Propensity score analysis with partially observed covariates: How should multiple imputation be used?

Handling missing data and measurement error for early-onset myopia risk prediction models

BACKGROUND

Early identification of children at high risk of developing myopia is essential to prevent myopia progression by introducing timely interventions. However, missing data and measurement error (ME) are common challenges in risk prediction modelling that can introduce bias in myopia prediction.

METHODS

We explore four imputation methods to address missing data and ME: single imputation (SI), multiple imputation under missing at random (MI-MAR), multiple imputation with calibration procedure (MI-ME), and multiple imputation under missing not at random (MI-MNAR). We compare four machine-learning models (Decision Tree, Naive Bayes, Random Forest, and Xgboost) and three statistical models (logistic regression, stepwise logistic regression, and least absolute shrinkage and selection operator logistic regression) in myopia risk prediction. We apply these models to the Shanghai Jinshan Myopia Cohort Study and also conduct a simulation study to investigate the impact of missing mechanisms, the degree of ME, and the importance of predictors on model performance. Model performance is evaluated using the receiver operating characteristic curve (AUROC) and the area under the precision-recall curve (AUPRC).

RESULTS

Our findings indicate that in scenarios with missing data and ME, using MI-ME in combination with logistic regression yields the best prediction results. In scenarios without ME, employing MI-MAR to handle missing data outperforms SI regardless of the missing mechanisms. When ME has a greater impact on prediction than missing data, the relative advantage of MI-MAR diminishes, and MI-ME becomes more superior. Furthermore, our results demonstrate that statistical models exhibit better prediction performance than machine-learning models.

CONCLUSION

MI-ME emerges as a reliable method for handling missing data and ME in important predictors for early-onset myopia risk prediction.

Addressing immortal time bias in precision medicine: Practical guidance and methods development

OBJECTIVE

To compare theoretical strengths and limitations of common immortal time adjustment methods, propose a new approach using multiple imputation (MI), and provide practical guidance for using MI in precision medicine evaluations centered on a real-world case study.

STUDY SETTING AND DESIGN

Methods comparison, guidance, and real-world case study based on previous literature. We compared landmark analysis, time-distribution matching, time-dependent analysis, and our proposed MI application. Guidance for MI spanned (1) selecting the imputation method; (2) specifying and applying the imputation model; and (3) conducting comparative analysis and pooling estimates. Our case study used a matched cohort design to evaluate overall survival benefits of whole-genome and transcriptome analysis, a precision medicine technology, compared to usual care for advanced cancers, and applied both time-distribution matching and MI. Bootstrap simulation characterized imputation sensitivity to varying data missingness and sample sizes.

DATA SOURCES AND ANALYTIC SAMPLE

Case study used population-based administrative data and single-arm precision medicine program data from British Columbia, Canada for the study period 2012 to 2015.

PRINCIPAL FINDINGS

While each method described can reduce immortal time bias, MI offers theoretical advantages. Compared to alternative approaches, MI minimizes information loss and better characterizes statistical uncertainty about the true length of the immortal time period, avoiding false precision. Additionally, MI explicitly considers the impacts of patient characteristics on immortal time distributions, with inclusion criteria and follow-up period definitions that do not inadvertently risk biasing evaluations. In the real-world case study, survival analysis results did not substantively differ across MI and time distribution matching, but standard errors based on MI were higher for all point estimates. Mean imputed immortal time was stable across simulations.

CONCLUSIONS

Precision medicine evaluations must employ immortal time adjustment methods for unbiased, decision-grade real-world evidence generation. MI is a promising solution to the challenge of immortal time bias.

Precision Antiplatelet Therapy after Percutaneous Coronary Intervention (Precision PCI) Registry - Informing optimal antiplatelet strategies

Dual antiplatelet therapy (DAPT) with aspirin and a P2Y12 receptor inhibitor (clopidogrel, prasugrel, or ticagrelor) is indicated after percutaneous coronary intervention (PCI) to reduce the risk of atherothrombotic events. Approximately 30% of the US population has a CYP2C19 no-function allele that reduces the effectiveness of clopidogrel, but not prasugrel or ticagrelor, after PCI. We have shown improved outcomes with the integration of CYP2C19 genotyping into clinical care to guide the selection of prasugrel or ticagrelor in CYP2C19 no-function allele carriers. However, the influence of patient-specific demographic, clinical, and other genetic factors on outcomes with genotype-guided DAPT has not been defined. In addition, the impact of genotype-guided de-escalation from prasugrel or ticagrelor to clopidogrel in patients without a CYP2C19 no-function allele has not been investigated in a diverse, real-world clinical setting. The Precision Antiplatelet Therapy after Percutaneous Coronary Intervention (Precision PCI) Registry is a multicenter US registry of patients who underwent PCI and clinical CYP2C19 testing. The registry is enrolling a diverse population, assessing atherothrombotic and bleeding events over 12 months, collecting DNA samples, and conducting platelet function testing in a subset of patients. The registry aims to define the influence of African ancestry and other patient-specific factors on clinical outcomes with CYP2C19-guided DAPT, evaluate the safety and effectiveness of CYP2C19-guided DAPT de-escalation following PCI in a real-world setting, and identify additional genetic influences of clopidogrel response after PCI, with the ultimate goal of establishing optimal strategies for individualized antiplatelet therapy that improves outcomes in a diverse, real-world population.

Effect of Direct Acting Antiviral Drugs on the Occurrence and Recurrence of Intra- and Extra-Hepatic Malignancies in Patients with Chronic Hepatitis C Virus Infection

Introduction: The use of direct-acting antivirals (DAAs) has drastically changed the management of HCV-infected patients by achieving a 95-98% sustained virologic response (SVR) and reducing morbidity and mortality in this population. However, despite their effectiveness, controversy exists concerning the occurrence of oncologic events following DAA therapy. Aims and Methods: A retrospective analysis was conducted on data from the Swiss Hepatitis C Cohort Study, a prospective cohort involving patients with positive HCV viremia upon inclusion, enrolled in various Swiss centers from September 2000 to November 2021. To examine potential differences in the risk of intrahepatic tumor (IHT) occurrence and death among patients treated with direct-acting antivirals (DAAs), untreated patients, and those receiving interferon (IFN)-based therapy, a semiparametric competing risk proportional hazards regression model was used. Results: Among 4082 patients (63.1% male, median age 45 years; genotype 1: 54.1%; cirrhosis: 16.1%), 1026 received exclusive treatment with IFN-based regimens, and 1180 were treated solely with DAAs. Over a median follow-up of 7.8 years (range: 3.8-11.9), 179 patients (4.4%) developed intrahepatic tumors (IHT), and 168 (4.1%) experienced extrahepatic tumors (EHT). The 5-year cumulative incidence of IHT was 1.55% (95% CI 0.96-2.48) for IFN-based therapy, 4.27% (95% CI 2.93-6.2) for DAA and 0.89% (95% CI 0.4-1.99) for untreated patients. There was no statistically significant difference in the risk of developing IHT (HR = 1.34; 95% CI = [0.70; 2.58]; p = 0.380) or death (HR = 0.66; 95% CI = [0.43; 1.03]; p = 0.066) between patients treated with DAAs and those treated with IFN. Conclusions: The DAAs reduced the risk of death and were not associated with an increased risk of extrahepatic tumors (EHT). In the adjusted model, accounting for cirrhosis and high liver stiffness, the DAA treatment was associated with a higher risk of IHT occurrence compared with untreated patients, emphasizing the relevance of implementing standardized hepatocellular carcinoma (HCC) screening post-DAA treatment.

Contemporary Prestroke Dual Antiplatelet Use and Symptomatic Intracerebral Hemorrhage Risk After Thrombolysis

Importance

Intravenous alteplase (IV-tPA) can be administered to patients with acute ischemic stroke but is associated with symptomatic intracerebral hemorrhage (sICH). It is unclear if patients taking prestroke dual antiplatelet therapy (DAPT) are at higher risk of sICH.

Objective

To determine the associated risk of sICH in patients taking prestroke dual antiplatelet therapy receiving alteplase for acute ischemic stroke using propensity score matching analysis.

Design, Setting, and Participants

This cohort study used data from the American Heart Association and American Stroke Association Get With The Guidelines-Stroke (GWTG-Stroke) registry between 2013 and 2021. Data were obtained from hospitals in the GWTG-Stroke registry. This study included patients hospitalized with acute ischemic stroke and treated with IV-tPA. Data were analyzed from January 2013 to December 2021.

Exposures

Prestroke DAPT before treatment with IV-tPA for acute ischemic stroke.

Main Outcome Measures

sICH, In-hospital death, discharge modified Rankin scale score, and other life-threatening systemic hemorrhages.

Results

Of 409 673 participants, 321 819 patients (mean [SD] age, 68.6 [15.1] years; 164 587 female [51.1%]) who were hospitalized with acute ischemic stroke and treated with IV-tPA were included in the analysis. The rate of sICH was 2.9% (5200 of 182 344), 3.8% (4457 of 117 670), and 4.1% (893 of 21 805) among patients treated with no antiplatelet therapy, single antiplatelet therapy (SAPT), and DAPT, respectively (P < .001). In adjusted analyses after propensity score subclassification, both SAPT (odds ratio [OR], 1.13; 95% CI, 1.07-1.19) and DAPT (OR, 1.28; 95% CI, 1.14-1.42) were associated with increased risks of sICH. Prestroke antiplatelet medications were associated with lower odds of discharge mRS score of 2 or less compared with no medication (SAPT OR, 0.92; 95% CI, 0.90-0.95; DAPT OR, 0.94; 95% CI, 0.88-0.98). Results of a subgroup analysis of patients taking DAPT exposed to aspirin-clopidogrel vs aspirin-ticagrelor combination therapy were not significant (OR, 1.35; 95% CI, 0.84-1.86).

Conclusions and Relevance

Prestroke DAPT was associated with a significantly elevated risk of sICH among patients with ischemic stroke who were treated with thrombolysis; however, the absolute increase in risk was small. Patients exposed to antiplatelet medications did not have excess sICH compared with landmark trials, which demonstrated overall clinical benefit of thrombolysis therapy for acute ischemic stroke.

Systemic Fluoroquinolone Use and Risk of Uveitis or Retinal Detachment

Importance

Fluoroquinolone use has been associated with increased risk of uveitis and retinal detachment in noninterventional studies, but the findings have been conflicting and causality is unclear.

Objective

To estimate the association of systemic fluoroquinolone use with acute uveitis or retinal detachment, using multiple analyses and multiple databases to increase the robustness of results.

Design, Setting, and Participants

This cohort study used data from the Clinical Practice Research Datalink Aurum and GOLD UK primary care records databases, which were linked to hospital admissions data. Adults prescribed a fluoroquinolone or a comparator antibiotic, cephalosporin, between April 1997 and December 2019 were included. Adults with uveitis or retinal detachment were analyzed in a separate self-controlled case series. Data analysis was performed from May 2022 to May 2023.

Exposures

Systemic fluoroquinolone or comparator antibiotic.

Main Outcomes and Measures

The primary outcome was a diagnosis of acute uveitis or retinal detachment. Hazard ratios (HRs) were estimated in the cohort study for the association of fluoroquinolone prescription with either uveitis or retinal detachment, using stabilized inverse probability of treatment weighted Cox regression. Rate ratios (RRs) were estimated in the self-controlled case series, using conditional Poisson regression. Estimates were pooled across databases using fixed-effects meta-analysis.

Results

In total, 3 001 256 individuals in Aurum (1 893 561 women [63.1%]; median [IQR] age, 51 [35-68] years) and 434 754 in GOLD (276 259 women [63.5%]; median [IQR] age, 53 [37-70] years) were included in the cohort study. For uveitis, the pooled adjusted HRs (aHRs) for use of fluoroquinolone vs cephalosporin were 0.91 (95% CI, 0.72-1.14) at first treatment episode and 1.07 (95% CI, 0.92-1.25) over all treatment episodes. For retinal detachment, the pooled aHRs were 1.37 (95% CI, 0.80-2.36) at first treatment episode and 1.18 (95% CI, 0.84-1.65) over all treatment episodes. In the self-controlled case series, for uveitis, the pooled adjusted RRs (aRRs) for fluoroquinolone use vs nonuse were 1.13 (95% CI, 0.97-1.31) for 1 to 29 days of exposure, 1.16 (95% CI, 1.00-1.34) for 30 to 59 days, and 0.98 (95% CI, 0.74-1.31) for 60 days for longer. For retinal detachment, pooled aRRs for fluoroquinolone use vs nonuse were 1.15 (95% CI, 0.86-1.54) for 1 to 29 days of exposure, 0.94 (95% CI, 0.69-1.30) for 30 to 59 days, and 1.03 (95% CI, 0.59-1.78) for 60 days or longer.

Conclusions and Relevance

These findings do not support an association of systemic fluoroquinolone use with substantively increased risk of uveitis or retinal detachment. Although an association cannot be completely ruled out, these findings indicate that any absolute increase in risk would be small and, hence, of limited clinical importance.

Screening women in young adulthood for disabling dysmenorrhoea: a nationwide cross-sectional study from the CONSTANCES cohort

RESEARCH QUESTION

How do different warning indicators help to identify disabling dysmenorrhoea among women in young adulthood?

DESIGN

A nationwide cross-sectional study of women aged 18-25 years from the CONSTANCES cohort was constructed. Disability was assessed with the Global Activity Limitation Indicator question 'For the past 6 months, have you been limited in routine activities?Yes, severely limited/Yes, limited/ No, not limited'. Dysmenorrhoea pain intensity and other chronic pelvic pain symptoms (dyspareunia and non-menstrual pain) were evaluated according to questions from a specific questionnaire. Probability of disability was estimated using a logistic prediction model according to dysmenorrhoea intensity, other indicators of pelvic pain symptoms and other obvious covariates. The results of the predictive model of disabling dysmenorrhoea were presented on a nomogram.

RESULTS

Among 6377 women, the rate of disability was estimated at 7.5%. Increased intensity of dysmenorrhoea (odds ratio [OR] 1.08, 95% confidence interval [CI] 1.04-1.13), increased frequency of dyspareunia (from OR 1.69, 95% CI 1.33-2.14 up to OR 3.41, 95% CI 2.16-5.38) non-menstrual chronic pelvic pain (OR 1.75, 95% CI 1.40-2.19), body mass index over 25 kg/m2 (OR 1.45, 95% CI 1.17-1.80) and non-use of the hormonal contraceptive pill (OR 1.29, 95% CI 1.05-1.59) were significantly associated with disability. According to the nomogram, a predicted probability of 15% or more could be chosen as a threshold. This represents almost 4.6% of young women in this sample being classified at risk of disabling dysmenorrhoea.

CONCLUSIONS

Dysmenorrhoea pain intensity and associated pelvic pain symptoms are warning indicators that can be measured to help screen young women who may suffer from disabling dysmenorrhoea.

Observational study of frailty in older Japanese patients with non-valvular atrial fibrillation receiving anticoagulation therapy

The number of patients with atrial fibrillation is increasing, and frailty prevalence increases with age, posing challenges for physicians in prescribing anticoagulants to such patients because of possible harm. The effects of frailty on anticoagulant therapy in older Japanese patients with nonvalvular atrial fibrillation (NVAF) are unclear. Herein, we prescribed rivaroxaban to Japanese patients with NVAF and monitored for a mean of 2.0 years. The primary endpoint was stroke or systemic embolism. The secondary endpoints were all-cause or cardiovascular death, composite endpoint, and major or non-major bleeding. Frailty was assessed using the Japanese long-term care insurance system. A multiple imputation technique was used for missing data. The propensity score (PS) was obtained to estimate the treatment effect of frailty and was used to create two PS-matched groups. Overall, 5717 older patients had NVAF (mean age: 73.9 years), 485 (8.5%) were classified as frail. After PS matching, background characteristics were well-balanced between the groups. Rivaroxaban dosages were 10 and 15 mg/day for approximately 80% and the remaining patients, respectively. Frailty was not associated with the primary endpoint or secondary endpoints. In conclusion, frailty does not affect the effectiveness or safety of rivaroxaban anticoagulant therapy in older Japanese patients with NVAF.Trial registration: UMIN000019135, NCT02633982.

Potential interactions between medications for rate control and direct oral anticoagulants: Population-based cohort and case-crossover study

BACKGROUND

Direct oral anticoagulants (DOACs) are commonly co-prescribed with amiodarone/diltiazem/verapamil, but whether there is a drug interaction between these drugs is unclear.

OBJECTIVE

The purpose of this study was to investigate the risk of clinical outcomes associated with concomitant use of DOACs and amiodarone/diltiazem/verapamil.

METHODS

We identified DOAC users in the Clinical Practice Research Datalink Aurum from January 1, 2011, to December 31, 2019. We used a cohort design to estimate hazard ratios for ischemic stroke, myocardial infarction, venous thromboembolism, intracranial bleeding, gastrointestinal bleeding, other bleeding, cardiovascular mortality, and all-cause mortality, comparing DOACs + amiodarone/diltiazem/verapamil users and DOACs + beta-blocker users. A case-crossover design comparing odds of exposure to different drug initiation patterns for all outcomes in hazard window vs referent window within an individual also was conducted.

RESULTS

Of 397,459 DOAC users, we included 9075 co-prescribed amiodarone, 9612 co-prescribed diltiazem, and 2907 co-prescribed verapamil. There was no difference in risk of any outcomes between DOACs + amiodarone/diltiazem/verapamil users vs DOACs + beta-blocker users in the cohort design. However, in the case-crossover design, we observed an odds ratio (OR) of 2.09 (99% confidence interval [CI] 1.37-3.18) for all-cause mortality associated with initiation of a DOAC while taking amiodarone, which was greater than that observed for DOAC monotherapy (OR 1.30; 99% CI 1.25-1.35). Similar findings were observed for cardiovascular mortality and all-cause mortality respectively with diltiazem.

CONCLUSION

Our study showed no evidence of higher bleeding or cardiovascular risk associated with co-prescribed DOACs and amiodarone, diltiazem, or verapamil. Elevated risks of cardiovascular and all-cause mortality were only observed during DOAC initiation when diltiazem/amiodarone were being taken.

On variance estimation of the inverse probability-of-treatment weighting estimator: A tutorial for different types of propensity score weights

Propensity score methods, such as inverse probability-of-treatment weighting (IPTW), have been increasingly used for covariate balancing in both observational studies and randomized trials, allowing the control of both systematic and chance imbalances. Approaches using IPTW are based on two steps: (i) estimation of the individual propensity scores (PS), and (ii) estimation of the treatment effect by applying PS weights. Thus, a variance estimator that accounts for both steps is crucial for correct inference. Using a variance estimator which ignores the first step leads to overestimated variance when the estimand is the average treatment effect (ATE), and to under or overestimated estimates when targeting the average treatment effect on the treated (ATT). In this article, we emphasize the importance of using an IPTW variance estimator that correctly considers the uncertainty in PS estimation. We present a comprehensive tutorial to obtain unbiased variance estimates, by proposing and applying a unifying formula for different types of PS weights (ATE, ATT, matching and overlap weights). This can be derived either via the linearization approach or M-estimation. Extensive R code is provided along with the corresponding large-sample theory. We perform simulation studies to illustrate the behavior of the estimators under different treatment and outcome prevalences and demonstrate appropriate behavior of the analytical variance estimator. We also use a reproducible analysis of observational lung cancer data as an illustrative example, estimating the effect of receiving a PET-CT scan on the receipt of surgery.

Anti-IL-4R versus anti-IL-5/5R after anti-IL-5/5R failure in asthma: An emulated target trial

BACKGROUND

Switching biologics is now common practice in severe eosinophilic asthma. After insufficient response to anti-IL-5 or 5 receptor (anti-IL-5/5R), the optimal switch between an anti-IL-4R mAb (interclass) or another anti-IL-5/5R drug (intraclass) remains unknown.

OBJECTIVE

We sought to compare the effectiveness of these 2 strategies in asthma control in patients with severe eosinophilic asthma and insufficient response to an anti-IL-5/5R mAb.

METHODS

We emulated a target randomized trial using observational data from the Recherche sur les AsthMes SEvèreS (RAMSES) cohort. Eligible patients were switched to an anti-IL-4R mAb or another anti-IL-5/5R drug after insufficient response to an anti-IL-5/5R mAb. The primary outcome was the change in Asthma Control Test score at 6 months.

RESULTS

Among the 2046 patients in the cohort, 151 were included in the study: 103 switched to an anti-IL-4R mAb and 48 to another anti-IL-5/5R. At 6 months, the difference in Asthma Control Test score improvement was not statistically significant (mean difference groups, 0.82 [-0.47 to 2.10], P = .213). The interclass group exhibited greater cumulative reduction in oral corticosteroid dose (Pinter-intra, -1.05 g [-1.76 to -0.34], P = .041). The interclass group had a better effect, although not significantly, on reducing exacerbations (Δinter-intra, -0.37 [-0.77 to 0.02], P = .124) and increasing lung function (FEV1) (126.8 mL [-12.7 to 266.4], P = .124).

CONCLUSIONS

After anti-IL-5/5R mAb insufficient response, switching to dupilumab demonstrated similar improvement in Asthma Control Test scores compared with intraclass switching. However, it appeared more effective in reducing oral corticosteroid use. Larger studies are warranted to confirm these results.

Erythropoiesis-Stimulating Agents and the Risk of Vision-Threatening Diabetic Retinopathy

PURPOSE

Animal studies have suggested that Erythropoiesis-Stimulating Agents (ESAs) may increase vascular endothelial growth factor (VEGF)-related retinopathies, but this effect is unclear in humans. This study evaluates the risk of vision-threatening diabetic retinopathy (VTDR), defined as either diabetic macular edema (DME) or proliferative diabetic retinopathy (PDR), in patients exposed to an ESA.

METHODS

Two analyses were performed. First, a retrospective matched-cohort study was designed using a de-identified commercial and Medicare Advantage medical claims database. The ESA cohort of non-proliferative diabetic retinopathy patients who were new users of an ESA from 2000 to 2022 was matched to controls up to a 3:1 ratio. Exclusion criteria included less than 2 years in the plan, history of VTDR or history of other retinopathy. Multivariable Cox proportional hazards regression with inverse proportional treatment weighting (IPTW) was used to assess the hazard of developing VTDR, DME, and PDR. The second analysis was a self-controlled case series (SCCS) evaluating the incidence rate ratios (IRR) of VTDR during 30-day periods before and after initiating an ESA.

RESULTS

After inclusion of 1502 ESA-exposed patients compared with 2656 controls, IPTW-adjusted hazard ratios found the ESA cohort had an increased hazard of progressing to VTDR (HR = 3.0 95%CI:2.3-3.8;p < .001) and DME (HR = 3.4,95%CI:2.6-4.4,p < .001), but not PDR (HR = 1.0,95%CI:0.5-2.3,p = .95). Similar results were found within the SCCS which demonstrated higher IRRs for VTDR (IRRs = 1.09-1.18;p < .001) and DME (IRRs = 1.16-1.18;p < .001), but not increased IRRs in PDR (IRR = 0.92-0.97,p = .02-0.39).

CONCLUSION

ESAs are associated with higher risks for VTDR and DME, but not PDR. Those studying ESAs as adjunctive therapy for DR should be cautious of possible unintended effects.

Impact of antitumour necrosis factor therapy on surgery in inflammatory bowel disease: a population-based study

OBJECTIVE

It is unclear whether widespread use of biologics is reducing inflammatory bowel disease (IBD) surgical resection rates. We designed a population-based study evaluating the impact of early antitumour necrosis factor (TNF) on surgical resection rates up to 5 years from diagnosis.

DESIGN

We evaluated all patients with IBD diagnosed in Cardiff, Wales 2005-2016. The primary measure was the impact of early (within 1 year of diagnosis) sustained (at least 3 months) anti-TNF compared with no therapy on surgical resection rates. Baseline factors were used to balance groups by propensity scores, with inverse probability of treatment weighting (IPTW) methodology and removing immortal time bias. Crohn's disease (CD) and ulcerative colitis (UC) with IBD unclassified (IBD-U) (excluding those with proctitis) were analysed.

RESULTS

1250 patients were studied. For CD, early sustained anti-TNF therapy was associated with a reduced likelihood of resection compared with no treatment (IPTW HR 0.29 (95% CI 0.13 to 0.65), p=0.003). In UC including IBD-U (excluding proctitis), there was an increase in the risk of colectomy for the early sustained anti-TNF group compared with no treatment (IPTW HR 4.6 (95% CI 1.9 to 10), p=0.001).

CONCLUSIONS

Early sustained use of anti-TNF therapy is associated with reduced surgical resection rates in CD, but not in UC where there was a paradoxical increased surgery rate. This was because baseline clinical factors were less predictive of colectomy than anti-TNF usage. These data support the use of early introduction of anti-TNF therapy in CD whereas benefit in UC cannot be assessed by this methodology.

Propensity Score Weighting with Missing Data on Covariates and Clustered Data Structure

Propensity score (PS) analyses are increasingly popular in behavioral sciences. Two issues often add complexities to PS analyses, including missing data in observed covariates and clustered data structure. In previous research, methods for conducting PS analyses with considering either issue alone were examined. In practice, the two issues often co-occur; but the performance of methods for PS analyses in the presence of both issues has not been evaluated previously. In this study, we consider PS weighting analysis when data are clustered and observed covariates have missing values. A simulation study is conducted to evaluate the performance of different missing data handling methods (complete-case, single-level imputation, or multilevel imputation) combined with different multilevel PS weighting methods (fixed- or random-effects PS models, inverse-propensity-weighting or the clustered weighting, weighted single-level or multilevel outcome models). The results suggest that the bias in average treatment effect estimation can be reduced, by better accounting for clustering in both the missing data handling stage (such as with the multilevel imputation) and the PS analysis stage (such as with the fixed-effects PS model, clustered weighting, and weighted multilevel outcome model). A real-data example is provided for illustration.

Balance diagnostics in propensity score analysis following multiple imputation: A new method

The combination of propensity score analysis and multiple imputation has been prominent in epidemiological research in recent years. However, studies on the evaluation of balance in this combination are limited. In this paper, we propose a new method for assessing balance in propensity score analysis following multiple imputation. A simulation study was conducted to evaluate the performance of balance assessment methods (Leyrat's, Leite's, and new method). Simulated scenarios varied regarding the presence of missing data in the control or treatment and control group, and the imputation model with/without outcome. Leyrat's method was more biased in all the studied scenarios. Leite's method and the combine method yielded balanced results with lower mean absolute difference, regardless of whether the outcome was included in the imputation model or not. Leyrat's method had a higher false positive ratio and Leite's and combine method had higher specificity and accuracy, especially when the outcome was not included in the imputation model. According to simulation results, most of time, Leyrat's method and Leite's method contradict with each other on appraising the balance. This discrepancy can be solved using new combine method.

Estimating the Effect of PrEP in Black Men Who Have Sex with Men: A Framework to Utilize Data from Multiple Non-Randomized Studies to Estimate Causal Effects

Black men who have sex with men (MSM) are disproportionately burdened by the HIV epidemic in the US. The effectiveness of pre-exposure prophylaxis (PrEP) in preventing HIV infection has been demonstrated through randomized placebo-controlled clinical trials in several populations. Importantly, no such trial has been conducted exclusively among Black MSM in the US, and it would be unethical and infeasible to do so now. To estimate the causal effects of PrEP access, initiation, and adherence on HIV risk, we utilized causal inference methods to combine data from two non-randomized studies that exclusively enrolled Black MSM. The estimated relative risks of HIV were: (i) 0.52 (95% confidence interval: 0.21, 1.22) for individuals with versus without PrEP access, (ii) 0.48 (0.12, 0.89) for individuals who initiated PrEP but were not adherent versus those who did not initiate, and (iii) 0.23 (0.02, 0.80) for individuals who were adherent to PrEP versus those who did not initiate. Beyond addressing the knowledge gap around the effect of PrEP in Black MSM in the US, which may have ramifications for public health, we have provided a framework to combine data from multiple non-randomized studies to estimate causal effects, which has broad utility.

Evaluating the safety profile of calcineurin inhibitors: cancer risk in patients with systemic lupus erythematosus from the LUNA registry-a historical cohort study

BACKGROUND

Previous studies have shown conflicting evidence regarding the incidence of cancer in patients with systemic lupus erythematosus (SLE) compared with that in healthy individuals. Calcineurin inhibitors (CNIs) such as cyclosporine and tacrolimus have been widely used to treat SLE; however, their effects on cancer risk remain unclear. We aimed to investigate the incidence of cancer in patients with SLE and determine the potential association between CNI use and cancer risk.

METHODS

The standardized incidence ratio (SIR) of cancer among patients with lupus in the Lupus Registry of Nationwide Institutions (LUNA) was calculated based on the age-standardized incidence rate of cancer reported by Japan's Ministry of Health, Labour and Welfare. We also examined the association between CNI exposure and cancer risk, while considering potential confounding factors. The analysis accounted for confounding variables such as age, sex, smoking history, maximum glucocorticoid dose, treatment history with cyclophosphamide, ongoing hydroxychloroquine, Systemic Lupus International Collaboration Clinics/American College of Rheumatology Damage Index (SDI) value (excluding cancer occurrence), comorbidity of diabetes mellitus, and smoking history.

RESULTS

The study included 704 patients with SLE (625 females; 88.8%) with a median age of 44 years [interquartile range (IQR) = 34-55] years. The median past maximum glucocorticoid dose was 40 mg/day [IQR = 30-60 mg/day], and the SDI at registration was 1 [IQR = 0-2]. Among the patients, 246 (35.1%) had smoking histories, and 38 (5.4%) experienced cancer complications. Gynecological malignancies accounted for 63.2% of all cancers. The SIR of cancer in the LUNA cohort was 1.08 (95% confidence interval [CI] = 0.74-1.43). No statistically significant risks of cancer were found in relation to CNI treatment history; the odds ratio using multiple logistic regression was 1.12 (95% CI = 0.42-3.00), the risk ratio using standardization was 1.18 (95% CI = 0.47-2.16), and the risk ratio using inverse probability weighting was 1.8 (95% CI = 0.41-4.66).

CONCLUSIONS

The incidence of cancer in patients with SLE in the LUNA cohort did not significantly differ from that in the general population. These findings suggest that CNI treatment in this cohort did not pose a risk factor for cancer development.

Oral Nutritional Supplement Prescription and Patient-Reported Symptom Burden Among Patients With Late-Stage Non-Dialysis Chronic Kidney Disease

Background

Malnutrition and protein-energy wasting (PEW) are nutritional complications of advanced chronic kidney disease (CKD) that contribute to morbidity, mortality, and decreased quality of life. No previous studies have assessed the effect of oral nutritional supplements (ONSs) on patient-reported symptom burden among patients with non-dialysis CKD (CKD-ND) who have or are at risk of malnutrition/PEW.

Objective

The objective of this study was (1) to quantify the associations between baseline nutritional parameters and patient-reported symptom scores for wellbeing, tiredness, nausea, and appetite and (2) to compare the change in symptom scores among patients prescribed ONS with patients who did not receive ONS in a propensity-score-matched analysis.

Design

This study conducted observational cohort analysis using provincial registry data.

Setting

This study was done in multidisciplinary CKD clinics in British Columbia.

Patients

Adult patients >18 years of age with CKD-ND entering multidisciplinary CKD clinics between January 1, 2010-July 31, 2019 who had at least 2 Edmonton Symptom Assessment System Revised: Renal (ESASr:Renal) assessments.

Measurements

The measurements include nutrition-related parameters such as body mass index (BMI), serum albumin, serum phosphate, serum bicarbonate, neutrophil-to-lymphocyte ratio (NLR), and ESASr:Renal scores (overall and subscores for wellbeing, tiredness, nausea, and appetite).

Methods

Multivariable linear regression was applied to assess associations between nutritional parameters and ESASr:Renal scores. Propensity-score matching using the greedy method was used to match patients prescribed ONS with those not prescribed ONS using multiple demographic, comorbidity, health care utilization, and temporal factors. Linear regression was used to assess the association between first ONS prescription and change in ESASr:Renal overall score and subscores for wellbeing, tiredness, nausea, and appetite.

Results

Of total, 2076 patients were included. Higher baseline serum albumin was associated with lower overall ESASr:Renal score (-0.20, 95% confidence interval [CI] = -0.40 to -0.01 per 1 g/L increase in albumin), lower subscores for tiredness (-0.04, 95% CI = -0.07 to -0.01), nausea (-0.03, 95% CI = -0.04 to -0.01), and appetite (-0.03, 95% CI = -0.06 to -0.01). Higher BMI was associated with higher overall ESASr:Renal score (0.32, 95% CI = 0.16 to 0.48 per 1 kg/m2 increase in BMI), higher symptom subscores for wellbeing (0.02, 95% CI = 0.00 to 0.04) and tiredness (0.05, 95% CI = 0.02 to 0.07). Higher baseline NLR was associated with higher overall score (0.21, 95% CI = 0.03 to 0.39 per 1 unit increase in NLR), higher symptom subscores for wellbeing (0.03, 95% CI = 0.01 to 0.05) and nausea (0.03, 95% CI = 0.02 to 0.05). In the propensity-score-matched analysis, there were no statistically significant associations between ONS prescription and change in overall ESASr:Renal (beta coefficient for change in ESASr:Renal = 0.17, 95% CI = -2.64 to 2.99) or for subscores for appetite, tiredness, nausea, and wellbeing.

Limitations

Possible residual confounding. The ESASr:Renal assessments were obtained routinely only in patients with G5 CKD-ND and/or experiencing significant CKD-related symptoms.

Conclusions

This exploratory observational analysis of patients with advanced non-dialysis CKD demonstrated BMI, serum albumin, and NLR were modestly associated with patient-reported symptoms, but we did not observe an association between ONS use and change in ESASr:Renal scores.

Use of Analog and Human Insulin in a European Hemodialysis Cohort With Type 2 Diabetes: Associations With Mortality, Hospitalization, MACE, and Hypoglycemia

RATIONALE & OBJECTIVE

Poor glycemic control may contribute to the high mortality rate in patients with type 2 diabetes receiving hemodialysis. Insulin type may influence glycemic control, and its choice may be an opportunity to improve outcomes. This study assessed whether treatment with analog insulin compared with human insulin is associated with different outcomes in people with type 2 diabetes and kidney failure receiving hemodialysis.

STUDY DESIGN

Retrospective cohort study.

SETTING & PARTICIPANTS

People in the Analyzing Data, Recognizing Excellence and Optimizing Outcomes (AROii) study with kidney failure commencing hemodialysis and type 2 diabetes being treated with insulin within 288 dialysis facilities between 2007 and 2009 across 7 European countries. Study participants were followed for 3 years. People with type 1 diabetes were excluded using an established administrative data algorithm.

EXPOSURE

Treatment with an insulin analog or human insulin.

OUTCOME

All-cause mortality, major adverse cardiovascular events (MACE), all-cause hospitalization, and confirmed hypoglycemia (blood glucose<3.0mmol/L sampled during hemodialysis).

ANALYTICAL APPROACH

Inverse probability weighted Cox proportional hazards models to estimate hazard ratios for analog insulin compared with human insulin.

RESULTS

There were 713 insulin analog and 733 human insulin users. Significant variation in insulin type by country was observed. Comparing analog with human insulin at 3 years, the percentage of patients experiencing end points and adjusted hazard ratios (AHR) were 22.0% versus 31.4% (AHR, 0.808 [95% CI, 0.66-0.99], P=0.04) for all-cause mortality, 26.8% versus 35.9% (AHR, 0.817 [95% CI, 0.68-0.98], P=0.03) for MACE, and 58.2% versus 75.0% (AHR, 0.757 [95% CI, 0.67-0.86], P<0.001) for hospitalization. Hypoglycemia was comparable between insulin types at 14.1% versus 15.0% (AHR, 1.169 [95% CI, 0.80-1.72], P=0.4). Consistent strength and direction of the associations were observed across sensitivity analyses.

LIMITATIONS

Residual confounding, lack of more detailed glycemia data.

CONCLUSIONS

In this large multinational cohort of people with type 2 diabetes and kidney failure receiving maintenance hemodialysis, treatment with analog insulins was associated with better clinical outcomes when compared with human insulin.

PLAIN-LANGUAGE SUMMARY

People with diabetes who are receiving dialysis for kidney failure are at high risk of cardiovascular disease and death. This study uses information from 1,446 people with kidney failure from 7 European countries who are receiving dialysis, have type 2 diabetes, and are prescribed either insulin identical to that made in the body (human insulin) or insulins with engineered extra features (insulin analog). After 3 years, fewer participants receiving analog insulins had died, had been admitted to the hospital, or had a cardiovascular event (heart attack, stroke, heart failure, or peripheral vascular disease). These findings suggest that analog insulins should be further explored as a treatment leading to better outcomes for people with diabetes on dialysis.

Dazhu Hongjingtian Injection for Ischemic Stroke: Protocol for a Prospective, Multicenter Observational Study

BACKGROUND

Although results from in vitro studies and small randomized controlled trials have shown positive effects of Dazhu hongjingtian injection (DZHJTI) on acute ischemic stroke (AIS), their generalizability to routine clinical practice remains to be established.

OBJECTIVE

The primary aim of this study is to evaluate the effectiveness of DZHJTI treatment for AIS with regard to changes in the stroke-related neurological deficit from baseline to outpatient follow-up, mortality, subsequent vascular events, disability, and traditional Chinese medicine syndrome in real-world clinical settings. By monitoring for adverse events or significant changes in vital signs and laboratory parameters, we also aim to assess the safety of DZHJTI.

METHODS

This prospective, multicenter cohort study plans to enroll 2000 patients with AIS within 14 days of symptom onset from 30 hospitals across China. Eligible patients will be followed up for 6 months after initiating medication treatments. The primary outcome will be the change in the National Institute of Health Stroke Scale score from baseline to outpatient follow-up. The secondary outcomes include overall mortality, stroke recurrence, new-onset major vascular events, global disability, and improvement of traditional Chinese medicine syndrome in 6 months. Adverse events or clinically significant changes in vital signs and laboratory parameters, regardless of the severity, will be recorded during the trial to assess the safety of DZHJTI. An augmented inverse propensity weighted estimator will be used to reduce variability and improve accuracy in average treatment effects estimation.

RESULTS

The clinical trial registration was approved in October 2022, and the recruitment and enrollment of participants started in November 2022. The study's outcomes are expected to be published in 2025 in reputable, peer-reviewed health-related research journals.

CONCLUSIONS

This real-world cohort study is the first to assess the effectiveness and safety of DZHJTI in treating AIS. It may provide additional clinical evidence, including the duration of response, long-term drug effectiveness, and subgroup efficacy data. The study results will be valuable for clinicians and patients seeking optimal treatment for AIS and could lead to better use of DZHJTI and improved patient outcomes.

TRIAL REGISTRATION

ITMCTR ITMCTR2022000005; http://tinyurl.com/554ns8m5.

INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID)

DERR1-10.2196/52447.

Clinical Impact of a Protocol Involving Cone-Beam CT (CBCT), Fusion Imaging and Ablation Volume Prediction in Percutaneous Image-Guided Microwave Ablation in Patients with Hepatocellular Carcinoma Unsuitable for Standard Ultrasound (US) Guidance

PURPOSE

to evaluate the clinical impact of a protocol for the image-guided percutaneous microwave ablation (MWA) of hepatocellular carcinoma (HCC) that includes cone-beam computed tomography (CBCT), fusion imaging and ablation volume prediction in patients with hepatocellular carcinoma unsuitable for standard ultrasound (US) guidance.

MATERIALS AND METHODS

this study included all patients with HCC treated with MWA between January 2021 and June 2022 in a tertiary institution. Patients were divided into two groups: Group A, treated following the protocol, and Group B, treated with standard ultrasound (US) guidance. Follow-up images were reviewed to assess residual disease (RD), local tumor progression (LTP) and intrahepatic distant recurrence (IDR). Ablation response at 1 month was also evaluated according to mRECIST. Baseline variables and outcomes were compared between the groups. For 1-month RD, propensity score weighting (PSW) was performed.

RESULTS

80 consecutive patients with 101 HCCs treated with MWA were divided into two groups. Group A had 41 HCCs in 37 patients, and Group B had 60 HCCs in 43 patients. Among all baseline variables, the groups differed regarding their age (mean of 72 years in Group A and 64 years in Group B, respectively), new vs. residual tumor rates (48% Group A vs. 25% Group B, p < 0.05) and number of subcapsular tumors (56.7% Group B vs. 31.7% Group A, p < 0.05) and perivascular tumors (51.7% Group B vs. 17.1% Group A, p < 0.05). The protocol led to repositioning the antenna in 49% of cases. There was a significant difference in 1-month local response between the groups measured as the RD rate and mRECIST outcomes. LTP rates at 3 and 6 months, and IDR rates at 1, 3 and 6 months, showed no significant differences. Among all variables, logistic regression after PSW demonstrated a protective effect of the protocol against 1-month RD.

CONCLUSIONS

The use of CBCT, fusion imaging and ablation volume prediction during percutaneous MWA of HCCs provided a better 1-month tumor local control. Further studies with a larger population and longer follow-up are needed.

Bayesian causal inference for observational studies with missingness in covariates and outcomes

Missing data are a pervasive issue in observational studies using electronic health records or patient registries. It presents unique challenges for statistical inference, especially causal inference. Inappropriately handling missing data in causal inference could potentially bias causal estimation. Besides missing data problems, observational health data structures typically have mixed-type variables - continuous and categorical covariates - whose joint distribution is often too complex to be modeled by simple parametric models. The existence of missing values in covariates and outcomes makes the causal inference even more challenging, while most standard causal inference approaches assume fully observed data or start their works after imputing missing values in a separate preprocessing stage. To address these problems, we introduce a Bayesian nonparametric causal model to estimate causal effects with missing data. The proposed approach can simultaneously impute missing values, account for multiple outcomes, and estimate causal effects under the potential outcomes framework. We provide three simulation studies to show the performance of our proposed method under complicated data settings whose features are similar to our case studies. For example, Simulation Study 3 assumes the case where missing values exist in both outcomes and covariates. Two case studies were conducted applying our method to evaluate the comparative effectiveness of treatments for chronic disease management in juvenile idiopathic arthritis and cystic fibrosis.

Addressing missing data in the estimation of time-varying treatments in comparative effectiveness research

Comparative effectiveness research is often concerned with evaluating treatment strategies sustained over time, that is, time-varying treatments. Inverse probability weighting (IPW) is often used to address the time-varying confounding by re-weighting the sample according to the probability of treatment receipt at each time point. IPW can also be used to address any missing data by re-weighting individuals according to the probability of observing the data. The combination of these two distinct sets of weights may lead to inefficient estimates of treatment effects due to potentially highly variable total weights. Alternatively, multiple imputation (MI) can be used to address the missing data by replacing each missing observation with a set of plausible values drawn from the posterior predictive distribution of the missing data given the observed data. Recent studies have compared IPW and MI for addressing the missing data in the evaluation of time-varying treatments, but they focused on missing confounders and monotone missing data patterns. This article assesses the relative advantages of MI and IPW to address missing data in both outcomes and confounders measured over time, and across monotone and non-monotone missing data settings. Through a comprehensive simulation study, we find that MI consistently provided low bias and more precise estimates compared to IPW across a wide range of scenarios. We illustrate the implications of method choice in an evaluation of biologic drugs for patients with severe rheumatoid arthritis, using the US National Databank for Rheumatic Diseases, in which 25% of participants had missing health outcomes or time-varying confounders.

Perioperative Use of Intravenous Levodopa as an Anti-Parkinsonian Drug: A Propensity Score Analysis

Background

Perioperative discontinuation of oral anti-parkinsonian medication can negatively impact the prognosis of abdominal surgery in patients with Parkinson's disease. Although intravenous levodopa may be an alternative, its efficacy has not yet been investigated.

Objectives

To determine the efficacy of intravenous levodopa as an alternative to oral anti-Parkinsonian drugs during gastric or colorectal cancer surgery.

Methods

We identified patients with Parkinson's disease who underwent surgery for gastric or colorectal cancer between April 2010 and March 2020, using the Diagnosis Procedure Combination database, a nationwide inpatient database in Japan. Patients were divided into two groups: those who received intravenous levodopa during the perioperative period and those who did not. We compared in-hospital mortalities, major complications, and postoperative length of stay between the groups after adjusting for background characteristics with overlap weights based on propensity scores.

Results

We identified 648 patients who received intravenous levodopa and 1207 who did not receive levodopa during the perioperative period. In the adjusted cohort, the mean postoperative length of stay was 24.7 and 29.0 days (percent difference, -7.7%; 95% confidence interval, -13.1 to -1.5); in-hospital death was 3.2% and 3.3% (adjusted odds ratio, 0.95; 95% CI: 0.54-1.67); and incidence of major complications were 21.4% and 19.3% (adjusted odds ratio, 0.89; 95% confidence interval, 0.70-1.13) in those with and without intravenous levodopa, respectively.

Conclusions

Intravenous levodopa was associated with a shorter postoperative length of stay, but not with mortality or morbidity. Intravenous levodopa may improve perioperative care in patients with Parkinson's disease.

SGLT2i reduces risk of developing HCC in patients with co-existing type 2 diabetes and hepatitis B infection: A territory-wide cohort study in Hong Kong

BACKGROUND AND AIMS

Type 2 diabetes (T2D) and chronic hepatitis B infection (CHB) are risk factors of HCC. Sodium glucose co-transporter 2 inhibitors (SGLT2i) inhibit HCC oncogenesis in preclinical studies. However, clinical studies are lacking. This study aimed to evaluate the impact of SGLT2i use on incident HCC using a territory-wide cohort of exclusively patients with co-existing T2D and CHB.

APPROACH AND RESULTS

Patients with co-existing T2D and CHB between 2015 and 2020 were identified from the representative electronic database of the Hong Kong Hospital Authority. Patients with and without SGLT2i use were 1:1 matched by propensity score for their demographics, biochemistry results, liver-related characteristics, and background medications. Cox proportional hazards regression model was used to assess the association between SGLT2i use and incident HCC. A total of 2,000 patients with co-existing T2D and CHB (1,000 in each SGLT2i and non-SGLT2i group; 79.7% on anti-HBV therapy at baseline) were included after propensity-score matching. Over a follow-up of 3,704 person-years, the incidence rates of HCC were 1.39 and 2.52 cases per 100 person-year in SGLT2i and non-SGLT2i groups, respectively. SGLT2i use was associated with a significantly lower risk of incident HCC (HR 0.54, 95%CI: 0.33-0.88, p =0.013). The association remained similar regardless of sex, age, glycemic control, diabetes duration, presence of cirrhosis and hepatic steatosis, timing of anti-HBV therapy, and background antidiabetic agents including dipeptidyl peptidase-4 inhibitors, insulin, or glitazones (all p interaction>0.05).

CONCLUSIONS

Among patients with co-existing T2D and CHB, SGLT2i use was associated with a lower risk of incident HCC.

Initiation of continuous renal replacement therapy versus intermittent hemodialysis in critically ill patients with severe acute kidney injury: a secondary analysis of STARRT-AKI trial

BACKGROUND

There is controversy regarding the optimal renal-replacement therapy (RRT) modality for critically ill patients with acute kidney injury (AKI).

METHODS

We conducted a secondary analysis of the STandard versus Accelerated Renal Replacement Therapy in Acute Kidney Injury (STARRT-AKI) trial to compare outcomes among patients who initiated RRT with either continuous renal replacement therapy (CRRT) or intermittent hemodialysis (IHD). We generated a propensity score for the likelihood of receiving CRRT and used inverse probability of treatment with overlap-weighting to address baseline inter-group differences. The primary outcome was a composite of death or RRT dependence at 90-days after randomization.

RESULTS

We identified 1590 trial participants who initially received CRRT and 606 who initially received IHD. The composite outcome of death or RRT dependence at 90-days occurred in 823 (51.8%) patients who commenced CRRT and 329 (54.3%) patients who commenced IHD (unadjusted odds ratio (OR) 0.90; 95% confidence interval (CI) 0.75-1.09). After balancing baseline characteristics with overlap weighting, initial receipt of CRRT was associated with a lower risk of death or RRT dependence at 90-days compared with initial receipt of IHD (OR 0.81; 95% CI 0.66-0.99). This association was predominantly driven by a lower risk of RRT dependence at 90-days (OR 0.61; 95% CI 0.39-0.94).

CONCLUSIONS

In critically ill patients with severe AKI, initiation of CRRT, as compared to IHD, was associated with a significant reduction in the composite outcome of death or RRT dependence at 90-days.

Dealing with missing data in laboratory test results used as a baseline covariate: results of multi-hospital cohort studies utilizing a database system contributing to MID-NET® in Japan

BACKGROUND

To evaluate missing data methods applied to laboratory test results used for confounding adjustment, utilizing data from 10 MID-NET®-collaborative hospitals.

METHODS

Using two scenarios, five methods dealing with missing laboratory test results were applied, including three missing data methods (single regression imputation (SRI), multiple imputation (MI), and inverse probability weighted (IPW) method). We compared the point estimates of adjusted hazard ratios (aHRs) and 95% confidence intervals (CIs) between the five methods. Hospital variability in missing data was considered using the hospital-specific approach and overall approach. Confounding adjustment methods were propensity score (PS) weighting, PS matching, and regression adjustment.

RESULTS

In Scenario 1, the risk of diabetes due to second-generation antipsychotics was compared with that due to first-generation antipsychotics. The aHR adjusted by PS weighting using SRI, MI, and IPW by the hospital-specific-approach was 0.61 [95%CI, 0.39-0.96], 0.63 [95%CI, 0.42-0.93], and 0.76 [95%CI, 0.46-1.25], respectively. In Scenario 2, the risk of liver injuries due to rosuvastatin was compared with that due to atorvastatin. Although PS matching largely contributed to differences in aHRs between methods, PS weighting provided no substantial difference in point estimates of aHRs between SRI and MI, similar to Scenario 1. The results of SRI and MI in both scenarios showed no considerable changes, even upon changing the approaches considering hospital variations.

CONCLUSIONS

SRI and MI provide similar point estimates of aHR. Two approaches considering hospital variations did not markedly affect the results. Adjustment by PS matching should be used carefully.

End-of-life care costs and place of death across health and social care sectors

OBJECTIVES

This study explores the relationship between end-of-life care costs and place of death across different health and social care sectors.

METHODS

We used a linked local government and health data of East London residents (n=4661) aged 50 or over, deceased between 2016 and 2020. Individuals who died in hospital were matched to those who died elsewhere according to a wide range of demographic, socioeconomic and health factors. We reported mean healthcare costs and 95% CIs by care sectors over the 12-month period before death. Subgroup analyses were conducted to investigate if the role of place of death differs according to long-term conditions and age.

RESULTS

We found that mean difference in total cost between hospital and non-hospital decedents was £4565 (95% CI £3132 to £6046). Hospital decedents were associated with higher hospital cost (£5196, £4499 to £5905), higher mental healthcare cost (£283, £78 to £892) and lower social care cost (-£838, -£1,209 to -£472), compared with individuals who died elsewhere. Subgroup analysis shows that the association between place of death and healthcare costs differs by age and long-term conditions, including cancer, mental health and cardiovascular diseases.

CONCLUSION

This study suggests that trajectories of end-of-life healthcare costs vary by place of death in a differential way across health and social care sectors. High hospital burden for cancer patients may be alleviated by strengthening healthcare provision in less cost-intensive settings, such as community and social care.

Do patients with prediabetes managed with metformin achieve better glycaemic control? A national study using primary care medical records

AIMS

To estimate the effectiveness of metformin on glycaemic parameters among participants with incident prediabetes attending Australian general practices.

METHODS

This retrospective cohort study used electronic health records of regular participants (3+ visits in two consecutive years) attending 383 Australian general practices (MedicineInsight). Participants with 'incident' prediabetes (newly recorded diagnosis between 2012 and 2017) and their glycaemic parameters (haemoglobin A1c [HbA1c] or fasting blood glucose [FBG]) at 6-, 12-, and 18-24 months post diagnosis (unexposed) or post-management with metformin (treatment) were identified from the database. We estimated the average treatment effect (ATE) of metformin management on glycaemic parameters using both linear regression and augmented inverse probability weighting.

RESULTS

Of the 4770 investigated participants with 'incident' prediabetes, 10.2% were managed with metformin. Participants on metformin had higher HbA1c levels at the baseline than those unexposed (mean 45 mmol/mol [6.2%] and 41 mmol/mol [5.9%], respectively), but no differences were observed at 6-12 months (mmol/mol ATE 0.0, 95% CI -0.4; 0.7) or 12-18 months (ATE -0.3, 95% CI -1.2; 0.3). However, participants on metformin had lower mean HbA1c mmol/mol at 18-24 months (ATE -1.1, 95% CI -2.0; 0.1) than those unexposed. Consistent results were observed for FBG (ATE at 6-12 months -0.14 [95% CI -0.25; -0.04], 12-18 months 0.02 [95% CI -0.08; 0.13] and 18-24 months -0.07 [95% CI -0.25; 0.12]).

CONCLUSION

The higher HbA1c and FBG baseline levels among participants with 'incident' prediabetes managed with metformin improved after 6-12 months of starting pharmacological management, and the effect persisted for up to 24 months. Management with metformin could prevent further deterioration of glycaemic levels.

Combining meta-analysis with multiple imputation for one-step, privacy-protecting estimation of causal treatment effects in multi-site studies

Missing data complicates statistical analyses in multi-site studies, especially when it is not feasible to centrally pool individual-level data across sites. We combined meta-analysis with within-site multiple imputation for one-step estimation of the average causal effect (ACE) of a target population comprised of all individuals from all data-contributing sites within a multi-site distributed data network, without the need for sharing individual-level data to handle missing data. We considered two orders of combination and three choices of weights for meta-analysis, resulting in six approaches. The first three approaches, denoted as RR + metaF, RR + metaR and RR + std, first combined results from imputed data sets within each site using Rubin's rules and then meta-analyzed the combined results across sites using fixed-effect, random-effects and sample-standardization weights, respectively. The last three approaches, denoted as metaF + RR, metaR + RR and std + RR, first meta-analyzed results across sites separately for each imputation and then combined the meta-analysis results using Rubin's rules. Simulation results confirmed very good performance of RR + std and std + RR under various missing completely at random and missing at random settings. A direct application of the inverse-variance weighted meta-analysis based on site-specific ACEs can lead to biased results for the targeted network-wide ACE in the presence of treatment effect heterogeneity by site, demonstrating the need to clearly specify the target population and estimand and properly account for potential site heterogeneity in meta-analyses seeking to draw causal interpretations. An illustration using a large administrative claims database is presented.

Real-world effectiveness of intensive chemotherapy with 7&3 versus venetoclax and hypomethylating agent in acute myeloid leukemia

Intensive chemotherapy with cytarabine and anthracycline (7&3) remains the standard therapy for patients medically fit for induction, but the assessment of fitness remains controversial. Venetoclax and hypomethylating agent (ven/HMA) combination therapy has improved outcomes in unfit patients but no prospective study has assessed ven/HMA versus 7&3 as initial therapy in older, fit patients. Given no studies and expectation of ven/HMA use in patients outside of trial criteria, we evaluated retrospective outcomes among newly diagnosed patients. A nationwide electronic health record (EHR)-derived database and the University of Pennsylvania EHR identified 312 patients receiving 7&3 and 488 receiving ven/HMA who were 60-75 years old without history of organ failure. Ven/HMA patients were older and more likely to have secondary AML, adverse cytogenetics, and adverse mutations. Median overall survival (OS) for patients receiving intensive chemotherapy was 22 versus 10 months for ven/HMA (HR 0.53, 95% CI 0.40-0.60). Controlling for measured baseline characteristic imbalances reduced survival advantage by half (HR 0.71, 95% CI 0.53-0.94). A sub-group of patients with equipoise, likelihood at least 30%-70% of receiving either treatment, had similar OS outcomes (HR 1.10, 95% CI 0.75-1.6). Regarding safety outcomes, 60-day mortality was higher for ven/HMA (15% vs. 6% at 60 days) despite higher documented infections and febrile neutropenia for 7&3. In this multicenter real-word dataset, patients selected for intensive chemotherapy had superior OS but a large group had similar outcomes with ven/HMA. Prospective randomized studies, controlling for both measured and unmeasured confounders, must confirm this outcome.

Inverse Probability Weights for Quasicontinuous Ordinal Exposures With a Binary Outcome: Method Comparison and Case Study

Inverse probability weighting (IPW), a well-established method of controlling for confounding in observational studies with binary exposures, has been extended to analyses with continuous exposures. Methods developed for continuous exposures may not apply when the exposure is quasicontinuous because of irregular exposure distributions that violate key assumptions. We used simulations and cluster-randomized clinical trial data to assess 4 approaches developed for continuous exposures-ordinary least squares (OLS), covariate balancing generalized propensity scores (CBGPS), nonparametric covariate balancing generalized propensity scores (npCBGPS), and quantile binning (QB)-and a novel method, a cumulative probability model (CPM), in quasicontinuous exposure settings. We compared IPW stability, covariate balance, bias, mean squared error, and standard error estimation across 3,000 simulations with 6 different quasicontinuous exposures, varying in skewness and granularity. In general, CBGPS and npCBGPS resulted in excellent covariate balance, and npCBGPS was the least biased but the most variable. The QB and CPM approaches had the lowest mean squared error, particularly with marginally skewed exposures. We then successfully applied the IPW approaches, together with missing-data techniques, to assess how session attendance (out of a possible 15) in a partners-based clustered intervention among pregnant couples living with human immunodeficiency virus in Mozambique (2017-2022) influenced postpartum contraceptive uptake.

The Association of Oxaliplatin-Containing Adjuvant Chemotherapy Duration with Overall and Cancer-Specific Mortality in Individuals with Stage III Colon Cancer: A Population-Based Retrospective Cohort Study

PURPOSE

Few studies have examined the relationship between duration of oxaliplatin-containing adjuvant chemotherapy for stage III colon cancer and mortality in routine practice. We examined the association between treatment with 50% versus >85% of a maximal course of adjuvant therapy (eight cycles of CAPOX, twelve cycles of FOLFOX) and mortality in stage III colon cancer.

METHODS

Using linked databases, we identified Ontarians aged ≥18 years at diagnosis of stage III colon cancer between 2007 and 2019. In the primary comparison, we compared patients who received 50% or >85% of a maximal course of adjuvant therapy; in a secondary comparison, we evaluated a dose effect across patients who received FOLFOX in one-cycle increments from six to ten cycles against >85% (more than ten cycles) of a maximal course of FOLFOX. The main outcomes were overall and cancer-specific mortality. Follow-up began 270 days after adjuvant treatment initiation and terminated at the first of the outcome of interest, loss of eligibility for Ontario's Health Insurance Program, or study end. Overlap propensity score weights accounted for baseline between-group differences. We determined the hazard ratio, estimating the association between mortality and treatment. Non-inferiority was concluded in the primary comparison for either outcome if the upper limit of the two-sided 95% CI was ≤1.11, which is the margin used in the International Duration Evaluation of Adjuvant Chemotherapy Collaboration.

RESULTS

We included 3546 patients in the analysis of overall mortality; 486 (13.7%) received 50% and 3060 (86.3%) received >85% of a maximal course of therapy. Median follow-up was 5.4 years, and total follow-up was 20,510 person-years. There were 833 deaths. Treatment with 50% of a maximal course of adjuvant therapy was associated with a hazard ratio of 1.13 (95% CI 0.88 to 1.47) for overall mortality and a subdistribution hazard ratio of 1.31 (95% CI 0.91 to 1.87) for cancer-specific mortality versus >85% of a maximal course of therapy. In the secondary comparison, there was a trend toward higher overall mortality in patients treated with shorter durations of therapy, though confidence intervals overlapped considerably.

CONCLUSION

We could not conclude that treatment with 50% of a maximal course is non-inferior to >85% of a maximal course of adjuvant therapy for mortality in stage III colon cancer. Clinicians and patients engaging in decision-making around treatment duration in this context should carefully consider the trade-off between treatment effectiveness and adverse effects of treatment.

The association between dementia and the risk of hypoglycaemia events among patients with diabetes mellitus: a propensity-score matched cohort analysis

Background

Hypoglycaemia commonly occurs in patients diagnosed with diabetes mellitus (DM) and dementia. The impact of dementia on hypoglycaemic events is controversial. Thus, we evaluated whether dementia increases the risk of hypoglycaemic events in older patients diagnosed with DM.

Design

A retrospective cohort study.

Setting

We used the IQVIA Medical Research Data (IMRD-UK) database (formerly known as the THIN database).

Participants

All patients aged ≥55 years and diagnosed with DM who were prescribed at least two prescriptions of antidiabetic medication between 2000 and 2017. Two groups of patients, dementia and non-dementia group, were propensity-score (PS) matched at 1:2. The risk of hypoglycaemia was assessed through a Cox regression analysis.

Main outcome and measures

Hypoglycaemic events were determined during the follow-up period by Read codes.

Results

From the database, 133,664 diabetic patients were identified, with a mean follow-up of 6.11 years. During the study period, 7,762 diabetic patients diagnosed with dementia were matched with 12,944 diabetic patients who had not been diagnosed with dementia. The PS-matched Cox regression analysis showed that patients diagnosed with dementia were at a 2-fold increased risk for hypoglycaemic events compared with those not diagnosed with dementia (hazard ratio [HR], 2.00; 95% CI, 1.63-2.66). A similar result was shown for a multivariable analysis using all patient data (adjusted HR, 2.25; 95% CI, 2.22-2.32).

Conclusion

Our findings suggest that diabetic patients with a diagnosis of dementia have a statistically significant higher risk of experiencing hypoglycaemia.

Inverse Probability of Treatment Weighting and Confounder Missingness in Electronic Health Record-based Analyses: A Comparison of Approaches Using Plasmode Simulation

BACKGROUND

Electronic health record (EHR) data represent a critical resource for comparative effectiveness research, allowing investigators to study intervention effects in real-world settings with large patient samples. However, high levels of missingness in confounder variables is common, challenging the perceived validity of EHR-based investigations.

METHODS

We investigated performance of multiple imputation and propensity score (PS) calibration when conducting inverse probability of treatment weights (IPTW)-based comparative effectiveness research using EHR data with missingness in confounder variables and outcome misclassification. Our motivating example compared effectiveness of immunotherapy versus chemotherapy treatment of advanced bladder cancer with missingness in a key prognostic variable. We captured complexity in EHR data structures using a plasmode simulation approach to spike investigator-defined effects into resamples of a cohort of 4361 patients from a nationwide deidentified EHR-derived database. We characterized statistical properties of IPTW hazard ratio estimates when using multiple imputation or PS calibration missingness approaches.

RESULTS

Multiple imputation and PS calibration performed similarly, maintaining ≤0.05 absolute bias in the marginal hazard ratio even when ≥50% of subjects had missing at random or missing not at random confounder data. Multiple imputation required greater computational resources, taking nearly 40 times as long as PS calibration to complete. Outcome misclassification minimally increased bias of both methods.

CONCLUSION

Our results support multiple imputation and PS calibration approaches to missingness in missing completely at random or missing at random confounder variables in EHR-based IPTW comparative effectiveness analyses, even with missingness ≥50%. PS calibration represents a computationally efficient alternative to multiple imputation.

Generalizing treatment effects with incomplete covariates: Identifying assumptions and multiple imputation algorithms

We focus on the problem of generalizing a causal effect estimated on a randomized controlled trial (RCT) to a target population described by a set of covariates from observational data. Available methods such as inverse propensity sampling weighting are not designed to handle missing values, which are however common in both data sources. In addition to coupling the assumptions for causal effect identifiability and for the missing values mechanism and to defining appropriate estimation strategies, one difficulty is to consider the specific structure of the data with two sources and treatment and outcome only available in the RCT. We propose three multiple imputation strategies to handle missing values when generalizing treatment effects, each handling the multisource structure of the problem differently (separate imputation, joint imputation with fixed effect, joint imputation ignoring source information). As an alternative to multiple imputation, we also propose a direct estimation approach that treats incomplete covariates as semidiscrete variables. The multiple imputation strategies and the latter alternative rely on different sets of assumptions concerning the impact of missing values on identifiability. We discuss these assumptions and assess the methods through an extensive simulation study. This work is motivated by the analysis of a large registry of over 20,000 major trauma patients and an RCT studying the effect of tranexamic acid administration on mortality in major trauma patients admitted to intensive care units. The analysis illustrates how the missing values handling can impact the conclusion about the effect generalized from the RCT to the target population.

Neutralizing Monoclonal Antibody Use and COVID-19 Infection Outcomes

Importance

Evidence on the effectiveness and safety of COVID-19 therapies across a diverse population with varied risk factors is needed to inform clinical practice.

Objective

To assess the safety of neutralizing monoclonal antibodies (nMAbs) for the treatment of COVID-19 and their association with adverse outcomes.

Design, Setting, and Participants

This retrospective cohort study included 167 183 patients from a consortium of 4 health care systems based in California, Minnesota, Texas, and Utah. The study included nonhospitalized patients 12 years and older with a positive COVID-19 laboratory test collected between November 9, 2020, and January 31, 2022, who met at least 1 emergency use authorization criterion for risk of a poor outcome.

Exposure

Four nMAb products (bamlanivimab, bamlanivimab-etesevimab, casirivimab-imdevimab, and sotrovimab) administered in the outpatient setting.

Main Outcomes and Measures

Clinical and SARS-CoV-2 genomic sequence data and propensity-adjusted marginal structural models were used to assess the association between treatment with nMAbs and 4 outcomes: all-cause emergency department (ED) visits, hospitalization, death, and a composite of hospitalization or death within 14 days and 30 days of the index date (defined as the date of the first positive COVID-19 test or the date of referral). Patient index dates were categorized into 4 variant epochs: pre-Delta (November 9, 2020, to June 30, 2021), Delta (July 1 to November 30, 2021), Delta and Omicron BA.1 (December 1 to 31, 2021), and Omicron BA.1 (January 1 to 31, 2022).

Results

Among 167 183 patients, the mean (SD) age was 47.0 (18.5) years; 95 669 patients (57.2%) were female at birth, 139 379 (83.4%) were White, and 138 900 (83.1%) were non-Hispanic. A total of 25 241 patients received treatment with nMAbs. Treatment with nMAbs was associated with lower odds of ED visits within 14 days (odds ratio [OR], 0.76; 95% CI, 0.68-0.85), hospitalization within 14 days (OR, 0.52; 95% CI, 0.45-0.59), and death within 30 days (OR, 0.14; 95% CI, 0.10-0.20). The association between nMAbs and reduced risk of hospitalization was stronger in unvaccinated patients (14-day hospitalization: OR, 0.51; 95% CI, 0.44-0.59), and the associations with hospitalization and death were stronger in immunocompromised patients (hospitalization within 14 days: OR, 0.31 [95% CI, 0.24-0.41]; death within 30 days: OR, 0.13 [95% CI, 0.06-0.27]). The strength of associations of nMAbs increased incrementally among patients with a greater probability of poor outcomes; for example, the ORs for hospitalization within 14 days were 0.58 (95% CI, 0.48-0.72) among those in the third (moderate) risk stratum and 0.41 (95% CI, 0.32-0.53) among those in the fifth (highest) risk stratum. The association of nMAb treatment with reduced risk of hospitalizations within 14 days was strongest during the Delta variant epoch (OR, 0.37; 95% CI, 0.31-0.43) but not during the Omicron BA.1 epoch (OR, 1.29; 95% CI, 0.68-2.47). These findings were corroborated in the subset of patients with viral genomic data. Treatment with nMAbs was associated with a significant mortality benefit in all variant epochs (pre-Delta: OR, 0.16 [95% CI, 0.08-0.33]; Delta: OR, 0.14 [95% CI, 0.09-0.22]; Delta and Omicron BA.1: OR, 0.10 [95% CI, 0.03-0.35]; and Omicron BA.1: OR, 0.13 [95% CI, 0.02-0.93]). Potential adverse drug events were identified in 38 treated patients (0.2%).

Conclusions and Relevance

In this study, nMAb treatment for COVID-19 was safe and associated with reductions in ED visits, hospitalization, and death, although it was not associated with reduced risk of hospitalization during the Omicron BA.1 epoch. These findings suggest that targeted risk stratification strategies may help optimize future nMAb treatment decisions.

Propensity score matching after multiple imputation when a confounder has missing data

One of the main challenges when using observational data for causal inference is the presence of confounding. A classic approach to account for confounding is the use of propensity score techniques that provide consistent estimators of the causal treatment effect under four common identifiability assumptions for causal effects, including that of no unmeasured confounding. Propensity score matching is a very popular approach which, in its simplest form, involves matching each treated patient to an untreated patient with a similar estimated propensity score, that is, probability of receiving the treatment. The treatment effect can then be estimated by comparing treated and untreated patients within the matched dataset. When missing data arises, a popular approach is to apply multiple imputation to handle the missingness. The combination of propensity score matching and multiple imputation is increasingly applied in practice. However, in this article we demonstrate that combining multiple imputation and propensity score matching can lead to over-coverage of the confidence interval for the treatment effect estimate. We explore the cause of this over-coverage and we evaluate, in this context, the performance of a correction to Rubin's rules for multiple imputation proposed by finding that this correction removes the over-coverage.

Tocilizumab Outcomes in Critically Ill COVID-19 Patients Admitted to the ICU and the Role of Non-Tocilizumab COVID-19-Specific Medical Therapeutics

Background: Tocilizumab is a monoclonal antibody proposed to manage cytokine release syndrome (CRS) associated with severe COVID-19. Previously published reports have shown that tocilizumab may improve the clinical outcomes of critically ill patients admitted to the ICU. However, no precise data about the role of other medical therapeutics concurrently used for COVID-19 on this outcome have been published. Objectives: We aimed to compare the overall outcome of critically ill COVID-19 patients admitted to the ICU who received tocilizumab with the outcome of matched patients who did not receive tocilizumab while controlling for other confounders, including medical therapeutics for critically ill patients admitted to ICUs. Methods: A prospective, observational, multicenter cohort study was conducted among critically ill COVID-19 patients admitted to the ICU of 14 hospitals in Saudi Arabia between 1 March 2020, and October 31, 2020. Propensity-score matching was utilized to compare patients who received tocilizumab to patients who did not. In addition, the log-rank test was used to compare the 28 day hospital survival of patients who received tocilizumab with those who did not. Then, a multivariate logistic regression analysis of the matched groups was performed to evaluate the impact of the remaining concurrent medical therapeutics that could not be excluded via matching 28 day hospital survival rates. The primary outcome measure was patients’ overall 28 day hospital survival, and the secondary outcomes were ICU length of stay and ICU survival to hospital discharge. Results: A total of 1470 unmatched patients were included, of whom 426 received tocilizumab. The total number of propensity-matched patients was 1278. Overall, 28 day hospital survival revealed a significant difference between the unmatched non-tocilizumab group (586; 56.1%) and the tocilizumab group (269; 63.1%) (p-value = 0.016), and this difference increased even more in the propensity-matched analysis between the non-tocilizumab group (466.7; 54.6%) and the tocilizumab group (269; 63.1%) (p-value = 0.005). The matching model successfully matched the two groups’ common medical therapeutics used to treat COVID-19. Two medical therapeutics remained significantly different, favoring the tocilizumab group. A multivariate logistic regression was performed for the 28 day hospital survival in the propensity-matched patients. It showed that neither steroids (OR: 1.07 (95% CI: 0.75–1.53)) (p = 0.697) nor favipiravir (OR: 1.08 (95% CI: 0.61–1.9)) (p = 0.799) remained as a predictor for an increase in 28 day survival. Conclusion: The tocilizumab treatment in critically ill COVID-19 patients admitted to the ICU improved the overall 28 day hospital survival, which might not be influenced by the concurrent use of other COVID-19 medical therapeutics, although further research is needed to confirm this.

Causal inference in survival analysis under deterministic missingness of confounders in register data

Long-term register data offer unique opportunities to explore causal effects of treatments on time-to-event outcomes, in well-characterized populations with minimum loss of follow-up. However, the structure of the data may pose methodological challenges. Motivated by the Swedish Renal Registry and estimation of survival differences for renal replacement therapies, we focus on the particular case when an important confounder is not recorded in the early period of the register, so that the entry date to the register deterministically predicts confounder missingness. In addition, an evolving composition of the treatment arms populations, and suspected improved survival outcomes in later periods lead to informative administrative censoring, unless the entry date is appropriately accounted for. We investigate different consequences of these issues on causal effect estimation following multiple imputation of the missing covariate data. We analyse the performance of different combinations of imputation models and estimation methods for the population average survival. We further evaluate the sensitivity of our results to the nature of censoring and misspecification of fitted models. We find that an imputation model including the cumulative baseline hazard, event indicator, covariates and interactions between the cumulative baseline hazard and covariates, followed by regression standardization, leads to the best estimation results overall, in simulations. Standardization has two advantages over inverse probability of treatment weighting here: it can directly account for the informative censoring by including the entry date as a covariate in the outcome model, and allows for straightforward variance computation using readily available software.

How Strong Is the Evidence for a Causal Reciprocal Effect? Contrasting Traditional and New Methods to Investigate the Reciprocal Effects Model of Self-Concept and Achievement

The relationship between students’ subject-specific academic self-concept and their academic achievement is one of the most widely researched topics in educational psychology. A large proportion of this research has considered cross-lagged panel models (CLPMs), oftentimes synonymously referred to as reciprocal effects models (REMs), as the gold standard for investigating the causal relationships between the two variables and has reported evidence of a reciprocal relationship between self-concept and achievement. However, more recent methodological research has questioned the plausibility of assumptions that need to be satisfied in order to interpret results from traditional CLPMs causally. In this substantive-methodological synergy, we aimed to contrast traditional and more recently developed methods to investigate reciprocal effects of students’ academic self-concept and achievement. Specifically, we compared results from CLPMs, full-forward CLPMs (FF-CLPMs), and random intercept CLPMs (RI-CLPMs) with two weighting approaches developed to study causal effects of continuous treatment variables. To estimate these different models, we used rich longitudinal data of N = 3757 students from lower secondary schools in Germany. Results from CLPMs, FF-CLPMs, and weighting methods supported the reciprocal effects model, particularly when math self-concept and grades were considered. Results from the RI-CLPMs were less consistent. Implications from our study for the interpretation of effects from the different models and methods as well as for school motivation theory are discussed.

Should multiple imputation be stratified by exposure group when estimating causal effects via outcome regression in observational studies?

BACKGROUND

Despite recent advances in causal inference methods, outcome regression remains the most widely used approach for estimating causal effects in epidemiological studies with a single-point exposure and outcome. Missing data are common in these studies, and complete-case analysis (CCA) and multiple imputation (MI) are two frequently used methods for handling them. In randomised controlled trials (RCTs), it has been shown that MI should be conducted separately by treatment group. In observational studies, causal inference is now understood as the task of emulating an RCT, which raises the question of whether MI should be conducted by exposure group in such studies.

METHODS

We addressed this question by evaluating the performance of seven methods for handling missing data when estimating causal effects with outcome regression. We conducted an extensive simulation study based on an illustrative case study from the Victorian Adolescent Health Cohort Study, assessing a range of scenarios, including seven outcome generation models with exposure-confounder interactions of differing strength.

RESULTS

The simulation results showed that MI by exposure group led to the least bias when the size of the smallest exposure group was relatively large, followed by MI approaches that included the exposure-confounder interactions.

CONCLUSIONS

The findings from our simulation study, which was designed based on a real case study, suggest that current practice for the conduct of MI in causal inference may need to shift to stratifying by exposure group where feasible, or otherwise including exposure-confounder interactions in the imputation model.

Guideline-Directed Medical Therapy and Long-Term Mortality and Amputation Outcomes in Patients Undergoing Peripheral Vascular Interventions

BACKGROUND

Lack of guideline-directed medical therapy (GDMT) in patients undergoing peripheral vascular interventions (PVIs) may increase mortality and amputation risk.

OBJECTIVES

The authors sought to study the association between GDMT and mortality/amputation and to examine GDMT variability among providers and health systems.

METHODS

We performed an observational study using patients in the Vascular Quality Initiative registry undergoing PVI between 2017 and 2018. Two-year all-cause mortality and major amputation data were derived from Medicare claims data. Compliance with GDMT was defined as receiving a statin, antiplatelet therapy, and angiotensin-converting enzyme inhibitor/angiotensin receptor blocker if hypertensive. Propensity 1:1 matching was applied for GDMT vs no GDMT and survival analyses were performed to compare outcomes between groups.

RESULTS

Of 15,891 patients undergoing PVIs, 48.8% received GDMT and 6,120 patients in each group were matched. Median follow-up was 9.6 (IQR: 4.5-16.2) months for mortality and 8.4 (IQR: 3.5-15.4) for amputation. Mean age was 72.0 ± 9.9 years. Mortality risk was higher among patients who did not receive GDMT versus those on GDMT (31.2% vs 24.5%; HR: 1.37, 95% CI: 1.25-1.50; P < 0.001), as well as, risk of amputation (16.0% vs 13.2%; HR: 1.20; 95% CI: 1.08-1.35; P < 0.001). GDMT rates across sites and providers ranging from 0% to 100%, with lower performance translating into higher risk.

CONCLUSIONS

Almost one-half of the patients receiving PVI in this national quality registry were not on GDMT, and this was associated with increased risk of mortality and major amputation. Quality improvement efforts in vascular care should focus on GDMT in patients undergoing PVI.

Effectiveness of active occupational therapy in patients with acute stroke: A propensity score-weighted retrospective study

Background and purpose

The effects of therapy and patient characteristics on rehabilitation outcomes in patients with acute stroke are unclear. We investigated the effects of intensive occupational therapy (OT) on patients with acute stroke.

Methods

We performed a retrospective cohort study using the 2005-2016 Japan Rehabilitation Database, from which we identified patients with stroke (n = 10,270) who were admitted to acute care hospitals (n = 37). We defined active OT (AOT) and non-AOT as OT intervention times (total intervention time/length of hospital stay) longer or shorter than the daily physical therapy intervention time, respectively. The outcomes assessed were the Functional Independence Measure (FIM) and National Institutes of Health Stroke Scale (NIHSS) scores, duration of hospitalization, and rate of discharge. Propensity scores and inverse probability of treatment weighting analyses adjusted for patient characteristics were performed to investigate the effects of AOT on patient outcomes.

Results

We enrolled 3,501 patients (1,938 and 1,563 patients in the AOT and non-AOT groups, respectively) in the study. After inverse probability of treatment weighting, the AOT group had a shorter length of hospitalization (95% confidence interval: -3.7, -1.3, p < 0.001), and the FIM (95% confidence interval: 2.0, 5.7, p < 0.001) and NIHSS (95% confidence interval; 0.3, 1.1, p < 0.001) scores improved significantly. Subgroup analysis showed that lower NHISS scores for aphasia, gaze, and neglect and lower overall NIHSS and FIM scores on admission led to a greater increase in FIM scores in the AOT group.

Conclusions

AOT improved the limitations in performing activities of daily living (ADL) and physical function in patients with acute stroke and reduced the length of hospitalization. Additionally, subgroup analysis suggested that the increase in FIM score was greater in patients with severe limitations in performing ADLs and worse cognitive impairment, such as neglect, on admission.

Cortisol and periodontitis: Prospective observational and Mendelian randomization studies

PURPOSE

Cortisol has obesogenic, hyperglycemic and immunomodulating effects. Preclinical and observational research suggested that it is associated with periodontitis but the evidence for potential causality in humans is sparse. We triangulated results from prospective observational and Mendelian randomization (MR) analyses to further explore this.

METHODS

Using pooled data from 3,388 participants of two population cohort studies embedded in the Study of Health in Pomerania (SHIP) project, we associated serum cortisol levels with periodontal outcomes measured after a median follow-up time of 6.9 years, adjusting for confounding and selection bias using propensity score weighting and multiple imputation. We further examined the effect of genetically proxied plasma morning cortisol levels on periodontitis using two-sample MR of 17,353 cases and 28,210 controls.

RESULTS

In SHIP, we found that cortisol levels were positively associated with follow-up levels of mean clinical attachment level (CAL), deep interdental CAL and bleeding on probing but were unrelated to mean probing pocket depth and deep periodontal pockets. In MR analysis, cortisol was not associated with periodontitis.

CONCLUSION

The observational study revealed a prospective association of spot cortisol with makers of periodontitis. Contrary to observational studies, genetically instrumented, long-term cortisol was unrelated to periodontitis. Our results find no univocal evidence that cortisol plays a role in periodontitis pathology, casting doubt on cortisol-related pathways.

Do Patients With Intersectional Disadvantage Have Poorer Outcomes From Osteoarthritis Management Programs? A Tapered Balancing Study of Patient Outcomes From the Good Life With Osteoarthritis in Denmark Program

OBJECTIVE

To investigate whether adults with potential multiple social disadvantage have poorer outcomes following attendance in an osteoarthritis (OA) management program (OAMP), and if so, what might determine this result.

METHODS

Among consecutive knee OA attendees of the Good Life With Osteoarthritis in Denmark (GLA:D) OAMP in Denmark we defined a group with potential "intersectional disadvantage" based on self-reported educational attainment, country of birth, and citizenship. Outcomes of this group were compared with GLA:D participants who were native Danish citizens with higher educational attainment. Outcomes were pain intensity, Knee Injury and Osteoarthritis Outcome Score (KOOS) quality of life subscale score, and the EuroQol 5-domain instrument in 5 levels (EQ-5D-5L) score at 3 and 12 months. After data preprocessing, we used entropy balancing to sequentially control for differences between the groups in baseline covariates. Mean between-group differences in outcomes were estimated by weighted linear regression.

RESULTS

Of 18,448 eligible participants, 250 (1.4%) were nonnative/foreign citizens with lower education. After balancing for differences in baseline score and in administrative and demographic characteristics, they had poorer outcomes than higher-educated native Danish citizens on pain intensity and EQ-5D-5L score at both follow-up points (e.g., between-group mean differences in pain visual analog scale [0-100] at 3 and 12 months: 3.4 [95% confidence interval (95% CI) -0.5, 7.3] and 6.2 [95% CI 1.7, 10.7], respectively). Differences in KOOS quality of life subscale score, were smaller or absent. Balancing for differences on baseline score, comorbidity, self-efficacy, and depression had the greatest effect on reducing observed outcome inequalities.

CONCLUSION

Outcome inequalities widened following OAMP attendance, particularly at longer-term follow-up, but the magnitude of differences was generally modest and inconsistent across outcome measures. Tailoring content to reduce outcome inequalities may be indicated, but improving access appears the greater priority.

A meta-analytic framework to adjust for bias in external control studies

While randomized controlled trials (RCTs) are the gold standard for estimating treatment effects in medical research, there is increasing use of and interest in using real-world data for drug development. One such use case is the construction of external control arms for evaluation of efficacy in single-arm trials, particularly in cases where randomization is either infeasible or unethical. However, it is well known that treated patients in non-randomized studies may not be comparable to control patients-on either measured or unmeasured variables-and that the underlying population differences between the two groups may result in biased treatment effect estimates as well as increased variability in estimation. To address these challenges for analyses of time-to-event outcomes, we developed a meta-analytic framework that uses historical reference studies to adjust a log hazard ratio estimate in a new external control study for its additional bias and variability. The set of historical studies is formed by constructing external control arms for historical RCTs, and a meta-analysis compares the trial controls to the external control arms. Importantly, a prospective external control study can be performed independently of the meta-analysis using standard causal inference techniques for observational data. We illustrate our approach with a simulation study and an empirical example based on reference studies for advanced non-small cell lung cancer. In our empirical analysis, external control patients had lower survival than trial controls (hazard ratio: 0.907), but our methodology is able to correct for this bias. An implementation of our approach is available in the R package ecmeta.

Econometric Issues in Prospective Economic Evaluations Alongside Clinical Trials: Combining the Nonparametric Bootstrap With Methods That Address Missing Data

Prospective economic evaluations conducted alongside clinical trials have become an increasingly popular approach in evaluating the cost-effectiveness of a public health initiative or treatment intervention. These types of economic studies provide improved internal validity and accuracy of cost and effectiveness estimates of health interventions and, compared with simulation or decision-analytic models, have the advantage of jointly observing health and economics outcomes of trial participants. However, missing data due to incomplete response or patient attrition, and sampling uncertainty are common concerns in econometric analysis of clinical trials. Missing data are a particular problem for comparative effectiveness trials of substance use disorder interventions. Multiple imputation and inverse probability weighting are 2 widely recommended methods to address missing data bias, and the nonparametric bootstrap is recommended to address uncertainty in predicted mean cost and effectiveness between trial interventions. Although these methods have been studied extensively by themselves, little is known about how to appropriately combine them and about the potential pitfalls and advantages of different approaches. We provide a review of statistical methods used in 29 economic evaluations of substance use disorder intervention identified from 4 published systematic reviews and a targeted search of the literature. We evaluate how each study addressed missing data bias, whether the recommended nonparametric bootstrap was used, how these 2 methods were combined, and conclude with recommendations for future research.

Transporting observational study results to a target population of interest using inverse odds of participation weighting

Inverse odds of participation weighting (IOPW) has been proposed to transport clinical trial findings to target populations of interest when the distribution of treatment effect modifiers differs between trial and target populations. We set out to apply IOPW to transport results from an observational study to a target population of interest. We demonstrated the feasibility of this idea with a real-world example using a nationwide electronic health record derived de-identified database from Flatiron Health. First, we conducted an observational study that carefully adjusted for confounding to estimate the treatment effect of fulvestrant plus palbociclib relative to letrozole plus palbociclib as a second-line therapy among estrogen receptor (ER)-positive, human epidermal growth factor receptor (HER2)-negative metastatic breast cancer patients. Second, we transported these findings to the broader cohort of patients who were eligible for a first-line therapy. The interpretation of the findings and validity of such studies, however, rely on the extent that causal inference assumptions are met.