Standardizing to specific target populations in distributed networks and multisite pharmacoepidemiologic studies.

Distributed network studies and multisite studies assess drug safety and effectiveness in diverse populations by pooling information. Targeting groups of clinical or policy interest (including specific sites or site combinations) and applying weights based on effect measure modifiers (EMMs) prior to pooling estimates within multisite studies may increase interpretability and improve precision. We simulated a 4-site study, standardized each site using inverse odds weights (IOWs) to resemble the 3 smallest sites or the smallest site, estimated IOW-weighted risk differences (RDs), and combined estimates with inverse variance weights (IVWs). We also created an artificial distributed network in the Clinical Practice Research Datalink (CPRD) Aurum consisting of 1 site for each geographic region. We compared metformin and sulfonylurea initiators with respect to mortality, targeting the smallest region. In the simulation, IOWs reduced differences between estimates and increased precision when targeting the 3 smallest sites or the smallest site. In the CPRD Aurum study, the IOW + IVW estimate was also more precise (smallest region: RD = 5.41% [95% CI, 1.03-9.79]; IOW + IVW estimate: RD = 3.25% [95% CI, 3.07-3.43]). When performing pharmacoepidemiologic research in distributed networks or multisite studies in the presence of EMMs, designation of target populations has the potential to improve estimate precision and interpretability. This article is part of a Special Collection on Pharmacoepidemiology.

Variable Selection When Estimating Effects in External Target Populations.

External validity is an important part of epidemiologic research. To validly estimate effects in specific external target populations using a chosen effect measure (i.e., "transport"), some methods require that one account for all effect measure modifiers [EMMs]. However, little is known about how including other variables that are not EMMs (i.e., non-EMMs) in adjustment sets impacts estimates. Using simulations, we evaluated how inclusion of non-EMMs affected estimation of the transported risk difference (RD) by assessing impacts of covariates that A) differ (or not) between the trial and the target, B) are associated with the outcome (or not), and C) modify the RD (or not). We assessed variation and bias when covariates with each possible combination of these factors were used to transport RDs using outcome modeling or inverse odds weighting. Including variables that differed in distribution between the populations but were non-EMMs reduced precision, regardless of whether they were associated with the outcome. However, non-EMMs associated with selection did not amplify bias resulting from omitting necessary EMMs. Including all variables associated with the outcome may result in unnecessarily imprecise estimates when estimating treatment effects in external target populations.

Glucagon like peptide-1 receptor agonists and the risk of skin cancer among patients with type 2 diabetes: Population-based cohort study.

AIMS: The objective of this study was to determine whether the use of glucagon-like peptide-1 receptor agonists (GLP-1 RAs) is associated with an increased risk of melanoma and nonmelanoma skin cancer, separately, compared with the use of sulfonylureas among patients with type 2 diabetes. METHODS: Using the United Kingdom Clinical Practice Research Datalink (2007-2019), we assembled two new-user active comparator cohorts. In the first cohort assessing melanoma as the outcome, 11,786 new users of GLP-1 RAs were compared with 208,519 new users of sulfonylureas. In the second cohort assessing nonmelanoma skin cancer as the outcome, 11,774 new users of GLP-1 RAs were compared with 207,788 new users of sulfonylureas. Cox proportional hazards models weighted using propensity score fine stratification were fit to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) of melanoma and nonmelanoma skin cancer, respectively. RESULTS: Compared with sulfonylureas, GLP-1 RAs were not associated with an increased risk of either melanoma (42.6 vs. 43.9 per 100,000 person-years, respectively; HR 0.96, 95% CI 0.53-1.75) or nonmelanoma skin cancer (243.9 vs. 229.9 per 100,000 person-years, respectively; HR 1.03, 95% CI 0.80-1.33). There was no evidence of an association between cumulative duration of use with either melanoma or nonmelanoma skin cancer. Consistent results were observed in secondary and sensitivity analyses. CONCLUSIONS: In this population-based cohort study, GLP-1 RAs were not associated with an increased risk of melanoma or nonmelanoma skin cancer, compared with sulfonylureas.

Fluoroquinolones and the risk of severe hypoglycaemia among sulphonylurea users: Population-based cohort study.

AIM: Fluoroquinolone-related hypoglycaemia is rare but may become clinically relevant in individuals at high baseline hypoglycaemic risk, such as patients with diabetes using sulphonylureas. Our population-based cohort study assessed whether fluoroquinolones are associated with an increased risk of severe hypoglycaemia compared with amoxicillin among patients treated with sulphonylureas. MATERIALS AND METHODS: Using the UK's Clinical Practice Research Datalink Aurum linked to hospitalization and vital statistics data, we assembled a base cohort of patients who initiated second-generation sulphonylureas (1998-2020). The study cohort included patients initiating either fluoroquinolones or amoxicillin while on sulphonylureas. Using an intent-to-treat exposure definition, we assessed the 30-day risk of severe hypoglycaemia (hospitalization with or death because of hypoglycaemia) associated with fluoroquinolones compared with amoxicillin. Cox models estimated hazard ratios (HRs) with 95% confidence intervals (CIs) of severe hypoglycaemia after 1:5 matching on previous sulphonylurea use and propensity scores. Secondary analyses were stratified by demographics and glycated haemoglobin. RESULTS: Overall, 143 417 patients initiated fluoroquinolones (n = 13 123) or amoxicillin (n = 130 294) while on sulphonylureas. Compared with amoxicillin, fluoroquinolones were not associated with the risk of severe hypoglycaemia (HR, 1.17; 95% CI, 0.91-1.50). Fluoroquinolones were associated with an increased risk in patients <65 years (HR, 2.90; 95% CI, 1.41-5.97) but not in those ≥65 years (HR, 1.03; 95% CI, 0.79-1.35) in stratified analyses. There was no evidence of effect modification by sex or glycated haemoglobin. CONCLUSIONS: In patients using second-generation sulphonylureas, fluoroquinolones were not associated with an increased risk of severe hypoglycaemia compared with amoxicillin. An increased risk among younger adults is possible.

Structured learning in time-dependent Cox models.

Cox models with time-dependent coefficients and covariates are widely used in survival analysis. In high-dimensional settings, sparse regularization techniques are employed for variable selection, but existing methods for time-dependent Cox models lack flexibility in enforcing specific sparsity patterns (ie, covariate structures). We propose a flexible framework for variable selection in time-dependent Cox models, accommodating complex selection rules. Our method can adapt to arbitrary grouping structures, including interaction selection, temporal, spatial, tree, and directed acyclic graph structures. It achieves accurate estimation with low false alarm rates. We develop the sox package, implementing a network flow algorithm for efficiently solving models with complex covariate structures. sox offers a user-friendly interface for specifying grouping structures and delivers fast computation. Through examples, including a case study on identifying predictors of time to all-cause death in atrial fibrillation patients, we demonstrate the practical application of our method with specific selection rules.

Type of infertility and prevalence of congenital malformations.

BACKGROUND: Children conceived with assisted reproductive technologies (ART) or after a long waiting time have a higher prevalence of congenital malformations, but few studies have examined the contribution of type of infertility. OBJECTIVES: To quantify the association between causes of infertility and prevalence of malformations. METHODS: We compared the prevalence at birth of all and severe malformations diagnosed up to age 2 between 6656 children born in 1996-2017 to parents who had previously been assessed for infertility a an academic fertility clinic ("exposed") and 10,382 children born in the same period to parents with no recent medical history of infertility ("reference"). We estimated prevalence ratios (PR) and prevalence differences (PD), by infertility status, type of treatment (non-ART, ART), and infertility diagnosis, in all children and among singletons. RESULTS: Compared with children of parents with no infertility, children of parents with infertility had a higher prevalence of malformations (both definitions), particularly following ART conceptions. After accounting for treatment, ovulatory disorders were associated with a higher prevalence of both all (PR 1.49, 95% confidence interval (CI) 1.15, 1.93; PD 3.8, 95% CI 1.0, 6.6) and severe (PR 1.53, 95% CI 1.02, 2.29; PD 1.8, 95% CI -0.2, 3.7) malformations (the estimates refer to exposed children conceived without treatment). Unexplained and male factor infertility were associated with all and severe malformations, respectively. Estimates among singletons were similar. A diagnosis of ovulatory disorders was associated with all malformations also in analyses restricted to exposed children, regardless of treatment (we did not examine severe malformations, due to limited power). CONCLUSIONS: In this study, ovulatory disorders were consistently associated with a higher prevalence of congenital malformations (including severe malformations) among live births, regardless of mode of conception.

Prescribing trends of proton pump inhibitors and histamine blockers among children in the United Kingdom (1998-2019): A population-based assessment.

PURPOSE: To describe the prescribing trends of proton pump inhibitors (PPIs) and H2 receptor antagonists (H2 RAs) among children with gastroesophageal reflux in the United Kingdom between 1998 and 2019. METHODS: We conducted a population-based retrospective cohort study using data from the Clinical Practice Research Datalink that included all children aged ≤18 years with a first ever diagnosis of gastroesophageal reflux between 1998 and 2019. Using negative binomial regression, we estimated crude and adjusted annual prescription rates per 1000 person-years and corresponding 95% confidence intervals (CIs) for PPIs and H2 RAs. We also assessed rate ratios of PPIs and H2 RAs prescription rates to examine changes in prescribing over time. RESULTS: Our cohort included 177 477 children with a first ever diagnosis of gastroesophageal reflux during the study period. The median age was 13 years (IQR: 1, 17) among children prescribed PPIs and 0.2 years (IQR: 0.1, 0.6) among those prescribed H2 RAs. The total prescription rate of all GERD drugs was 1468 prescriptions per 1000 person-years (PYs) (95% CI 1463-1472). Overall, PPIs had a higher prescription rate (815 per 1000 PYs, 95% CI 812-818) than H2 RAs (653 per 1000 PYs 95% CI 650-655). Sex- and age-adjusted rate ratios of 2019 versus 1998 demonstrated a 10% increase and a 76% decrease in the prescription rates of PPIs and H2 RAs, respectively. CONCLUSIONS: Prescription rates for PPIs increased, especially during the first half of the study period, while prescription rates for H2 RA decreased over time.

Evaluating hybrid controls methodology in early-phase oncology trials: A simulation study based on the MORPHEUS-UC trial.

Phase Ib/II oncology trials, despite their small sample sizes, aim to provide information for optimal internal company decision-making concerning novel drug development. Hybrid controls (a combination of the current control arm and controls from one or more sources of historical trial data [HTD]) can be used to increase statistical precision. Here we assess combining two sources of Roche HTD to construct a hybrid control in targeted therapy for decision-making via an extensive simulation study. Our simulations are based on the real data of one of the experimental arms and the control arm of the MORPHEUS-UC Phase Ib/II study and two Roche HTD for atezolizumab monotherapy. We consider potential complications such as model misspecification, unmeasured confounding, different sample sizes of current treatment groups, and heterogeneity among the three trials. We evaluate two frequentist methods (with both Cox and Weibull accelerated failure time [AFT] models) and three different commensurate priors in Bayesian dynamic borrowing (with a Weibull AFT model), and modifications within each of those, when estimating the effect of treatment on survival outcomes and measures of effect such as marginal hazard ratios. We assess the performance of these methods in different settings and the potential of generalizations to supplement decisions in early-phase oncology trials. The results show that the proposed joint frequentist methods and noninformative priors within Bayesian dynamic borrowing with no adjustment on covariates are preferred, especially when treatment effects across the three trials are heterogeneous. For generalization of hybrid control methods in such settings, we recommend more simulation studies.

Invited Commentary: Aligning Methodological Research on Pregnancy Weight Gain With the Questions That Matter Most for Public Health Guidelines.

The inherent correlation between the total amount of weight gained in pregnancy and the duration of pregnancy creates major methodological challenges in the study of pregnancy weight gain. In this issue (Am J Epidemiol. 2022;191(10):1687-1699), Richards et al. examine the extent to which different measures of pregnancy weight gain (including covariate adjustment for gestational age and standardizing weight gain for gestational duration using a pregnancy weight gain chart) are able to disentangle the effects of low weight gain on perinatal health from the role of younger gestational age at delivery for 3 outcomes: small-for-gestational-age birth, cesarean delivery, and low birth weight. While methodological research to understand how to best disentangle the effects of gestational weight gain from pregnancy duration is valuable, we argue that the practical utility of this type of research would be increased by aligning the specific research questions more closely with health outcomes on which evidence is most needed-those not considered in current weight gain guidelines due to lack of high-quality evidence (such as pre-eclampsia and stillbirth). Further, evaluations of weight gain charts should separate out the potential for bias introduced by the use of a normative chart per se from the use of a chart unsuitable for the study population.

Use of a Statistical Adaptive Treatment Strategy Approach for Emulating Randomized Controlled Trials Using Observational Data: The Example of Blood-Pressure Control Strategies for the Prevention of Cardiovascular Events Among Individuals With Hypertension at High Cardiovascular Risk.

Statistical approaches to adaptive treatment strategies (ATS) can be used to mimic the sequential decision-making inherently found in clinical practice. To illustrate the use of a statistical ATS approach, we emulated a target trial of different blood pressure (BP) control plans for the prevention of cardiovascular events among individuals with hypertension at high cardiovascular risk, inspired by the Systolic Blood Pressure Intervention Trial (SPRINT). We included 103,708 patients with hypertension and a "QRISK3" estimated 10-year risk of cardiovascular disease of ≥20% who initiated an antihypertensive drug between 1998 and 2018. Dynamic marginal structural models estimated the comparative effects of treating patients with intensive (target BP: 130/80 mm Hg), standard (140/90 mm Hg), and conservative (150/90 mm Hg) BP control strategies. The adjusted hazard ratios (HRs) for the intensive versus standard strategy were 0.96 (95% confidence interval (CI): 0.92, 1.00) for major adverse cardiovascular events and 0.93 (95% CI: 0.88, 0.97) for death from cardiovascular causes. For the conservative versus standard strategy, they were 1.06 (95% CI: 1.02, 1.10) and 1.08 (95% CI: 1.03, 1.13), respectively. These results are largely compatible with SPRINT. ATS can be used to emulate randomized controlled trials of complex treatment strategies in an observational setting and represents an alternative approach for situations where randomized controlled trials are not feasible.

Investigating the Shape and Strength of the Relationship Between Maternal Weight Gain and Gestational Age at Delivery in Twin and Singleton Pregnancies.

Both inadequate and excessive maternal weight gain are correlated with preterm delivery in singleton pregnancies, yet this relationship has not been adequately studied in twins. We investigated the relationship between time-varying maternal weight gain and gestational age at delivery in twin pregnancies and compared it with that in singletons delivered in the same study population. We used serial weight measurements abstracted from charts for twin and singleton pregnancies delivered during 1998-2013 in Pittsburgh, Pennsylvania. Our exposure was time-varying weight gain z score, calculated using gestational age-standardized and prepregnancy body mass index-stratified twin- and singleton-specific charts, and our outcome was gestational age at delivery. Our analyses used a flexible extension of the Cox proportional hazards model that allowed for nonlinear and time-dependent effects. We found a U-shaped relationship between weight gain z score and gestational age at delivery among twin pregnancies (lowest hazard of delivery observed at z score = 1.2), which we attributed to increased hazard of early preterm spontaneous delivery among pregnancies with low weight gain and increased hazard of late preterm delivery without labor among pregnancies with high weight gain. Our findings may be useful for updating provisional guidelines for maternal weight gain in twin pregnancies.

Causal inference for recurrent event data using pseudo-observations.

Recurrent event data are commonly encountered in observational studies where each subject may experience a particular event repeatedly over time. In this article, we aim to compare cumulative rate functions (CRFs) of two groups when treatment assignment may depend on the unbalanced distribution of confounders. Several estimators based on pseudo-observations are proposed to adjust for the confounding effects, namely inverse probability of treatment weighting estimator, regression model-based estimators, and doubly robust estimators. The proposed marginal regression estimator and doubly robust estimators based on pseudo-observations are shown to be consistent and asymptotically normal. A bootstrap approach is proposed for the variance estimation of the proposed estimators. Model diagnostic plots of residuals are presented to assess the goodness-of-fit for the proposed regression models. A family of adjusted two-sample pseudo-score tests is proposed to compare two CRFs. Simulation studies are conducted to assess finite sample performance of the proposed method. The proposed technique is demonstrated through an application to a hospital readmission data set.

Use of metformin and insulin among pregnant women with gestation diabetes in the United Kingdom: A population-based cohort study.

AIMS: The contemporary prescription patterns of antidiabetic drugs following guideline changes recommending metformin as first-line gestational diabetes (GDM) pharmacotherapy is underexplored. We aimed to examined use of metformin and insulin during pregnancy among women with GDM over 20 years in the United Kingdom. METHODS: We conducted a population-based cohort study using linked data from the Clinical Practice Research Datalink, its pregnancy register and Hospital Episode Statistics from 1998 to 2017. We included pregnancies of women without prior diabetes history who received GDM diagnosis or initiated an antidiabetic drug after 20 weeks gestation. Patient-level and practice-level characteristics were compared between metformin initiators and insulin initiators. We described trends of initiating metformin as first-line treatment and described time to initiation of rescue insulin overall, and by body mass index among metformin initiators. RESULTS: Our cohort included 5633 pregnancies from 5393 women with GDM, of whom 38.9% initiated pharmacotherapy (41% insulin, 59% metformin). Metformin prescriptions (as opposed to insulin) increased substantially, from <5% of pregnancies before 2007 to 42.5% in 2008. Over 85% of pregnancies that were prescribed pharmacotherapy were prescribed metformin as first-line treatment in 2015. Among metformin initiators, 16% initiated rescue insulin, typically occurring within 40 days of metformin initiation. Choice of GDM pharmacotherapy varied by characteristics, including smoking, obesity, race/ethnicity and general practice regions. CONCLUSIONS: Metformin was the most prescribed medication for GDM, with large increases over the past 2 decades. The increasing use of oral-antidiabetic drugs during pregnancy, consistent with other regions, highlights the need for future studies examining effectiveness and safety of antidiabetic drug use during pregnancy.

Risk of recurrent severe maternal morbidity: a population-based study.

BACKGROUND: Severe maternal morbidity is a composite indicator of maternal health and obstetrical care. Little is known about the risk of recurrent severe maternal morbidity in a subsequent delivery. OBJECTIVE: This study aimed to estimate the risk of recurrent severe maternal morbidity in the next delivery after a complicated first delivery. STUDY DESIGN: We analyzed a population-based cohort study of women with at least 2 singleton hospital deliveries between 1989 and 2021 in Quebec, Canada. The exposure was severe maternal morbidity in the first hospital-recorded delivery. The study outcome was severe maternal morbidity at the second delivery. Log-binomial regression models adjusted for maternal and pregnancy characteristics were used to generate relative risks and 95% confidence intervals comparing women with and without severe maternal morbidity at first delivery. RESULTS: Among 819,375 women, 43,501 (3.2%) experienced severe maternal morbidity in the first delivery. The rate of severe maternal morbidity recurrence at second delivery was 65.2 vs 20.3 per 1000 in women with and without previous severe maternal morbidity (adjusted relative risk, 3.11; 95% confidence interval, 2.96-3.27). The adjusted relative risk for recurrence of severe maternal morbidity was greatest among women who had ≥3 different types of severe maternal morbidity at their first delivery, relative to those with none (adjusted relative risk, 5.50; 95% confidence interval, 4.26-7.10). Women with cardiac complication at first delivery had the highest risk of severe maternal morbidity in the next delivery. CONCLUSION: Women who experience severe maternal morbidity have a relatively high risk of recurrent morbidity in the subsequent pregnancy. In women with severe maternal morbidity, these study findings have implications for prepregnancy counseling and maternity care in the next pregnancy.

Long-acting insulin analogues and the risk of diabetic retinopathy among patients with type 2 diabetes: A population-based cohort study.

AIM: To determine whether the use of long-acting insulin analogues is associated with an increased risk of incident diabetic retinopathy (DR) among patients with type 2 diabetes. METHODS: Using data from the Clinical Practice Research Datalink Aurum, this retrospective, population-based cohort study included patients with type 2 diabetes who initiated a long-acting insulin analogue (glargine, detemir, degludec) or Neutral Protamine Hagedorn (NPH) insulin. The primary outcome was incident DR. We used Cox proportional hazards models with inverse probability of treatment weighting to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for incident DR with insulin analogues versus NPH insulin. RESULTS: There were 66 280 new users of long-acting insulin analogues and 66 173 new users of NPH insulin. The incidence rate of DR was 101.7 per 1000 person-years (95% CI, 98.7-104.8) for insulin analogues and 93.2 (95% CI, 90.0-96.5) per 1000 person-years for NPH insulin. Compared with the current use of NPH insulin, insulin analogues were not associated with the risk of incident DR (HR 1.04, 95% CI, 0.99-1.09). The adjusted HRs were 0.84 (95% CI, 0.66-1.07) for proliferative DR and 1.02 (95% CI, 0.97-1.08) for non-proliferative DR. CONCLUSIONS: Compared with NPH insulin, long-acting insulin analogues were not associated with the risk of incident DR among patients with type 2 diabetes. This finding provides important reassurance regarding the safety of long-acting insulin analogues with respect to incident DR.

Levothyroxine initiation and the risk of pregnancy loss among pregnant women with subclinical hypothyroidism: An observational study emulating a target trial.

BACKGROUND: While the benefits of levothyroxine are well-established for overt hypothyroidism, they are unclear for subclinical hypothyroidism (SCH) among pregnant women. OBJECTIVE: To estimate the effect of initiation of levothyroxine on pregnancy loss among women with SCH with an emulated target trial using observational data. METHODS: We emulated a target trial using the United Kingdom's Clinical Practice Research Datalink to account for the staggered timing of diagnosis and treatment of SCH and the time of entry of women into prenatal care. We emulated multiple nested trials (at each gestational week) and used an intention-to-treat approach to define levothyroxine use (≥1 prescription in the 7 days prior to trial entry), with eligible users matched to non-users (1:4) on time of diagnosis, gestational week of the first eligible trial and high-dimensional propensity score. Pregnancy losses included spontaneous abortion and stillbirth. A pooled logistic regression model with bootstrap resampling was used to estimate the hazard ratios (HR) and 95% confidence intervals (CI). RESULTS: Based on 159,177 eligible person-trials (5781 women), the matched cohort included 181 initiators and 640 non-initiators of levothyroxine, with 57 pregnancy losses occurring during follow-up. Overall, the mean age of women was 32.2 years (SD 5.4), 25% were obese, 8% had type 2 diabetes and about 50% were nulliparous. After matching, women who initiated levothyroxine versus not had higher thyroid-stimulating levels during pregnancy and were more likely to have a history of hypothyroidism. The cumulative incidence of pregnancy loss was lower in initiators versus non-initiators of levothyroxine. The adjusted HR for pregnancy loss was 0.87 (95% CI 0.22, 1.56). CONCLUSIONS: Although our assessment of the effect of initiation of levothyroxine for SCH in pregnancy precludes any definitive conclusions due to wide confidence intervals, this study illustrates the feasibility of using the target trial emulation framework to examine the effectiveness of medication use in pregnancy.

Hydrochlorothiazide use and risk of keratinocyte carcinoma and melanoma: A multisite population-based cohort study.

BACKGROUND: The association between hydrochlorothiazide (HCTZ) and skin cancer remains controversial. OBJECTIVE: To determine whether HCTZ is associated with an increased risk of skin cancer compared with angiotensin-converting enzyme inhibitors and calcium channel blockers. METHODS: Two new-user, active comparator cohorts were assembled using 6 Canadian databases. Site-specific hazard ratios (HRs) with 95% CIs were estimated using standardized morbidity ratio weighted Cox proportional hazard models and pooled using random-effects meta-analysis. RESULTS: HCTZ was not associated with an overall increased risk of keratinocyte carcinoma compared with angiotensin-converting enzyme inhibitors or calcium channel blockers, although increased risks were observed with longer durations (≥10 years; HR: 1.12; 95% CI: 1.03-1.21) and higher cumulative doses (≥100,000 mg; HR: 1.49; 95% CI: 1.27-1.76). For melanoma, there was no association with angiotensin-converting enzyme inhibitors, but a 32% increased risk with calcium channel blockers (crude incidence rates: 64.2 vs 58.4 per 100,000 person-years; HR: 1.32; 95% CI: 1.19-1.46; estimated number needed to harm at 5 years of follow-up: 1627 patients), with increased risks with longer durations and cumulative doses. LIMITATIONS: Residual confounding due to the observational design. CONCLUSIONS: Increased risks of keratinocyte carcinoma and melanoma were observed with longer durations of use and higher cumulative doses of HCTZ.

Meta-analysis with sample-standardization in multi-site studies.

PURPOSE: To conceptualize a particular target population and estimand for multi-site pharmacoepidemiologic studies within data networks and to analytically examine sample-standardization as a meta-analytic method compared with inverse-variance weighted meta-analyses. METHODS: The target population of interest is all and only all individuals from the data-contributing sites. Standardization, a general conditioning technique frequently employed for confounding control, was adopted to estimate the network-wide causal treatment effect. Specifically, the proposed sample-standardization yields a meta-analysis estimator, that is, a weighted summation of site-specific results, where the weight for a site is the proportion of its size in the entire network. This sample-standardization estimator was evaluated analytically in comparison to estimators from inverse-variance weighted fixed-effect and random-effects meta-analyses in terms of statistical consistency. RESULTS: A proof is reported to justify the consistency of the sample-standardization estimator with and without treatment effect heterogeneity by site. Both inverse-variance weighted fixed-effect and random-effects meta-analyses were found to generally result in inconsistent estimators in the presence of treatment effect heterogeneity by site for this particular target population and estimand. CONCLUSIONS: Sample-standardization is a valid approach to generate causal inference in multi-site studies when the target population comprises all and only all individuals within the network, even in the presence of heterogeneity of treatment effect by site. Multi-site studies should clearly specify the target population and estimand to help select the most appropriate meta-analytic methods.

High-dimensional propensity scores for empirical covariate selection in secondary database studies: Planning, implementation, and reporting.

Real-world evidence used for regulatory, payer, and clinical decision-making requires principled epidemiology in design and analysis, applying methods to minimize confounding given the lack of randomization. One technique to deal with potential confounding is propensity score (PS) analysis, which allows for the adjustment for measured preexposure covariates. Since its first publication in 2009, the high-dimensional propensity score (hdPS) method has emerged as an approach that extends traditional PS covariate selection to include large numbers of covariates that may reduce confounding bias in the analysis of healthcare databases. hdPS is an automated, data-driven analytic approach for covariate selection that empirically identifies preexposure variables and proxies to include in the PS model. This article provides an overview of the hdPS approach and recommendations on the planning, implementation, and reporting of hdPS used for causal treatment-effect estimations in longitudinal healthcare databases. We supply a checklist with key considerations as a supportive decision tool to aid investigators in the implementation and transparent reporting of hdPS techniques, and to aid decision-makers unfamiliar with hdPS in the understanding and interpretation of studies employing this approach. This article is endorsed by the International Society for Pharmacoepidemiology.

Effect of reduction in brain amyloid levels on change in cognitive and functional decline in randomized clinical trials: An instrumental variable meta-analysis.

INTRODUCTION: Whether the reduction in brain amyloid beta (Aß) plaque alone may substantially slow cognitive and functional decline in patients with dementia or mild cognitive impairment due to Alzheimer's disease (AD) remains debated. METHODS: An instrumental variable meta-analysis was performed to infer the effect of change in positron emission tomography (PET)-measured Aß standardized uptake value ratio (SUVR) on cognitive and functional decline. RESULTS: Pooling data from 16 randomized trials demonstrates that each 0.1-unit decrease in PET Aß SUVR is associated with a reduction (95% confidence interval) by 0.09 (0.034-0.15), 0.33 (0.12-0.55), and 0.13 (0.017-0.24) point in the average change of the Clinical Dementia Rating-Sum of Boxes, the Alzheimer's Disease Assessment Scale-Cognitive Subscale, and the Mini-Mental State Examination, respectively. DISCUSSION: This meta-analysis provides statistically significant evidence of a likely causal relationship between a reduction in Aß plaque and a reduction in cognitive and functional decline in patients with AD. HIGHLIGHTS: A widely cited meta-analysis article concluded amyloid beta reduction does not substantially improve cognition. We identified data inconsistencies in the initial publication and found new trial data. We repeated the meta-analysis after correcting data inconsistencies and adding new trial data. Updated results suggested statistically significant clinical benefit of amyloid beta reduction. Amyloid beta is a viable biological target for the treatment and prevention of AD.