoperative complications were significantly higher in the bony-ankylosed group than in the control group ( p < 0.0001, p < 0.0001, p < 0.0001, p < 0.0001, p = 0.0003, p < 0.0001, p = 0.0015, and p = 0.0021, respectively; Table ). Complications in the bony-ankylosed group included eight cases of heterotopic ossification (HO), with four grade 1, three grade 2 and one grade 3 in Brooker's HO grade ; and loosening of the acetabular side prosthesis; destruction of the femoral side prosthesis; periprosthetic joint infection; and dislocation in one case each. Complications in the control group included periprosthetic joint infection and dislocation in one case each.
In both groups, the postoperative JOA hip scores at the last follow-up had significantly improved compared to the preoperative scores for all components of “total”, “pain”, “range of motion”, “gait ability”, and “activities of daily living” (bony-ankylosed group: p < 0.0001, p < 0.0001, p < 0.0001, p = 0.0038, and p = 0.0001, respectively; control group: p < 0.0001, p < 0.0001, p < 0.0001, p < 0.0001, and p < 0.0001, respectively; Table ).
the bony-ankylosed group on postoperative day 1, the platelet and creatinine phosphokinase levels were significantly lower and higher, respectively, than those in the control group ( p < 0.0209 and p < 0.0034, respectively; Table ). A positive correlation was found between operative time and intraoperative blood loss; a negative correlation was found between age at primary THA and intraoperative blood loss, and between the duration of ankylosis before primary THA and intraoperative blood loss ( p = 0.0001, p < 0.0080, and p < 0.0289, respectively; Table ).
To study the additional outcomes, CT was performed in 11 and eight patients in the bony-ankylosed and control groups, respectively (19 cases). The mean postoperative CSA of the gluteus medius in the bony-ankylosed group was 1363.3.9 ± 495.1 mm 2 , which was significantly lower than that in the control group (2377.5 ± 580.2 mm 2 ; p = 0.0030). The mean postoperative CT values of the gluteus medius at the CSA in the bony-ankylosed group was −4.4 ± 28.6 Hounsfield unit, which was significantly lower than that in the control group (31.8 ± 23.0 mm 2 ; p = 0.0039). The mean time of postoperative CT imaging in the bony-ankylosed and control groups were 7.6 ± 3.5 and 10.1 ± 2.6 years, respectively, which were not significantly different ( p = 0.2437).
is study revealed two important clinical findings. First, except for pain, patients with bony ankylosed hips had improved JOA hip scores after THA; however, the pain scores were worse. Furthermore, the postoperative JOA hip scores in all subcategories and the rate at which the use of walking aids was not required in the bony-ankylosed group were significantly lower than those in the control group. Second, tThe results of the current study showed that although the total JOA hip score of the bony-ankylosed group improved, the pain score worsened: this disadvantage indicated some discomfort and fatigue in the hip that was originally pain-free. Compared to the control group, JOA hip scores were lower in the bony-ankylosed group in all categories. More than half the patients (52.5%, 21/40) with bony ankylosed hips used a walking aid postoperatively (odds ratio, 7.7) compared to 12.5% (5/40) of those in the control group. These results suggest that THA for patients with bony ankylosed hips can somewhat improve the JOA hip score, but not to the same extent as THA for patients without ankylosed hips. This should be preoperatively communicated to patients with bony ankylosed hips who undergo THA. Herein, a greater number of hips in the bony-ankylosed group had postoperative complications than those in the control group and the complications were characterised by HO. HO may be a consequence of the large number of bone fragments produced during osteotomy . Another possible reason is soft tissue damage during surgery. Soft tissue damage due to surgery provides an environment for osteoblasts to develop from mesenchymal cells, causing HO . We observed that blood creatine phosphokinase levels on the first postoperative day in the bony-ankylosed group were higher than those in the control group; creatine phosphokinase levels increase with soft tissue—including skeletal muscle—damage . We hypothesise that damage to the soft tissue during surgery was greater in the bony-ankylosed group than in the control group because of the significant adhesions and anatomical variations around the hip joint, which required a more extensive approach than in normal THA procedures. THA for an ankylosed hip requires a higher level of surgical skill for osteotomy and soft tissue approach and a longer operative time than normal THA . In the current study, intraoperative blood loss, operative time, and blood transfusion were higher in the bony-ankylosed group than in the control group, and the platelet count on postoperative day 1 was lower. Additionally, intraoperative blood loss and operative time were positively correlated. Age at the time of THA and the duration of ankylosis also showed negative correlations with intraoperative blood loss. Skeletal muscle is a vascular-rich tissue; however, muscle mass and vascularity decrease with age and prolonged lack of use . We believe that this negative correlation occurred because the amount of muscle approached at the time of surgery differed depending on the age and duration of ankylosis. Therefore, adequate blood transfusion preparation should be ensured prior to THA for ankylosed hips, especially in young patients or patients with short duration ankylosis. The gluteus medius is one of the major muscles of the hip abductor group and an important determinant of hip function: it ensures hip stability and controls pelvic posture during standing and walking . Decreased gluteus medius volume is the main cause of limp gait in patients with hip joint osteoarthritis . A decrease in the CT values of the gluteus medius represents an atrophic change in the nature of the muscle and muscle weakness in patients with gait disturbance . Herein, the bony-ankylosed group had significantly lower CSA and CT values of the gluteus medius than the control group. Thus, gluteus medius atrophy may have contributed to lower JOA hip scores and longer hospital stays in the bony-ankylosed group. Several studies have reported implant survival rates for THA of ankylosed hips. Hamadouche et al. reported an 8-year survival rate of 96.7% and Joshi et al. reported a 10-year survival rate of 96.1% . Paxton et al. conducted an international comparison of implant survival rates of primary THA for osteoarthritis and reported survival rates of approximately 93–95% . In the current study, the 10-year survival rate of implants in the bony-ankylosed group was 95%, which is comparable with the three aforementioned reports and the current study’s control group. In this study, the THAs performed on the patients with bony ankylosed hips had high survival rates. In contrast, Richards and Duncan reported a 10-year survival rate of 74.2% for conversion of hip arthrodesis to THA . They reported lower survival rates than those of our study. Although a statistical examination was not possible, the mean age of the patients in their study (50 years) was lower than that in our study (60.2 years). The relatively older age of our patients at the time of surgery and the different generations of implants may have affected the survival rate of the implants. This study had four limitations. First, this was a retrospective, single-centre study. However, we used propensity score matching to minimise confounding factors. Second, due to inadequate data recording, we could not assess the outcomes of adjacent joint disorders. Third, only a small number of patients underwent postoperative CT. However, the statistical power of the outcomes was sufficient. Fourth, many patients were excluded due to follow-up periods of less than 10 years. The reason for this is that many patients came to the hospital from far away and the follow-up may have been carried out by the patient's neighbouring doctor, making the follow-up untraceable. In summary, THA for patients with bony ankylosed hips achieved positive results, such as increased JOA hip scores for all items except pain; however, these scores were inferior to those observed in patients with non-ankylosed hips. The number of hips with postoperative complications was significantly higher in the bony-ankylosed group than in the control group.
|
Nationwide survey on the usage of ovulation-induction agents among obstetricians and gynecologists in China | 43be6b14-9a18-4666-bf31-ae61a8925766 | 6831081 | Gynaecology[mh] | The authors are grateful to the Chinese obstetrician and gynecologists for participating in the study. They also thank the Gynecological Endocrinology Committee of the Chinese Maternal and Child Health Association for helping to carry out the survey.
This study was supported by a grant from The National Natural Science Foundation of China (No. 11471024).
None |
Association of sociodemographic factors with the prescription pattern of opioids for dental patients: A systematic review protocol | 54793020-244d-44e2-b651-401f098ed1ab | 8341526 | Dental[mh] | Opioids have long since been used for pain relief, with a key role in modern anesthesia, as well as in postoperative, palliative, and emergency care [ – ]. The analgesia produced by opioids is derived by its complex interaction with receptors of the Central Nervous System . Though effective and largely used worldwide [ – ], opioids are associated with a wide variety of pernicious side effects . When administered at higher doses, the probability of addiction and abuse increases over the long-term . The growth in opioid use has led to a health crisis in some countries . In 2017, the United States (US) declared the opioid epidemic a public health emergency. Opioid misuse has been recognized as a national health issue in Australia . The roots of this epidemic rely on the overall recognition that pain has been underestimated by healthcare providers in the past. In 1990, Max stated that pain was being poorly managed and suggested that authorities should encourage the therapeutic use of opioids. In addition, in 1995, pain was described as the fifth vital sign . Associated with pharmaceutical companies’ aggressive marketing of new opioid formulations and other initiatives that came thereafter, healthcare providers became more sensitive in the treatment of pain . Consequently, opioid sales skyrocketed . The health and social burdens of an opioid epidemic are significant. From 2005 to 2015, there was a 22.3% increase in disorders associated with opioid use worldwide . In the US, data from 2016 estimated 42,245 deaths provoked by opioid overdose (a mean of 118 deaths daily) . Young adults between 25 and 34 years were the most affected (n = 11,552; 20%), and among individuals within this age range, one in five deaths were associated with opioids . In this scenario, deliberate attempts to mitigate the number of prescriptions of opioids by healthcare providers, including oral health practitioners, have been made [ , , ]. Oral health practitioners accounted for 8.6% of all providers who prescribe opioids in the US . The prescription patterns of these drugs in Dentistry seems to be influenced by many factors, including the type of procedure performed, the patient’s threshold for pain, and sociodemographic characteristics [ – ]. It seems that women are more likely to receive an opioid prescription . However, one systematic review reported that an individual’s sex was not identified as an associated factor for opioid prescriptions after surgery or trauma . Compared to white patients, African-Americans are more likely to receive an opioid prescription provided by a dentist . Conversely, another study reported that race was not a predictor for opioid prescriptions in Dentistry . In Brazil, the more privileged the area, the higher the opioid sales . However, in Australia, living in lower socioeconomic conditions was associated with a higher risk of receiving an opioid prescription . Regarding dental insurance, the association with opioid use is still unclear [ , , , ]. In view of the opioid epidemic, awareness of the underlying risk factors for opioid prescriptions could contribute to the rational use of this medication . Recently, protocols to aid health professionals in opioid prescriptions, such as the Centers for Disease Control and Prevention (CDC) guideline , have been provided, but there is still a consensus that a multi-faceted approach is needed to address this issue . Hence, a better knowledge of sociodemographic determinants, associated with the prescription pattern of opioids, may be helpful in the development of interventions to tackle the problem and in the reduction of the perpetuated disparities in the use of pain medications between privileged and underprivileged groups . Primary studies have assessed the possible influence of sociodemographic factors on the prescription pattern of opioids in Dentistry; however, the evidence produced by such studies seems controversial [ , , , , ]. A preliminary search was conducted in the PROSPERO, MEDLINE (PubMed), and Web of Science databases, and no systematic review on this issue was identified. Therefore, the objective of this systematic review is to identify if patients’ sociodemographic factors are in fact associated with the prescription pattern of opioids in Dentistry.
The proposed systematic review was registered in the International Prospective Register Of Systematic Reviews (PROSPERO): CRD42020211226. The Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) checklist is presented ( ). This study is supported by the Coordination for the Improvement of Higher Education Personnel (CAPES, in Portuguese). This funding agency played no role in the study design, decision to publish, or preparation of the manuscript. Review question Inclusion criteria Search strategy Study selection Data extraction Assessment of methodological quality Data synthesis Certainty of evidence If there is sufficient evidence to make recommendations, the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) will be used to rank the certainty of evidence. The GRADEpro (McMaster University, ON, Canada) will be used to construct a Summary of Findings. GRADE is based on five domains: the risk of bias within individual studies, the study design, the indirectness of the evidence, the inconsistency, the imprecision of the effect size estimates, and the risk of publication bias. The certainty of the evidence for each outcome may be ‘very low’, ‘low’, ‘moderate’, or ‘high’ .
Are sociodemographic factors associated with the prescription pattern of opioids for dental patients?
Participants Exposure Outcome Types of studies This systematic review will consider studies with individuals at any age, in which the influence of patients’ sociodemographic factors on receiving an opioid prescription provided by an oral health practitioner was assessed. Studies investigating opioid prescriptions for dental conditions delivered by healthcare providers other than oral health practitioners (e.g., physicians, emergency room [ED] doctors) will be excluded.
In this systematic review, the exposure will consist of dental patients’ sociodemographic characteristics, including sex, age, race, income, educational level, living environment, and dental insurance. Data sources for the assessment of sociodemographic factors will include validated questionnaires, dental charts/records, and oral healthcare system databases.
This systematic review will consider the prescription pattern of opioids provided by oral health practitioners as an outcome.
Observational studies: cross-sectional, case-control, and cohort studies will be included. Letters to the editor, editorials, ecological studies, case reports, case series, and literature reviews will be excluded.
For the definition of the search strategy for this systematic review, three steps were taken. First, a preliminary search was performed in MEDLINE (PubMed) to identify articles that meet the inclusion criteria. Thereafter, the titles and abstracts of these studies were used to identify keywords and indexing terms to develop the final search strategy. Second, with the assistance of a librarian, the final search strategy was tailored for MEDLINE (PubMed) using MeSH terms, entry terms, and synonyms linked with Boolean operators ( ). The search was adapted for each of the other databases, using controlled vocabulary (MeSH, EMTREE, and others), as well as entry terms. Finally, the list of references of all studies included in the systematic review will be screened. Before final analysis, the searches will be updated. No restrictions will be applied to language and year of publication, or geographic limits. Information sources Searches will be conducted in the following electronic databases: MEDLINE (PubMed), EMBASE, Scopus, Web of Science, LILACS and SciELO. A search in Google Scholar, limited to the first 200 most relevant studies , will also be conducted. Grey literature will be searched in OpenGrey.
All identified citations will be uploaded into EndNote 20 (Clarivate Analytics, PA, USA) and duplicates removed. Two reviewers will select the studies independently. They will begin by assessing the titles/abstracts. Those that meet the inclusion criteria will be selected. If the title/abstract does not provide sufficient information for a decision, the full text of the reference will be evaluated. References whose full text fulfills the eligibility criteria will be included. Any disagreements arising during the study selection will be discussed and resolved by consensus. If an agreement is not reached, a third reviewer will decide. Interrater reliability will be estimated with the kappa test.
Two reviewers will extract all data from the included studies independently and in duplicate. If divergences between reviewers occur, a discussion will be set in place until a consensus is reached. Data extracted will include study details (author, year, journal), study methods (design, setting, sample, recruitment process, exposure), prescription of opioids (frequency, dose, duration of prescribed opioid), and the results of evaluation of the association of exposures (sociodemographic factors: sex, age, race, income, educational level, living environment, and dental insurance) with opioid prescriptions. The form developed for data extraction is displayed as supporting information ( ).
Two reviewers will assess the methodological quality of the included studies independently. Disagreements arising during this process will be discussed and resolved by consensus. In the cases in which agreement is unattainable, a third reviewer will be consulted. Depending on each study design, the following JBI’s critical appraisal tools will be deployed: Checklist for Analytical Cross-Sectional Studies, Checklist for Case Control Studies, and Checklist for Cohort Studies will be used, depending on each study design . In the cross-sectional studies, eight items will be assessed: definition of the inclusion criteria; depicted information of participants, study´s setting, and time period; use of valid and reliable methods for the assessment of the exposure; if objective and standard criteria were employed to assess the condition; awareness of confounders; reliable measurement of the outcome; and use of adequate statistical analysis . In the case control studies, 11 items will be assessed: similarity between groups, presence of disease in cases or the absence of disease in controls; matching of cases and controls; adoption of the same criteria to identify cases and controls; adoption of a valid and reliable strategy for the measurement of the exposure; same measurement of the exposure for cases and controls; awareness of confounders; strategies to handle confounders; adoption of a valid and reliable method for the measurement of the outcome for cases and controls; period of interest of the exposure, if this period was long enough to be significant; and adoption of adequate statistical analysis . In the cohort studies, 11 items will also be assessed: recruitment of participants of the two groups from the same population; similar measurement of the exposure to assign participants to both exposed and unexposed groups; use of a valid and reliable instrument for the measurement of the exposure; identification of confounders; strategies to deal with confounders; groups of individuals without the outcome at the study’s onset; adoption of a valid and reliable strategy for the measurement of the outcome; report of the follow-up period and whether this period is long enough for the occurrence of the outcome; information whether follow-up was complete or at least a statement on the reasons for follow-up losses; strategies to handle incomplete follow-up; and adoption of adequate statistical analysis . In each study, three ratings may be assigned to the items: ‘low risk of bias’ (if the answer to the question is ‘Yes’), ‘high risk of bias’ (if the answer to the question is ‘No’), and ‘unclear risk of bias’ . All studies will be submitted to data extraction and synthesis. The impact of risk of bias will be considered when developing conclusions and recommendations (if feasible).
In the first step of data synthesis, we will present the results of the study selection process using the PRISMA statement flowchart. The interrater reliability (agreement between the two reviewers) estimate will be given with the kappa test. In the second step of data synthesis, homogeneous data will be aggregated into meta-analyses. Effect measures will be expressed as either odds ratio/risk ratio (for dichotomous data), or mean differences (for continuous data). In the case of meta-analyses of continuous outcomes, for which included studies used different scales, the standardized mean difference will be determined. The 95% confidence interval (CI) will be calculated as well. The possibility of combining studies of continuous and dichotomous outcomes through transformations will be assessed. Heterogeneity will be assessed using chi-square and the I 2 statistic . The confidence interval of the I 2 will be determined as well. The following formula will be used: exp (ln I 2 ±1.96× SE [ ln ( I 2 )]) . The random effect model will be deployed in the meta-analysis . We will also check whether the results of the meta-analyses remain even if undetected heterogeneity is assumed. The random effect model will be implemented through the DerSimonian and Laird inverse variance . Meta-analyses will be developed at RevMan 5.4 (Copenhagen, The Nordic Cochrane Centre, Cochrane). Sensitivity analysis might be conducted with the removal of studies one by one in an attempt to reduce heterogeneity. If possible, subgroup analyses will be performed according to the different sociodemographic factors assessed (sex, age, race, income, educational level, living environment, and dental insurance). The design of the included studies will also be considered during the analysis of subgroups. The results of meta-analysis will be presented as forest plots. Publication bias will be addressed by developing a funnel plot . The RevMan 5.4 (Copenhagen, The Nordic Cochrane Centre, Cochrane) will be used. Considering that the visual evaluation of the funnel plot might be subjective, we will run a test for funnel plot asymmetry, so long as at least ten studies are incorporated into the meta-analysis . The regression method for the detection of funnel plot asymmetry proposed by Harbord and colleagues (2006) will be used for dichotomous outcomes. For the meta-analysis of continuous outcomes, the assessment of funnel plot asymmetry will be performed with the Egger test . Moreover, if less than ten studies are incorporated into the meta-analysis, no funnel plot will be constructed, and this will be reported and discussed as a limitation of the systematic review. Bias assessment, considering the year of publication of the articles, will also be performed, as publication bias in systematic reviews of more recent studies is lower .
S1 Checklist PRISMA-P 2015 checklist: Recommended items to address in a systematic review protocol. (PDF) Click here for additional data file. S1 File Search conducted in MEDLINE (PubMed) on September 28th, 2020. (PDF) Click here for additional data file. S2 File Form to extract data of included studies. (PDF) Click here for additional data file.
|
Correlation between marginal bone loss around dental implants and various systemic diseases: a cross-sectional study | 9a15e94e-c66d-4be8-b4c4-47d6536aa3e3 | 11502616 | Dentistry[mh] | Dental Implants are a widely utilized option for patients suffering from tooth loss that can help restore both function and esthetics. The long-term survival of dental implant therapy is reported widely in various studies . A dental implant failure can occur in the early stages due to a lack of osseointegration or in the late stage due to peri-implantitis . Marginal Bone Loss [MBL]There is a bidirectional relationship between systemic and oral health, which may compromise the survival of the dental implant . Hypertension is defined as a systolic value over 140 mmHg and/or a diastolic reading of over 90 mmHg and is considered the main cause of premature diseases and death in the world . Blood pressure readings over time are valuable predictors of an individual’s risk of developing cardiovascular disease (CVD). Not only does it impose a risk of CVD, it can lead to comorbidities such as chronic renal failure and stroke. Bone abnormalities have also been linked to hypertension including a decrease in the regeneration, density, and quality of the alveolar bone due to impaired calcium metabolism and delayed healing . Hyperlipidemia occurs as a result of abnormal lipid levels within the blood. This is characterized by increased levels of total cholesterol, triglycerides and low-density lipoproteins (LDL) as well as a decrease in the levels of “good cholesterol” known as high density lipoproteins (HDL) . Currently, nearly 86 million adults in the United States 20 years and older have total cholesterol levels above 200 mg/dL . A high level of cholesterol is a major controllable risk factor for other health complications including heart disease, myocardial infarction, and cerebrovascular accidents. Several studies have investigated the effects of hyperlipidemia on bone and have found that elevated amounts of plasma lipoproteins may increase the number of osteoclasts in the alveolar bone . This phenomenon will lead to the inhibition of osteoblastic activity needed for osseointegration of the dental implant . Diabetes mellitus is one of the most common chronic health issues within the world and is directly correlated with impaired wound healing due to an overactive immune response to pathogens. Inflammatory mediators (IL-1, IL-6, IL-8) and tumor necrosis factor (TNF-ɑ) associated with diabetes are released into the oral tissues. These factors can lead to increased inflammation and a reduction in collagen synthesis which ultimately affects formation of bone and the healing capability of oral tissues . Hyperglycemic conditions over the long-term may degrade the vascularity within the oral cavity and that supplying the alveolar bone . Just as in hyperlipidemia, the differentiation of osteoclasts is promoted and osteoblast formation is inhibited. Due to these risk factors, dental implants may not be considered on these individuals. While complications are evident among diabetic individuals, they have been seen to diminish in those with properly controlled diabetes mellitus . In 2016, a systematic review narrated a significant delay in the osseointegration of dental implants in poorly controlled diabetic patients . A recent meta-analysis indicated a direct association between hyperglycemia and the risk of peri‐implant diseases, and there is a high risk for MBL of dental implants for type 2 diabetes mellitus (T2DM) control patients . It has been suggested that successful implants must present ≤ 2 mm of MBL during the first year after placement, followed by ≤ 0.2 mm per subsequent year . It was indicated that there is an increased risk of implant failure if the MBL was 0.44 mm, at six months post-loading . In 2018, Neves et al. concluded that rheumatologic and cardiovascular disorders are associated with an increased risk of peri-implant pathology . A cross-sectional biochemical study comparing the inflammatory and lipid profiles of patients with and without peri-implantitis suggested that even healthy individuals with peri-implantitis exhibited a low-grade systemic inflammatory state, evidenced by elevated circulating white blood cell levels, as well as dyslipidemia, characterized by increased LDL cholesterol and total cholesterol levels . A recent umbrella review emphasized the need for further studies to assess the long-term effects of cardiovascular disease, neurological disorders, and the use of certain medications on dental implant survival rates . Another systematic review also highlighted that more research is required before drawing definitive conclusions about the association between cardiovascular disease and peri-implantitis, as the current body of literature contains too few studies to establish a clear link . The present study aimed to evaluate marginal bone loss (MBL) around dental implants in patients with hypertension, hyperlipidemia, and diabetes mellitus attending the University of Nevada, Las Vegas (UNLV) dental clinics. The null hypothesis proposed that there would be no significant difference in MBL among patients with hypertension, hyperlipidemia, and diabetes.
Study design and population Search strategy Inclusion and exclusion criteria Data extraction Ethical approval Radiographic analysis This study is a cross-sectional study of patients who attended the University of Nevada, Las Vegas dental clinics.
Clinical notes from patients at the University of Nevada, Las Vegas (UNLV) dental clinics were analyzed using AxiUm™ software. AxiUm™ is a comprehensive dental practice management software for educational institutions, particularly dental schools. It helps streamline various administrative, clinical, and financial operations by integrating tools and features that manage patient records, treatment planning, billing, scheduling, and academic workflows. A search was conducted using keywords such as ‘systemic disease,’ ‘marginal bone loss,’ ‘dental implant,’ ‘high cholesterol,’ ‘hypertension,’ and ‘diabetes’ within the system (Fig. ).
The study included patients with dental implants diagnosed with hypertension, diabetes mellitus, and hyperlipidemia who attended the UNLV School of Dental Medicine clinics from 2012 to 2022. Exclusions were made for patients with acquired immune deficiency syndrome and those with a limited number of radiographs.
All patients are informed during their intake to the UNLV dental clinics that they are being treated at an educational institution and their information may be used for research purposes. Their information is protected according to regulations dictated by the Federal Health Insurance Portability and Accountability Act of 1996 (HIPAA). Socio-demographic data including the patient’s age, gender, ethnicity, smoking habits (yes/no and type), and alcohol consumption (yes/no) were recorded. Dental and medical histories were collected from the existing chart records and evaluated. Data collected from these forms included systemic health issues and prescribed medications. Data were extracted by AC and NHA.
As no supplementary radiographs or examinations were conducted specifically for the study, and the data collected were analyzed and presented anonymously, the study was granted exempt status by the Institutional Review Board (IRB) of the University of Nevada at Las Vegas (UNLV; #UNLV-2022-256). The study adhered to the principles of good clinical practice in accordance with the World Medical Association (WMA) Declaration of Helsinki (1975), revised in 2013.
All radiographs were obtained using a long-cone technique at 70 kV and exposure time of 0.16 s. Only periapical radiographs were considered for the study, no CBCT imaging was included. All periapical radiographs were standardized using sensor holders and the parallel technique. Additionally, experienced dental technicians supervised all radiographic images taken at the radiology clinic to ensure quality and consistency. If the implant was placed at the UNLV dental clinic, peri-implant bone measurements were obtained from the most recent radiographs and radiographs at the time of placement. If the patient presented to the UNLV clinics with an existing implant, new radiographs taken over the course of at least 6 months were included in the data. The marginal bone level is defined as the vertical distance from the tip of the implant body to the coronal edge of the first bone-to-implant contact . Marginal bone level was evaluated based on this principle as well as methods used in previous studies. Measurements were taken from the implant-abutment junction to the crest of the bone at both mesial and distal sides of each implant following the methodology described by Shi et al. (Fig. ). MBL was confirmed by comparing subsequent bone-to-implant contact levels to the initial radiographs. The MiPACS (Medical Imaging Picture Archiving and Communication System) is a comprehensive digital imaging system fully integrated with the axiUm™ dental software, and it was used for all radiographic measurements. MiPACS is the primary imaging module for capturing, viewing, and managing dental radiographs such as intraoral radiographs, panoramic images, and CBCT scans. MiPACS enhances clinical diagnostics with advanced image enhancement tools, including zoom, contrast adjustment, annotations, and measurement tools. One investigator (first author), conducted the measurement in order to eliminate inter-examiner variation. Using Cohen’s kappa in SPSS (version 28), the intra-examiner reliability was calculated to be 85%, indicating a strong level of agreement between repeated measurements.
After screening the 1,310 potentially eligible electronic Axium records, 57 patients with 165 implants met the selected inclusion criteria (Fig. ). Table shows the descriptive statistics of the sample. The majority of the patients were aged 65 and older (79%) followed by those between 55 and 64 years of age (18%). Social factors such as smoking and alcohol consumption accounted for < 50% of the sample (Table ). The more systemic health disease a subject presented with, the more likely they had bone loss surrounding an implant as 45.6% of patients reported having more than four systemic diseases and 67% were taking four or more prescription medications (Fig. ). A decrease in the marginal bone surrounding dental implants in this sample was strongly correlated with patients diagnosed with hypertension (78.95%) or hyperlipidemia (73.68%) compared to those with diabetes mellitus (40.35%). Patients diagnosed with both hypertension and hyperlipidemia comprised 29.82% of the sample data and held statistical significance ( p < 0.05) (Fig. ). Overall, varying combinations of these systemic diseases and comorbidities were more closely associated with peri-implant MBL than patients with a single systemic disease diagnosis. Prescribed medications to combat these health issues, such as statins and antihypertensive, also showed the same trends and corresponded to a higher prevalence of MBL (Table ). Over 90% of patients in each category reported taking the daily doses of medications on a regular basis.
The aim of the present cross-sectional study was to investigate the correlation between hypertension, diabetes mellitus and hyperlipidemia, on the marginal bone loss (MBL) surrounding dental implants among patients attending the University of Nevada, Las Vegas dental clinics. After analyzing the results, it was found that there was a statistically significant difference in MBL between patients with cardiovascular diseases (HTN, hyperlipidemia) and diabetic patients. This study has successfully demonstrated that implant failure and peri-implant bone loss occur more in individuals that present with cardiovascular disease. Therefore, the null hypothesis was rejected. Several studies have been published that report a direct relationship between diabetes and peri-implantitis. They concluded that diabetic patients, particularly those diagnosed with type 1, have a greater estimated MBL over time . This has been attributed to the reduced angiogenesis due to hypercoagulation as well as a hindrance of bone formation markers . For Diabetic patients, it is important to note that the type and stability of diabetes is a determinant in the overall success of the implant. Deeper pocket depths, bleeding on probing and increased MBL were more commonly found among those with poor metabolic control compared to those with ‘well controlled’ diabetes . This data correlates with the present study as diabetic patients presented with Type 2 diabetes had varying level of glycemic controls. Bone metabolism is affected by hyperlipidemia through both osteoclasts and osteoblasts which may promote bone loss and inhibited osseointegration of dental implants . In an experimental study on rats, Teken and Toker found that the rats that were fed a high cholesterol diet showed significantly lower bone-to-implant contact values than the control group . The findings of their study give support to the hypothesis that hyperlipidemia can lead to a decrease in implant osseointegration and implant stability. Statin medications used to treat hyperlipidemia were also correlated to peri-implant bone loss. Behrami et al. conducted a retrospective study on the influence of statin use on the severity of peri-implantitis and the incidence of peri-implant bone loss . However, several studies reported an increase in osseointegration of the implant body into the alveolar bone when coupled with Simvastatin. Implants were either coated with the Simvastatin or a gel containing the medication was placed into the alveolar socket at the time of placement . These findings illustrate the need for further investigation into the effects of hyperlipidemia and statins on endosteal implants. Hypertension has a positive association with moderate to severe periodontitis as bleeding on probing, CAL, and pocket depths were poorer on hypertensive patients . Singh et al. found that hypertension led to a 20.8% increase in the failure of dental implants. This phenomenon may be due to impaired calcium metabolism and delayed healing associated with higher arterial pressure . In a cohort study by Wu et al., of the 1,449 implants, a failure rate of 0.6% was observed in people using antihypertensive drugs and 4.1% for nonusers . The results of these two studies show a possible failure of compliance in hypertensive patients taking their medications. This should be accounted for in the present study as the data relies on self-reporting from patients. Future studies should also evaluate the influence of time and follow-up periods on MBL which was not considered in this study. There are several limitations to the present study. First, this is a cross-sectional dental-record based study that relied on the accuracy of the examination and documentation. Incomplete documentation was anticipated as patients are seen by a variety of students at University Dental Clinics. Second, there was a lack of quantifiable measurements of MBL radiographically and also no CBCTs were used in the investigation. The correlation was related to the main and mostly reported systemic diseases and not all systemic diseases. The measurements were performed as a visual representation of the effects associated with specific health issues and medications rather than a diagnostic tool. Third, oral hygiene and parafunctional habits could not be properly assessed due to the nature of this study; therefore the oral environment and the forces applied to the prosthesis might have an impact on the clinical outcomes. Lastly, a limited number of subjects, implants and restoration types were identified within the study.
Within the limitations of the present investigation, patients diagnosed with hyperlipidemia and hypertension were more likely to exhibit MBL surrounding dental implants. Further investigation is required to correlate the presence of MBL around dental implants with antihypertensive and statin medication use.
|
SOCIAL DETERMINANTS OF MORTALITY FROM COVID-19, A RETROSPECTIVE STUDY OF 6,000 PATIENTS | c718c3a0-ec83-4251-860b-f83f7d1fb422 | 8972426 | Internal Medicine[mh] | Health disparities and Coronavirus disease 2019 (COVID-19) mortality is an evolving topic. This study sought to explore the relationship between patient ethnicity, annual household income, and COVID-19 mortality.
A chart review was conducted of 6,000 hospitalized patients with positive COVID-19 polymerase chain reaction (PCR) tests from March 2020 to June 2021 at Methodist Health System in Dallas, Texas. Patient age, gender, ethnicity, and zip code were collected. The sample included 3,114 males and 2,886 females with a mean age of 61.6±17.1 years. Ethnicity selected by the patient was used. Median annual income by zip code was obtained from the 2020 U.S. Census Data. A Chi-square test was used to calculate p-values.
No statistically significant difference in mortality based on ethnicity or median annual income by zip code was found. Asian American patients had the lowest mortality rate, while Hispanic, Latino, or Spanish origin had the highest mortality rate with p(0.92), independent of other factors. Patients living in a zip code with a median household annual income of $40-80K had the lowest mortality rate while those with ≥$80K had the highest mortality rate with p(0.12), independent of other factors as shown in Table 1.
Our study suggested that neither ethnicity nor income predicted mortality in admitted COVID-19 patients, independent of age and gender.
|
Low Skeletal Muscle Mass: A Strong Predictive Factor for Surgical Complications After Free Forearm Flap Reconstruction in Oral Cancer Patients | cb29d944-aaa9-4b76-8a51-f4f4831188f2 | 11907681 | Musculoskeletal System[mh] | Introduction Free forearm flaps (FFAF), that is, the free radial forearm flap (FRFF) and free ulnar forearm flap (FUFF), are frequently used free vascularised flaps for reconstruction of large head and neck soft tissue defects following ablative cancer surgery. A number of preoperative risk factors are known to be associated with postoperative complications after free tissue transfer, which include age, gender, tobacco use, diabetes, hypertension, body mass index (BMI) and prior radiotherapy . Sarcopenia, a condition characterized by loss of skeletal muscle mass (SMM) and low muscle strength, has been found to influence both treatment outcomes and survival in head and neck cancer (HNC) patients . Low SMM alone has been associated with a higher incidence of postoperative complications, chemotherapy related toxicity, longer hospital stays and diminished disease‐free and overall survival in HNC patients . Patients with HNC are more likely to develop sarcopenia because of swallowing disorders caused by the localization of the primary tumor, decreased nutritional intake and cancer‐induced catabolism . Some studies found that sarcopenia is a significant independent risk factor for free flap complications, surgical site infection and other postoperative complications in patients with HNC [ , , , ]. The aim of this article is to analyze the association of low SMM with free flap related and other postoperative complications in a subgroup of HNC patients who underwent an FFAF reconstruction following resection of an oral cancer. Furthermore, the relationship between low SMM and the duration of hospital stay is investigated.
Material and Methods 2.1 2.3 2.4 Statistical Analysis The data analyses were performed using IBM SPSS Statistics version 25.0 (IBM Corp., Armonk, NY, USA). The baseline characteristics were presented as frequencies and percentages. Correlation analysis was performed by use of Pearson's correlation analysis for variables with a normal distribution and Spearman's correlation analysis was used for non‐normally distributed variables. Logistic regression was used for univariate and multivariate analysis of surgical complications. Covariates used in the multivariate analysis were selected based on clinical significance or selected based on statistical significance ( p < 0.20) in univariate analysis. A test of normality (One sample Kolgomov–Smirnov) was performed for the duration of hospital stay. If hospital stay was normally distributed, an independent sample t ‐test or a linear regression model was performed if the assumptions of linear regression were met. Statistical significance in this analysis was evaluated at the p < 0.05 level using two‐sided tests.
Ethical Approval The design of this study was approved by the Medical Ethical Research Committee of the University Medical Center Utrecht (approval ID 17‐365/C). All proceduresPatients and Study Design A retrospective study was performed of consecutive patients who underwent reconstruction of oral cavity defects with FFAF after resection of a malignancy between 2003 and 2020 at the Departments of Oral and Maxillofacial Surgery, Otorhinolaryngology and Head and Neck Surgery and Head and Neck Surgical Oncology of the University Medical Center Utrecht, Utrecht, the Netherlands. All surgical procedures were performed by experienced microvascular head and neck surgeons. The choice of an ulnar or radial forearm flap was based on the surgeon's preference and experience with raising the specific flap. Patients were included if they had a recent (less than 1 month before surgery) CT or MRI scan of the head and neck. Clinical and demographic data were collected from the medical records. Data collected included age at surgery, gender, BMI, alcohol consumption (defined as drinking more than 2 units alcohol per day), smoking history (categorized as current smoker or having stopped more than 12 months), diagnosis, TNM stage (pathological), localization of the defect, comorbidity as expressed by the Adult Comorbidity Evaluation‐27 (ACE‐27) score, duration of hospital stay and occurrence of postoperative complications Patients with prior chemoradiation therapy to the head and neck were not included. Neither were patients with recurrences or preoperative nasogastric tube insertion included. All postoperative complications were scored according to the Clavien–Dindo classification of surgical complications . Patients with multiple complications were scored according to their highest grade of complication. Complications with a Clavien–Dindo grade III–V were graded as severe complications. Postoperative complications specifically related to the free flap were also analyzed and scored. These were categorized as congestion or thrombosis, partial skin paddle necrosis or dehiscence, donor site morbidity and flap failure.
Body Composition Measurement SMM was measured as muscle cross‐sectional area (CSA) on pre‐treatment CT or MRI imaging of the head and neck area at the level of the third cervical vertebrae (C3). The first axial slide of the imaging when scrolling from cranially to caudally, which showed both transverse processes and the entire vertebral arc, was selected for segmentation of muscle tissue. For CT imaging, muscle area was defined as the pixel area between the radiodensity range of −29 and +150 Hounsfield Units (HU), which is specific for muscle tissue. For MRI, muscle area was manually segmented, and fatty tissue was manually excluded. The CSA was calculated as the sum of the delineated areas of the paravertebral muscles and both sternocleidomastoid muscles. Segmentation of muscle tissue was manually performed using the commercially available software package SliceOmatic (Tomovision, Canada) by a single researcher (NCM) who was blinded for patient outcomes. An example of segmentation at the level of C3 is shown in Figure . CSA at the level of C3 was converted to CSA at the level of L3 using a previously published formula 1 (23). The lumbar skeletal muscle index (LSMI) was calculated by correcting SMM at the level of L3 for squared height as shown in formula 2. Low SMM was defined as a LSMI below 43.2 cm 2 /m 2 , a cutoff value which was determined in a separate cohort of head and neck cancer patients . Formula 1 : CSA at L3 cm 2 = 27.304 + 1.363 * CSA at C3 cm 2 – 0.67 1 * Age years + 0.640 * Weight kg + 26.44 2 * Sex Sex = 1 for female and 2 for male Formula 2 : Lumbar SMI cm 2 / m 2 = CSA at L 3 / length m 2
Results 3.1 3.2 3.3 3.4 Patient Characteristics Descriptive data are presented in Table . In total, 174 patients were included. Low SMM was identified in 115 (66.1%) patients. One hundred and thirty‐eight (79.3%) patients underwent a FRFF and 36 (20.7%) patients underwent a FUFF.
All Complications Table shows the non‐flap complications. Clavien–Dindo scores of all postoperative complications are described in Table . In total, 117 (67.2%) patients had any postoperative complication, of whom 77 (65.8%) had low SMM. Forty‐one patients (23.6%) had severe complications (Clavien–Dindo III–V), of whom 27 (65.9%) had low SMM. Four patients (2.3%), all with low SMM, died in hospital within 1 month due to a complication. The results of the univariate analyses on potential risk factors for any postoperative complications are shown in Table . Age, alcohol use, smoking, BMI, ACE‐27 score and SMM were included in the multivariate logistic analysis. The results of the multivariate analysis are shown in Table . Low SMM was not associated with any postoperative complications (OR 1.18; 95% CI 0.58–2.57, p = 0.64). Furthermore low SMM was significantly associated with severe postoperative complications (Clavien–Dindo III–V) (OR 1.46; 95% CI 1.20–2.09, p = 0.02).
FFAF Related Complications FFAF‐related complications are described in Table . Complications related to the FFAF occurred in 47 (27.0%) patients, of which 25 (53.2%) occurred in patients with low SMM. Five (3%) patients needed flap revision due to venous thrombosis or arterial occlusion. In 3 (1.7%) patients, the complete flap was lost. In the logistic regression, low SMM was associated with flap related complications (OR 2.14; 95% CI 1.02–4.39, p = 0.029).
Duration of Hospital Stay Median length of hospital stay was 14.3 days (95% CI 12.99–15.58, SD 8.54). Mean number of days in hospital was similar for patients with or without low SMM (14.40 days, SD 7.23 and 14.23 days, SD 9.05, respectively).
Discussion This study assessed risk factors for the occurrence of postoperative complications in patients who underwent reconstruction of head and neck defects with FFAF after resection of a malignant oral cavity tumor. We compared several potential perioperative predictive factors, of which low SMM was significantly likely to cause more postoperative FFAF related complications such as flap dehiscence, flap necrosis, thrombosis and flap failure. Postoperative complications of any type were not significantly higher in patients with low SMM. However low SMM was predictive for the occurrence of severe complications (Clavien–Dindo III–V). The finding of the current study is in line with previous studies. A systematic review with meta‐analysis demonstrated that low SMM was associated with the occurrence of severe postoperative complications in patients with head and neck squamous cell carcinoma (OR 4.79, 95% CI: 2.52–9.11) . Previously we evaluated postoperative complications among patients who had undergone reconstruction of segmental mandibular defects with free fibula flaps . In another article low SMM in FRFF was found to be a predictor for postoperative complications (OR 2.0, 95% CI 1.1–3.8, p = 0.03) . We observed that low SMM was a negative predictive factor for postoperative flap complications. A retrospective case–control study by Alwani et al. evaluated complications among 168 HNC patients who received autologous free vascularized tissue reconstruction. Patients with low SMM had higher rates of complications, including pneumonia, venous thromboembolism, longer mechanical ventilation times, delirium, wound disruptions/fistula, and intensive care unit stays. Overall, these patients had higher rates of any postoperative complications and also flap‐specific complications . In another study of 239 HNC patients who underwent free flap reconstruction, low SMM was a predictor of perioperative blood transfusion requirements . In a cohort of 206 HNC patients 30.1% were discharged after free flap reconstruction to post–acute care facilities, including skilled nursing facilities, in‐patient rehabilitation facilities, and long‐term care hospitals, for extended support and recuperation beyond the immediate postoperative setting. Low SMM was found to be independently associated with discharge to the above mentioned post–acute care facilities . In this cohort flap, dehiscence was particularly present in the low SMM group 44.7% versus 31.9% in the normal SMM group. In line with this finding is a study of patients undergoing total laryngectomy, where sarcopenia was found to be the sole predictive factor of any wound complication (OR, 7.54; 95% CI, 1.56–36.4) . The pathophysiological mechanisms underlying the association between preoperative sarcopenia and the risk of postoperative complications have not been elucidated. SMM depletion is associated the production of anti‐inflammatory cytokines and adiponectin decreases and the production of pro‐inflammatory molecules, such as leptin, chemerin, resistin, tumor necrosis factor‐α, interleukin‐1 and ‐6 increases . Based on this mechanism, patients with sarcopenia are considered to be in a pro‐inflammatory state. The pro‐inflammatory state leads to a weakening of the immune system and poor wound healing after surgery, thereby exerting an impact on the risk of postoperative complications . Postoperative complications can have devastating consequences for both functional and cosmetic outcomes and can have a serious psychological impact on patients. Assessment of SMM is an objective measure that can provide valuable information in the clinical setting. It can be used to predict postoperative outcomes and consequently aid in surgical decision making. For instance, as in whether or not to opt for a microvascular reconstruction or a reconstruction with a local flap, which entails less risk of wound dehiscence and shorter operative time. Selecting patients with low SMM for a local flap reconstruction can potentially reduce postoperative complications. Several methods to increase skeletal muscle mass and to decrease systemic inflammation have been reported in the literature . In a double‐blind, randomized placebo‐controlled trial by Rooks et al. bimagrumab treatment, a monoclonal antibody that blocks activin type II receptor (ActRII) to inhibit myostatin signaling and stimulate protein anabolism was added to optimized standard of care in community‐dwelling older adults with sarcopenia . The results showed no difference in the improvement of physical function between bimagrumab versus placebo, although participants who received bimagrumab had an increased lean body mass and reduced fat mass versus participants who received placebo. Physical activity is also effective at mitigating sarcopenia. Several studies have demonstrated that prehabilitation in patients undergoing major abdominal surgery, particularly with exercise programs, results in lower rates of postoperative morbidity . However, these studies did not stratify for sarcopenia. Optimizing nutritional status may be valuable in patients prior to surgery because this can potentially decrease systemic inflammation and promote better wound healing. An example is the area of immunonutrition supplements that contain arginine, omega‐3 fatty acids, and dietary nucleotides that modulate prostaglandin E2 production, decrease IL‐6 production, and promote T‐cell differentiation. Mueller et al. demonstrated that the use of immunonutrition for 5 days prior to salvage surgery among patients with HNC compared to a control group, caused a significant reduction in overall complications (35% vs. 58%, p = 0.027) and in overall length of hospital stay (median 6 vs. 17 days, p = < 0.001) . A study by Aeberhard et al. found that the use of immunonutrition for 5 days prior to surgery was associated with a significant decrease in the occurrence of wound abscesses and orocutaneous or pharyngocutaneous fistulas compared to the control group (7.4% vs. 15.3%, OR = 0.30, p = 0.006) . The combination of nutritional intervention and exercise has also been reported. A recent meta‐analysis of 42 RCTs compared multiple exercise intervention arms in 3728 older people with sarcopenia . This analysis found that adding nutritional interventions to exercise had a larger effect on handgrip strength than exercise alone while showing a similar effect on other physical function measures to exercise alone. Our study was limited due to its single‐institution and retrospective nature and therefore limited by the available medical documentation. Heterogeneity in terms of types of flaps, that is, only FRFF and FUFF, was limited (particularly when compared to other studies), but may have introduced bias with respect to flap complications. Although we found an association between low SMM and severe complications and also flap complications, we could not determine the odds of specific complications such as pneumonia, wound infections, delirium and flap failure due to the limited number of events. SMM was assessed on cross‐sectional imaging at the level of C3 because of the availability of routinely performed head and neck CT and MRI and its high correlation with the most often used SMM assessment at the third lumbar vertebra (L3) [ , , ]. Changes in SMM occur over time since cancer‐related skeletal muscle depletion is a continuous process. This study is limited to preoperative SMM at a single point in time, although the degree and effect of eventual perioperative change of SMM is unknown. In conclusion, low preoperative SMM has a negative impact on the occurrence of FFAF‐related complications in patients undergoing an FAFF reconstruction for oral cavity cancer. It is also a negative impact factor for severe (CD ≥ III) postoperative complications. Identification of high‐risk patients by SMM assessment allows for alternative surgical treatment planning, for example, less extensive surgery and less complex reconstructions, pre‐ and perioperative management and counseling.
|
A feasibility study of DNA ploidy analysis, HPV, and TCT for screening of cervical cancer: A retrospective study | 3e00b45f-17b7-4de7-9d1c-87b5329e3452 | 11630942 | Biopsy[mh] | Cervical cancer (CC) is the second most prevalent malignant tumor in females ; it is the only disease with a clear pathogenetic mechanism and can be prevented by vaccination and screening. However, because of outdated screening technology and a lack of awareness regarding prevention, there is still high morbidity and mortality due to CC in some areas. According to a recent estimation in 2018, the number of new global cases of CC and the number of deaths due to CC were 569,847 and 311,365, respectively. Precancerous and early invasive CC can be detected by CC screening, which effectively reduces the associated morbidity and mortality. Presently, the “three-step” diagnostic method is widely used to diagnose cervical diseases; this method includes cytology and/or human papillomavirus (HPV-DNA) test, colposcopy, and histopathological diagnosis. HPV-DNA testing is a pivotal element in CC screening, focusing on the identification of high-risk HPV types that are strongly linked to the development of CC. This method has significantly augmented the capability of screening programs to predict and prevent CC by targeting the primary cause of the disease before significant cellular changes occur. Concurrently, the thin-prep cytologic test (TCT) has refined the traditional Pap smear approach by employing liquid-based cytology. This advancement enhances the quality of cell samples and the accuracy of the results, allowing for better detection of cellular abnormalities. TCT categorizes cells more precisely, which significantly decreases the rate of false negatives and increases the sensitivity of detecting precancerous states. However, the efficacy of TCT largely depends on the subjective assessment of cytologists, which can introduce variability in diagnosis. In recent years, many new CC screening methods have emerged. Deoxyribonucleic acid (DNA) ploidy analysis for cervical epithelial cells is a new method of cellular DNA quantitative diagnosis, and it is used to estimate the physiological status and pathological changes in cells by detecting the ploidy in the cell nucleus. An advantage of this method is that it can be combined with HPV tests to enhance the accuracy of CC screening. Despite the effectiveness of HPV testing and TCT, there are inherent limitations that affect their diagnostic precision and ability to detect all cases of precancer and early invasive CC. Variability in human interpretation of TCT results, and the limited ability of HPV testing alone to predict which infections might progress to cancer, highlight the need for improved screening methods. This study proposes an innovative approach by integrating DNA ploidy analysis with HPV testing and TCT. This combination aims to leverage the strengths of each method to enhance the overall accuracy and predictive capability of CC screening. Therefore, in the present study, random samples were collected from 14,019 patients who underwent both TCT and HPV-DNA test in Shandong Provincial Hospital, Liaocheng People’s Hospital, and First People’s Hospital of Tancheng County from June 2021 to June 2022. The results of DNA ploidy analysis were compared with those of TCT, HPV-DNA, and biopsy, and samples with inconsistent results were rechecked to investigate the application of DNA ploidy analysis in the preliminary screening of HPV (+) CC.
2.1. Sample source 2.2. HPV-DNA test 2.3. Thin-prep cytologic test 2.4. DNA ploidy analysis 2.5. Evaluation criterion for biopsy 2.6. Inclusion and exclusion criteria 2.7. Statistical analysis The study included 14,019 patients who underwent both TCT and HPV-DNA tests from June 2021 to June 2022 (10,199 patients from Shandong Provincial Hospital, 2450 patients from First People’s Hospital of Tancheng County, and 1370 patients from Liaocheng People’s Hospital). Before conducting DNA ploidy analysis, informed consent was obtained from all patients. This study was approved by the Ethics Committee of Shandong Province Hospital.
Cells were collected from the cervical transformation zones of the patients by using a special cervical brush and stored in a bottle containing a special storage solution. HPV type 16 and 18 and other 12 high-risk subtypes (type 31, 33, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68) were detected by PCR and relevant reagents within 3 days of sample collection. In accordance with the method protocol, the hybridization test showed a clear bluish-violet spot that indicated a positive result. The results of typing were determined according to the HPV scattergram (a clear bluish-violet spot in the hybridization test indicated a positive result for the genotype being tested, while no clear bluish-violet spot indicated a negative result).
By using a sterile cotton swab, the patient’s cervical surface mucus secretion was cleaned, and the cells were collected from the cervical transformation zones of the patient by using a special cervical brush. The collected cells were immediately rinsed in a bottle containing a special cell preservation solution, and thin-layer slide staining was subsequently performed. All TCT slides were diagnosed by professional pathologists by using an optical microscope. TCT results were classified according to TBS classification standards 2001 recommended by “The International Agency for Research on Cancer” as follows: negative for intraepithelial lesion or malignancy (NILM), atypical squamous cells-cannot exclude high-grade squamous intraepithelial lesion (ASC-H), atypical squamous cells of undetermined significance (ASC-US), atypical glandular cells (AGC, and squamous cell carcinoma (SCC). Non-NILM was considered a positive result in TCT classification.
The automatic cellular DNA image analysis system, abbreviated as TAD system (Registration No.: Lu Xie registration 20192220050), is controlled by a fully automated scanning platform; it includes a high-definition digital camera with software to automatically read and quantitatively analyze the exfoliated cell smear stained with Feulgen under a microscope. Based on the DNA index, a single cell with DI = 1 to 2 (i.e., 2C–4C) is considered a normal cell, while those with DI ≥ 2.5 (i.e., 5C) are considered cells with abnormal ploidy. If no abnormal ploidy cells were observed in the cell smear, no visible aneuploid cell peak was observed, or the number of proliferating cells accounted for <5%, the result was considered negative. If 1 to 2 cells with abnormal ploidy were observed or the number of proliferating cells accounted for 5% to 10%, the result was considered suspiciously positive. If 3 or more cells with abnormal ploidy were observed, if a visible aneuploid cell peak appeared, or if the number of proliferating cells accounted for more than 10%, the result was considered positive.
According to the pathological diagnosis results, the patients were assigned to the nonlesion group (NILM, including chronic cervicitis with or without squamous hyperplasia, condyloma-like lesions, papillary erosion, and polypoid hyperplasia) and the lesion group (LSIL and other high-grade lesions).
The inclusion criteria were as follows: patients who consulted a doctor because of vaginal bleeding, abnormal leucorrhea, and other symptoms; and patients who were willing to undergo DNA ploidy analysis, TCT, and HPV-DNA test. The exclusion criteria were as follows: patients with dysfunction of the heart, liver, and kidney or other severe underlying diseases; patients with malignant tumors; and patients with psychological diseases.
Statistical analysis was performed by SPSS 26.0 statistical software. Enumeration data were represented by [case (%)]. The χ 2 test was used for group comparison, the Kruskal–Wallis test was used for ordinal data, and Kappa test was used for consistency analysis; the biopsy result was considered as the golden standard. In the Kappa test, a score of <0.4 indicated low consistency, 0.4 to 0.6 indicated medium consistency, and >0.6 indicated high consistency. The significance level was set at .05 ( P < .05).
3.1. Features of lesions in DNA ploidy analysis and TCT cytodiagnosis 3.2. Status of infection in different grades of cervical lesions (TCT cytodiagnosis) 3.3. Distribution and correlation of DNA ploidy analysis in HPV-DNA (+) patients 3.4. Analysis of different diagnostic methods and biopsy results 3.5. Comparison of the diagnostic value of TCT, HPV test, DNA ploidy analysis, and biopsy The DNA ploidy analysis showed the highest sensitivity and significant differences ( P < .05) in diagnosing LSIL (+) and HSIL (+) patients. Tables and show the results of the combination of TCT and HPV-DNA test, TCT and DNA ploidy analysis, and HPV-DNA test and DNA ploidy analysis. The combination of the HPV-DNA test and DNA ploidy analysis exhibited the highest sensitivity. A significant difference was observed in sensitivity between the TCT + HPV-DNA test and HPV-DNA + DNA ploidy analysis ( P < .05). In contrast, the results of HPV-DNA + DNA ploidy analysis and biopsy showed the best agreement.
First, the features of lesion in DNA ploidy analysis and TCT cytodiagnosis were analyzed. The results (Table ) showed that in the DNA ploidy test, positive, suspected, and negative results accounted for 9.22%, 19.3%, and 71.48%, respectively. Positive results for DNA ploidy were 4.27% in NILM, 41.08% in ASC/AGC, 80.33% in LSIL, 88.89% in HSIL, and 100% in SCC. With the increase in TBS grading, the positive rate of DNA ploidy also gradually increased. A Kappa value of 0.230 ( P < .05) was achieved in the Kappa test. The χ 2 value was 3874.973 ( P < .01).
Next, we analyzed the features of lesions in the HPV-DNA test and TCT cytodiagnosis. The results (Table ) showed that the HPV-DNA infection rates in HPV16/18 type and other types were 4.90% and 14.85%, respectively, and the negative infection rate was 80.25%. For HSIL and SCC, the HPV-DNA infection rate in HPV16/18 type was higher than that for other types. However, for NILM, ASC/AGC, and LSIL, the HPV-DNA infection rate in HPV16/18 type was lower than that for other types. With the increase in TBS grading, the positive rate of HPV16/18 type also gradually increased [Kappa = 0.263, P < .05; Kruskal–Wallis test (χ 2 = 2122.202, P < .01)].
A correlation analysis was performed for the results of the HPV-DNA test and DNA ploidy analysis. As shown in Table , among HPV-DNA (−) patients, the results of the DNA ploidy analysis were negative, suspected positive, and positive for 77.24%, 18.15%, and 4.60% patients, respectively. Similarly, for HPV-DNA (+) patients, the results of DNA ploidy analysis were negative, suspected positive, and positive for 48.08%, 23.91%, and 28.01% patients, respectively (Kappa = 0.173).
A total of 979 patients underwent cervical biopsy examination; of these, 25, 181, 234, and 539 patients were diagnosed to have SCC, HSIL, LSIL, and cervicitis, respectively. Among the 979 patients, 408 had normal TCT results, while 571 had abnormal results. A total of 277 patients had negative results for high-risk HPV detection, while 702 patients showed positive results (including 203 patients with HPV16/18 positivity). The DNA ploidy analysis showed normal results for 235 patients and abnormal results for 744 patients (including 458 patients with DNA ≥ 3). The details are provided in Table .
In China, approximately 131,500 new cases of CC are recorded each year, which accounts for 26.3% of the global cases ; this situation reflects the inadequate screening and detection of CC in China. According to previous studies, only 1-fifth of Chinese women undergo the Pap smear screening test for CC. In recent years, the incidence and mortality of CC have been steadily increasing, and CC is being increasingly detected in young patients. A reduction in the incidence and mortality of CC largely depends on the CC screening test. Following the introduction of CC screening in Europe and America in the mid-20th century, the incidence and mortality of CC have significantly decreased. In China, CC screening started late, and it was not until 2009 that the government organized a CC screening project in rural areas. According to previous research in 2010, the coverage of CC screening in China was only 20%, while the coverage in urban areas did not exceed 30%. This is also 1 of the main reasons for the continuous increase in the incidence of CC in China. For a long time, TCT has been used as the primary screening method for CC in China, as this test is easy to conduct and is easily accepted by patients. TCT has remarkably enhanced the quality of smears and adopts the TBS system (a scientifically rigorous system) for classification. This has led to a decrease in the false negative rate and an increase in the positive rate. This technique, however, has a limitation because the diagnostic criteria are mainly based on the differences in cellular morphology, and the outcome mainly depends on the experience and subjective judgment of the laboratory personnel. DNA ploidy analysis is a computer-based automated method that can offset the disadvantage of TCT. DNA ploidy analysis has been widely used as an auxiliary method for CC screening overseas, particularly in North America and Europe, where it has been used as 1 of the routine clinical examinations. DNA ploidy analysis began to be used in China in the year 2000. This test detects the genetic material in the patient’s cell nucleus and determines the extent of changes in nuclear DNA content that precede morphological changes in cells. During the carcinogenesis process, most cells in the malignant solid tumor are aneuploid cells. If this type of cell is detected, it indicates early malignant lesions. DNA ploidy analysis can detect early precancerous cells and cancer cells. It may play a role in triaging patients to determine whether they require further examination by colposcopy or biopsy. Thus, DNA ploidy analysis can also detect lesions. DNA ploidy analysis is used to evaluate the physiological status and pathological changes in cells by detecting the DNA content or ploidy in the cell nucleus. Some studies have suggested that the accuracy of results can be increased by combining DNA ploidy analysis and TCT for screening CC and precancerous lesions. In our present study, we found that the positive rate of DNA ploidy analysis was 37.14%, and the probability of having abnormal ploidy in patients with positive results for squamous intraepithelial lesion (ASC/AGC, LSIL, HSIL, and SCC) in TCT was higher than that for no abnormal ploidy. With the increase in DNA ploidy, the incidence of squamous intraepithelial lesions also increased; this finding indicates a correlation between the number of cells with abnormal DNA ploidy and the degree of squamous intraepithelial lesion. In a previous study, DNA ploidy analysis was combined with the HPV test, and the results showed that the detection rate of cervical precancerous lesions and CC positivity was consistent with cervical biopsy results. In the present study, DNA ploidy analysis showed a positive correlation with the results of the HPV-DNA test. In patients with positive DNA ploidy results, the proportion of HPV (+) was significantly higher than that for HPV (-); this finding indicates that HPV-DNA plays an important role in auxiliary diagnosis in CC screening. The HPV-DNA and TCT combination test is the primary method for CC screening. In the present study, we analyzed the correlation of HPV infection with TCT. The results showed that among LSIL and other high-grade lesions evaluated by TCT, the number of patients with HPV (+) was more than that for HPV (−), thus, indicating that HPV infection is 1 of the causes of squamous intraepithelial lesions. In patients with HSIL and SCC, the HPV infection rate in HPV16/18 type was higher than that in other types; however, an opposite result was noted for patients with NILM, ASC/AGC, and LSIL. With the increase in TBS grading, the positive rate of HPV16/18 also gradually increased; this finding suggests that HPV16 and 18 types are the key factors that lead to the occurrence of CC. A total of 979 patients underwent cervical biopsy examination, and the results showed that there were 539, 234, 181, and 25 patients with chronic inflammation, LSIL, HSIL, and SCC, respectively. The results of TCT showed that there were 440, 384, 151, 36, and 187 patients with chronic inflammation, ASC/AGC, LSIL, HSIL, and LSIL, and high-grade lesions, respectively. The detection rate of TCT was significantly lower than that of pathological screening. A total of 702 patients were positive in HPV-DNA screening, with a positivity rate of 71.71%. Furthermore, a total of 744 patients were positive in DNA ploidy screening, with a positivity rate of 76.00%. A single screening test has its own advantages and disadvantages. Regarding the screening of high-grade cervical lesions (pathological results showing HSIL (+)), the sensitivity of the results shows as ASC-US and higher-grade lesions, high-risk HPV positive and DNA ploidy suspected is 66.99%, 79.13%, and 81.55%, respectively. T86%, 90.91%, and 97.95%, respectively. For screening high-grade cervical lesions (pathological results show HSIL (+)), the sensitivity of the TCT and HPV-DNA test combination, TCT and DNA ploidy analysis combination, and HPV-DNA test and DNA ploidy analysis combination was 88.35%, 85.92%, and 98.06%, respectively. The positive and negative predictive values of the HPV-DNA test and DNA ploidy analysis combination were relatively high, and the comprehensive detection showed a similar outcome as that for a histopathology examination. In this study, DNA ploidy analysis was employed alongside traditional methods such as the HPV-DNA test and TCT to improve the predictive accuracy for precancerous cervical lesions. We consider patients to be eligible for DNA ploidy analysis in the following situations. Firstly, patients showing positive results for high-risk HPV types, especially HPV16 and HPV18, also tended to have abnormal DNA ploidy outcomes. This correlation suggests that DNA ploidy analysis could be particularly insightful in cases of high-risk HPV positivity, providing a stronger basis for the predictive assessment of cervical lesions. Additionally, abnormal TCT results, especially those classified as ASC-H, HSIL, or greater, indicate another scenario where DNA ploidy analysis adds significant diagnostic value. Lastly, in instances of discrepancy between TCT and HPV-DNA test results, implementing DNA ploidy analysis offers a critical triage tool. It aids in resolving diagnostic uncertainties and directs the clinical pathway towards further investigation through colposcopy or biopsy where necessary.
The findings of this study underscore the significant clinical utility of DNA ploidy analysis as a complementary diagnostic tool in the mass screening for CC. By integrating DNA ploidy analysis with the HPV-DNA test and TCT, we observed a marked improvement in the diagnostic accuracy for detecting precancerous lesions. This was particularly evident in patients who tested positive for high-risk HPV types and those presenting with abnormal TCT results, where DNA ploidy analysis contributed to a more definitive interpretation of the potential for disease progression.
Conceptualization: Jin Li, Qing-Feng Bu, Ming-Li Zuo, Jia Wang. Data curation: Jin Li, Qing-Feng Bu, Ming-Li Zuo, Jia Wang. Formal analysis: Xin-Yi Bi, Zheng-Wu Pan, Shu-Lan Liu, Xiao-Ming Chen. Investigation: Wen-Ping Sun, Yan Zhang, Wei Liu, Fei Wang, Chang-Zhong Li. Methodology: Xin-Yi Bi, Zheng-Wu Pan, Shu-Lan Liu, Xiao-Ming Chen. Resources: Jin Li, Qing-Feng Bu, Ming-Li Zuo, Jia Wang. Software: Wen-Ping Sun, Yan Zhang, Wei Liu. Supervision: Wen-Ping Sun, Yan Zhang, Fei Wang, Chang-Zhong Li. Writing – original draft: Xin-Yi Bi, Zheng-Wu Pan, Shu-Lan Liu, Xiao-Ming Chen. Writing – review & editing: Wei Liu, Fei Wang, Chang-Zhong Li.
|
Association of self-reported periodontal disease and inequities with long haul COVID-19 | c64736dc-e46e-485b-9bde-23a990bb18c9 | 11469594 | Dentistry[mh] | In 2000, the Surgeon General’s Report on Oral Health clearly stated that oral health is connected to overall health and well-being . The most prevalent oral diseases are dental caries and periodontal diseases, which are largely preventable. Populations at higher risk for developing medical conditions are the same as populations at higher risk for developing oral diseases . Socioeconomic status refers to the absolute or relative levels of economic resources, power, and prestige closely associated with wealth of an individual, community, or country . Populations of lower socioeconomic status have increased prevalence of comorbid conditions, generally poorer oral health, and more limited access to health care services, all summing up to health inequities . While the COVID-19 pandemic had an impact that could be felt worldwide, populations that most experience oral health inequities disproportionately felt its effects . This is due to many risk factors, a number of which were heightened by the COVID-19 pandemic: stress, alcohol use, tobacco use, poor diet, domestic violence issues, behavioral health problems, and poverty . Among patients hospitalized due to COVID-19, two of the most prevalent comorbidities reported are diabetes and cardiovascular disease . There is enough literature to support the association of these two conditions with periodontal disease . According to the Institute of Medicine and National Research Council, “poor and minority children are substantially less likely to have access to oral health care than their nonpoor and nonminority peers .” These populations are also more likely to lack dental insurance or depend on Medicaid. The Centers for Disease Control and Prevention (CDC) notes that “non-Hispanic Band the same populations have been found to have significantly higher incidence of COVID-19–related infection and death . A meta-analysis by Magesh et al. concluded that members of racial and ethnic minority groups had higher risks of testing positive for COVID-19 and of having a more severe disease course. They also determined that socioeconomic determinants were strongly associated with COVID-19 outcomes in racial and ethnic minority populations . For example, African American and Hispanic individuals were most likely to test positive for COVID-19. Asian American individuals had the highest risk of intensive care unit admission. Decreased access to clinical care was positively associated with COVID-19 positivity in Hispanic and African American individuals . Early in the COVID-19 pandemic, the American Dental Association recommended postponement of elective dental procedures, with the provision of only urgent or emergency care if necessary. In March 2020, 95% of surveyed dental offices were either entirely closed, or closed and seeing emergency patients only. This also resulted in patients’ lack of prioritization of oral health care, and delay in obtaining care . These changes caused an exacerbation of disparities already present amongst communities already at elevated risk . While several systemic issues exist to prevent marginalized communities from receiving equal access to care, reduced emphasis on the importance of dental care during the COVID-19 pandemic resulted in deeper oral health disparities within the population . Additionally, with shrinking of public health insurance benefits like the Medicaid in several states, suspension of preventive dental care programs such as free clinics and school sealant programs in areas with a shortage of dental providers, thousands if not millions of adults and children were left without the care they may otherwise not have access to . In 2020, Northridge et al. reported that “in response to fiscal challenges, many states have reduced or eliminated Medicaid dental coverage over the past decade, with a concurrent 10% decline in oral health care utilization among low-income adults” . Regarding at-risk populations who have dental benefits under Medicaid, they further report that there is often “difficulty finding Medicaid-contracted dental providers, because only 20% of dentists nationwide accept Medicaid .” Literature notes, in respiratory conditions such as COVID-19, potential mechanisms of pathogenesis include aspiration of oral pathogens into the lungs, alteration of respiratory tract mucosal surfaces to favor adhesion of pathogens, and secretion of hydrolytic enzymes from pathogens that inhibit the innate immune response within the respiratory tract . Additionally, several studies have demonstrated a connection between poor oral hygiene and conditions such as pneumonia, or good oral hygiene and reduced incidence of respiratory disease . There is also evidence that the virus may reside and replicate within periodontal pockets . The symbiotic relationships between microorganisms in the oral cavity are disrupted by poor oral hygiene and periodontal disease. Bacteria in a disturbed biofilm further stimulates cytokine release, in addition to those triggered by the condition of periodontitis alone. These cytokines, upon aspiration, may induce infection and inflammation in the lungs. In early stages of infection, throat is a key area of replication of the virus . It has been shown that within the first week of infection, patients infected with SARS-CoV-2 had elevated concentrations of viral RNA in oropharyngeal swabs, indicating active replication in the region . Severity of infection with SARS-CoV-2 appears to be amplified as a result of comorbidities such as diabetes, hypertension, and cardiovascular disease. These comorbid conditions also have a connection with periodontal disease . While a direct causal relationship cannot be established, it is possible that periodontal disease can intensify the severity of a COVID-19 infection through mechanisms such as enhancing inflammatory responses, causing microbial dysbiosis, and immune system overstimulation . A study performed by Larvin et al. investigated a potential impact of periodontal disease on hospital admission and mortality associated with COVID-19 . While the study could not conclusively link periodontal disease with an increased risk of infection, it was found that within the sample population of patients infected with COVID-19, there was significantly higher mortality amongst participants with periodontal disease . The primary objective of this study is to assess the disproportionate impact of the COVID-19 pandemic (long haul COVID) on populations from lower socioeconomic status in the state of Indiana. The study also aims to assess the association of self-reported periodontal disease and COVID-19 disease course and severity. Considering all the above factors, it is hypothesized that the COVID-19 pandemic did in fact heighten oral health disparities.
The Indiana Clinical and Translational Sciences Institute (CTSI) was contacted to identify a sample population with the inclusion criteria of 1) resident of the state of Indiana and 2) positive diagnosis of SARS-CoV-2 in the past year. The sociodemographic distribution (age, gender, race/ethnicity) and information about the social determinants of health for this cohort (income, zip code/neighborhoods and education levels) were also requested. Information about the research study was shared with the cohort, and participants were given a study information sheet detailing its purpose, procedures, risks, and benefits. After reviewing the information sheet, participants were asked to indicate their consent electronically before proceeding to the survey. This method ensured that consent was documented and stored securely within the REDCap platform. This study was IRB exempt. The IRB number associated with the research is #15239. Upon completion of this portion of the project, a questionnaire was sent to the cohort via the CTSI inquiring whether they are experiencing any symptoms of “long haul COVID.” The survey tool was developed from validated questionnaire available from literature . Questions relevant to long haul COVID included inquiring about symptoms such as cough, muscle pain, and loss of taste/smell after the resolution of the infection. The survey tool also inquired about the course of COVID-19 disease progression (mild/moderate/severe ‐ requiring hospitalization), symptoms of gum disease, oral hygiene habits, and presence of co-morbid chronic conditions (such as diabetes and cardiovascular issues). Example questions are, “ How do you feel about your sense of taste compared with the status prior to COVID-19 ?” to determine if a participant is still experiencing symptoms after a positive COVID diagnosis, and “ Have you visited an oral healthcare center (dental clinic) for the following in the past 12 months for treatment for gum disease such as scaling and deep cleaning ?” to determine if the participant has a history of periodontal disease and received any treatment for it previously. The survey can be found in , and was administered via REDCap, a secure virtual platform. Once completed surveys were returned, the data was consolidated and analyzed to best align it with the research question. Associations of patient characteristics and self-reported periodontal-related survey items with COVID-related survey items were evaluated using chi-square tests, considering the ordinal response categories when appropriate. Spearman correlation coefficients also were calculated when both sets of response categories were ordinal. A two-sided 5% significance level was used for all tests. Analyses were performed using SAS version 9.4 (SAS Institute, Inc., Cary, NC, USA).
The results of the study underscore the intricate link between periodontal disease, socioeconomic factors, and long-term effects of COVID-19. The percentages represent the proportion of respondents within each category who reported a specific outcome, while ’n’ values denote the actual number of individuals within those percentages. When reporting multiple categories, the proportion was calculated out of our total sample (209) respondents. The comprehensive analysis included a cohort of 209 individuals from Indiana, who in the past year, tested positive for COVID-19. The demographic data, from “ ”, reveal a predominant representation of females (n = 154; 74%), while male respondents accounted for a smaller percentage (n = 55; 26%). The age distribution showcased a relatively balanced spread across the different age brackets, with a marginally higher prevalence in the 30–39 years (n = 48; 23%) and 50–64 years (n = 52; 25%) age groups. In terms of racial demographics, most participants were Whites (n = 18; 87%), followed by other racial categories comprising remaining 13% of the participant population. The educational background of the participants was notably high, with most of the participants (n = 122; 59%) holding a college or a postgraduate degree (n = 73; 35%). Employment status predominantly included employed for wages (n = 135; 65%) individuals. The income distribution indicated that a significant portion of the cohort, (n = 93; 44%), had an annual income of $75,000 or more. COVID vaccine uptake among the surveyed individuals shows that the majority (n = 131; 63%), reported having received the initial vaccine series plus one booster shot. “ ” reveals a significant correlation between past COVID-19 hospitalization and oral health practices; Past COVID-19 hospitalization was associated with more frequent dental floss use (p = 0.049), more frequent rinsing mouth (p = 0.041), and treatment of gum diseases within the past 12 months (p<0.001). It must be noted, our sample only had 11 participants (5%) who had reported hospitalization due to COVID-19. Among participants who reported a history of hospitalization, a higher percentage used dental floss once or more in a day (n = 9; 4%). Similarly, most participants with past hospitalization rinsed their mouth once or more in a day (n = 10; 5%). And lastly, many participants who reported hospitalization, visited a clinic in the past 12 months for the treatment of gum disease (n = 7; 22%). There was no statistically significant association between self-reported periodontal disease factors (detected loss of bone around teeth, permanent teeth lost due to gum disease, bleeding gums and previous diagnosis of gum disease), socio-demographic factors, or oral hygiene variables, with past COVID-19 hospitalization. “ ” shows the association of sense of smell with socio-demographic and oral factors. Reports of a worsened sense of smell were associated with lower education levels (p = 0.121). A greater proportion of individuals with Grade 12 or GED level of education reported worsened sense of smell (n = 6, 43%) than individuals with higher levels of education such as post-graduate (n = 19, 26%). In addition, lowered sense of smell was significantly more common amongst participants who were unemployed or disabled (p = 0.008). The survey results indicated that a greater percentage of unemployed individuals (n = 12; 75%) had a worsened sense of smell post COVID-19 infection compared to their employed counterparts. Furthermore, there was a significant association observed between the participants having a worse general health status (p<0.001), bleeding gums (p = 0.031), history of toothaches (p = 0.000) and a lower oral health rating (p = 0.002). Among the respondents, a significant number of those who reported a total loss of or sense of smell worse than before (n = 18; 9%), also reported having fair to poor general health status. Among the same group, a significant number also reported having bleeding gums (n = 33, 16%) and a history of toothaches (n = 38, 18%). A majority of participants with a self-assessed oral health rating of ’poor,’ (n = 4; 67%) reported a worsened sense of smell post-COVID when compared with those who had an oral health rating of ‘very good’ (n = 16; 27%) or ‘excellent’ (n = 8; 24%). Lastly, a significant association was also noted between COVID-19 vaccination status and sense of smell (p = 0.011). A greater percentage of participants reported a worsened sense of smell if they had not received the vaccine (n = 7, 64%). A significantly lower percentage of participants reported worsened sense of smell if they had received a booster (n = 41, 31% and n = 7, 24%). “ ” shows association between sense of taste post COVID-19 and socio-demographic and oral factors. Lower sense of taste ratings were associated with older age (p = 0.018). Participants within the age group of 40–49 years reported a worsened sense of taste at a higher frequency than others (n = 19; 46%); followed by the >65 years age group (n = 12; 39%). The least affected group was those under the age of thirty (n = 7; 19%). Lower sense of taste was also associated with worse overall health status (p<0.001), mobile or loose teeth (p = 0.010), tooth loss (p = 0.013), bleeding gums (p<0.001), possibility of gum disease (p<0.001), permanent teeth lost due to gum disease (p = 0.006), toothache (p<0.001), and lower oral health ratings (p = 0.001). Among the respondents, a significant number of those who reported a total loss of or sense of tastemobile or loose teeth (n = 12, 6%) a history of tooth loss (n = 11, 5%), bleeding gums (n = 35, 17%), possibility of gum disease (n = 26, 12%), permanent tooth loss due to gum disease (n = 10, 5%), and a history of toothaches (n = 34, 16%). Mosttaste post-COVID when compared with those who had an oral health rating of ‘very good’ (n = 14; 23%) or ‘excellent’ (n = 10; 29%). And lastly, a significant association was noted between COVID-19 vaccine and sense of taste (p = 0.001). A greater percentage of participants reported a worsened sense of taste post COVID-19 if they had not received the vaccine (n = 8, 73%). A significantly lower percentage of participants reported worsened sense of taste if they had received the vaccine and booster(s) (n = 35, 27% and n = 9, 31%).
The study’s primary focus was to investigate the impact of COVID-19 on lower socioeconomic populations and its association with self-reported periodontal disease. It has been demonstrated that the COVID-19 pandemic and all of the changes that came with it–dental offices closing temporarily, free health care programs being suspended, and patient lack of pursuit of dental treatment for a variety of reasons–largely stood in the way of patients obtaining dental care necessary for their well-being. As a result, the oral health status of marginalized populations deteriorated. Literature also notes the association between income loss during the COVID-19 pandemic and unmet dental care for children, emphasizing the financial barriers to dental care during the pandemic . Oral health is connected to overall health, and a worsening condition of the oral cavity, potentially combined with comorbid conditions, can impact the course of an individual’s COVID-19 infection, potentially increasing the severity and increasing the probability for symptoms remaining long term. While some results were statistically significant, with only eleven hospitalizations (5%) observed in the sample, some frequencies reported were higher in opposing categories. It is important to note that while these statistically significant associations between variables provide valuable insights, they may not necessarily indicate direct causal relationships due to the limitations of a small sample size and data scope. Therefore, it is also important to emphasize that the study’s results are not entirely generalizable. When considering demographic factors, age was the sole variable that displayed a significant association with changes in gustatory function following a COVID-19 infection. This is consistent with a recent study by Perlis et al. which found that long haul COVID was associated with older age and female sex . Per our study results, for changes in olfactory function, education and employment were the only socio-demographic factors that showed statistical significance in relation to changes in sense of smell compared to pre-COVID. Individuals with lower levels of education seemed to report a worsened sense of smell, along with individuals that are unemployed as opposed to those who are not. These results align with our study’s overarching theme of addressing disparities in oral health, supporting the notion that socioeconomic factors impacted access to dental care during the COVID-19 pandemic. This may be contrasting with a cross-sectional study by Mahmoodi et al. found that individuals with higher education levels and underlying comorbid conditions were at greater risk for having symptoms of long haul COVID . Despite this contrast between the two studies, one portion of Mahmoodi et al.’s findings does align with results found in our data; individuals with lower general health status were more likely to report having symptoms of long haul COVID such as worsened sense of smell or taste. Periodontal disease factors such as bleeding gums, previous diagnosis of gum disease, and permanent teeth lost due to gum disease exhibited significant association with a worsened sense of taste post COVID-19. As for worsened sense of smell post COVID-19, only bleeding gums of all the self-reported gum disease symptoms was found to be associated. Similar to our results, a case control study performed by Sari et al. concluded that poor periodontal health and higher incidence of periodontitis was observed in populations that was recently infected by COVID-19 . While our study cannot attribute a direct association between self-reported symptoms of periodontal disease and history of hospitalization, a case control study by Marouf et al. suggests periodontal disease was associated with COVID-19 complications including admission to the hospital, need for assisted ventilation, and even death . This is due to a possible mechanism of increased local and systemic inflammatory responses . Lastly, a study by Costa et al. determined that there was a positive association between oral health conditions such as periodontitis and severe COVID-19 outcomes such as hospitalization . This study focused on immediate consequences within a clinical setting, while our study uniquely explores post-COVID-19 sensory changes in a broader demographic context . This comparison highlights the complementary nature of the two studies, offering a more comprehensive understanding of the diverse impacts of COVID-19 on oral health outcomes. Interestingly, despite the expected potential benefits of oral hygiene practices, our study revealed that oral hygiene practices such as brushing teeth, using dental floss, and mouth rinsing were not significantly associated with reports of altered sense of taste or smell. In contrast to this, a study by Catton et al. indicates a link between flossing and rapid taste recovery, suggesting that oral hygiene practices may actually reduce COVID-19 viral entry and dissemination . As mentioned previouslyOur results in this category may not indicate direct causal relationships as a result of the limitations of a small sample size and data scope. With regards to vaccination status, individuals who had not been vaccinated for COVID-19 generally reported worsened levels of both sense of smell and taste. Additionally, those who had not received a COVID-19 booster also reported worsened levels of sense of smell and taste. A recent study by Perlis et al. found that the risk of long-term COVID was reduced in populations that received the primary vaccination series . This complements results from other studies–vaccination generally results in a less severe disease course. And a less severe disease course is associated with a lower incidence of long haul COVID symptoms, such as worsened smell or taste. Research suggests that vaccination may alleviate certain long-COVID symptoms. This potential improvement could be due to an enhanced immune response triggered by the vaccine. Conversely, per a recent study, those who received at least one dose of a COVID-19 vaccine were more likely to report prolonged long-COVID symptoms more than a year after infection . A study by Strain et al., noted that the Moderna mRNA-1273 vaccine showed the most significant improvement in long COVID symptoms (66% improvement vs. 12% deterioration), followed by Pfizer-BioNTech BNT162b2 (56% improvement vs. 18% deterioration) and Oxford-AstraZeneca ChAdOx1 nCoV-19 (58% improvement vs. 19% deterioration) . In contrast, some studies indicated that the overall impact of vaccine type on symptom changes was minimal, with no significant differences reported between the types of vaccines . However, many studies have had small sample sizes, lacked diverse representation, and didn’t account for pre-pandemic symptoms or include a never-infected comparison group, which could show similar nonspecific symptoms . Despite these mixed findings, vaccines are known to significantly reduce the severity of COVID-19, including the risk of hospitalization and death . The observed association between vaccination and prolonged symptoms does not diminish these protective benefits but underscores the need for further research. Specifically, future studies should explore how different vaccines might influence long-COVID persistence and impact long-term oral health outcomes. Understanding these interactions, especially in relation to pre-existing conditions like periodontal disease, could provide valuable insights into post-COVID recovery. Strengths and weaknesses Future implications Future research should aim at expanding on these findings with a larger, more diverse cohort and multivariable analysis to dissect the complex relationships between sociodemographic variables, oral health and COVID-19 outcomes. Additionally, further investigation into the role of oral health practices and their potential impact on the severity of COVID-19 symptoms could provide valuable insights for public health interventions. Identifying these disparities is crucial for informing policy changes aimed at improving access to oral health care for marginalized groups.
Our study provides valuable insights into the progression and long-term effects of COVID-19, especially in relation to oral health. This approach is complemented by our attention to diverse demographic and socioeconomic factors, enriching our understanding of the pandemic’s impact across different population groups. These findings shed light on the complex interplay between oral health, socio-demographic factors, and COVID-19 outcomes, highlighting the relevance of oral health in understanding the impact of the pandemic and inform future public health strategies and research directions. Conversely, there are several limitations present in our study. The study data was cross sectional, which does not confer causality. The reliance on self-reported data for periodontal disease diagnosis is a limitation, as it lacks the clinical validation that a comprehensive periodontal examination, including radiographs and clinical assessments, would provide. In addition, self-reported data is often masked by recall bias which again is a big limitation. The small sample size, particularly in subgroups such as those hospitalized due to COVID-19, limits the generalizability of our results. This is evident in the inverse frequency of certain outcomes, such as more hospitalizations among participants who followed oral hygiene practices more religiously, which may be attributed to the small sample size and may not accurately reflect a larger trend. The survey did not collect specific data on the types of COVID-19 vaccines received by participants, which may limit the understanding of the impact of different vaccine types on long-term COVID-19 symptoms. Additionally, because our study population was geographically limited to Indiana, our results may not be generalizable.
This study reveals complex links between periodontal disease, socioeconomic factors, and long-term COVID-19 impacts. While associations were identified, limitations such as a small sample size and geographical focus exist. The findings underscore the importance of oral health in the context of the pandemic, suggesting potential connections with COVID-19 outcomes. However, the study also highlights the need for more comprehensive research, emphasizing the intricate nature of these relationships. Addressing oral health disparities, particularly among marginalized groups, is crucial for informed policy changes and improved health outcomes in future public health strategies.
S1 Appendix Long COVID-19 oral health survey. (PDF) S1 File (XLSX)
|
A 40-Year-Old With Prior Stem Cell Transplant for Chronic Myeloid Leukemia Presents With Dyspnea and Respiratory Failure | 5c463d25-a482-424f-bd78-d5399fe647ca | 11867892 | Surgical Procedures, Operative[mh] | On arrival at the ICU, his temperature was 36.3 °C, heart rate 82 beats/min, blood pressure 109/65 mm Hg, respiratory rate 19 breaths/min, and pulse oxygen saturation 98% on 8 L nasal cannula. The patient was somnolent but awoke easily to voice and followed simple commands. He was oriented to person and place. Pupils were equal and reactive to light. There was no jugular venous distension. The cardiovascular examination was notable for no murmurs, normal rate and rhythm, and normal pulses. Diffuse wheezes and crackles were present bilaterally. He had 1-2+ pitting edema of the lower extremities. He had a mild erythematous papular rash on the bilateral lower extremities below the knee.
WBC count was 13.6 × 10 9 /L (normal, 3.8-10.6), hemoglobin was 8.5 g/dL (normal, 12.9-16.9), platelets were 22 × 10 9 /L (normal, 156-369), absolute neutrophil count was 16.7 × 10 9 /L (normal, 2.24-7.68), blast forms were 4%, creatinine 1.3 mg/dL, brain natriuretic peptide 16 pg/mL. Other electrolyte levels were normal. Previous Hospitalization Studies Current Hospitalization Studies A respiratory viral panel from nasal swab did not detect respiratory viruses. Legionella urinary antigen was negative. Blood cultures remained no growth at day +5. Sputum culture obtained on admission was pending. Cryptococcal antigen was negative. Urine culture showed no growth. Methicillin-resistant Staphylococcus aureus nasal screen was negative. Chest radiography demonstrated increasing indistinct opacity in lungs, more diffuse on the left but somewhat confined to the inferomedial right ( A). A chest CT showed increasing ground-glass opacities and interstitial thickening, right middle lobe atelectasis, and ill-defined reticulonodular pattern in the bilateral lower lobes ( B). Repeat transthoracic echocardiogram showed normal structure and function of the right and left ventricle, with an estimated pulmonary arterial systolic pressure of 32 mm Hg. A bedside right heart catheterization showed a right atrial pressure of 7 mm Hg, pulmonary arterial pressures of 30/15 mm Hg, pulmonary capillary wedge pressure of 6 mm Hg, and cardiac output of 10.1 L/min by thermal technique. Lung biopsy with video-assisted thoracoscopic surgery was initially attempted, but the patient was unable to tolerate single-lung ventilation. The procedure was converted to open thoracotomy, with successful biopsy without complication. The biopsy showed intra-alveolar exudates ( A) with periodic acid-Schiff (PAS)-positive globules ( B). Infectious pathogen stains from the biopsy specimen were negative. What is the diagnosis? Diagnosis: Pulmonary alveolar proteinosis secondary to hematologic malignancy
BAL performed 6 weeks before admission showed no bacterial, fungal, nocardia, or acid-fast bacilli pathogen growth. Herpes simplex virus polymerase chain reaction was negative. Galactomannan was 0.08 (normal, < 0.5) from blood and 0.14 (normal, < 0.5) from alveolar lavage. Blood BD-glucan was 211 pg/mL (normal, < 60). Blood Epstein-Barr virus and cytomegalovirus polymerase chain reaction were negative.
In patients with hematologic malignancies and allogeneic or haploidentical hematopoietic cell transplant, a differential diagnosis for acute hypoxemic respiratory failure can be summarized in broad categories depending on whether the complication occurs within the first 100 days of transplantation, including (1) infectious (bacterial, viral, or fungal, depending on presence of neutropenia); (2) noninfectious, noninflammatory (pulmonary edema, alveolar hemorrhage from thrombocytopenia, renal dysfunction, direct involvement from hematologic malignancy); and (3) noninfectious, inflammatory (graft-vs-host disease, drug pneumonitis, radiation pneumonitis, idiopathic pneumonia syndrome, alveolar proteinosis, veno-occlusive disease). Pulmonary alveolar proteinosis (PAP) is a rare disease, defined by protein (surfactant) accumulation within alveoli impairing gas exchange. Prevalence is between 3.7 to 40 cases per million, and incidence is 0.2 cases per million. Although the most common cause of this rare disorder is an autoimmunity to granulocyte-macrophage colony-stimulating factor (GM-CSF), which accounts for 90% of all cases, PAP also may be attributable to secondary causes, such as infectious pathogens (eg, Pneumocystis ), hematological malignancies, occupational exposures, and lung transplantation. In neonates, congenital defects in surfactant protein also can cause PAP. This discussion focuses on secondary PAP caused by hematologic malignancy. Secondary PAP accounts for less than 10% of cases and is caused by a reduction in the number or function of alveolar macrophages, not because of GM-CSF antibodies seen in the autoimmune PAP. The resulting buildup of surfactant in alveoli therefore causes the typical symptoms of cough and dyspnea. Secondary PAP can be associated with hematopoietic cell transplantation, medications such as sirolimus and ruxolitinib, or opportunistic pathogens. Considering hematologic disorders, myelodysplastic syndrome and chronic myeloid leukemia are often associated with PAP. The mechanism of PAP development is poorly understood in hematologic malignancy but may have to do with monocytopenia or GATA2 deficiency. GATA2, a hematopoietic differentiation transcription factor, has been implicated in rare cases of PAP. GATA2 deficiency can be seen associated with hematologic malignancies such as chronic myeloid leukemia, as well as viral or fungal infections. Whether there are specific treatments to alter the course of PAP in this population is unknown. The diagnosis of secondary PAP is determined with radiographic findings, bronchoalveolar lavage cytology, or lung biopsy findings. Lactate dehydrogenase may be elevated but is nonspecific. Chest radiography findings include bilateral alveolar opacities, usually in a perihilar and basilar distribution. The “crazy paving” pattern is observable on CT scan and is attributable to a combination of intralobular thickening and diffuse ground-glass opacities. Only 33% of cases have this described pattern. Bronchoalveolar lavage or lung biopsy is the confirmatory diagnostic test(s) of choice. Cytology will reveal large foamy macrophages and PAS staining. In the absence of crazy paving, PAP can be misdiagnosed, and it also can have overlap with infectious causes, particularly in the setting of immunosuppression. Treatment for secondary PAP is reliant on the underlying condition. This patient had already had a bone marrow transplant in the setting of his leukemia, and this was a recurrence. Whole-lung lavage can be used as supportive treatment in addition to GM-CSF. Apart from symptomatic and supportive respiratory care and therapy directed toward the underlying hematologic malignancy, no specific therapy has been identified for the treatment of PAP associated with hematologic malignancy. Clinical Course Vancomycin and diuretics were administered initially in addition to his prior antimicrobials, including posaconazole, amphotericin, and acyclovir. Following the diagnosis of PAP, GM-CSF was administered pending GM-CSF antibody testing, which was not detected. The patient subsequently underwent whole lung lavage for 3 sequential days. Cloudy fluid, characteristic of PAP, was obtained during the lavage, as depicted in . The patient’s oxygenation briefly improved, but unfortunately, his respiratory failure persisted. In the context of his multiple comorbidities, multi-organ dysfunction, and refractory hematologic malignancy, he and his family transitioned to comfort measures only. The patient died days later while hospitalized.
1. The differential diagnosis for acute hypoxemic respiratory failure in hematopoietic cell transplant patients is vast, although it can be broadly categorized as infectious (viral, fungal, bacterial), inflammatory (drug pneumonitis, idiopathic pneumonia syndrome, graft-vs-host disease), and noninfectious, noninflammatory (pulmonary edema, renal dysfunction, congestive heart failure). 2. Autoimmunity to GM-CSF accounts for 90% of PAP cases in adults, although secondary cases occur from infection, hematologic malignancy, post-solid organ and hematopoietic cell transplantation, or occupational exposures. 3. Bronchoscopy and BAL plus or minus biopsy are the mainstay of diagnosing PAP, often with proteinaceous milky return and PAS-positive material on cytology or biopsy. 4. Treatment often includes GM-CSF administration if the cause is autoimmune, whereas whole lung lavage is supportive to improve oxygenation while pursuing treatment, depending on the cause. 5. GATA2 deficiency has been associated with PAP, although treatment remains unchanged for this rare entity.
None declared.
|
Transformation of dental services from a governmental model to a revenue-generation model of operation in a tertiary care hospital: a health economics assessment | f2a4b2fb-77ce-4389-b9d9-6e16bfdd9501 | 9706714 | Dental[mh] | Healthcare in Saudi Arabia is undergoing a major transformation. King Faisal Specialist Hospital and Research Centre-Riyadh (KFSHRC-R) will transform its model of operation from a governmental hospital that delivers free service to its patients to a model where the service will be paid either by insurance or a governmental fund. It is hoped that this model of care will improve efficiency and productivity. The Department of Dentistry will consequently bill for its services and it is imperative to assess one's readiness for such a transformation in terms of efficiency, timely delivery of care and sustainable revenue streams. There is no precedent for this model in Saudi Arabia and KFSHRC-R is the first hospital to undergo such a transformation in the model of operation. The aim of this study was to identify how much the Department of Dentistry would generate in terms of revenue under such a model. Included in objectives are whether there might be any losses and to identify which specialties would incur such losses and the reasons behind such losses.
Anonymized data with regard to dental practice at KFSHRC-R were extracted from Cerner Millennium by Cerner, R4 Clinical+ v 4.7.0 by Carestream, Venus Billing System and Revenue Cycle Management System by Data Ocean for the period from January 2015 to December 2019. Visits to the emergency department were excluded from the analysis. The department employed on average 41 dentists as full-time employees and 8.6 hygienists on average as full-time employees. Each dental full-time employee is paid an average of 700 000 SAR (186 667 USD) per year. All visits for dentists and hygienists were included in the analysis. The demographics of the sample collected in terms of gender and age were analyzed. A descriptive analysis was carried out of the type of outpatient appointments and the category of current procedure terminology (CPT) code. Each CPT code had a billed rate as per the KFSHRC-R billing system. The billing rate was compared to a benchmark cost without profit that was provided through previous consulting work handled through a billing consulting company. If a CPT code was being billed at a rate less than the cost rate, then it was considered a CPT code that was incurring a negative margin of loss. If it was billed at a rate equal or higher than the cost rate then it was a CPT code that was incurring a positive margin of profit. Moreover, the billing rate was compared to an average calculated from three market leaders providing comparable services to that provided by KFSHRC-R to show if the billing rate was higher or lower than the prevailing market price for that CPT code. Data was collected from 835 individual data entries from 660 different CPT codes. Of these, 474 entries from 352 CPT codes had no volume in the year and thus were excluded from the analysis. In total 361 individual entries from 292 CPT codes were included for analysis. Data was collected detailing revenue for different types of dental work for a one-year period. The aim of the analysis was to detail the revenue for each type of work, and to categorize the amount of revenue in work with different profit levels. Revenue for each type of dental work was collected, with each type of work categorized into CPT codes. Most CPT codes consisted of one type of work only, but some codes had more than one individual type of work associated with them. For the purposes of analysis, one observation per CPT code was included in the analysis. Where there was more than one set of values per CPT code, the total revenue for the code as a whole was calculated. Where the margin and the difference with the benchmark price varied within a CPT code, a single value per CPT was calculated by weighting the results for each type of work by the revenue of the work. All analyses were descriptive in nature using the Dental Health Record C4, Kodak Dental Systems version 3.1.8 and analysis with IBM SPSS Statistics 29. Categorical variables were summarized by the number and percentage in each category. Continuous variables were summarized by the mean and standard deviation or median and data range. The main outcomes of interest were total revenue, and the percentage of all revenue for each type of work, or groups of work. Each type of work was categorized in three main ways: 1) based on whether the price of work was higher or lower than the ‘benchmark’ price, 2) whether there was a positive margin (profit) or a negative margin (loss), and 3) the speciality of the work. Summaries of revenue in each of these categories, or combinations of categories were quantified.
During the period of the study from 2015 to 2019, 11 214 (50.8%) and 10 840 (49.2%) female and male were treated, respectively ( ). They paid 179 554 (52.9%) and 159 858 (47.1%) outpatients visits, respectively, with an average 16 visits per female patient and 14.7 visits per male patient. Saudis constituted 89.4% of the patients and 93.7% of the visits. The number of patients treated and number of patients visits increased from 7294 and 51 505 to 9722 and 80 319 with an average annual increase of 486 (6.7%) and 5763 (11.2%), respectively, with a median number of visits per patient of 5 in 2015 to 6 in 2019, which equates to care delivered by fewer visits per patient since the number of patients increased from 2015 to 2019 ( ). Patients younger than 14 years of age and seniors older than 65 years comprised only 17.3% and 8.5% of all the visits and 18.6% and 8.95% of all the patients treated during the period 2015-2019, respectively. ( ). Distribution of visits showed that dental nursing clinic and hygienist clinic appointments constituted 23.5% and 7.2% of the total visits for the whole period of 2015-2019, respectively. During the same period, general dentistry, pedodontics, orthodontics, endodontics, oral and maxillofacial, prosthodontics, periodontics made up 28.1%, 12.5%, 7.1%, 7.0%, 6.9%, 4.4% and 3.3%, respectively, of the total number of visits ( ). On average each dentist and hygienist saw 2263 and 760 visits per year. The most common charge description master (CDM) codes are shown in for each year. Most were for x-rays or oral hygiene instructions. The revenue for all dental specialties for the period of the study was 51 922 240 Saudi Arabian Riyals (SAR) (13 864 597 USD) ( ). The revenue for the nursing and hygienist clinics was 7 167 530 SAR (1 911 341 USD). Thus, the total revenue for all the appointments was 59 089 770 SAR (15 757 272 USD). shows the revenue per year of study. Approximately 131 (44.9%) CPT codes incurred a negative margin of loss when compared to the benchmark cost of the CPT code while 161 (55.1%) incurred a positive margin compared to the benchmark cost of the CPT code. Eighty CPT codes (27.4%) incurred a positive profit margin and were higher in price than the average of the market leaders. Eighty-one (27.7%) CPT codes incurred a positive profit margin, but were lower in price than the average of the market leaders. On the other hand, 83 (28.4%) CPT codes incurred a negative margin of loss, but were higher in price than the average of the market leaders. Forty-eight (16.4%) CPT codes incurred a negative margin of loss and at the same time were lower in price than the average of the market leaders. shows the distribution of the CPT codes that incurred a negative margin of loss whether higher or lower than the average price of the market leaders among the different specialties. The estimated loss of revenue due to pricing that was lower than the actual cost was estimated at 29 922 900 SAR (7 979 440 USD).
There is a debate worldwide about who should cover dental care: the patient alone, the government alone, or by the patient with subsidization from the government. - This debate is ongoing in Saudi Arabia, which currently has a national health service covered by the government. Other countries have already started to reform. This research attempted to quantify the cost of care and the expected revenue along with suggestions to improve efficiency of dental services in an academic tertiary care hospital. During a period of 5 years, each patient paid 11 visits excluding hygienists and nursing appointments. However, the number increased from a median of 5 visits in 2015 to 6 visits in 2019 ( ). The study could not be extended to encompass the period 2020-2021 due to the reduction in delivery of dental services because of the COVID-19 pandemic. During that period, only emergency and necessary care was provided). The high number of visits probably reflects the nature of patients, many of whom were being treated for tertiary complex conditions like cancer and organ transplantation that required complex medical consultations prior to any dental intervention. But, more importantly, the increasing number of visits (27.7% in number of visits per patient per year), would have also meant a delay in execution of care, which would reduce patient satisfaction in a private model of operation. The average number of patient visits per year per dentist in the US was reported at 3566.4 in 2018 and has remained stable over the period 2015-2018 in a national study of private dental clinics. The average benchmark in the US fluctuates between 1600 and 2300 patient visits per year per dentist in contracted salary-based practice. The average number of outpatient visits per year per dentist at KFSHRC-R was 1256 in 2015 and reached 1959 in 2019 which is comparable to the benchmarks for a salary-based practice. This indicates that dentists at KFSHRC-R are performing on par with their colleagues in the US, but are underperforming by 45% in comparison to those performing in a feefor-service practice. The benchmark from the US for the hygienist should be 7.5-8.5 patients per an 8-hour day. This amounts to 1600 visits per year. Hygienists at KFSHRC-R see only 530 per year which means they are underperforming by 67%. As expected, nursing clinic and general dentistry clinic visits comprised 51.6% of the visits. The initial encounter for any patient is usually through these clinics, where any need for more specialized dental care is determined. Pedodontic visits comprised only 17.3% of all visits, which is comparable to a national study from Canada were pedodontic visits reached 18% of all visits. Total revenue for all the clinic appointments was 59 089 770 SAR (15 757 272 USD). There was an average increase in revenue of more than 10% each year, but it dropped to only around 2% in 2019. That drop coincided with a plateau in the number of visits in the same year at 80 319. This indicated a limited capacity to expand in terms of appointments and may be related to issues pertaining to infrastructure like the lack of more dental chairs and probably longer turnaround times between patients. The average salary for a dentist in the department was 700 000 SAR (186 667 USD) per year. The sum of salaries for the same period 2015-2019 was 143 500 000 SAR (38 266 667 USD). The revenue generated was only 36.2% of the salaries paid (51 922 240 SAR; 13 845 931 USD), which means that the dentists were underperforming in relation to salaries paid. Productivity should be increased to reach a sustainable budget in the long run. It was not possible to identify the risks for low productivity in a statistical analysis, but a trend was noticed in the data: the sections with the lowest productivity had a higher portion of dentists who worked in private practice, were within 5 years of the retirement age of 60 years, have been in practice more than 20 years, were administrative leaders or past leaders or were not North American board certified. This indicated a few things: local board training probably led to less experienced dentists who tended to have difficulty in delivering care in a timely fashion. Dentists near retirement or who had been in practice for 20 years or more were probably less motivated to engage in higher output due to their high salaries. Those working in private practice were probably accruing more income than their salaries, and were thus less inclined to increase productivity. Individuals in administrative positions spent more time than required on administrative duties. A solution to the low productivity could be a movement to a fee-for-service model, for dentists in practice for 20 years or more or near retirement. A downside of a fee-for-service model is the possibility of providing unwarranted care and overbilling, a practice that should be monitored and discouraged. The pricing of the CPT codes showed that 131 codes (44.9%) were incurring a negative margin of loss compared to the calculated cost of the CPT code. In fact out of these 48 CPT codes (16.4%) were priced even lower than the average of the market leaders. This indicated two things. First, the CPT codes at KFSHRC-R must be re-priced as the current prices will create a huge loss for the hospital when it moves into a private sector type of billing for services. Second, 83 CPT codes (28.4%) were incurring a negative margin of loss, but were higher than the average price of the market leaders, indicating a huge gap for improvement in terms of cost efficiency with regard to these CPT codes. The hospital must revisit the practice of the dentists to see if there was wastage in terms of materials used in the treatment plan or overutilization of laboratory tests and x-rays. Incorrect use of CPT codes does not explain the negative margin of loss as this analysis compared the registered billing price of the CPT code and the actual cost of that code and the average price of the market leaders regardless of what was actually delivered to the patient. Therefore, additional training about CPT codes would probably not resolve this problem but may still be useful. The three specialties that really needed to revise the prices of their CPT codes were general dentistry, endodontics and orthodontics. The two specialties that needed to look into their practice and improve efficiency and show more cost containment were prosthodontics and oral and maxillofacial surgery. This cost-containment can be approached through a reduction in total number of visits to deliver the care and also by a more judicial use of supplies. The loss of revenue due to the use of a billing price lower than the actual cost was estimated at 13 315 970 SAR (3 550 925 USD), which amounted to an annual loss of revenue of 10.1%. A market concentrated with dental providers will likely lead to less reimbursement in the insurance market as evidenced from a national study from US. Saudi Arabia is in a similar situation and this will compound the loss for revenue for KFSHRC-R. In conclusion, the analysis identified some delay in delivery of treatment, which might be expected due to treating complex tertiary conditions. Moreover, there was limited capacity to expand and meet the demand as shown by the plateau in the number of appointments. The productivity was low as might be expected, from a salaried model of practice, which could be improved by switching to a fee-for-service model. There was underproductivity among the hygienists. Furthermore, certain specialties must revisit their practice and improve cost-containment through a reduction in number of visits and judicial use of supplies. Last, the CDM price list needs revision to cover the cost of the services delivered.
|
Using strain-resolved analysis to identify contamination in metagenomics data | 9a5fb524-2cb4-434c-824b-e8716b950338 | 9979413 | Forensic Medicine[mh] | The advancement of sequencing technologies has enabled researchers to investigate microbial communities at high resolution and throughput. However, contamination poses challenges for data analysis. Contamination refers to DNA within a sample that did not originate from that specific biological sample. Recognizing contamination, followed by appropriate decontamination, should be a critical first step for all microbiome analyses. Skipping this step can easily result in confounded results and false conclusions. Contamination can originate outside of a study. Microbial DNA from the surrounding environment, native microbiomes of researchers, and microbial DNA present in DNA extraction and library preparation kits are all considered external contaminants [ – ]. Detection of such contamination is enabled by the addition of negative controls (i.e., blank reagent controls) during sample collection, preparation, and/or sequencing. To date, strategies have been devised to minimize, detect, and/or remove externally derived contamination in silico [ , – ]. Cross-sample contamination is less well studied . Contaminants that originate within a study can be introduced during DNA extraction when DNA from one sample spills over into another. It can also occur during sequencing as a result of index switching or sample bleeding . Since the contaminating DNA in this case originates from microorganisms present in samples from the study, one cannot decontaminate by removing “contaminant species” present in controls from the actual dataset. While strategies have been developed for solving index switching and sample bleeding arising during sequencing , much less is known about the well-to-well contamination occurring prior to sequencing in metagenomics data and how to detect it. In recent work that characterized early-life intestinal strain colonization, we collected a large clinical dataset consisting of over 400 fecal samples from infants and their mothers . Unfortunately, we detected clear evidence of cross-sample contamination that required us to exclude one-half of the samples from the study. This motivated an in-depth investigation of the signals used to detect contamination in this and a second clinical dataset of preterm infants. Our approach relied upon a strain-resolved workflow. Strain-level analyses have been increasingly applied in microbiome studies, for example, to study mother-infant gut microbiome transmission [ – ] and conduct epidemiological surveillance . The high specificity and sensitivity of strain tracking methods are powerful, yet conclusions from these methods can also be confounded by cross-sample contamination . Here, we provide a framework for the detection of cross-sample contamination using strain-based surveillance, and in two case studies, we show how such a framework can be used to detect contamination in large-scale metagenomics datasets.
Case study 1: longitudinal preterm and full-term infant fecal samples Case study 2: longitudinal preterm infant fecal, mouth, and skin samples Study overview Conclusions from case study 2 Study overview Conclusions from case study 1 The first case study consists of 402 fecal samples collected from 19 preterm and 23 full-term infants from birth to age 1 and their mothers around time of birth . This study was designed to investigate strain persistence within infants, strain sharing between infants and their mothers, and strain sharing among different infants. DNA extraction was primarily achieved using 96-well plates (“ ”), and there were a total of five extraction plates, labeled P1 to P5. There were five reagent-only-negative controls, one for each DNA extraction plate, and they were labeled by the plate number (i.e., NC1 refers to the negative control on P1). One ZymoBIOMICS Microbial Community Standard (catalog no. D6300; termed “Zymo”) was included as a DNA extraction-positive control on P5. Following DNA extraction, all samples including controls were subjected to metagenomics sequencing and de novo genome reconstruction. Dereplication of the genomes constructed from fecal samples resulted in 1005 representative genomes, as previously described . Reads from all samples were then mapped to this dereplicated genome set for organism detection (“ ”). To detect potential sources of contamination, we examined strain sharing among unrelated samples within and across extraction plates (“ ”). Samples are considered unrelated if they are from different infants that are not biologically related or if one sample is a negative control. Evaluation of extraction negative and positive controls Evaluation of cross-sample contamination: index hoping and sample bleeding Evaluation of cross-sample contamination: well-to-well contamination Evaluation of underlying biological signals after removing contaminants The identification of well-to-well contaminated samples allowed us to assess strain sharing among supposedly uncontaminated infant samples on originally discarded P3 and P4. On P3, samples from six preterm infants shared a total of four strains. These four are strains of Clostridioides difficile , Clostridium paraputrificum , Clostridium butyricum , and Lactobacillus rhamnosus (Fig. ). Samples that shared these four strains were not often adjacent to one another. Furthermore, for each of these four organisms, near-identical strains were also found among samples that were extracted on different plates. Notably, C. difficile , C. paraputrificum , and C. butyricum strains were shared among preterm infants only. The pattern of strain sharing on P4 was similar to that on P3, except that there were more shared strains that were shared by fewer samples (11 strains were shared among samples from four infant pairs). As for P3, we detected no strong signal for contamination via dispersal of strains into all or most surrounding wells from single-source wells. Given minimal evidence of well-to-well contamination and since all infants in this study were born in the same hospital, we speculate that most of the strains shared by the infants whose samples were extracted on P3 and P4 were probably hospital derived, a phenomenon that has been reported previously .
No organism was detected in negative controls NC1 and NC5. However, negative controls NC2, NC3, and NC4 had at least one read mapped to ≥ 50% of at least one of the 1005 representative genomes (this value served as our threshold for detection; Fig. ). No contaminants were detected in the Zymo-positive control. The only genome detected in NC2 was Cutibacterium acnes . This organism was also found in nine fecal samples from five infants. These samples were extracted on P1, P2, and P4. Strain-to-strain comparisons indicate that all strains are sample-specific and different from that in NC2. Therefore, the presence of C. acnes in NC2 was not considered to be due to cross-sample contamination. As C. acnes is a common skin commensal bacterium , and is often detected in laboratory reagents and kits , we suspect this organism is an externally derived contaminant. NC3 and NC4 each had ~60 genomes that were above our read-mapping detection threshold. The organisms represented by these genomes could be externally derived and/or originated within the study via index switching, sample bleeding, and/or well-to-well DNA contamination. If any of these genomes were derived from external sources (e.g., from DNA extraction kits), we would expect the strains to be in the majority of samples from the same plate (because they were processed simultaneously) and perhaps across extraction plates. However, no strain was shared among negative controls. Furthermore, the strains found in NC3 and NC4 were only shared by a maximum of 7.7% and 6.5% of the unrelated samples from PC3 and PC4, respectively. Thus, we conclude that the genomes in NC3 and NC4 were unlikely due to external contamination and likely originated from other samples from this study.
The majority of strains found in NC3 and NC4 were only shared with samples from the same extraction plate. This observation rules out index switching and sample bleeding, as these phenomena should lead to contamination of samples from other plates because DNA from different plates was pooled for sequencing (P1 and P2 were pooled and sequenced at a different time than samples from P3-P5). Both index switching and sample bleeding refer to the misassignment of reads to samples. However, index switching results from indices being similar in multiplexing sequencing and can be essentially prevented by using unique dual indexes, which was what we used for sequencing in this study. Sample bleeding, on the other hand, occurs due to the close proximity of sample read clusters on the flow cell during sequencing . We confirmed that this is not the main explanation for the contamination in NC3 and NC4 by resequencing these two controls and finding that the community compositions were essentially the same as the originally sequenced NC3 and NC4.
To test for the remaining possible explanation for the contamination in NC3 and NC4, well-to-well contamination, we visualized strain sharing patterns in the context of the extraction plates. Based on the observation made by Minich et al. that neighboring wells are more prone to well-to-well contamination than distant wells , we hypothesized that, if between-sample contamination occurred within an extraction plate, nearby unrelated sample pairs would more likely to share strains than those that were farther apart. Distance-correlated strain sharing was not seen on P1, P2, and P5 ( p = 0.18, 0.31, and 0.32, respectively; Wilcoxon rank-sum test). This finding is consistent with the plate-based strain sharing visualization, which shows that unrelated samples on P1, P2, and P5 rarely shared any strains, and for pairs that did share strains, the majority of them were not immediately adjacent to each other (Fig. ). On P3 and P4, however, nearby samples were significantly more likely to share strains than those that were farther apart ( p = 2.3e-3 and 4.7e-3, respectively; Wilcoxon rank-sum test). Plate-based strain sharing visualization indicates that a few samples including NC3 and NC4 in particular (pink circles in Fig. ; see also Fig. S ) exhibited location-specific sharing patterns, consistent with well-to-well contamination. For instance, on P3, NC3 located on column L primarily shared strains with samples from infants #82 and #83 that were loaded onto columns K and L for extraction. This suggests that the fecal samples from infants #82 and #83 were potential sources of the contamination seen in NC3. NC3 also shared strains with a sample from infant #83 that was loaded onto column C for extraction. As expected, samples from infant #83 share strains, so strains in NC3 could have come from any of the samples from infant #83. Likely, the contaminant strains were derived from the samples adjacent to the negative control. Thus, we do not attribute this instance of sharing to long distance well-to-well contamination. NC4 on P4 exhibited a similar proximity-based strain sharing pattern to NC3 (Figs. and S ), and we deduce that NC4 was primarily cross-contaminated by adjacent samples. In addition to NC3 and NC4, four preterm infant samples displayed plate location-specific strain sharing patterns (unlabeled four pink circles in Fig. ). The first instance involved a sample from infant #98 (termed #98D4). #98D4 was extracted on P4, and it shared strains with nearly half of the samples extracted from the same plate, including those on columns J and K that were nine and ten columns away (Fig. ). Shared strains were from six infants, including a pair of twins (#122 and #123). Strains shared by each of the twins and #98D4 overlapped completely, and these strains were not found in any other infants. The other four of these six infants also shared strains that were otherwise unique to them with #98D4. #98D4 did not share strains with samples extracted from different plates. We therefore deduce that #98D4 was contaminated by multiple wells on P4. Besides #98D4, three other samples, each derived from a different infant (#63, #114 and #128), primarily shared strains with neighboring samples, similar to the patterns seen in NC3 and NC4. We confirmed that these four infant samples were indeed cross-contaminated by re-extracting and sequencing three of the four samples (the original stool sample from #128 and its close-by-date replacement were unavailable). For the sample from infant #63, a close-by-date sample (day of life 6 instead of 9) was selected as the original stool sample was unavailable. For all three re-extracted samples, their DNA concentrations became ~0 (Table S ) and their location-based strain sharing patterns were not observed.
Using strain-resolved workflow, we identified well-to-well contamination to be the major source of contamination in this dataset. The six contaminated samples (two negative controls and four preterm infant samples) were all low in microbial biomass. If such contamination was not addressed, we would have falsely concluded that strain sharing among non-related infants was not rare, and that some non-related infant pairs could share as many strains as sibling pairs do.
We applied our strain-resolved workflow to a different clinical dataset consisting of 533 samples collected from the skin, mouth, and stool of 16 preterm infants. These preterm infants were born in the same hospital as the infants from case study 1. This dataset was part of a study designed to elucidate strain transmissions between the hospital environment and preterm infants. DNA extraction was primarily achieved using 96-well plates. One reagent-only-negative control was included in each extraction plate. There were six plates (labeled P1 to P6) and six negative controls, labeled NC1 to NC6. In addition, P4, P5, and P6 each had ~3 Zymo standards as DNA extraction-positive controls. Two-hundred thirty-six of the 533 samples (including 3 negative controls (NC1, NC3, and NC5) and 3 positive controls, one from P5 and two from P6; termed PC5, PC6_1, and PC6_2) were selected for metagenomics sequencing (colored circles except for those light blue ones in plate maps in Figs. , , 7). Before library preparation, DNA was transferred from the extraction plates to three new 96-well plates, one for each sample type (“ ”). Following sequencing, de novo genome reconstruction and dereplication yielded 152 representative genomes, which served as the reference genomes for read-mapping based organism detection for this dataset (“ ”). To detect potential sources of contamination, we examined strain sharing among all unrelated samples, as described in case study 1. Evaluation of extraction negative and positive controls Evaluation of Zymo contamination in infant samples Evaluation of additional contamination not present in negative controls Using the strain-resolved approach developed in case study 1, we evaluated strain sharing among infants after excluding Zymo and Burkholderia strains. Five bacterial strains were widely shared by samples from different infants. Specifically, for each of these five strains, at least half of the infants had one sample that shared such a strain with a sample from an unrelated infant. Of these five strains, two are Staphylococcus epidermidis strains A and B, and the other three are Staphylococcus M0480, Corynebacterium aurimucosum , and Cutibacterium acnes . All five are common members of the healthy skin microbiome . S. epidermidis strain A and the Staphylococcus M0480 strain were shared among all sample types (skin, mouth and stool), and S. epidermidis strain B, the Corynebacterium aurimucosum strain, and the C. acnes strain were shared among skin samples only. Additionally, a near-identical S. epidermidis strain A was found in fecal samples from 16 out of 42 infants from case study 1. It is uncommon to find a single strain of each of these organisms in the majority of infants of a single dataset; we therefore identify these five strains to be externally derived contaminants (e.g., from staff who handled the samples). We re-examined strain sharing among samples of this dataset after excluding all identified external contaminants (Zymo strains, a Burkholderia strain, and five skin-associated strains). While most of the extraction plates did not exhibit location-based strain sharing, samples from one infant pair on P4 did, suggesting that there might be well-to-well contamination (pink circle in Fig. S ). A sample from infant #12 shared up to 3 strains with neighboring skin and oral samples of infant #13. These 3 shared strains were not shared by infant #13 and any other infants. In addition, none of the other samples from infant #12 shared strains with infant #13. This suspected cross-contaminated infant #12 sample was collected from skin, and its strain sharing pattern was similar to those of well-to-well contaminated samples in case study 1.
Of the three sequenced negative controls, one genome was detected in NC1 and NC3, and 9 genomes were detected in NC5 (Fig. ). No contaminants were detected in the Zymo-positive controls. Burkholderia sp. was the only organism detected in NC1 and NC3. The Burkholderia strain in NC1 was not detected in any sequenced samples. However, a different Burkholderia strain was found in NC3 and 12 infant skin samples that were extracted from four extraction plates (Fig. ). No Burkholderia was found in fecal or mouth samples, both of which were higher in biomass than skin samples ( p = 5.38e-37 and 1.58e-40, respectively; Wilcoxon rank-sum test). The skin samples that contained the Burkholderia strain had a significantly lower biomass than the skin samples that did not ( p = 2.79e-24; Wilcoxon rank-sum test). Burkholderia is not part of the normal skin microbiome , but it has often been reported as a reagent contaminant [ , , , ]. Identifying one single Burkholderia strain among infant and negative samples suggests this strain is a result of external contamination. Interestingly, this strain is also in one low-biomass gut sample from case study 1. This contaminant was likely introduced prior to library preparation as it was not detected in NC1 and NC5, both of which were on the same library preparation plate as those Burkholderia -containing skin samples (Fig. C). For NC5, one organism, Klebsiella pneumoniae , was present at an extremely low abundance (< 0.1%) and was not detected in any other samples. We cannot determine if adjacent samples on the extraction plate were possible sources of this organism because those samples were not sequenced. The remaining 8 organisms in NC5 were all bacterial members of the Zymo community (Fig. ). NC5 was adjacent to PC5, a Zymo-positive control, on the extraction plate. Thus, Zymo organisms in NC5 could be attributed to well-to-well contamination.
To further evaluate contamination by the Zymo strains, we searched for these strains in infant-derived samples. During DNA extraction, six skin and oral samples were deliberately spiked with 75 μL of the Zymo standard, four of which were sequenced (“ ”). By examining strains shared between positive controls and biological samples, we found 12 additional infant samples containing at least the four most abundant Zymo members (Fig. ). All but one of these samples were from skin or mouth, which had lower biomass than gut samples ( p = 1.58e-40 and 5.57e-9, respectively; Wilcoxon rank-sum test). Of these 12 infant samples, 9 were adjacent to a Zymo-spiked infant sample or a positive control. Since the contaminated samples generally only shared Zymo strains with neighboring Zymo-spiked samples (and not the other organisms in those samples), we conclude that the observed Zymo contaminants were more likely to be introduced accidentally, possibly via aerosolization or mis-pipetting, rather than via well-to-well contamination.
ur strain-resolved workflow identified external contamination to be the major source of contamination in this dataset. Suspected contaminants include Burkholderia strains, Zymo DNA, and five skin-associated strains. Our approach also suggested one skin sample to be cross-contaminated by adjacent samples from the same extraction plate. Notably, most of these contaminants were found in skin samples, which had lower biomass than oral and gut samples.
Contamination is an insidious and potentially unavoidable problem in metagenomics-based microbiome research. If not appropriately addressed, extraneous DNA sequences can skew conclusions, resulting in potentially false statements. Here, we devised a workflow that utilizes strain resolution to detect contamination based on the unexpected sharing of essentially identical strains. We demonstrate its usefulness in two clinical metagenomics datasets. By examining strain sharing based on genotype distribution across samples, and considering sample proximity on DNA extraction plates, our work is the first to show how to identify and differentiate, using non-synthetic microbiome data, contamination that derived from external sources and that which originated within the sample set. In a recent review, it was noted that negative controls have been included in ~30% of prior microbiome studies, although only a subset of studies sequenced and analyzed data from the negative controls . Although negative controls can identify foreign DNA that does not belong to the study, they only offer a limited view of contamination and do not constrain the contamination source or the number of samples impacted. In our first case study, analysis of the pattern of shared strains in the context of sample location on DNA extraction plates indicated that the DNA in negative controls was mostly derived from neighboring biological samples. Detection of contamination motivated a more complete analysis of well-to-well contamination in all biological samples in our study. In four cases, genotypes found in many samples from one infant were shared by only one of the samples from another infant. In each such case, the contaminated sample was located adjacent to the putative contamination source on the DNA extraction plate. This conclusion was verified, as the strains were no longer shared when the DNA was re-extracted and resequenced. In our second case study, two externally derived contaminants in the negative controls were identified: a Burkholderia strain and DNA from the Zymo-positive control. We thus investigated strains that were apparently shared by samples from unrelated infants and identified DNA from five additional contaminants in samples from the majority of infants yet absent in the negative controls. It may be possible to find true biological signals if the signal from contamination can be removed. In case study 1, after removing the well-to-well contaminated samples, we identified three Clostridia strains that were shared only among preterm infant samples. Since preterm infants in our study spent their first 2–3 months in the hospital, they have a higher chance to acquire hospital-associated strains than full-term infants. Given that strains of these bacteria have been found previously in preterm infants that were born in the same hospital and in neonatal intensive care unit room microbiomes , we hypothesize that these three strains may be circulating among infants in the hospital. In cases where samples from the same subject or sample set are loaded onto the same DNA extraction plate, the possibility that random and independent well-to-well transfer events could introduce strains that are not recognizable as contaminants should be considered. In our case study 1, although only ~3.3% of samples on P3 and P4 were identified to be contaminated by well-to-well transfer, these two plates had a much higher number of within-plate strain sharing than the other three plates. This raises the possibility of undetected well-to-well transfer events that may have artificially inflated strain sharing among samples. Our study provides a detailed workflow for contamination identification. Based on our observations and previous contamination-related studies, we list several suggestions for minimizing contamination in metagenomics-based studies. First, we encourage others to minimize plate-based extraction if possible. If plate-based extraction is a must, one could consider taking additional precautions such as limiting the number of open wells when extracting, using individual caps for covering wells on the plate, and fully spinning down the samples before removing the caps to prevent well-to-well contamination from occurring. In addition, one should consider including the DNA extraction plate maps in their published work. Second, we urge others to randomize samples when extracting DNA. Specifically, one should avoid loading samples from the same individual or experimental group or biologically related individuals adjacent to one another for extraction. Third, sequencing of sampling negative controls (i.e., empty swab that is used during sample collection) is recommended in addition to extraction negative controls. This should identify contamination introduced during sampling, which is important because such contaminants will likely not display extraction-plate-based sharing patterns. Removal of strains seemingly shared but introduced during sampling will clarify strains truly shared among samples.
Contamination may be unavoidable in high-throughput sequencing, and our results suggest that it can be particularly problematic for low-biomass samples. Genotype-level surveillance has the advantage that it does not require additional expenditures related to library construction and sequencing. Our work emphasizes the importance of routinely assessing contamination prior to data analysis so as to avoid incorrect findings. Harder to detect than external contamination, well-to-well contamination can be a concern of the microbiome field. Random mixtures of samples from the same extraction plate can result in distorted community diversity metrics and inflated strain sharing rate. This is particularly alarming for microbiome-based clinical studies since well-to-well contamination can inflate differences between the treatment and the control groups. As microbiome-based analysis and diagnosis are becoming more popular, we conclude that the use of genotype-specific surveillance methods as well as negative controls are recommended to ensure the integrity and reproducibility of the results.
Sample collection DNA extraction Metagenomic sequencing Metagenomic assembly, de novo binning, and taxonomy assignment Genome dereplication Detection of subspecies and identification of strains using inStrain Statistical analysis Infant fecal samples from case study 1 were collected either at UPMC Magee-Womens Hospital by trained nurses or at home by parents provided with detailed collection instructions. For sample collection and storage details, see Lou et al. . For case study 2, all infant samples (skin, oral, and stool samples) were collected at UPMC Magee-Womens Hospital by trained nurses. Skin and oral samples were obtained by the charge nurse using a BD BBL CultureSwab EZ under supervision of study personnel. Skin and oral samples were collected in duplicate at each timepoint for each preterm infant in order to increase the biomass available for DNA extraction. For details of sample collection and storage, see Olm et al. .
DNA was extracted using either the Qiagen QIAamp PowerFecal Pro DNA Isolation Kit (single-tube extractions; used for 14 of 402 samples in case studies 1 and 7 of 533 samples in case study 2) or Qiagen DNeasy PowerSoil HTP 96 DNA Isolation Kit (96-well plate extractions; used in the majority of samples in both case studies) with modifications to the manufacturer’s protocol. To minimize cross-plate contamination, no plates were extracted at the same time. For each 96-well DNA extraction plate, a reagent-only negative control was included. ZymoBIOMICS Microbial Community Standard (catalog no. D6300) was included as a positive control on one extraction plate from case study 1 and three extraction plates from case study 2. When loading samples into the wells of DNA extraction plates, samples were not randomly distributed among the plates, and often, samples from the same infant were present next to each other sequentially along columns on the same extraction plate. For DNA extracted from stool using the single-tube format, the manufacturer’s protocol was followed except for a heating step at 65 °C for 10 min before the homogenization step. For DNA extracted from stool with the 96-well kit, fecal samples were added to individual wells of the bead plate and stored overnight at −80 °C. The following day, the Bead Solution and Solution C1 were added, and the plates were incubated at 65 °C for 10 min. The plates were shaken on a Retsch Oscillating Mill MM400 with 96-well plate adaptors for 10 min at speed 20. The plates were rotated 180° and shaken again for 10 min at speed 20. All remaining steps followed the manufacturer’s centrifugation protocol. All skin and oral samples from case study 2 were extracted using 96-well plates. Specifically, for skin and oral swab samples, the swab head was cut off directly into the5 min at speed 20. The plates were rotated 180° and shaken again for 5 min at speed 20. The Solution C2 and C3 steps were combined (200 μl of each added) to improve DNA yield. All remaining steps followed the manufacturer’s centrifugation protocol. For six selected skin and oral samples, 75 μl of ZymoBIOMICS Microbial Community Standard (catalog no. D6300) was added to the wells of these six samples prior to the heating step during DNA extraction. Extracted DNA was quantitated using the Quant-iT High-Sensitivity dsDNA Assay Kit (Thermo Fisher Scientific) in 96-well plates, and measurements were made on a SpectraMax M2 microplate reader. DNA yields were used as proxies for sample biomasses. All extracted samples from case study 1 were sequenced, whereas only about half of the extracted samples from case study 2 were sequenced. DNA extractions and quantifications were performed at the University of Pittsburgh School of Medicine, Pittsburgh, PA, USA. Once completed, the extracted DNA was shipped to the QB3 Vincent J. Coates Genomics Sequencing Laboratory at UC Berkeley for library preparation and sequencing. For case study 1, the extracted DNA was sent in the same plates in which the DNA was eluted in the final step of the DNA extraction protocol. For case study 2, the DNA from selected samples was transferred from the extraction plates to three new 96-well plates (one for skin samples, one for oral samples, and one for stool samples) before shipping to Berkeley. For each pair of the duplicated skin and oral samples, their extracted DNA was combined into a single volume on the new 96-well plates.
Samples from case studies 1 and 2 had separate library preparation and sequencing runs. However, the overall sequencing workflow is essentially identical. Metagenomic sequencing of all samples was performed in collaboration with the California Institute for Quantitative Biosciences at UC Berkeley (QB3-Berkeley). Library preparation on all samples from each study was performed as previously described . Final sequence ready libraries were visualized and quantified on the Advanced Analytical Fragment Analyzer. All libraries were then evenly pooled into a single pool and sequenced on individual Illumina NovaSeq 6000 150 paired-end sequencing lanes with 2% PhiX v3 spike-in controls. Post-sequencing bcl files were converted to demultiplexed fastq files per the original sample count with Illumina’s bcl2fastq v2.20 software.
Sequencing reads from case studies 1 and 2 were assembled separately. However, the overall workflow of metagenomics assembly, de novo binning, and taxonomy assignment was essentially identical. See Lou et al. for details on read assembly, de novo binning, and taxonomy assignment on resulting genome bins .
To generate a set of study-specific, high-quality, and nonredundant reference genomes, all de novo assembled genome bins were dereplicated at 98% whole-genome average nucleotide identity (gANI) via dRep (v2.6.2) , using a minimum completeness of 75%, maximum contamination of 10%, the ANImf algorithm, 98% secondary clustering threshold, and 25% minimum coverage overlap. Since biological samples from case study 2 were deliberately spiked with Zymo standard (catalog no. D6300), 10 publicly available Zymo genomes ( https://s3.amazonaws.com/zymo-files/BioPool/ZymoBIOMICS.STD.refseq.v2.zip ) were added to the original genome set of case study 2 before dereplication. Genomes with gANI ≥ 98% were classified as the same “subspecies,” and the genome with the highest score (as determined by dRep) was chosen as the representative genome from each subspecies.
Reads from each individual sample were mapped to study-specific representative subspecies (generated via dRep as described above) concatenated together using Bowtie2 under default settings. inStrain (v1.3.4) profile was run on all resulting mapping files using a minimum mapQ score of 0 and insert size of 160. Genomes with ≥ 0.5 breadth (meaning at least half of the nucleotides of the genome are covered by ≥ 1 read) in samples were considered to be present. inStrain compare was used under default settings to compare read mappings to the same genome in different pairs of samples. In case study 1, samples were considered to share the same strain of the examined genome if the compared region of the genome from samples shared ≥ 99.999% population-level ANI ( popANI ), whereas in case study 2, the popANI threshold was set to be 99.995%. Only genomic areas with at least 5× coverage in samples were compared, and sample pairs with less than 50% of comparable regions of the genome were often excluded (≥ 0.5 percent_genome_compared ). For edge cases, such as when popANI values were within 0.005% of the threshold or when percent_genome_compared values were within 0.2% of the threshold, inStrain compare results were manually assessed to determine whether the sample pairs shared the same or different strains.
Statistical significance for two-group univariate comparisons was calculated using Wilcoxon rank-sum test (as implemented using the SciPy module “scipy.stats.ranksums”) as reported in the main text. For instance, to assess whether strain sharing was correlated with sample pair distance on each DNA extraction plate, we compared within-plate Euclidean distances of sample pairs that did not share strains to those that shared at least one strain using Wilcoxon rank-sum test.
Additional file 1: Figure S1. Details of strain sharing on P3 and P4 from case study I. Rectangular areas represent plates (P3 and P4) and circles show sample placements within each plate. Infant samples are named by infant number and infant day of life (i.e., #63D9 refers to infant #63 and this sample was collected when the infant was 9-day-old). If it is a maternal sample, such a sample is named by the infant number with a letter “M” in the end (i.e., #40M refers to the maternal fecal sample collected from infant #40). A line was drawn between unrelated samples if they shared ≥1 strain. The more strains a sample pair shared, the thicker and brighter the line. If a sample did not share any strains with other unrelated samples, its corresponding circle is colorless. Pink circles represent samples that were likely cross-contaminated. Additional file 2: Figure S2. Detection of one cross-contaminated sample on P4 from case study II. Merged circles represent duplicated samples that were extracted adjacent to each other and were merged before being transferred to the library preparation plates. Infant samples are named by infant number, infant day of life and sample type (“M” refers to mouth samples, “S” refers to skin samples, and “G” refers to gut samples). If a sample pair from unrelated infants shared ≥1 strain, the corresponding samples circles were colored gray and a line was drawn between them. The more strains a sample pair shared, the thicker and brighter the line. Additional file 3: Table S1*: The re-extracted stool sample (#63D6) is not the same as the original one (#63D9) as the original stool sample is unavailable. This alternative stool sample was collected 3 days earlier than the original sample. The original and the re-extracted DNA concentrations of four cross-contaminated samples from case study I.
|
End of preview. Expand
in Data Studio
README.md exists but content is empty.
- Downloads last month
- 51