Reducing Bias in Observational Studies Using Subclassification on the Propensity Score

Propensity score analysis for a semi-continuous exposure variable: a study of gestational alcohol exposure and childhood cognition

Propensity score methodology has become increasingly popular in recent years as a tool for estimating causal effects in observational studies. Much of the related research has been directed at settings with binary or discrete exposure variables with more recent work involving continuous exposure variables. In environmental epidemiology, a substantial proportion of individuals is often completely unexposed while others may experience heavy exposure leading to an exposure distribution with a point mass at zero and a heavy right tail. We suggest a new approach to handle this type of exposure data by constructing a propensity score based on a two-part model and show how this model can be used to more reliably adjust for covariates of a semi-continuous exposure variable. We also consider the case when a misspecified propensity score is used in a regression adjustment and derive an explicit form of the bias. We show that the potential bias gets smaller as the estimated propensity score gets closer to the true expectation of the exposure variable given a set of observed covariates. While this result pertains to a more general setting, we use it to evaluate the potential bias in settings in which the true exposure has a semi-continuous structure. We also evaluate and compare the performance of our proposed method through simulation studies relative to a simpler linear regression-based propensity score for a continuous exposure variable as well as through direct covariate adjustment. Overall, we find that using a propensity score constructed via a two-part model significantly improves the regression estimate when the exposure variable is semi-continuous in nature. Specifically when the proportion of non-exposed subjects is high and the effects of covariates on exposure and outcome are strong, the proposed two-part propensity score method outperforms the more standard competing methods. We illustrate our method using data from the Detroit Longitudinal Cohort Study in which the exposure variable reflects gestational alcohol exposure featuring zero values and a long tail.

Effectiveness of topical vancomycin in the prevention of spinal surgical site infections: a retrospective cohort study

BACKGROUND

The risk of surgical site infections (SSIs), particularly methicillin-resistant Staphylococcus aureus (MRSA) SSIs, after spinal surgeries is one of the most daunting experiences to patients and surgeons. Some authors suggest applying vancomycin powder on the wound before skin closure to minimize the risk of SSIs; however, this practice is not supported by well-established evidence. This study sought to assess the effectiveness of topical (i.e. intra-wound) vancomycin in minimizing the risk of SSIs in patients who underwent spinal surgeries at a Saudi hospital.

METHODS

A retrospective cohort study was conducted using the hospital database. Patients who underwent spinal surgeries from the period of 09/2013 to 09/2019 were included and followed up (observed from the time of the surgery) to 30 days (surgeries without implants) or 90 days (with implants). The odds ratio (OR) of the primary outcome between vancomycin treated versus non-treated patients was estimated using a logistic regression model adjusting for the measured confounders. A sensitivity analysis was conducted using propensity score analysis (inverse probability of treatment weighting [IPTW] with stabilized weights) to control for confounding by indication. All study analyses were completed using RStudio Version 1.2.5033.

RESULTS

We included 81 vancomycin treated vs. 375 untreated patients with 28 infections (8/81 vs. 20/375; respectively). The adjusted OR of SSIs between the two groups was 0.40 (95% confidence interval [CI] 0.11 to 1.34). The result of the propensity score analysis was consistent (OR: 0.97 [95% CI 0.35 to 2.68]).

CONCLUSIONS

We could not find a lower association of SSIs with intra-wound vancomycin in patients who underwent spinal surgeries. Further studies are needed to assess benefits of using topical vancomycin for this indication vs. the risk of antimicrobial resistance.

Data Consult Service: Can we use observational data to address immediate clinical needs?

OBJECTIVE

A number of clinical decision support tools aim to use observational data to address immediate clinical needs, but few of them address challenges and biases inherent in such data. The goal of this article is to describe the experience of running a data consult service that generates clinical evidence in real time and characterize the challenges related to its use of observational data.

MATERIALS AND METHODS

In 2019, we launched the Data Consult Service pilot with clinicians affiliated with Columbia University Irving Medical Center. We created and implemented a pipeline (question gathering, data exploration, iterative patient phenotyping, study execution, and assessing validity of results) for generating new evidence in real time. We collected user feedback and assessed issues related to producing reliable evidence.

RESULTS

We collected 29 questions from 22 clinicians through clinical rounds, emails, and in-person communication. We used validated practices to ensure reliability of evidence and answered 24 of them. Questions differed depending on the collection method, with clinical rounds supporting proactive team involvement and gathering more patient characterization questions and questions related to a current patient. The main challenges we encountered included missing and incomplete data, underreported conditions, and nonspecific coding and accurate identification of drug regimens.

CONCLUSIONS

While the Data Consult Service has the potential to generate evidence and facilitate decision making, only a portion of questions can be answered in real time. Recognizing challenges in patient phenotyping and designing studies along with using validated practices for observational research are mandatory to produce reliable evidence.

Fracture risks in patients with atopic dermatitis: A nationwide matched cohort study

BACKGROUND

The risk of osteoporosis has been explored in atopic dermatitis (AD). The long-term risk of fractures in patients with AD and the effects of various AD treatments on bone health remain to be elucidated.

OBJECTIVE

To evaluate the long-term risk of fractures in patients with AD.

METHODS

This nationwide matched cohort study was conducted using the National Health Insurance Research Database of Taiwan for the period 1997 to 2013. A total of 36,855 patients with AD and 147,420 reference subjects without AD were identified. Demographic characteristics and comorbidities were compared, and cumulative incidence of fractures was evaluated. Adjusted hazard ratios for fracture risks of AD and various AD treatments were calculated using the Cox proportional hazards model.

RESULTS

A total of 1518 patients (4.12%) in the AD cohort and 5579 patients (3.78%) in the reference cohort had fractures (P = .003). The mean ages were 22.6 years in both groups. The 16-year cumulative incidence of fractures in the AD cohort (8.043%) was significantly higher than that in the reference cohort (7.366%) (P = .002). Severe AD (adjusted hazard ratio [aHR], 1.31; 95% confidence interval [CI], 1.08-1.59) was independently associated with fractures. Other independent risk factors included exposure to topical (aHR, 1.21; 95% CI, 1.05-1.39) or systemic (≥10 mg/d; aHR, 1.62; 95% CI, 1.38-1.91) corticosteroids. Use of disease-modifying antirheumatic drugs (aHR, 0.71; 95% CI, 0.53-0.90) and phototherapy (aHR, 0.73; 95% CI, 0.56-0.95) was associated with a lower risk of fractures. The results were consistent across sensitivity analyses.

CONCLUSION

Patients with AD have a higher incidence of fractures. Severe AD is independently associated with fractures.

App Usage Associated With Full Redemption of WIC Food Benefits: A Propensity Score Approach

OBJECTIVE

To examine the relationship between WICShopper application (app) usage and full redemption of Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) food benefits.

DESIGN

A cross-sectional study.

SETTING

West Virginia WIC program.

PARTICIPANTS

A total of 23,050 West Virginia WIC households in 2019.

MAIN OUTCOME MEASURES

Full redemptions were defined as the redemption rate of ≥ 90% for a given food benefit.

ANALYSIS

App/non-app users were defined as households with someone using/not using the WICShopper app. Multivariate logistic regressions were applied. To address the potential self-selection bias in app usage, the propensity score (PS) of app usage was estimated. The regressions were rerun with the balanced sample by matching the PS.

RESULTS

With PS matching, the prevalence of full redemption at the household level was 7.2% for app users vs 4.7% for nonapp users (odds ratio, 1.64; 95% confidence interval, 1.40-1.92; P < 0.001). App users had a higher prevalence of full redemption in most food categories, even with PS matching.

CONCLUSIONS AND IMPLICATIONS

Use of the WICShopper app was associated with a higher prevalence of full redemptions in most food benefits after controlling the self-selection bias.

Effect of a lay counselor delivered integrated maternal mental health and early childhood development group-based intervention in Siaya County, Kenya: A quasi-experimental longitudinal study

BACKGROUND

Maternal mental health is linked to early childhood development; yet there is a gap in evidence-based interventions for low-resource settings. This study estimates the impact of 'Integrated Mothers and Babies Course and Early Childhood Development' (iMBC/ECD), a cognitive-behavioral, group-based intervention, on maternal depression and early childhood social-emotional development in Siaya County, Kenya.

METHODS

This quasi-experimental study enrolled 417 pregnant women and mothers of children under age 2 across two sub-counties in Siaya County. The intervention area had 193 women in 23 groups implementing iMBC/ECD and the control area had 224 women in 30 groups exposed to ECD only content. Mother/index child dyads were followed for two years. To estimate the causal treatment effect from the non-randomized design, we implemented the propensity score weighting method with inverse probability weights.

RESULTS

At baseline, 10.2% of participants endorsed moderate/severe depressive symptoms. At 14-months post-intervention, 7.4% endorsed moderate/severe depression. Overall, iMBC/ECD intervention did not have a significant impact on reducing maternal depression or improving children's social and emotional development. However, sub-group analyses revealed that iMBC/ECD was associated with lowered depressive symptoms among women with no/low education, four or more children and/or no experience of intimate partner violence in the past year. Women with high program attendance (more than half of 14 sessions) also experienced consistently fewer depressive symptoms compared to those with lower attendance.

LIMITATIONS

Non-randomized study, sub-group analyses are exploratory.

CONCLUSIONS

The iMBC/ECD program may have the potential to improve maternal mental health and early child development for more targeted vulnerable populations.

Association between nutritional guidance or ophthalmological examination and discontinuation of physician visits in patients with newly diagnosed diabetes: A retrospective cohort study using a nationwide database

AIMS/INTRODUCTION

Discontinuation of diabetes care has been studied mostly in patients with prevalent diabetes and not in patients with newly diagnosed diabetes, whose dropout risk is highest. Because enrolling patients in a prospective study will influence adherence, we retrospectively examined whether guideline-recommended practices, defined as nutritional guidance or ophthalmological examination, can prevent patient discontinuation of diabetes care after its initiation.

MATERIALS AND METHODS

We retrospectively identified adults with newly screened diabetes during checkups using a large Japanese administrative claims database (JMDC, Tokyo, Japan) that contains laboratory data and lifestyle questionnaires. We defined discontinuation of physician visits as a follow-up interval exceeding 6 months. We divided the patients into those who received guideline-recommended practices (nutritional guidance or ophthalmology consultation) within the same month as the first visit and those who did not. We calculated propensity scores and carried out inverse probability of treatment weighting analyses to compare discontinuation between the two groups.

RESULTS

We identified 6,508 patients with at least one physician consultation for diabetes care within 3 months after their checkup, including 4,574 patients without and 1,934 with guideline-recommended practices. After inverse probability of treatment weighting, patients with guideline-recommended practices had a significantly lower proportion of discontinuation than those without (17.2% vs 21.8%; relative risk 0.79, 95% confidence interval 0.69-0.91).

CONCLUSIONS

This study is the first to show that after adjustment for both patient and healthcare provider factors, guideline-recommended practices within the first month of physician consultation for diabetes care can decrease subsequent discontinuation of physician visits in patients with newly diagnosed diabetes.

The impact of hydroxychloroquine on obstetric outcomes in refractory obstetric antiphospholipid syndrome

BACKGROUND

The use of low-dose aspirin (LDA) and heparin has improved pregnancy outcomes in women with antiphospholipid syndrome (APS). However, 20-30% still have adverse outcomes despite treatment. Recent retrospective studies showed a beneficial effect of hydroxychloroquine (HCQ) in APS due to its anti-inflammatory, immunomodulatory and antithrombotic properties. Data in refractory obstetric APS (OAPS) remain scarce and include heterogeneous populations with various concomitant treatments.

OBJECTIVE

The objective of this study was to assess the impact on the obstetric outcomes of adding HCQ to classical treatments for women with refractory primary obstetric APS.

METHODS

In a retrospective single-centre cohort study, we compared pregnancy outcomes in women with refractory primary OAPS (2004-2019) who received two different treatments in subsequent pregnancies. Group A received 400 mg HCQ + 60 mg enoxaparin + LDA, while Group B received 60 mg enoxaparin + LDA. The main outcome was live birth rates, while pregnancy complications (early and late pregnancy losses and placental-mediated complications) were the secondary outcome.

RESULTS

A total of 101 pregnancies in 87 refractory primary OAPS patients were included. The rate of live-born babies in Group A (HCQ) was 97.1% (67/69) vs. 62.5% (20/32) in Group B (RR: 1.55 [95% CI, 1.19-2.1]; p < 0.001). Pregnancy complications in Group A were 8.7% (6/69) vs. 37.5% (12/32) in Group B (RR 0.22 [95% CI, 0.15-0.30]; p < 0.001).

CONCLUSION

Hydroxychloroquine was associated with a higher rate of live births and a lower prevalence of pregnancy complications in refractory primary obstetric APS. The addition of HCQ to classical treatment may present a promising approach that needs to be confirmed with prospective studies.

Estimating adjusted risk differences by multiply-imputing missing control binary potential outcomes following propensity score-matching

We describe a new method to combine propensity‐score matching with regression adjustment in treatment‐control studies when outcomes are binary by multiply imputing potential outcomes under control for the matched treated subjects. This enables the estimation of clinically meaningful measures of effect such as the risk difference. We used Monte Carlo simulation to explore the effect of the number of imputed potential outcomes under control for the matched treated subjects on inferences about the risk difference. We found that imputing potential outcomes under control (either single imputation or multiple imputation) resulted in a substantial reduction in bias compared with what was achieved using conventional nearest neighbor matching alone. Increasing the number of imputed potential outcomes under control resulted in more efficient estimation, with more efficient estimation of the estimated risk difference when increasing the number of the imputed potential outcomes. The greatest relative increase in efficiency was achieved by imputing five potential outcomes; once 20 outcomes under control were imputed for each matched treated subject, further improvements in efficiency were negligible. We also examined the effect of the number of these imputed potential outcomes on: (i) estimated standard errors; (ii) mean squared error; (iii) coverage of estimated confidence intervals. We illustrate the application of the method by estimating the effect on the risk of death within 1 year of prescribing beta‐blockers to patients discharged from hospital with a diagnosis of heart failure.

Variance estimators for weighted and stratified linear dose-response function estimators using generalized propensity score

Propensity score methods are widely used in observational studies for evaluating marginal treatment effects. The generalized propensity score (GPS) is an extension of the propensity score framework, historically developed in the case of binary exposures, for use with quantitative or continuous exposures. In this paper, we proposed variance estimators for treatment effect estimators on continuous outcomes. Dose-response functions (DRFs) were estimated through weighting on the inverse of the GPS, or using stratification. Variance estimators were evaluated using Monte Carlo simulations. Despite the use of stabilized weights, the variability of the weighted estimator of the DRF was particularly high, and none of the variance estimators (a bootstrap-based estimator, a closed-form estimator especially developed to take into account the estimation step of the GPS, and a sandwich estimator) were able to adequately capture this variability, resulting in coverages below the nominal value, particularly when the proportion of the variation in the quantitative exposure explained by the covariates was large. The stratified estimator was more stable, and variance estimators (a bootstrap-based estimator, a pooled linearized estimator, and a pooled model-based estimator) more efficient at capturing the empirical variability of the parameters of the DRF. The pooled variance estimators tended to overestimate the variance, whereas the bootstrap estimator, which intrinsically takes into account the estimation step of the GPS, resulted in correct variance estimations and coverage rates. These methods were applied to a real data set with the aim of assessing the effect of maternal body mass index on newborn birth weight.

Rituximab versus mitoxantrone: comparing effectiveness and safety in advanced relapsing multiple sclerosis

Background:

Rituximab (RTX), a CD20 depleting agent, is a frequently used off-label treatment for multiple sclerosis (MS), while mitoxantrone (MTX) is approved, albeit rarely used for active relapsing MS (RMS). However, observational data comparing RTX and MTX effectiveness and safety are scarce.

Objective:

We aimed to compare effectiveness and safety of MTX and RTX in patients with active RMS.

Methods:

From combined retrospective clinical data of three MS centers, we selected patients who had received at least one infusion of RTX or MTX and had at least a 6-month clinical follow-up available. Treatment groups were compared by propensity score (PS)-adjusted regression and inverse PS-weighted generalized estimated equation models regarding disability progression, relapse activity, and adverse events (AEs).

Results:

We included 292 RMS patients (mean age 41.8 years, 71.6% female) who received RTX (119 patients, mean age 36.8 years, 74.8% female) or MTX (173 patients mean age 45.3 years, 69.4% female). Using both PS methods, we did not find a significant effect favoring RTX or MTX treatment regarding the probability of disability worsening or relapse occurrence. However, RTX treatment was associated with a significantly lower probability of severe AEs and AEs.

Conclusions:

RTX shows comparable effectiveness but a favorable safety profile compared with MTX in active RMS.

COVID-19 severity and mortality in multiple sclerosis are not associated with immunotherapy: Insights from a nation-wide Austrian registry

BACKGROUND

The COVID-19 pandemic challenges neurologists in counselling patients with multiple sclerosis (pwMS) regarding their risk by SARS-CoV-2 and in guiding disease-modifying treatment (DMT).

OBJECTIVE

To characterize the prevalence and outcome of COVID-19 in pwMS specifically associated with different DMT in a nationwide population-based study.

METHODS

We included patients aged ≥18 years with a confirmed diagnosis of MS and a diagnosis of COVID-19 established between January 1, 2020 and December 31, 2020. We classified COVID-19 course as either mild, severe or fatal. Impact of DMT and specifically immunosuppressants (alemtuzumab, cladribine, fingolimod, ocrelizumab or rituximab) on COVID-19 outcome was determined by multivariable models, adjusted for a-priori-risk.

RESULTS

Of 126 MS patients with COVID-19 (mean age 43.2 years [SD 13.4], 71% female), 86.5% had a mild course, 9.5% a severe course and 3.2% died from COVID-19. A-priori-risk significantly predicted COVID-19 severity (R2 0.814; p<0.001) and mortality (R2 0.664; p<0.001). Adjusting for this a-priori-risk, neither exposure to any DMT nor exposure to specific immunosuppressive DMT were significantly associated with COVID-19 severity (odds ratio [OR] 1.6; p = 0.667 and OR 1.9; p = 0.426) or mortality (OR 0.5; p = 0.711 and 2.1; 0.233) when compared to no DMT.

CONCLUSIONS

In a population-based MS cohort, COVID-19 outcome was not associated with exposure to DMT and immunosuppressive DMT when accounting for other already known risk factors. This provides reassuring evidence that COVID-19 risk can be individually anticipated in MS and-except for a very small proportion of high-risk patients-treatment decisions should be primarily focused on treating MS rather than the pandemic.

A large-scale multi-institutional study evaluating prognostic aspects of positive ascites cytology and effects of therapeutic interventions in epithelial ovarian cancer

Positive ascites cytology is a strong prognostic factor in patients with early-stage ovarian cancer (OvCa). However, limited information is currently available on the impact of positive ascites cytology on patient prognoses under each clinical background. We herein investigated the comprehensive impact of positive ascites cytology on patients with epithelial OvCa and the effectiveness of additional therapeutic interventions, including complete staging surgery and chemotherapy. Among 4730 patients with malignant ovarian neoplasms, retrospectively identified in multiple institutions, 1906 with epithelial OvCa were included. In the investigation of its effects on clinical factors using a multivariate analysis, positive ascites cytology correlated with a poor prognosis. Positive ascites cytology had a significantly worse prognosis than those with negative cytology in all subgroups except for patients with stage IV tumors and a mucinous histology. Chemotherapy may be effective in reducing the negative impact of positive ascites cytology on the prognosis of patients in terms of progression-free and overall survivals, while complete staging surgery did not improve the prognosis of patients with positive ascites cytology. Collectively, our findings suggested that positive ascites cytology had a negative impact on the prognosis of patients with epithelial OvCa, but not those with stage IV tumors or a mucinous histology.

A new three-step method for using inverse propensity weighting with latent class analysis

Bias-adjusted three-step latent class analysis (LCA) is widely popular to relate covariates to class membership. However, if the causal effect of a treatment on class membership is of interest and only observational data is available, causal inference techniques such as inverse propensity weighting (IPW) need to be used. In this article, we extend the bias-adjusted three-step LCA to incorporate IPW. This approach separates the estimation of the measurement model from the estimation of the treatment effect using IPW only for the later step. Compared to previous methods, this solves several conceptual issues and more easily facilitates model selection and the use of multiple imputation. This new approach, implemented in the software Latent GOLD, is evaluated in a simulation study and its use is illustrated using data of prostate cancer patients.

Subclassification estimation of the weighted average treatment effect

Weighting and subclassification are popular approaches using propensity scores (PSs) for estimation of causal effects. Weighting is appealing in that it gives consistent estimators for various causal estimands if appropriate weights are well defined and the PS model is correctly specified. Subclassification is known to be more robust to model misspecification than weighting, but its application to diverse causal estimands is limited. In this article, we propose generalized stratum weights to implement subclassification estimators for various causal estimands. These weights include stratum weights for the average treatment effect (ATE) of the overall population and those for the ATE of the treated as special cases. For this, we incorporate strata into the expression of the weighted average treatment effect (WATE). Particularly, we identify stratum weights for the ATE for the overlap population (ATO), for which the weighting estimator is known to be most efficient among the class of WATE estimators. We show that the identified stratum weights for ATO are equivalent to the optimal stratum weights, which are the inverse variances of the stratum-specific estimators. Simulation studies demonstrate that the proposed subclassification estimator for ATO is more robust to model misspecification than the weighting estimator for ATO. We also propose augmented subclassification estimators, which are shown to be less biased than the subclassification estimators when only the outcome model is correctly specified. The practical utility of the proposed methods is illustrated in a study of right heart catheterization.

A Prospective Cohort Study Comparing Long-Term Outcomes with and without Palifermin in Patients Receiving Hematopoietic Cell Transplantation for Hematologic Malignancies

The incidence of debilitating oral mucositis (OM) can be as high as 99% after myeloablative conditioning regimens preparing patients with hematologic malignancies for hematopoietic cell transplantation (HCT). Palifermin (KGF) is a recombinant human keratinocyte growth factor that reduces the incidence and duration of severe OM. The long-term safety of KGF has not been well established, however. In this long-term prospective matched-cohort study, patients who received KGF (cases) and underwent autologous or allogeneic HCT for hematologic malignancies between 2006 and 2013 were matched 1:1 to patients who did not receive KGF (controls). The primary outcome was overall survival (OS). Other outcomes were disease relapse, new malignancies, pancreatitis, renal failure requiring dialysis, pulmonary complications, cataract surgery, and acute and chronic graft-versus-host disease (GVHD). The analysis population comprised 2191 matched pairs with a wide range of diseases and donor types that received diverse conditioning and GVHD preventive regimens, representing contemporary practice patterns. The median duration of follow-up was 8 years (range, 1 to 12.5 years). In multivariate analyses, the probabilities of OS (relative risk [RR], 1.01; 95% confidence interval [CI], 0.91 to 1.12), relapse (RR, 1.06; 95% CI, 0.94 to 1.18), new malignancies (RR, 0.89; 95% CI, 0.67 to 1.18), and cataract surgery (RR, 1.05; 95% CI, 0.74 to 1.50) were not statistically significantly different between cases and controls. In univariate analyses, no increased risks were observed for renal failure requiring dialysis, pancreatitis, acute GVHD, chronic GVHD, interstitial pneumonitis/acute respiratory distress syndrome/idiopathic pneumonia syndrome, or bronchiolitis obliterans/cryptogenic organizing pneumonia/bronchiolitis obliterans organizing pneumonia among cases compared with controls. This long-term prospective safety cohort study demonstrates that the KGF group had no increased risk of overall mortality, relapse, new malignancies, or any other key outcome. The broad inclusion criteria allow the results to be generalized to contemporary practice for patients with a wide range of diseases and receiving a wide range of HCT conditioning regimens and graft sources from diverse donor types.

Estimating the Impact of In-Hospital Infant Formula Supplementation on Breastfeeding Success

Objective: To assess whether in-hospital infant formula supplementation impacts later successful breastfeeding among healthy mother-infant dyads in the United States who are not intending to exclusively use infant formula. Study Design: Using secondary analysis of a national longitudinal survey (Infant Feeding Practices Study II, n = 2,399), we estimated effects of in-hospital infant formula supplementation on later breastfeeding success by matching mothers whose infants received in-hospital formula supplementation with mothers whose infants did not. Estimates were compared across four matching methods. Outcomes of breastfeeding success included likelihood of following a sustained breastfeeding trajectory for the first year postpartum; feelings of favorability and breastfeeding as long as desired postweaning; and breastfeeding intention, initiation, and duration for subsequent children. Results: In-hospital formula supplementation halved the likelihood of following a breastfeeding trajectory characterized by sustained exclusive breastfeeding. Supplementation decreased feelings of favorability toward breastfeeding postweaning but did not impact the likelihood of feeling that one breastfed as long as desired. Supplementation did not impact intention to breastfeed a future child; it did, however, decrease the likelihood of breastfeeding initiation with a subsequent child by >66% and reduced average duration of breastfeeding any subsequent children by >6 weeks. Conclusion: A lack of experimental methodologies in previous studies makes it difficult to determine a causal link between infant formula in the hospital and less breastfeeding success. Assuming we have accounted for all appropriate confounders, this study provides evidence for such a causal link. Birth hospital policies and practices should speak of this risk of harm.

High-Dimensional Mediation Analysis With Confounders in Survival Models

Mediation analysis is a common statistical method for investigating the mechanism of environmental exposures on health outcomes. Previous studies have extended mediation models with a single mediator to high-dimensional mediators selection. It is often assumed that there are no confounders that influence the relations among the exposure, mediator, and outcome. This is not realistic for the observational studies. To accommodate the potential confounders, we propose a concise and efficient high-dimensional mediation analysis procedure using the propensity score for adjustment. Results from simulation studies demonstrate the proposed procedure has good performance in mediator selection and effect estimation compared with methods that ignore all confounders. Of note, as the sample size increases, the performance of variable selection and mediation effect estimation is as well as the results shown in the method which include all confounders as covariates in the mediation model. By applying this procedure to a TCGA lung cancer data set, we find that lung cancer patients who had serious smoking history have increased the risk of death via the methylation markers cg21926276 and cg20707991 with significant hazard ratios of 1.2093 (95% CI: 1.2019-1.2167) and 1.1388 (95% CI: 1.1339-1.1438), respectively.

Exploring the Effect of Social Support and Empathy on User Engagement in Online Mental Health Communities

It is known that social support and empathy are beneficial for mental health. As a result of the widespread development of social media, online social support and empathy could also influence user behaviors during the development of online communities. However, few studies have examined these effects from the perspective of online mental health communities. These communities appear to be a crucial source for mental health related support, but the spread of online empathy in these communities is not well-understood. This study focused on 22 mental health related subreddits, and matched and compared users (1) who received social support with those who did not receive social support, and users (2) who received more empathic social support with those who received less empathic social support. The results showed that social support and empathy are "contagious". That is, users who received social support at their first post would be more likely to post again and provide support for others; in addition, users who received more empathic support would subsequently express a higher level of empathy to others in the future. Our findings indicate the potential chain reaction of social support and empathy in online mental health communities. Our study also provides insights into how online mental health communities might better assist people to deliver social support that can help others to deal with mental problems.

Long-term effectiveness of catheter ablation in patients with atrial fibrillation and heart failure

AIMS

Randomized trials suggest reductions in all-cause mortality and heart failure (HF) rehospitalizations with catheter ablation (CA) in patients with atrial fibrillation (AF) and HF. Whether these results can be replicated in a real-world population with long-term follow-up or varies over time is unknown. We sought to evaluate the long-term effectiveness of CA in reducing the incidence of all-cause mortality, HF hospitalizations, stroke, and major bleeding in AF-HF patients.

METHODS AND RESULTS

In a cohort of patients newly diagnosed with AF-HF in Quebec, Canada (2000-2017), CA patients were matched 1:2 to controls on time and frequency of hospitalizations. Confounders were controlled for using inverse probability of treatment weighting. Multivariable Cox models adjusted for the presence of cardiac electronic implantable devices and medication use during follow-up, and the effect of time since CA was modelled with B-splines. For non-fatal outcomes, the Lunn-McNeil approach was used to account for the competing risk of death. Among 101 933 AF-HF patients, 451 underwent CA and were matched to 899 controls. Over a median follow-up of 3.8 years, CA was associated with a statistically significant reduction in all-cause mortality [hazard ratio 0.4 (95% confidence interval 0.2-0.7)], but no difference in stroke or major bleeding. The hazard of HF rehospitalization for CA patients, relative to non-CA patients, varied with time since CA (P = 0.01), with a reduction in HF rehospitalizations until approximately 3 years post-CA.

CONCLUSION

Compared with matched non-CA patients, CA was associated with a long-term reduction in all-cause mortality and a reduction in HF rehospitalizations until 3 years post-CA.

Estimation of causal effect in integrating randomized clinical trial and observational data - An example application to cardiovascular outcome trial

Safety evaluation of drug development is a comprehensive process across the product lifecycle. While a randomized clinical trial (RCT) can provide high-quality data to assess the efficacy and safety of a new intervention, the pre-marketing trials are limited in statistical power to detect causal elevation of rare but potentially serious adverse events. On the other hand, real-world data (RWD) sources play a critical role in further understanding the safety profile of the new intervention. Bringing together the breadth and strength of RWD and RCT data, we can maximize the utility of RWD and answer broader questions. In this manuscript, we propose a three-step statistical framework to corroborate findings from both RCT and RWD for evaluating important safety concerns identified in the pre-marketing setting. By the proposed approach, we first match the observational study to RCT, then the causal estimation is validated via the matched observational study with the target RCT by targeted maximum likelihood estimation (TMLE) method, and lastly the evidence from RCT and RWD can be combined in an integrative analysis. A potential application to cardiovascular outcome trials for type 2 diabetes mellitus is illustrated. Finally, simulation results suggest that the heterogeneity of patient population from RCT and RWD can lead to varying degrees of treatment effect estimation and the proposed approach may be able to mitigate such difference in the integrative analysis.

Propensity Score Analysis with Partially Observed Baseline Covariates: A Practical Comparison of Methods for Handling Missing Data

(1) Background: Propensity score methods gained popularity in non-interventional clinical studies. As it may often occur in observational datasets, some values in baseline covariates are missing for some patients. The present study aims to compare the performances of popular statistical methods to deal with missing data in propensity score analysis. (2) Methods: Methods that account for missing data during the estimation process and methods based on the imputation of missing values, such as multiple imputations, were considered. The methods were applied on the dataset of an ongoing prospective registry for the treatment of unprotected left main coronary artery disease. The performances were assessed in terms of the overall balance of baseline covariates. (3) Results: Methods that explicitly deal with missing data were superior to classical complete case analysis. The best balance was observed when propensity scores were estimated with a method that accounts for missing data using a stochastic approximation of the expectation-maximization algorithm. (4) Conclusions: If missing at random mechanism is plausible, methods that use missing data to estimate propensity score or impute them should be preferred. Sensitivity analyses are encouraged to evaluate the implications methods used to handle missing data and estimate propensity score.

Augmenting Both Arms of a Randomized Controlled Trial Using External Data: An Application of the Propensity Score-Integrated Approaches

Leveraging external data is a topic that have recently received much attention. The propensity score-integrated approaches are a methodological innovation for this purpose. In this paper we adapt these approaches, originally introduced to augment single-arm studies with external data, for the augmentation of both arms of a randomized controlled trial (RCT) with external data. After recapitulating the basic ideas, we provide a step-by-step tutorial of how to implement the propensity score-integrated approaches, from study design to outcome analysis, in the RCT setting in such a way that the study integrity and objectively are maintained. Both the Bayesian (power prior) approach and the frequentist (composite likelihood) approach are included. Some extensions and variations of these approaches are also outlined at the end of this paper.

Does coal-to-gas policy reduce air pollution? Evidence from a quasi-natural experiment in China

Whether the use of cleaner energy can reduce air pollution is the focus of debate among scholars, but there is still no unanimous conclusion. This study seeks to explore the net impact of coal-to-gas policy, an energy transition policy in China, on air pollution. Utilizing prefecture-level city data from 2003 to 2016, we apply the PSM-DID method to estimate the policy's net impact. Further, we examine the dynamic effects of coal-to-gas policy and its impact mechanism on air pollution. The results show that 1) The coal-to-gas policy has an average reduction effect of 31.3%, 36%, 0.3%, and 33.1% on industrial sulfur dioxide (SO₂), industrial Smoke (dust), fine particulate matter (PM_2.5), and air quality index (AQI). After eliminating the spreading interference of PM_2.5 in surrounding areas, the effect of this policy on PM_2.5 reduction is 7%; 2) the impact of the coal-to-gas policy is significant in 2012 and 2013, i.e. the second and third years after the implementation of the policy. Then, the reduction effect of the policy on air pollution began to decrease; 3) the coal-to-gas policy has led to the increase in the proportion of the tertiary industry and the decrease in the degree of industrialization. Since the development of the tertiary industry and the reduction of industrialization also led to a reduction in air pollution, the coal-to-gas policy can reduce air pollution through industrial structure upgrading and de-industrialization. The robustness test results support the above conclusions. Practicable policies to reduce air pollution in China are suggested and applicable to other developing countries with resource-scarce and serve air pollution.

Propensity-score-based meta-analytic predictive prior for incorporating real-world and historical data

As the availability of real-world data sources (eg, EHRs, claims data, registries) and historical data has rapidly surged in recent years, there is an increasing interest and need from investigators and health authorities to leverage all available information to reduce patient burden and accelerate both drug development and regulatory decision making. Bayesian meta-analytic approaches are a popular historical borrowing method that has been developed to leverage such data using robust hierarchical models. The model structure accounts for various degrees of between-trial heterogeneity, resulting in adaptively discounting the external information in the case of data conflict. In this article, we propose to integrate the propensity score method and Bayesian meta-analytic-predictive (MAP) prior to leverage external real-world and historical data. The propensity score methodology is applied to select a subset of patients from external data that are similar to those in the current study with regards to key baseline covariates and to stratify the selected patients together with those in the current study into more homogeneous strata. The MAP prior approach is used to obtain stratum-specific MAP prior and derive the overall propensity score integrated meta-analytic predictive (PS-MAP) prior. Additionally, we allow for tuning the prior effective sample size for the proposed PS-MAP prior, which quantifies the amount of information borrowed from external data. We evaluate the performance of the proposed PS-MAP prior by comparing it to the existing propensity score-integrated power prior approach in a simulation study and illustrate its implementation with an example of a single-arm phase II trial.

Functional connectivity between the caudate and medial prefrontal cortex reflects individual honesty variations in adults and children

Deception emerges in early childhood and prevails in adults. Activation patterns in previous adults' task-state functional magnetic resonance imaging (fMRI), though sensitive to state honesty on a specific decision, are less reliable reflecting trait honesty. Besides of state honesty, most previous neuroimaging studies about dishonesty suffer the generalization problem due to the major focus on adults with children unexplored. To investigate honesty associated functional brain networks variations, 98 healthy adults (Age: 18-28 y.o.; 49 males and 49 females) were invited to participate in a resting-state functional magnetic resonance imaging (rfMRI) study (Study 1). We investigated how functional connections between the caudate and the medial prefrontal cortex (mPFC) change among adults who differ in self-reported trait honesty. Results showed that adults with higher trait honesty have increased functional connectivity from the caudate to the mPFC, which is identified as an honesty-related hub region in global brain connectivity analysis and connects more tightly to a wide range of brain regions including the amygdala. Study 2 compared functional connectivity between children with high vs. low lying frequencies (Age: 6-16 y.o.; 61 males and 39 females) based on a publicly accessible database of rfMRI. Consistent with findings in adults, increased functional connectivity from the caudate to the mPFC was found in less frequently lying children. Despite different honesty indicators of self-reported honesty trait in adults and parent-reported lying patterns in children, consistent findings have been noted in the two samples with regards to functional connectivity variations between reward-related and self-related brain regions. These findings suggest functional connectivity alterations between the caudate and the mPFC contribute to honesty variations in both adults and children.

Matching Methods for Confounder Adjustment: An Addition to the Epidemiologist's Toolbox

Propensity score weighting and outcome regression are popular ways to adjust for observed confounders in epidemiological research. Here, we provide an introduction to matching methods, which serve the same purpose but can offer advantages in robustness and performance. A key difference between matching and weighting methods is that matching methods do not directly rely on the propensity score and so are less sensitive to its misspecification or to the presence of extreme values. Matching methods offer many options for customization, which allow a researcher to incorporate substantive knowledge and carefully manage bias/variance trade-offs in estimating the effects of nonrandomized exposures. We review these options and their implications, providing guidance for their use, and comparison with weighting methods. Because of their potential advantages over other methods, matching methods should have their place in an epidemiologist's methodological toolbox.

How to Apply Multiple Imputation in Propensity Score Matching with Partially Observed Confounders: A Simulation Study and Practical Recommendations

Propensity score matching (PSM) has been widely used to mitigate confounding in observational studies, although complications arise when the covariates used to estimate the PS are only partially observed. Multiple imputation (MI) is a potential solution for handling missing covariates in the estimation of the PS. However, it is not clear how to best apply MI strategies in the context of PSM. We conducted a simulation study to compare the performances of popular non-MI missing data methods and various MI-based strategies under different missing data mechanisms. We found that commonly applied missing data methods resulted in biased and inefficient estimates, and we observed large variation in performance across MI-based strategies. Based on our findings, we recommend 1) estimating the PS after applying MI to impute missing confounders; 2) conducting PSM within each imputed dataset followed by averaging the treatment effects to arrive at one summarized finding; 3) a bootstrapped-based variance to account for uncertainty of PS estimation, matching, and imputation; and 4) inclusion of key auxiliary variables in the imputation model.

Adherence to National Consensus Guidelines and Association with Clinical Outcomes in Patients with Candidemia

Background:

National consensus guidelines outline recommendations for best practices in treating patients with candidemia. This study evaluated the impact of receiving care adherent to the best practice recommendations on clinical outcomes in patients with candidemia.

Methods:

This retrospective, multicenter study included patients with candidemia from 2010 to 2015 at 9 hospitals. The primary outcome was the composite of 30-day in-hospital mortality and 90-day candidemia recurrence. Outcomes were compared between those receiving and not receiving care adherent to the guideline recommendations. Inverse probability weights with regression adjustment were utilized to determine the average treatment effect of adherent care on the composite outcome

Results:

295 patients were included with 14.2% meeting criteria for the composite outcome (11.9% mortality and 2.4% recurrence). The average treatment effect of adherent care was not significant (P = .75). However, receiving appropriate initial antifungal treatment and central venous catheter removal were both associated with the composite (average treatment effect of −17.5%, P = .011 and −8.8%, P = .013, respectively). In patients with a source of infection other than the central line, central venous catheter removal was not associated with the composite (P = .95). The most common reason for failure to receive appropriate initial antifungal treatment was omission of the loading dose.

Conclusions:

Central venous catheter removal and appropriate initial antifungal treatment were associated with a lower incidence of the composite of mortality and recurrence. Additional studies are needed to determine the optimal duration of therapy following candidemia clearance.

The effect of time to irrigation and debridement on the rate of reoperation in open fractures : a propensity score-based analysis of the Fluid Lavage of Open Wounds (FLOW) study

AIMS

Despite long-standing dogma, a clear relationship between the timing of surgical irrigation and debridement (I&D) and the development of subsequent deep infection has not been established in the literature. Traditionally, I&D of an open fracture has been recommended within six hours of injury based on animal studies from the 1970s, however the clinical basis for this remains unclear. Using data from a multicentre randomized controlled trial of 2,447 open fracture patients, the primary objective of this secondary analysis is to determine if a relationship exists between timing of wound I&D (within six hours of injury vs beyond six hours) and subsequent reoperation rate for infection or healing complications within one year for patients with open limb fractures requiring surgical treatment.

METHODS

To adjust for the influence of patient and injury characteristics on the timing of I&D, a propensity score was developed from the dataset. Propensity-adjusted regression allowed for a matched cohort analysis within the study population to determine if early irrigation put patients independently at risk for reoperation, while controlling for confounding factors. Results were reported as odds ratios (ORs), 95% confidence intervals (CIs), and p-values. All analyses were conducted using STATA 14.

RESULTS

In total, 2,286 of 2,447 patients randomized to the trial from 41 orthopaedic trauma centres across five countries had complete data regarding time to I&D. Prior to matching, the patients managed with early I&D had a higher proportion requiring reoperation for infection or healing complications (17% vs 13%; p = 0.019), however this does not account for selection bias of more severe injuries preferentially being treated earlier. When accounting for propensity matching, early irrigation was not associated with reoperation (OR 0.71 (95% CI 0.47 to 1.07); p = 0.73).

CONCLUSION

When accounting for other variables, late irrigation does not independently increase risk of reoperation. Cite this article: Bone Joint J 2021;103-B(6):1055-1062.

Thyroidectomy for Graves' Disease Predicts Postoperative Neck Hematoma and Hypocalcemia: A North American cohort study

Objective:

Examine the association of Graves’ disease with the development of postoperative neck hematoma.

Design:

A cohort of patients participating in the Thyroid Procedure-Targeted Database of the National Surgical Quality Improvement Program from January 1, 2016 to December 31, 2018.

Setting:

A North American surgical cohort study.

Methods:

17 906 patients who underwent thyroidectomy were included. Propensity score matching was performed to adjust for differences in baseline covariates. Multivariate logistic regression was used to ascertain the association between thyroidectomy for Graves’ disease and risk of postoperative adverse events within 30 days of surgery. The primary outcome was postoperative hematoma. Secondary outcomes were postoperative hypocalcemia and recurrent laryngeal nerve injury.

Results:

One-to-three propensity score matching yielded 1207 patients with mean age (SD) of 42.6 (14.9) years and 1017 (84.3%) female in the group with Graves’ disease and 3621 patients with mean age (SD) of 46.7 (15.0%) years and 2998 (82.8%) female in the group with indications other than Graves’ disease for thyroidectomy. The cumulative 30-day incidence of postoperative hematoma was 3.1% (38/1207) in the Graves’ disease group and 1.9% (70/3621) in other patients. The matched cohort showed that Graves’ disease was associated with higher odds of postoperative hematoma (OR 1.65, 95% CI 1.10-2.46) and hypocalcemia (OR 2.04, 95% CI 1.66-2.50) compared with other indications for thyroid surgery. There was no difference in recurrent laryngeal nerve injury among the 2 groups.

Conclusions:

Patients with Graves’ disease undergoing thyroidectomy are more likely to suffer from postoperative hematoma and hypocalcemia compared to patients undergoing surgery for other indications.

Design of the Association of Uterine Perforation and Expulsion of Intrauterine Device study: a multisite retrospective cohort study

BACKGROUND

Intrauterine devices are effective and safe, long-acting reversible contraceptives, but the risk of uterine perforation occurs with an estimated incidence of 1 to 2 per 1000 insertions. The European Active Surveillance Study for Intrauterine Devices, a European prospective observational study that enrolled 61,448 participants (2006-2012), found that women breastfeeding at the time of device insertion or with the device inserted at ≤36 weeks after delivery had a higher risk of uterine perforation. The Association of Uterine Perforation and Expulsion of Intrauterine Device (APEX-IUD) study was a Food and Drug Administration-mandated study designed to reflect current United States clinical practice. The aims of the APEX-IUD study were to evaluate the risk of intrauterine device-related uterine perforation and device expulsion among women who were breastfeeding or within 12 months after delivery at insertion.

OBJECTIVE

We aimed to describe the APEX-IUD study design, methodology, and analytical plan and present population characteristics, size of risk factor groups, and duration of follow-up.

STUDY DESIGN

APEX-IUD study was a retrospective cohort study conducted in 4 organizations with access to electronic health records: Kaiser Permanente Northern California, Kaiser Permanente Southern California, Kaiser Permanente Washington, and Regenstrief Institute in Indiana. Variables were identified through structured data (eg, diagnostic, procedural, medication codes) and unstructured data (eg, clinical notes) via natural language processing. Outcomes include uterine perforation and device expulsion; potential risk factors were breastfeeding at insertion, postpartum timing of insertion, device type, and menorrhagia diagnosis in the year before insertion. Covariates include demographic characteristics, clinical characteristics, and procedure-related variables, such as difficult insertion. The first potential date of inclusion for eligible women varies by research site (from January 1, 2001 to January 1, 2010). Follow-up begins at insertion and ends at first occurrence of an outcome of interest, a censoring event (device removal or reinsertion, pregnancy, hysterectomy, sterilization, device expiration, death, disenrollment, last clinical encounter), or end of the study period (June 30, 2018). Comparisons of levels of exposure variables were made using Cox regression models with confounding adjusted by propensity score weighting using overlap weights.

RESULTS

The study population includes 326,658 women with at least 1 device insertion during the study period (Kaiser Permanente Northern California, 161,442; Kaiser Permanente Southern California, 123,214; Kaiser Permanente Washington, 20,526; Regenstrief Institute, 21,476). The median duration of continuous enrollment was 90 (site medians 74-177) months. The mean age was 32 years, and the population was racially and ethnically diverse across the 4 sites. The mean body mass index was 28.5 kg/m², and of the women included in the study, 10.0% had menorrhagia ≤12 months before insertion, 5.3% had uterine fibroids, and 10% were recent smokers; furthermore, among these women, 79.4% had levonorgestrel-releasing devices, and 19.5% had copper devices. Across sites, 97,824 women had an intrauterine device insertion at ≤52 weeks after delivery, of which 94,817 women (97%) had breastfeeding status at insertion determined; in addition, 228,834 women had intrauterine device insertion at >52 weeks after delivery or no evidence of a delivery in their health record.

CONCLUSION

Combining retrospective data from multiple sites allowed for a large and diverse study population. Collaboration with clinicians in the study design and validation of outcomes ensured that the APEX-IUD study results reflect current United States clinical practice. Results from this study will provide valuable information based on real-world evidence about risk factors for intrauterine devices perforation and expulsion for clinicians.

Metformin use and risk of COVID-19 among patients with type II diabetes mellitus: an NHIS-COVID-19 database cohort study

AIMS

The relationship between metformin therapy and the risk of coronavirus disease (COVID-19) has not been reported among patients with type 2 diabetes mellitus (DM). We aimed to investigate whether metformin therapy was associated with the incidence of COVID-19 among type 2 DM patients in South Korea.

METHODS

The National Health Insurance Service-COVID-19 cohort database, comprising COVID-19 patients from 1 January 2020 to 4 June 2020, was used for this study. Among them, adult patients with type 2 DM were included in this study. Metformin users were defined as those who had been prescribed continuous oral metformin for over a period of ≥ 90 days, and the control group was defined as all other patients.

RESULTS

Overall, 27,493 patients with type 2 DM (7204, metformin user group; 20,289, control group) were included. After propensity score matching, 11,892 patients (5946 patients in each group) were included in the final analysis. In the logistic regression analysis, the odds of metformin users developing COVID-19 was 30% lower than that of the control group [odds ratio (OR): 0.70, 95% confidence interval (CI): 0.61-0.80; P < 0.001]. However, in the multivariate model, metformin use was not associated with hospital mortality when compared with that of the control group (OR: 1.26, 95% CI: 0.81-1.95; P = 0.301).

CONCLUSIONS

Metformin therapy might have potential benefits for the prevention of COVID-19 among patients with type 2 DM in South Korea. However, it did not affect the hospital mortality of type 2 DM patients diagnosed with COVID-19.

Propensity score matching versus coarsened exact matching in observational comparative effectiveness research

Aim & methods: We compared propensity score matching (PSM) and coarsened exact matching (CEM) in balancing baseline characteristics between treatment groups using observational data obtained from a pan-Canadian prostate cancer radiotherapy database. Changes in effect estimates were evaluated as a function of improvements in balance, using results from randomized clinical trials to guide interpretation. Results: CEM and PSM improved balance between groups in both comparisons, while retaining the majority of original data. Improvements in balance were associated with effect estimates closer to those obtained in randomized clinical trials. Conclusion: CEM and PSM led to substantial improvements in balance between comparison groups, while retaining a considerable proportion of original data. This could lead to improved accuracy in effect estimates obtained using observational data in a variety of clinical situations.

Value of comparative studies of "real clinical practice" in modern cardiology. Position paper based on the expert council discussion dated 12/18/2020

On December 18, 2020, an expert council was held with the participation of members of the Russian Society of Cardiology, the Eurasian Association of Ther-apists, the National Society for Atherothrombosis, the National Society for Evi-dence-Based Pharmacotherapy, and the Russian Heart Failure Society. The event was devoted to the discussion of the correct use of research data of "real clinical practice" in decision making.

Sensitivity analyses of unmeasured and partially-measured confounders using multiple imputation in a vaccine safety study

PURPOSE

Sensitivity analyses have played an important role in pharmacoepidemiology studies using electronic health records data. Despite the existence of quantitative bias analysis in pharmacoepidemiologic studies, simultaneously adjusting for unmeasured and partially measured confounders is challenging in vaccine safety studies. Our objective was to develop a flexible approach for conducting sensitivity analyses of unmeasured and partially-measured confounders concurrently for a vaccine safety study.

METHODS

We derived conditional probabilities for an unmeasured confounder based on bias parameters, used these conditional probabilities and Monte Carlo simulations to impute the unmeasured confounder, and re-constructed the analytic datasets as if the unmeasured confounder had been observed. We simultaneously imputed a partially measured confounder using a prediction model. We considered unmeasured breastfeeding and partially measured family history of Type 1 diabetes (T1DM) in a study examining the association between exposure to rotavirus vaccination and T1DM.

RESULTS

Before sensitivity analyses, the hazard ratios (HR) were 1.50 (95% CI, 0.81-2.77) for those partially exposed and 1.03 (95% CI, 0.62-1.72) for those fully exposed with unexposed children as the referent group. When breastfeeding and family history of T1DM were adjusted, the HR was 1.55 (95% CI, 0.84-2.87) for the partially exposed group; the HR was 0.98 (95% CI, 0.58-1.63) for the fully exposed group.

CONCLUSIONS

We conclude that adjusting for unmeasured breastfeeding and partially measured family history of T1DM did not alter the conclusion that there was no evidence of association between rotavirus vaccination and developing T1DM. This novel approach allows for simultaneous adjustment for multiple unmeasured and partially-measured confounders.

Pathway-Driven Coordinated Telehealth System for Management of Patients With Single or Multiple Chronic Diseases in China: System Development and Retrospective Study

Background

Integrated care enhanced with information technology has emerged as a means to transform health services to meet the long-term care needs of patients with chronic diseases. However, the feasibility of applying integrated care to the emerging “three-manager” mode in China remains to be explored. Moreover, few studies have attempted to integrate multiple types of chronic diseases into a single system.

Objective

The aim of this study was to develop a coordinated telehealth system that addresses the existing challenges of the “three-manager” mode in China while supporting the management of single or multiple chronic diseases.

Methods

The system was designed based on a tailored integrated care model. The model was constructed at the individual scale, mainly focusing on specifying the involved roles and responsibilities through a universal care pathway. A custom ontology was developed to represent the knowledge contained in the model. The system consists of a service engine for data storage and decision support, as well as different forms of clients for care providers and patients. Currently, the system supports management of three single chronic diseases (hypertension, type 2 diabetes mellitus, and chronic obstructive pulmonary disease) and one type of multiple chronic conditions (hypertension with type 2 diabetes mellitus). A retrospective study was performed based on the long-term observational data extracted from the database to evaluate system usability, treatment effect, and quality of care.

Results

The retrospective analysis involved 6964 patients with chronic diseases and 249 care providers who have registered in our system since its deployment in 2015. A total of 519,598 self-monitoring records have been submitted by the patients. The engine could generate different types of records regularly based on the specific care pathway. Results of the comparison tests and causal inference showed that a part of patient outcomes improved after receiving management through the system, especially the systolic blood pressure of patients with hypertension (P<.001 in all comparison tests and an approximately 5 mmHg decrease after intervention via causal inference). A regional case study showed that the work efficiency of care providers differed among individuals.

Conclusions

Our system has potential to provide effective management support for single or multiple chronic conditions simultaneously. The tailored closed-loop care pathway was feasible and effective under the “three-manager” mode in China. One direction for future work is to introduce advanced artificial intelligence techniques to construct a more personalized care pathway.

Assessment of heterogeneous treatment effect estimation accuracy via matching

We study the assessment of the accuracy of heterogeneous treatment effect (HTE) estimation, where the HTE is not directly observable so standard computation of prediction errors is not applicable. To tackle the difficulty, we propose an assessment approach by constructing pseudo-observations of the HTE based on matching. Our contributions are three-fold: first, we introduce a novel matching distance derived from proximity scores in random forests; second, we formulate the matching problem as an average minimum-cost flow problem and provide an efficient algorithm; third, we propose a match-then-split principle for the assessment with cross-validation. We demonstrate the efficacy of the assessment approach using simulations and a real dataset.

Angiotensin II receptor blocker or angiotensin-converting enzyme inhibitor use and COVID-19-related outcomes among US Veterans

BACKGROUND

Angiotensin II receptor blockers (ARBs) and angiotensin-converting enzyme inhibitors (ACEIs) may positively or negatively impact outcomes in severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection. We investigated the association of ARB or ACEI use with coronavirus disease 2019 (COVID-19)-related outcomes in US Veterans with treated hypertension using an active comparator design, appropriate covariate adjustment, and negative control analyses.

METHODS AND FINDINGS

In this retrospective cohort study of Veterans with treated hypertension in the Veterans Health Administration (01/19/2020-08/28/2020), we compared users of (A) ARB/ACEI vs. non-ARB/ACEI (excluding Veterans with compelling indications to reduce confounding by indication) and (B) ARB vs. ACEI among (1) SARS-CoV-2+ outpatients and (2) COVID-19 hospitalized inpatients. The primary outcome was all-cause hospitalization or mortality (outpatients) and all-cause mortality (inpatients). We estimated hazard ratios (HR) using propensity score-weighted Cox regression. Baseline characteristics were well-balanced between exposure groups after weighting. Among outpatients, there were 5.0 and 6.0 primary outcomes per 100 person-months for ARB/ACEI (n = 2,482) vs. non-ARB/ACEI (n = 2,487) users (HR 0.85, 95% confidence interval [CI] 0.73-0.99, median follow-up 87 days). Among outpatients who were ARB (n = 4,877) vs. ACEI (n = 8,704) users, there were 13.2 and 14.8 primary outcomes per 100 person-months (HR 0.91, 95%CI 0.86-0.97, median follow-up 85 days). Among inpatients who were ARB/ACEI (n = 210) vs. non-ARB/ACEI (n = 275) users, there were 3.4 and 2.0 all-cause deaths per 100 person months (HR 1.25, 95%CI 0.30-5.13, median follow-up 30 days). Among inpatients, ARB (n = 1,164) and ACEI (n = 2,014) users had 21.0 vs. 17.7 all-cause deaths, per 100 person-months (HR 1.13, 95%CI 0.93-1.38, median follow-up 30 days).

CONCLUSIONS

This observational analysis supports continued ARB or ACEI use for patients already using these medications before SARS-CoV-2 infection. The novel beneficial association observed among outpatients between users of ARBs vs. ACEIs on hospitalization or mortality should be confirmed with randomized trials.

Ischemic/bleeding event after short dual-antiplatelet therapy in patients with high bleeding risk: Sub-analysis of the MODEL U-SES study

BACKGROUND

This analysis aimed to evaluate the clinical impact of high bleeding risk (HBR) on adverse events after percutaneous coronary intervention (PCI).

METHODS

We retrospectively analyzed 1695 patients in the MODEL U-SES study, which was a multicenter, open-label, prospective observational study evaluating safety of 3-month dual antiplatelet therapy (DAPT) after Ultimaster stent (Terumo Corporation, Tokyo, Japan) implantation at 65 sites in Japan. Patients were divided into 2 groups (HBR/Non-HBR) according to modified Academic Research Consortium-HBR criteria. Ischemic/thrombotic event (cardiovascular death, myocardial infarction, ischemic stroke, and stent thrombosis) and bleeding event (Bleeding Academic Research Consortium 3 or 5) at 1 year were evaluated.

RESULTS

Of 1695 patients, 840 patients were categorized as HBR and 855 patients were Non-HBR. One-year follow-up was completed in 95.3%. During 1-year follow-up, ischemic/thrombotic events were observed in 31 cases (1.8%) and bleeding events occurred in 21 cases (1.2%). Presence of HBR was significantly associated with higher incidence of ischemic/thrombotic events as compared to Non-HBR (adjusted hazard ratio, 0.16; 95% confidence interval, 0.05 to 0.50), whereas the incidence of bleeding events did not reach statistical significance between HBR and Non-HBR. In comparison of monotherapy after DAPT, P2Y₁₂ inhibitor monotherapy after DAPT had comparable ischemic/thrombotic and bleeding events with aspirin monotherapy after DAPT in both HBR and Non-HBR.

CONCLUSION

In contemporary PCI practice, nearly half of patients had HBR and presence of HBR significantly increased risk of ischemic/thrombotic events. Both aspirin and P2Y₁₂ inhibitor monotherapy following short DAPT had low and comparable ischemic/bleeding events.

Leverage multiple real-world data sources in single-arm medical device clinical studies

The interest in utilizing real-world data (RWD) has been considerably increasing in medical product development and evaluation. With proper usage and analysis of high-quality real-world data, real-world evidence (RWE) can be generated to inform regulatory and healthcare decision-making. This paper proposes a study design and data analysis approach for a prospective, single-arm clinical study that is supplemented with patients from multiple real-world data sources containing patient-level covariate and outcome data. After the amount of information to be borrowed from each real-world data source is determined, the propensity score-integrated composite likelihood method is applied to obtain an estimate of the parameter of interest based on data from the prospective clinical study and this real-world data source. This method is applied to each real-world data source. The final estimate of the parameter of interest is then obtained by taking a weighted average of all these estimates. The performance of the proposed approach is evaluated via a simulation study. A hypothetical example is presented to illustrate how to implement the proposed approach.

When Can Nonrandomized Studies Support Valid Inference Regarding Effectiveness or Safety of New Medical Treatments?

The randomized controlled trial (RCT) is the gold standard for evaluating the causal effects of medications. Limitations of RCTs have led to increasing interest in using real-world evidence (RWE) to augment RCT evidence and inform decision making on medications. Although RWE can be either randomized or nonrandomized, nonrandomized RWE can capitalize on the recent proliferation of large healthcare databases and can often answer questions that cannot be answered in randomized studies due to resource constraints. However, the results of nonrandomized studies are much more likely to be impacted by confounding bias, and the existence of unmeasured confounders can never be completely ruled out. Furthermore, nonrandomized studies require more complex design considerations which can sometimes result in design-related biases. We discuss questions that can help investigators or evidence consumers evaluate the potential impact of confounding or other biases on their findings: Does the design emulate a hypothetical randomized trial design? Is the comparator or control condition appropriate? Does the primary analysis adjust for measured confounders? Do sensitivity analyses quantify the potential impact of residual confounding? Are methods open to inspection and (if possible) replication? Designing a high-quality nonrandomized study of medications remains challenging and requires broad expertise across a range of disciplines, including relevant clinical areas, epidemiology, and biostatistics. The questions posed in this paper provide a guiding framework for assessing the credibility of nonrandomized RWE and could be applied across many clinical questions.

Risk of Nonmelanoma Skin Cancer in Association With Use of Hydrochlorothiazide-Containing Products in the United States

Background

European studies reported an increased risk of nonmelanoma skin cancer associated with hydrochlorothiazide (HCTZ)-containing products. We examined the risks of basal cell carcinoma (BCC) and squamous cell carcinoma (SCC) associated with HCTZ compared with angiotensin-converting enzyme inhibitors (ACEIs) in a US population.

Methods

We conducted a retrospective cohort study in the US Food and Drug Administration's Sentinel System. From the date of HCTZ or ACEI dispensing, patients were followed until a SCC or BCC diagnosis requiring excision or topical chemotherapy treatment on or within 30 days after the diagnosis date or a censoring event. Using Cox proportional hazards regression models, we estimated the hazard ratios (HRs), overall and separately by age, sex, and race. We also examined site- and age-adjusted incidence rate ratios (IRRs) by cumulative HCTZ dose within the matched cohort.

Results

Among 5.2 million propensity-score matched HCTZ and ACEI users, the incidence rate (per 1000 person-years) of BCC was 2.78 and 2.82, respectively, and 1.66 and 1.60 for SCC. Overall, there was no difference in risk between HCTZ and ACEIs for BCC (HR = 0.99, 95% confidence interval [CI] = 0.97 to 1.00), but there was an increased risk for SCC (HR = 1.04, 95% CI = 1.02 to 1.06). HCTZ use was associated with higher risks of BCC (HR = 1.09, 95% CI = 1.07 to 1.11) and SCC (HR = 1.15, 95% CI = 1.12 to 1.17) among Caucasians. Cumulative HCTZ dose of 50 000 mg or more was associated with an increased risk of SCC in the overall population (IRR = 1.19, 95% CI = 1.05 to 1.35) and among Caucasians (IRR = 1.27, 95% CI = 1.10 to 1.47).

Conclusions

Among Caucasians, we identified small increased risks of BCC and SCC with HCTZ compared with ACEI. Appropriate risk mitigation strategies should be taken while using HCTZ.

Propensity score analysis methods with balancing constraints: A Monte Carlo study

The inverse probability weighting is an important propensity score weighting method to estimate the average treatment effect. Recent literature shows that it can be easily combined with covariate balancing constraints to reduce the detrimental effects of excessively large weights and improve balance. Other methods are available to derive weights that balance covariate distributions between the treatment groups without the involvement of propensity scores. We conducted comprehensive Monte Carlo experiments to study whether the use of covariate balancing constraints circumvent the need for correct propensity score model specification, and whether the use of a propensity score model further improves the estimation performance among methods that use similar covariate balancing constraints. We compared simple inverse probability weighting, two propensity score weighting methods with balancing constraints (covariate balancing propensity score, covariate balancing scoring rule), and two weighting methods with balancing constraints but without using the propensity scores (entropy balancing and kernel balancing). We observed that correct specification of the propensity score model remains important even when the constraints effectively balance the covariates. We also observed evidence suggesting that, with similar covariate balance constraints, the use of a propensity score model improves the estimation performance when the dimension of covariates is large. These findings suggest that it is important to develop flexible data-driven propensity score models that satisfy covariate balancing conditions.