Directed Acyclic Graph Assisted Method For Estimating Average Treatment Effect

Observational data, such as electronic clinical records and claims data, can prove invaluable for evaluating the Average Treatment Effect (ATE) and supporting decision-making, provided they are employed correctly. The Inverse Probability of Treatment Weighting (IPTW) method, based on propensity scores, has demonstrated remarkable efficacy in estimating ATE, assuming that the assumptions of exchangeability, consistency, and positivity are met. Directed Acyclic Graphs (DAGs) offer a practical approach to assess the exchangeability assumption, which asserts that treatment assignment and potential outcomes are independent given a set of confounding variables that block all backdoor paths from treatment assignment to potential outcomes. To ensure a consistent ATE estimator, one can adjust for a minimally sufficient adjustment set of confounding variables that block all backdoor paths from treatment assignment to the outcome. To enhance the efficiency of ATE estimators, our proposal involves incorporating both the minimally sufficient adjustment set of confounding variables and predictors into the propensity score model. Extensive simulations were conducted to evaluate the performance of propensity score-based IPTW methods in estimating ATE when different sets of covariates were included in the propensity score models. The simulation results underscored the significance of including the minimally sufficient adjustment set of confounding variables along with predictors in the propensity score models to obtain a consistent and efficient ATE estimator. We applied this proposed method to investigate whether tracheostomy was causally associated with in-hospital infant mortality, utilizing the 2016 Healthcare Cost and Utilization Project Kids' Inpatient Database. The estimated ATE was found to be approximately 2.30%-2.46% with p-value >0.05.

Estimating the causal effects of multiple intermittent treatments with application to COVID-19

To draw real-world evidence about the comparative effectiveness of multiple time-varying treatments on patient survival, we develop a joint marginal structural survival model and a novel weighting strategy to account for time-varying confounding and censoring. Our methods formulate complex longitudinal treatments with multiple start/stop switches as the recurrent events with discontinuous intervals of treatment eligibility. We derive the weights in continuous time to handle a complex longitudinal dataset without the need to discretize or artificially align the measurement times. We further use machine learning models designed for censored survival data with time-varying covariates and the kernel function estimator of the baseline intensity to efficiently estimate the continuous-time weights. Our simulations demonstrate that the proposed methods provide better bias reduction and nominal coverage probability when analyzing observational longitudinal survival data with irregularly spaced time intervals, compared to conventional methods that require aligned measurement time points. We apply the proposed methods to a large-scale COVID-19 dataset to estimate the causal effects of several COVID-19 treatments on the composite of in-hospital mortality and ICU admission.

How does debt shape health outcomes for older Americans?

The deleterious effects that debt can have on health outcomes are well documented, yet comprehensive studies of the debt-health link for older adults remain limited even as their indebtedness has increased dramatically in recent decades. Additionally, the literature cannot explain the causal pathway linking poor health with debt. Using data from the Health and Retirement Study (1998-2016), we examine a range of physical and mental health measures and assess how they may be shaped by the amount and type of debt held by older adults. To address the likely endogeneity of debt and health, we employ marginal structural models, developed specifically as an identification strategy in the presence of possible endogeneity, alongside population-averaged models that allow us to compare outcomes for populations with and without debt without relying on unverifiable assumptions regarding the underlying population distribution as is the case with random- and fixed-effects models. Findings indicate that carrying any debt has a negative effect on a range of health outcomes for older adults, including objective and subjective physical and mental health. In addition, the more debt older adults carry, the more detrimental it is for their health. Finally, the type of debt matters: while secured debt has a limited, if any, negative impact on health outcomes, unsecured debt has a substantial negative impact on health. Policymakers should design policies that promote the prudent use of debt and discourage carrying large debt burdens, especially unsecured debt, into retirement as this would promote better health outcomes for older Americans.

Distinct hyperuricemia trajectories are associated with different risks of incident diabetes: A prospective cohort study

BACKGROUND AND AIM

Conflicting results suggest a link between serum uric acid and diabetes and previous studies ignored the effect of continuous exposure of serum uric acid on diabetes risk. This study aims to characterize hyperuricemia trajectories in middle-aged adults and to examine its potential impact on diabetes risk, considering the role of obesity, dyslipidemia, and hypertension.

METHODS AND RESULTS

The cohort included 9192 participants who were free of diabetes before 2013. The hyperuricemia trajectories during 2009-2013 were identified by latent class growth models. Incident diabetes during 2014-2018 was used as the outcome. Modified Poisson regression models were used to assess the association of trajectories with diabetes. Furthermore, marginal structural models were used to estimate the mediating effects of the relationship between hyperuricemia trajectories and diabetes. We identified three discrete hyperuricemia trajectories: high-increasing (n = 5794), moderate-stable (n = 2049), and low-stable (n = 1349). During 5 years of follow-up, we documented 379 incident diabetes cases. Compared with the low-stable pattern, the high-increasing pattern had a higher risk of developing diabetes (RR, 1.42; 95% CI: 1.09-1.84). In addition, the percentages of total effect between the high-increasing hyperuricemia pattern and diabetes mediated by obesity, dyslipidemia, and hypertension were 24.41%, 18.26%, and 6.29%. However, the moderate-stable pattern was not associated with an increased risk of diabetes.

CONCLUSIONS

These results indicate that the high-increasing hyperuricemia trajectory is significantly associated with an increased risk of diabetes. Furthermore, obesity, dyslipidemia, and hypertension play mediating roles in the relationship between the high-increasing hyperuricemia pattern and increased diabetes risk.

Instrumental variable estimation of the marginal structural Cox model for time-varying treatments

Robins (1998) introduced marginal structural models, a general class of counterfactual models for the joint effects of time-varying treatments in complex longitudinal studies subject to time-varying confounding. Robins (1998) established the identification of marginal structural model parameters under a sequential randomization assumption, which rules out unmeasured confounding of treatment assignment over time. The marginal structural Cox model is one of the most popular marginal structural models for evaluating the causal effect of time-varying treatments on a censored failure time outcome. In this paper, we establish sufficient conditions for identification of marginal structural Cox model parameters with the aid of a time-varying instrumental variable, in the case where sequential randomization fails to hold due to unmeasured confounding. Our instrumental variable identification condition rules out any interaction between an unmeasured confounder and the instrumental variable in its additive effects on the treatment process, the longitudinal generalization of the identifying condition of Wang & Tchetgen Tchetgen (2018). We describe a large class of weighted estimating equations that give rise to consistent and asymptotically normal estimators of the marginal structural Cox model, thereby extending the standard inverse probability of treatment weighted estimation of marginal structural models to the instrumental variable setting. Our approach is illustrated via extensive simulation studies and an application to estimating the effect of community antiretroviral therapy coverage on HIV incidence.

Family structure and depressive symptoms among older adults in China: A marginal structural model analysis

BACKGROUND

Mental well-being in the aging population is inevitably linked to families due to the reliance of older adults on family members. This study investigates the causal relationship between family structure and depressive symptoms among Chinese older adults in general and between gender and residential areas.

METHODS

We used China Health and Retirement Longitudinal Study (CHARLS) panel data, covering four data collection rounds over seven years. Family structure was classified into single-member, couple, nuclear family, and extended family. Taking into account time-varying confounding, we estimated the causal effects of family structure on depressive symptoms using marginal structural models.

RESULTS

Older people with cumulative exposure to single-member family type had an increased odds of depressive symptoms by an average of 33 % (95 % CI: 1.22-1.44) than their counterparts who lived in the couple family. Additionally, older people living in extended families also had 6 % higher odds of experiencing depressive symptoms (95 % CI: 1.00, 1.11). The longitudinal associations were consistent across gender groups and residential areas (p-value for interaction is 0.6638 for gender and 0.7043 for the residential area).

LIMITATION

The time-varying confounders (e.g., chronic health conditions) included in the analysis are based on self-reported data, which may be subject to measurement errors.

CONCLUSION

The risk of depressive symptoms is greater for older individuals living alone and in extended families. Screening for depression in the older population, particularly those living in "at-risk" households, is recommended.

Childhood stunting and subsequent educational outcomes: a marginal structural model analysis from a South African longitudinal study

OBJECTIVE

To examine the association between childhood stunting and grade completion (as educational outcome) in South Africa.

DESIGN

Longitudinal study. Data were obtained using the National Income Dynamics Study over five waves (2008 to 2017). Children were tracked at wave 1 in 2008 until wave 5 in 2017 to determine their total years of schooling. We controlled for time-variant and time-varying confounding with a marginal structural model to estimate the associations between childhood stunting and subsequent grade completion.

SETTING

Nationally representative study of South African households.

PARTICIPANTS

A total of 2629 children aged 2 and 3 years in 2008.

RESULTS

We observed a substantial decrease in the prevalence of stunting between wave 1 (28·2 %) and wave 4 (8·6 %). Our marginal structural model results suggest that childhood stunting was significantly associated with decreased odds (22 % less likely) of grade completion (OR = 0·78; 95 % CI: 0·40, 0·86; P = 0·015), while those who were only stunted during early childhood had a 29 % reduction in the odds of grade completion (OR = 0·71; 95 % CI: 0·51, 0·82; P = 0·020).

CONCLUSION

These findings underscore the fact that stunting is a significant predictor of academic achievement, whose effects might be long-lasting.

Evaluating Adherence to Concomitant Diabetes, Hypertension, and Hyperlipidemia Treatments and Cardiovascular Outcomes Among Elderly Patients Using Marginal Structural Modeling

INTRODUCTION

Comorbid diabetes, hypertension, and hyperlipidemia is associated with an adverse effect on cardiovascular (CV) outcomes. Adherence to concurrent anti-diabetics, anti-hypertensives, and lipid-lowering therapies is essential to achieve therapeutic benefits.

AIM

The objective was to evaluate the association between adherence to concomitant oral antidiabetics, statins, and RAS antagonists (triple therapy) and CV outcomes, among elderly patients using marginal structural modeling (MSM).

METHODS

A retrospective study was conducted among patients on concurrent triple therapy from January 2016 until December 2019. Adherence to concurrent triple therapy was measured every 6 months using proportion of days covered (PDC) to determine the different adherence groups. CV outcomes were also measured every 6 months. A MSM controlling for baseline covariates and time-varying confounders affected by prior adherence was conducted to evaluate the association between adherence and CV outcomes. A sub-analysis was conducted among patients with prior CV events to evaluate the association between adherence to triple therapy and CV outcomes using MSMs.

RESULTS

The final cohort comprised of 7433 patients. The MSM model revealed no significant associations between adherence to triple/double therapies and cardiovascular outcomes. For sub-analysis, 471 patients with a prior CV event were identified. Results of the sub-analysis revealed no significant associations between adherence to triple/double therapies and CV outcomes among patients with prior CV events.

CONCLUSION

Future studies should evaluate the association with longer follow-up periods.

Statins was not associated with hepatocellular carcinoma after controlling for time-varying confounders in patients with diabetes

OBJECTIVE

We examined the association between statin use and hepatocellular carcinoma (HCC) incidence in patients with diabetes using marginal structural models (MSMs) estimated by inverse probability weight (IPW), which adjusts for time-varying confounders that are also mediators, and we compared the results with conventional regression methods.

STUDY DESIGN AND SETTING

This retrospective cohort study included 245,122 patients with type 2 diabetes who were new users of lipid-lowering drugs identified using the claims data of a universal health insurance program. Statin exposure was time-updated every 3 months during the follow-up period. Stabilized IPW was calculated and accounted for chronic liver diseases considering as time-dependent confounders affected by past statin exposure.

RESULTS

Over a median follow-up of 5.2 years, 1,694 patients developed HCC. In the conventional regression analysis, the hazard ratio of HCC associated with statin use was 0.88 (95% CI: 0.79-0.97) after adjusting for baseline covariates and 0.97 (95% CI: 0.87-1.08) after additionally adjusting for time-varying covariates. The hazard ratio increased to 1.11 (95% CI: 0.94-1.31) using the MSM approach.

CONCLUSION

Statins use was not associated with the risk of developing HCC in patients with diabetes. Our findings highlight the importance of controlling time-varying confounders in observational studies.

Application of Causal Inference Methods in the Analysis of Observational Neurosurgical Data: G-Formula and Marginal Structural Model

OBJECTIVE

When using observational data to estimate the causal effects of a treatment on clinical outcomes, we need to adjust for confounding. In the presence of time-dependent confounders that are affected by previous treatment, adjustments cannot be made via the conventional regression approach or propensity score-based methods, but requires sophisticated methods called g-methods. We aimed to introduce g-methods to estimate the causal effects of treatment strategies defined by treatment at multiple time points, such as treat 2 days versus treat only day 1 versus never-treat.

METHODS

Two g-methods were introduced: the g-formula and inverse probability-weighted marginal structural models. Under exchangeability, consistency, and positivity assumptions, they provide a consistent estimate of the causal effects of the treatment strategy.

RESULTS

Using a numeric example that mimics the observational study data, we presented how the g-formula and inverse probability-weighted marginal structural models can estimate the effect of the treatment strategy.

CONCLUSIONS

Both g-formula and inverse probability-weighted marginal structural models can correctly estimate the effect of the treatment strategy under 3 identifiability assumptions, which conventional regression analysis cannot. G-methods may assist in estimating the effect of treatment strategy defined by treatment at multiple time points.

Comparative Effectiveness of Dialysis Modality on Laboratory Parameters of Mineral Metabolism

INTRODUCTION

Chronic kidney disease-mineral and bone disorders (CKD-MBD) are prevalent in patients undergoing maintenance dialysis. Yet, there are limited and mixed evidence on the effects of different dialysis modalities involving longer treatment times or higher frequencies on CKD-MBD markers.

METHODS

This cohort study used data from 132,523 incident dialysis patients treated with any of the following modalities: conventional thrice-weekly in-center hemodialysis, nocturnal in-center hemodialysis (NICHD), home hemodialysis (HHD), or peritoneal dialysis (PD) from 2007 to 2011. We used marginal structural models fitted with inverse probability weights to adjust for fixed and time-varying confounding and informative censoring. We estimated the average effects of treatments with different dialysis modalities on time-varying serum concentrations of CKD-MBD markers: albumin-corrected calcium, phosphate, parathyroid hormone (PTH), and alkaline phosphatase (ALP) using pooled linear regression.

RESULTS

Most of the cohort were exclusively treated with conventional in-center hemodialysis, while few were ever treated with NICHD or HHD. At the baseline, PD patients had the lowest mean and median values of PTH, while NICHD patients had the highest median values. During follow-up, compared to hemodialysis patients, patients treated with NICHD had lower mean serum PTH (19.8 pg/mL [95% confidence interval: 2.8, 36.8] lower), whereas PD and HHD patients had higher mean PTH (39.7 pg/mL [31.6, 47.8] and 51.2 pg/mL [33.0, 69.3] higher, respectively). Compared to hemodialysis patients, phosphate levels were lower for patients treated with NICHD (0.44 mg/dL [0.37, 0.52] lower), PD (0.15 mg/dL [0.12, 0.19] lower), or HHD (0.33 mg/dL [0.27, 0.40] lower). There were no clinically meaningful associations between dialysis modalities and concentrations of calcium or ALP.

CONCLUSION

In incident dialysis patients, compared to treatment with conventional in-center hemodialysis, treatments with other dialysis modalities with longer treatment times or higher frequency were associated with different patterns of serum phosphate and PTH. Given the recent growth in the use of dialysis modalities other than hemodialysis, the associations between the treatment and the CKD-MBD markers warrant additional study.

Evaluating a Cox marginal structural model to assess the comparative effectiveness of inhaled corticosteroids versus no inhaled corticosteroid treatment in chronic obstructive pulmonary disease

PURPOSE

To evaluate the potential of a Cox marginal structural model (MSM) to estimate the time-varying causal inference of a known clinical trial association where the effectiveness of inhaled corticosteroid- (ICS-) versus non-ICS-containing treatments has been compared in patients with chronic obstructive pulmonary disease (COPD).

METHODS

This retrospective study from 2006 to 2016 used linked data from Clinical Practice Research Datalink-GOLD, Hospital Episode Statistics and Office for National Statistics mortality. A Cox MSM, incorporating a new-user design, was deemed capable of replicating a clinical trial-like pathway. Repeated outcomes for exacerbation events and stabilised weights were used to include time-varying and fixed covariate exposures.

RESULTS

Of 45,958 patients, 55% were male; 52% had moderate COPD. ICS-treated patients had a higher incidence of comorbid asthma than non-ICS-treated patients. Adjusted hazard risk ratios for any exacerbation event: ICS/long-acting β₂-agonist (LABA) versus long-acting muscarinic antagonist (LAMA), 1.07 (95% confidence interval 1.04-1.10); ICS/LABA versus LABA/LAMA, 1.05 (1.00-1.10); ICS/LABA/LAMA versus LAMA, 1.04 (1.01-1.06); ICS/LABA/LAMA versus LABA/LAMA 1.02 (0.97-1.07).

CONCLUSIONS

The Cox MSM was not able to fully demonstrate results consistent with the previously established benefits of ICS-containing treatments seen in clinical trials. Future studies should continue to investigate causal inference methods and their capability to estimate the long-term outcomes of treatment in COPD.

Estimating the causal effect of transient anemia status on renal and cardiovascular outcomes in community-dwelling patients in Japan at the beginning of impaired renal function using marginal structural modeling

Background

Anemia status may be transient. Causal associations between changes in anemia status over time and adverse outcome development are not well characterized in community-dwelling subjects at the beginning of impaired kidney function.

Methods

This retrospective cohort study used annual health checkup and medical and pharmacy claims data from the JMDC between January 2005 and June 2019. Community-dwelling subjects in Japan with a pre-index estimated glomerular filtration rate (eGFR) ≥ 60 mL/min/1.73 m² followed by a subsequent eGFR < 60 mL/min/1.73 m² (index) were included. The composite renal outcome was ≥ 30% eGFR reduction over 3 years from baseline, serum creatinine doubling, progression to chronic dialysis, kidney transplantation, or eGFR < 15 mL/min/1.73 m². The composite cardiovascular outcome was fatal and non-fatal unstable angina, myocardial infarction, heart failure, or cerebrovascular event. Time-dependent anemia risk was evaluated using Breslow’s estimator and marginal structural Cox models (MSM).

Results

In 32,870 included subjects, 1,396 had anemia at baseline. Adverse outcome incidence was higher in the baseline anemic group, but absolute differences in renal and cardiovascular outcomes between groups were diminished after adjusting for baseline characteristics. In MSM, time-dependent anemia status was associated with higher risk of renal (hazard ratio [95% confidence interval]; 2.6 [1.7–3.8]) and cardiovascular (1.6 [1.2–2.2]) outcomes and mortality (2.8 [1.8–4.3]). Absolute differences in survival probabilities were retained over time but were clinically marginal (1.1–2.7% over 6 years).

Conclusions

Even in subjects at the very early stage of impaired kidney function, early detection and treatment of anemia may help reduce the development of negative sequelae.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10157-021-02137-1.

Estimating the joint effect of diabetes and subsequent depressive symptoms on mortality among older latinos

PURPOSE

Diabetes and depression are risk factors for cardiovascular disease, but the evidence about their interaction effect on long-term health outcomes among Latinos is lacking. We aimed to investigate the joint association of diabetes and subsequent depressive symptoms with mortality among older Latinos, an understudied racial/ethnic group with high prevalence of diabetes.

METHODS

This study included 1,495 adults from the Sacramento Area Latino Study on Aging. We employed Cox proportional-hazards models to estimate the adjusted hazard ratios [aHRs] for cardiovascular and all-cause mortality according to diabetes status at enrollment and depressive symptoms a year after the enrollment. We used marginal structural models to adjust for time-varying confounders.

RESULTS

The mean age (standard deviation) of participants was 70 (6.6) years. Over follow-up (median 7.7 years), diabetes and depressive symptoms were individually associated with increased risk of cardiovascular mortality (diabetes, aHR[95% CI]=2.13[1.60-2.84]; depressive symptoms, aHR[95% CI]=1.62[1.09-2.39]) and all-cause mortality (diabetes, aHR[95% CI]=1.92[1.53-2.41]; depressive symptoms, aHR[95% CI]=1.41[1.02-1.94]). After adjusting for time-varying confounders, we found a multiplicative interaction between diabetes and subsequent depressive symptoms for cardiovascular mortality (aHR[95% CI]=2.94[1.07-8.39]), but not all-cause mortality (aHR[95% CI]=1.80[0.81-4.35]).

CONCLUSIONS

Using a longitudinal cohort of community-dwelling older Latinos, we found that diabetes and subsequent depressive symptoms were jointly associated with increased risk of cardiovascular mortality.

How Far Does Highly Active Antiretroviral Treatment Reduce TB Incidence among Children? A Marginal Structural Modeling Analysis, Southwest Ethiopia

Background

Children younger than 15 years, carry almost 80% of the global burden of HIV/AIDS. HIV worsens the progression of latent TB to active TB disease. Although antiretroviral treatment has shown marked reduction in Tuberculosis incidence, TB continues to occur in Sub-Saharan countries including Ethiopia. The aim of this study was to investigate the impact of HAART on the incidence of tuberculosis among children infected with HIV in Southwest Ethiopia.

Methods

A retrospective cohort study was conducted between 2009 to 2014. We used chi-square test, and Mann-Whitney U test to compare non-HAART and HAART cohort. We estimated the effect of HAART on TB incidence using marginal structural model after adjusting for time-dependent confounders affected by exposure.

Result

A total of 844 children were followed. We observed them for a median of 51 months (IQR 31) and a total of 2942.99 child-years. The overall TB incidence rate was 7.917 per 100 child years (95% CI, 6.933–9.002). TB incidence for specific HAART and non-HAART cohort were 7.67 per 100 child-years (95% CI, 6.318–9.217) and 8.17 per 100 child-years (95% CI, 6.772–9.767) respectively. From marginal structural modeling, children on HAART were 36% (HR=0.642, 95% CI 0.442–0.931, p<0.02) less likely to develop TB compared to those who were not.

Conclusion

HAART reduced the hazard of TB in HIV-infected children by 36%. This is by far less than what is expected.

Gender-specific effects of adverse childhood experiences on incidence of activities of daily life disability in middle-age and elderly Chinese population

BACKGROUND

Adverse Childhood Experiences (ACEs) may have long-lasting effects on late life health, probably through life-course mediators. However, whether such effects still exist when these mediators have been appropriately controlled is unclear.

OBJECTIVES

To estimate the controlled direct effect of ACEs on Activities of Daily Life (ADL) disability in middle-aged people and examine the gender-difference of this effect.

PARTICIPANTS AND SETTING

We used data from the China Health and Retirement Longitudinal Study, a nationally representative longitudinal survey of persons aged 45+ years.

METHODS

ACEs were measured by the Adverse Childhood Experiences International Questionnaire and number of ACEs was classified as 0, 1, 2 and 3+, while ADL disability was measured using the Katz Index. Gender-specific controlled direct effects of ACEs on the incidence of ADL disability were estimated by marginal structural model (MSM) with stabilized inverse-probability-of-treatment weights of mediators (unhealthy behaviors, chronic diseases and depression).

RESULTS

4,544 males and 4,767 females were included. Gender differences existed in most categories of ACEs, and about 10 % participants had 3+ ACEs. Participants who had 3+ ACEs had 39 % and 59 % higher risk of ADL disability than those with 0 ACEs among males and females, respectively. After controlling for the mediators, the direct effect was slightly increased in males (risk ratio (RR) = 1.45, p < 0.001) but decreased in females (RR=1.28, p < 0.05).

CONCLUSIONS

Precautions targeted in reducing ACEs may be beneficial in preventing ADL disability, but gender-specific prevention should be considered.

The effect of gestational diabetes mellitus on the risk of asthma in offspring

PURPOSE

To examine the effect of maternal gestational diabetes mellitus on the risk of asthma in the offspring.

METHODS

This cohort study used data from 19,933 children in the National Longitudinal Survey of Children and Youth (NLSCY), 1994/1995-2008/2009, Canada. Children were followed until the first-time report of having health professional-diagnosed asthma (hereafter incident asthma), loss to follow-up, or end of the NLSCY follow-up, whichever occurred first. As a surrogate for Cox proportional hazards regression, pooled logistic regression models, crude and adjusted for potential confounders, were fitted to estimate the effect of gestational diabetes mellitus on the risk of asthma in the offspring.

RESULTS

Among the 19,933 children, 1,178 (5.9%) had mothers with gestational diabetes mellitus. The median duration of follow-up was 4 (interquartile range: 4) years. A total of 1639 children in the cohort had reported incident asthma during the follow-up, and 119 of them had mothers with gestational diabetes mellitus. The adjusted hazard ratio for the association between gestational diabetes mellitus and incident asthma in offspring was 1.25 (95% confidence interval [CI] 1.03, 1.51).

CONCLUSIONS

Our findings suggest that gestational diabetes mellitus increases the risk of asthma in the offspring.

Diabetes, employment and behavioural risk factors in China: Marginal structural models versus fixed effects models

We use longitudinal data from the China Health and Nutrition Survey, covering the years 1997-2011, to estimate the effect of a diabetes diagnosis on an economic outcome (employment probabilities) and behavioural risk factors (alcohol consumption, smoking cessation, body mass index (BMI), physical activity and hypertension) for men and women. We apply two complementary statistical techniques-marginal structural models (MSMs) and fixed effects (FE) models-to deal with confounding. Both methods suggest, despite their different underlying assumptions, similar patterns that indicate important differences between men and women. Employment probabilities decline substantially after the diagnosis for women (-12.4 (MSM) and -15.5 (FE) percentage points), but do not change significantly for men. In particular, the MSM estimates indicate an increase in hypertension (13 percentage points) and a decrease in physical activity for women, while men have small and statistically insignificant changes in these outcomes. For BMI, the MSM results indicate statistically significant changes for men (-.76), but not for women, while the FE estimates show similar reductions for men and women (-.80 and -.73 respectively). Men also reduce their alcohol consumption, but do not cease to smoke. For women these risk factors have a prevalence close to zero to begin with, though women seem to still reduce alcohol consumption somewhat. These results suggest important gender differences in the impact of diabetes in China. To narrow these inequities policies supporting women to reduce diabetes related risk factors are likely important.

An assessment of the potential miscalibration of cardiovascular disease risk predictions caused by a secular trend in cardiovascular disease in England

Background

A downwards secular trend in the incidence of cardiovascular disease (CVD) in England was identified through previous work and the literature. Risk prediction models for primary prevention of CVD do not model this secular trend, this could result in over prediction of risk for individuals in the present day. We evaluate the effects of modelling this secular trend, and also assess whether it is driven by an increase in statin use during follow up.

Methods

We derived a cohort of patients (1998–2015) eligible for cardiovascular risk prediction from the Clinical Practice Research Datalink with linked hospitalisation and mortality records (N = 3,855,660). Patients were split into development and validation cohort based on their cohort entry date (before/after 2010). The calibration of a CVD risk prediction model developed in the development cohort was tested in the validation cohort. The calibration was also assessed after modelling the secular trend. Finally, the presence of the secular trend was evaluated under a marginal structural model framework, where the effect of statin treatment during follow up is adjusted for.

Results

Substantial over prediction of risks in the validation cohort was found when not modelling the secular trend. This miscalibration could be minimised if one was to explicitly model the secular trend. The reduction in risk in the validation cohort when introducing the secular trend was 35.68 and 33.24% in the female and male cohorts respectively. Under the marginal structural model framework, the reductions were 33.31 and 32.67% respectively, indicating increasing statin use during follow up is not the only the cause of the secular trend.

Conclusions

Inclusion of the secular trend into the model substantially changed the CVD risk predictions. Models that are being used in clinical practice in the UK do not model secular trend and may thus overestimate the risks, possibly leading to patients being treated unnecessarily. Wider discussion around the modelling of secular trends in a risk prediction framework is needed.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12874-020-01173-x.

Cognitively stimulating activities and risk of probable dementia or cognitive impairment in the English Longitudinal Study of Ageing

Objectives

To examine the association between cognitive stimulating activities (CSA) in later life (internet/email use, employment, volunteering, evening classes, social club membership and newspaper reading) and risk of cognitive impairment or dementia using marginal structural models to account for time-varying confounding affected by prior exposure.

Methods

Data were used from the English Longitudinal Study of Ageing waves 1 (2002) to 7 (2014), a nationally representative sample of adults in England aged ≥50. Self-reported participation in CSAs were measured as binary exposures from waves 2 (2004) to 6 (2012), with final sample sizes between n = 3937 and n = 2530 for different CSAs. Baseline exposure and covariates were used to create inverse probability of treatment and censoring weights (IPTCW). IPTCW repeated measures Poisson and linear regression were used to estimate each CSAs effect on risk of probable cognitive impairment or dementia at wave 7 (defined as a score of ≤11/27 on a modified telephone interview for cognitive status (TICS-27)). Results were compared to standard regression adjustment.

Results

Internet use at any wave (Risk ratios between 0.62 and 0.69) and volunteering in waves 3 to 6 (RRs between 0.516 and 0.633) were associated with reduced risk of cognitive impairment in IPTCW models. Standard estimates were similar for both internet use and volunteering.

Newspaper reading (RR 95% Confidence interval 0.74–0.99) and social club membership (RR 95% CI 0.54–0.86) at wave 6 were significantly associated with risk of cognitive impairment in standard models, but not in the IPTCW models (RR 95% CI 0.82–1.11 and 0.60–1.08 respectively). Employment and evening classes were not associated with cognitive impairment in either model.

Conclusions

We found that volunteering and internet use were associated with reduced risk of cognitive impairment. Associations between newspaper reading or social club membership and cognitive impairment may be due to time-varying confounding affected by prior exposure.

•

Confounding affected by past exposure is a problem in studies of cognitive function.

•

We addressed this using inverse probability weighted marginal structural models.

•

Volunteering and internet use were protective against cognitive impairment.

•

Other cognitively stimulating activities were protective with standard regression.

•

But these associations were non-significant in the marginal structural models.

Investigating the causal effect of socioeconomic status on quality of care under a universal health insurance system - a marginal structural model approach

Background

Social disparities in healthcare persist in the US despite the expansion of Medicaid under the Affordable Care Act. We investigated the causal impact of socioeconomic status on the quality of care in a setting with minimal confounding bias from race, insurance type, and access to care.

Methods

We designed a retrospective population-based study with a random 25% sample of adult Taiwan population enrolled in Taiwan’s National Health Insurance system from 2000 to 2016. Patient’s income levels were categorized into low-income group (<25th percentile) and high-income group (≥25th percentile). We used marginal structural modeling analysis to calculate the odds of hospital admissions for 11 ambulatory care sensitive conditions identified by the Agency for Healthcare Research and Quality and the odds of having an Elixhauser comorbidity index greater than zero for low-income patients.

Results

Among 2,844,334 patients, those in lower-income group had 1.28 greater odds (95% CI 1.24–1.33) of experiencing preventable hospitalizations, and 1.04 greater odds (95% CI 1.03–1.05) of having a comorbid condition in comparison to high-income group.

Conclusions

Income was shown to be a causal factor in a patient’s health and a determinant of the quality of care received even with equitable access to care under a universal health insurance system. Policies focusing on addressing income as an important upstream causal determinant of health to provide support to patients in lower socioeconomic status will be effective in improving health outcomes for this vulnerable social stratum.

Smoking status and cause-specific discontinuation of tumour necrosis factor inhibitors in axial spondyloarthritis

Background

The impact of smoking on TNF inhibition (TNFi) therapy is unclear. We examined the effect of smoking on all-cause and cause-specific TNFi discontinuation in axial spondyloarthritis (axSpA).

Methods

We used longitudinal data from the British Society for Rheumatology Biologics Register for Ankylosing Spondylitis (BSRBR-AS). Patients fulfilling the ASAS criteria for axSpA, who started their first TNFi, were eligible for analysis. Inverse-probability weights were used to balance differences in baseline disease severity and other confounders. We used marginal structural Cox proportional hazard models to estimate hazard ratios (HR) for TNFi discontinuation according to smoking status. In analyses of cause-specific discontinuation, competing risk events were considered as censoring, using inverse-probability weights.

Results

A total of 758 participants were included in the analysis (66% male, mean age 45 years), providing 954 patient-years of follow-up. TNFi was discontinued in 174 (23%) patients, among whom 26% stopped due to infections, 20% due to other adverse events and 44% due to inefficacy or other reasons. Thirty-four percent were current smokers and 30% ex-smokers. Compared to never smokers, current smokers’ risk of TNFi discontinuation was HR 0.79 (95%CI 0.53 to 1.20) and ex-smokers HR 0.68 (95%CI 0.45 to 1.04). Our data did not show evidence that current smoking influenced discontinuation due to infections (HR 0.79, 95%CI 0.40 to 1.54), other adverse events (HR 0.86, 95%CI 0.41 to 1.78) or inefficacy/other causes (HR 1.44, 95%CI 0.86 to 2.41).

Conclusion

Baseline smoking status did not impact TNFi discontinuation in this UK cohort of axSpA participants.

Electronic supplementary material

The online version of this article (10.1186/s13075-019-1958-z) contains supplementary material, which is available to authorized users.

The effect of delaying initiation with umeclidinium/vilanterol in patients with COPD: an observational administrative claims database analysis using marginal structural models

BACKGROUND

Chronic obstructive pulmonary disease (COPD) is associated with high clinical and economic burden. Optimal pharmacological therapy for COPD aims to reduce symptoms and the frequency and severity of exacerbations. Umeclidinium/vilanterol (UMEC/VI) is an approved combination therapy for once-daily maintenance treatment of patients with COPD. This study evaluated the impact of delaying UMEC/VI initiation on medical costs and exacerbation risk.

METHODS

A retrospective analysis of patients with COPD who initiated UMEC/VI between 4/28/2014 and 7/31/2016 was conducted using the Optum Research Database. The index date was the first COPD visit after UMEC/VI available on US formulary (Commercial 4/28/2014; Medicare Advantage 1/1/2015). Patients were followed for 12 months post-index, and categorized into 12 cohorts corresponding to month (30-day period) of UMEC/VI initiation (i.e. Months 1-12) post-index. The outcomes studied during the follow up period included COPD-related and all-cause medical costs, and risk of COPD exacerbations. Marginal structural models (MSM) were used to control for time-varying confounding due to changes in treatment and severity during follow up.

RESULTS

2,200 patients initiating UMEC/VI were included in the study sample. Patients' average age was 69.3 years, 49.9% were female and 69.7% were Medicare insured. Following MSM analysis, 12-month adjusted COPD-related medical costs increased by 2.9% (95% confidence interval [CI]: 0.1-5.9%; p = 0.044) for each monthly delay in UMEC/VI initiation, with a 37.4% higher adjusted cost for patients initiating UMEC/VI in Month 12 versus Month 1 ($13,087 vs. $9524). The 12-month adjusted all-cause medical costs increased by 2.8% (95% CI: 0.6-5.2%; p = 0.013) for each monthly delay, with a 36.1% higher adjusted cost for patients initiating UMEC/VI at Month 12 versus Month 1 ($22,766 vs. $16,727). The monthly risk of severe exacerbation was significantly higher in patients who had not yet initiated UMEC/VI than those who had (hazard ratio: 1.74; 95% CI: 1.35-2.23; p < 0.001).

CONCLUSIONS

Prompt use of UMEC/VI following a physician visit for COPD appears to result in economic and clinical benefits, with reductions in medical costs and exacerbation risk. Additional research is warranted to assess the benefits of initiating UMEC/VI as a first-line therapy compared with escalation to UMEC/VI from monotherapies.

Tracheostomy and mortality in patients with severe burns: A nationwide observational study

BACKGROUND

Tracheostomy is often performed in patients with severe burns who are undergoing prolonged mechanical ventilation. However, the appropriate timing of tracheostomy and its effect on mortality remain unknown. The aim of this study was to determine whether tracheostomy can reduce mortality in patients with severe burns.

METHODS

Using the Japanese Diagnosis Procedure Combination database from April 2010 to March 2014, we extracted data on adult patients with severe burns (burn index score of ≥15) who started mechanical ventilation within 3days of admission. We estimated the hazard ratio for 28-day in-hospital mortality associated with tracheotomy performed from day 5 to 28. We adjusted for baseline and time-dependent confounders using inverse probability of treatment weighting methods and fitted a marginal structural Cox proportional hazard model.

RESULTS

We identified 680 eligible patients (94 in the tracheostomy group, 2289 person-days; 586 in the non-tracheostomy group, 11,197 person-days). Patients who underwent a tracheostomy had worse prognostic factors for mortality. After adjustment for these factors, the hazard ratio for 28-day mortality associated with tracheostomy compared with non-tracheostomy was 0.73 (95% confidence interval, 0.39-1.34).

CONCLUSIONS

There was no significant association between 28-day in-hospital mortality and early tracheostomy in adult patients with severe burns.

Causal Effect of Donor Source on Survival of Renal Transplantation Using Marginal Structural Models

BACKGROUND

Marginal Structural Models (MSMs) are novel methods to estimate causal effect in epidemiology by using Inverse Probability of Treatment Weighting (IPTW) and Stabilized Weight to reduce confounding effects. This study aimed to estimate causal effect of donor source on renal transplantation survival.

METHODS

In this cohort study, 1354 transplanted patients with a median 42.55 months follow-up in Namazee Hospital Transplantation Center, Shiraz from Mar 1999 to Mar 2009, were included to use marginal structural Cox regression, binomial logistic regression model to estimate causal effect of donor source on the survival of renal transplantation. IPTW and stabilized inverse probability of treatment weighting are used as weights.

RESULTS

The un-weighted (crude) hazard ratios for live unrelated donor and deceased donor in comparison to live related donor as reference group was (HR: 1.03, 95% CI: 0.58-1.83, P=0.89) and (HR: 2.69, 95% CI: 1.67-4.31, P=0.001), respectively. Using a marginal structural Cox regression model and by stabilized weight, the hazard ratios for live-unrelated donor and cadaveric donor were (HR: 1.08, 95% CI: 0.47-2.45, P=0.84) and (HR: 3.63, 95% CI: 1.59-8.26, P=0.002), respectively. There was no difference between estimated effect size from marginal structural Cox regression, marginal structural logistic regression, and marginal structural Weibull regression model.

CONCLUSION

There is no difference between related and unrelated donor source hazard ratio; however, hazard ratio for cadaveric donor was 3.63 times of hazard ratio for related donor and 3.34 times of it for unrelated donor. Therefore, the live donor (related or unrelated) has a better survival of renal transplantation than cadaveric donor.

Utility of inverse probability weighting in molecular pathological epidemiology

As one of causal inference methodologies, the inverse probability weighting (IPW) method has been utilized to address confounding and account for missing data when subjects with missing data cannot be included in a primary analysis. The transdisciplinary field of molecular pathological epidemiology (MPE) integrates molecular pathological and epidemiological methods, and takes advantages of improved understanding of pathogenesis to generate stronger biological evidence of causality and optimize strategies for precision medicine and prevention. Disease subtyping based on biomarker analysis of biospecimens is essential in MPE research. However, there are nearly always cases that lack subtype information due to the unavailability or insufficiency of biospecimens. To address this missing subtype data issue, we incorporated inverse probability weights into Cox proportional cause-specific hazards regression. The weight was inverse of the probability of biomarker data availability estimated based on a model for biomarker data availability status. The strategy was illustrated in two example studies; each assessed alcohol intake or family history of colorectal cancer in relation to the risk of developing colorectal carcinoma subtypes classified by tumor microsatellite instability (MSI) status, using a prospective cohort study, the Nurses' Health Study. Logistic regression was used to estimate the probability of MSI data availability for each cancer case with covariates of clinical features and family history of colorectal cancer. This application of IPW can reduce selection bias caused by nonrandom variation in biospecimen data availability. The integration of causal inference methods into the MPE approach will likely have substantial potentials to advance the field of epidemiology.

Impact of Food Insecurity on Depressive Symptoms Among HIV-HCV Co-infected People

Food insecurity (FI) is associated with depressive symptoms among HIV mono-infected people. Our objective was to examine to what extent this association holds among HIV-hepatitis C virus (HCV) co-infected people. We used data from a prospective cohort study of HIV-HCV co-infected people in Canada. FI was measured using the ten-item adult scale of Health Canada's Household Food Security Survey Module and was classified into three categories: food secure, moderate FI, and severe FI. Depressive symptoms were measured using the Center for Epidemiologic Studies Depression Scale (CES-D-10) and was classified into absence or presence of depressive symptoms. FI, depressive symptoms, and other covariates were updated every 6 months. The association between FI and depressive symptoms was assessed using a stabilized inverse probability weighted marginal structural model. The study sample included 725 HIV-HCV co-infected people with 1973 person-visits over 3 years of follow up. At baseline, 23% of participants experienced moderate food insecurity, 34% experienced severe food insecurity and 52% had depressive symptoms. People experiencing moderate FI had 1.63 times (95% CI 1.44-1.86) the risk of having depressive symptoms and people experiencing severe FI had 2.01 times (95% CI 1.79-2.25) the risk of having depressive symptoms compared to people who were food secure. FI is a risk factor for developing depressive symptoms among HIV-HCV co-infected people. Food supplementation, psychosocial support and counseling may improve patient health outcomes.

Modeling the causal effect of treatment initiation time on survival: Application to HIV/TB co-infection

The timing of antiretroviral therapy (ART) initiation for HIV and tuberculosis (TB) co-infected patients needs to be considered carefully. CD4 cell count can be used to guide decision making about when to initiate ART. Evidence from recent randomized trials and observational studies generally supports early initiation but does not provide information about effects of initiation time on a continuous scale. In this article, we develop and apply a highly flexible structural proportional hazards model for characterizing the effect of treatment initiation time on a survival distribution. The model can be fitted using a weighted partial likelihood score function. Construction of both the score function and the weights must accommodate censoring of the treatment initiation time, the outcome, or both. The methods are applied to data on 4903 individuals with HIV/TB co-infection, derived from electronic health records in a large HIV care program in Kenya. We use a model formulation that flexibly captures the joint effects of ART initiation time and ART duration using natural cubic splines. The model is used to generate survival curves corresponding to specific treatment initiation times; and to identify optimal times for ART initiation for subgroups defined by CD4 count at time of TB diagnosis. Our findings potentially provide 'higher resolution' information about the relationship between ART timing and mortality, and about the differential effect of ART timing within CD4 subgroups.

Marginal structural model to evaluate the joint effect of socioeconomic exposures on the risk of developing end-stage renal disease in patients with type 1 diabetes: a longitudinal study based on data from the Swedish Childhood Diabetes Study Group

PURPOSE

Diabetic nephropathy is a severe complication of type 1 diabetes (T1D) that may lead to renal failure and end-stage renal disease (ESRD) demanding dialysis and transplantation. The etiology of diabetic nephropathy is multifactorial and both genes and environmental and life style-related factors are involved. In this study, we investigate the effect of the socioeconomic exposures, unemployment and receiving income support, on the development of ESRD in T1D patients, using a marginal structural model (MSM) in comparison with standard logistic regression models.

METHODS

The study is based on the Swedish Childhood Diabetes Register which in 1977 started to register patients developing T1D before 15 years of age. In the analyses, we include patients born between 1965 and 1979, developing diabetes between 1977 and 1994, and followed until 2013 (n = 4034). A MSM was fitted to adjust for both baseline and time-varying confounders.

RESULTS

The main results of the analysis indicate that being unemployed for more than 1 year and receiving income support are risk factors for the development of ESRD. Multiple exposures over time to these risk factors increase the risk associated with the disease.

CONCLUSIONS

Using a MSM is an advanced method well suited to investigate the effect of exposures on the risk of complications of a chronic disease with longitudinal data. The results show that socioeconomic disadvantage increases the risk of developing ESRD in patients with T1D.

Contours of a causal feedback mechanism between adaptive personality and psychosocial function in patients with personality disorders: a secondary analysis from a randomized clinical trial

BACKGROUND

Patients with personality disorders commonly exhibit impairment in psychosocial function that persists over time even with diagnostic remission. Further causal knowledge may help to identify and assess factors with a potential to alleviate this impairment. Psychosocial function is associated with personality functioning which describes personality disorder severity in DSM-5 (section III) and which can reportedly be improved by therapy.

METHODS

The reciprocal association between personality functioning and psychosocial function was assessed, in 113 patients with different personality disorders, in a secondary longitudinal analysis of data from a randomized clinical trial, over six years. Personality functioning was represented by three domains of the Severity Indices of Personality Problems: Relational Capacity, Identity Integration, and Self-control. Psychosocial function was measured by Global Assessment of Functioning. The marginal structural model was used for estimation of causal effects of the three personality functioning domains on psychosocial function, and vice versa. The attractiveness of this model lies in the ability to assess an effect of a time - varying exposure on an outcome, while adjusting for time - varying confounding.

RESULTS

Strong causal effects were found. A hypothetical intervention to increase Relational Capacity by one standard deviation, both at one and two time-points prior to assessment of psychosocial function, would increase psychosocial function by 3.5 standard deviations (95% CI: 2.0, 4.96). Significant effects of Identity Integration and Self-control on psychosocial function, and from psychosocial function on all three domains of personality functioning, although weaker, were also found.

CONCLUSION

This study indicates that persistent impairment in psychosocial function can be addressed through a causal pathway of personality functioning, with interventions of at least 18 months duration.

Renin-angiotensin II-aldosterone system blockers and time to renal replacement therapy in children with CKD

BACKGROUND

Clinical care decisions to treat chronic kidney disease (CKD) in a growing child must often be made without the benefit of evidence from clinical trials. We used observational data from the Chronic Kidney Disease in Children cohort to estimate the effectiveness of renin-angiotensin II-aldosterone system blockade (RAAS) to delay renal replacement therapy (RRT) in children with CKD.

METHODS

A total of 851 participants (median age: 11 years, median glomerular filtration rate [GFR]: 52 ml/min/1.73 m2, median urine protein to creatinine ratio: 0.35 mg/mg) were included. RAAS use was reported at annual study visits. Both Cox proportional hazards models with time-varying RAAS exposure and Cox marginal structural models (MSM) were used to evaluate the effect of RAAS use on time to RRT. Analyses were adjusted or weighted to control for age, male sex, glomerular diagnosis, GFR, nephrotic range proteinuria, anemia, elevated blood pressure, acidosis, elevated phosphate and elevated potassium.

RESULTS

There were 217 RRT events over a 4.1-year median follow-up. At baseline, 472 children (55 %) were prevalent RAAS users, who were more likely to be older, have a glomerular etiology, have higher urine protein, be anemic, have elevated serum phosphate and potassium, take more medications, but less likely to have elevated blood pressure, compared with non-users. RAAS use was found to reduce the risk of RRT by 21 % (hazard ratio: 0.79) to 37 % (hazard ratio: 0.63) from standard regression adjustment and MSM models, respectively.

CONCLUSIONS

These results support inferences from adult studies of a substantial benefit of RAAS use in pediatric CKD patients.

The Effect of Depressive Symptoms on Adherence to Daily Oral PrEP in Men who have Sex with Men and Transgender Women: A Marginal Structural Model Analysis of The iPrEx OLE Study

We assessed the role of depressive symptoms on adherence to daily oral FTC/TDF for HIV PrEP in cisgender men who have sex with men (MSM) and transgender women who have sex with men (TGW) using data from the iPrEx OLE study. A marginal structural logistic regression model was used to estimate the effect of time-varying CES-D scores on having protective levels of drug concentration, adjusting for confounding by sexual practices over time, prior adherence, and baseline demographic characteristics. We found a non-monotonic relationship between CES-D score and odds of protective FTC/TDF levels in MSM. We found evidence that the effect of depression on adherence varied between MSM and TGW, and that depressive symptoms did not contribute greatly to decreased adherence on a population scale. We recommend that depressive symptoms not preclude the prescription of PrEP, and that MSM and TGW be studied separately.

Continuous renal replacement therapy versus intermittent hemodialysis in intensive care patients: impact on mortality and renal recovery

PURPOSE

The best renal replacement therapy (RRT) modality remains controversial. We compared mortality and short- and long-term renal recovery between patients treated with continuous RRT and intermittent hemodialysis.

METHODS

Patients of the prospective observational multicenter cohort database OUTCOMEREA™ were included if they underwent at least one RRT session between 2004 and 2014. Differences in patients' baseline and daily characteristics between treatment groups were taken into account by using a marginal structural Cox model, allowing one to substantially reduce the bias resulting from confounding factors in observational longitudinal data analysis. The composite primary endpoint was 30-day mortality and dialysis dependency.

RESULTS

Among 1360 included patients with RRT, 544 (40.0 %) and 816 (60.0 %) were initially treated by continuous RRT and intermittent hemodialysis, respectively. At day 30, 39.6 % patients were dead. Among survivors, 23.8 % still required RRT. There was no difference between groups for the primary endpoint in global population (HR 1.00, 95 % CI 0.77-1.29; p = 0.97). In patients with higher weight gain at RRT initiation, mortality and dialysis dependency were significantly lower with continuous RRT (HR 0.54, 95 % CI 0.29-0.99; p = 0.05). Conversely, this technique appeared to be deleterious in patients without shock (HR 2.24, 95 % CI 1.24-4.04; p = 0.01). Six-month mortality and persistent renal dysfunction were not influenced by the RRT modality in patients with dialysis dependence at ICU discharge.

CONCLUSION

Continuous RRT did not appear to improve 30-day and 6-month patient outcomes. It seems beneficial for patients with fluid overload, but might be deleterious in the absence of hemodynamic failure.

Double robust and efficient estimation of a prognostic model for events in the presence of dependent censoring

In longitudinal data arising from observational or experimental studies, dependent subject drop-out is a common occurrence. If the goal is estimation of the parameters of a marginal complete-data model for the outcome, biased inference will result from fitting the model of interest with only uncensored subjects. For example, investigators are interested in estimating a prognostic model for clinical events in HIV-positive patients, under the counterfactual scenario in which everyone remained on ART (when in reality, only a subset had). Inverse probability of censoring weighting (IPCW) is a popular method that relies on correct estimation of the probability of censoring to produce consistent estimation, but is an inefficient estimator in its standard form. We introduce sequentially augmented regression (SAR), an adaptation of the Bang and Robins (2005. Doubly robust estimation in missing data and causal inference models. Biometrics 61, 962-972.) method to estimate a complete-data prediction model, adjusting for longitudinal missing at random censoring. In addition, we propose a closely related non-parametric approach using targeted maximum likelihood estimation (TMLE; van der Laan and Rubin, 2006. Targeted maximum likelihood learning. The International Journal of Biostatistics 2 (1), Article 11). We compare IPCW, SAR, and TMLE (implemented parametrically and with Super Learner) through simulation and the above-mentioned case study.

Identifying a set that contains the best dynamic treatment regimes

A dynamic treatment regime (DTR) is a treatment design that seeks to accommodate patient heterogeneity in response to treatment. DTRs can be operationalized by a sequence of decision rules that map patient information to treatment options at specific decision points. The sequential, multiple assignment, randomized trial (SMART) is a trial design that was developed specifically for the purpose of obtaining data that informs the construction of good (i.e. efficacious) decision rules. One of the scientific questions motivating a SMART concerns the comparison of multiple DTRs that are embedded in the design. Typical approaches for identifying the best DTRs involve all possible comparisons between DTRs that are embedded in a SMART, at the cost of greatly reduced power to the extent that the number of embedded DTRs (EDTRs) increase. Here, we propose a method that will enable investigators to use SMART study data more efficiently to identify the set that contains the most efficacious EDTRs. Our method ensures that the true best EDTRs are included in this set with at least a given probability. Simulation results are presented to evaluate the proposed method, and the Extending Treatment Effectiveness of Naltrexone SMART study data are analyzed to illustrate its application.

Estimating the causal effects of cumulative treatment episodes for adolescents using marginal structural models and inverse probability of treatment weighting

BACKGROUND

Substance use treatment is rarely a one-time event for individuals with substance use disorders. Sustained reductions in substance use and its related symptoms may result from multiple treatment episodes.

METHODS

We use a marginal structural model with inverse-probability-of-treatment weighting to estimate the causal effects of cumulative treatment experiences over a period of 9 months on drug use at the end of 1-year among 2870 adolescents receiving care in community-based treatment settings. During the 9 months, adolescents move in and out of outpatient and residential treatment with periods where they only receive biological drug screening (BDS) or no treatment at all. The use of inverse-probability-of-treatment weighting reduces confounding bias due to observed baseline and time-varying measures over the course of follow-up; weights were estimated using generalized boosted models.

RESULTS

Each additional period of treatment (representing at least one day, 1 session, or 1 BDS during the 90 day period between follow-up visits) yielded reductions in average substance use frequency at 1-year relative to no treatment during the 90-day period. For residential treatment it was a 16% decrease (95% CI=-27%, -7%), for outpatient treatment it was a 9% decrease (95% CI=-18%, -0%), and for BDS (with no additional outpatient or residential treatment) it was an 11% decrease (95% CI=-20%, -3%).

CONCLUSIONS

Using robust statistical methods, we find promising (albeit preliminary) evidence that additional periods of outpatient and residential treatment, as well as biological drug screening, lead to reductions in substance use outcomes at one year.

Modeling the impact of hepatitis C viral clearance on end-stage liver disease in an HIV co-infected cohort with targeted maximum likelihood estimation

Despite modern effective HIV treatment, hepatitis C virus (HCV) co-infection is associated with a high risk of progression to end-stage liver disease (ESLD) which has emerged as the primary cause of death in this population. Clinical interest lies in determining the impact of clearance of HCV on risk for ESLD. In this case study, we examine whether HCV clearance affects risk of ESLD using data from the multicenter Canadian Co-infection Cohort Study. Complications in this survival analysis arise from the time-dependent nature of the data, the presence of baseline confounders, loss to follow-up, and confounders that change over time, all of which can obscure the causal effect of interest. Additional challenges included non-censoring variable missingness and event sparsity. In order to efficiently estimate the ESLD-free survival probabilities under a specific history of HCV clearance, we demonstrate the double-robust and semiparametric efficient method of Targeted Maximum Likelihood Estimation (TMLE). Marginal structural models (MSM) can be used to model the effect of viral clearance (expressed as a hazard ratio) on ESLD-free survival and we demonstrate a way to estimate the parameters of a logistic model for the hazard function with TMLE. We show the theoretical derivation of the efficient influence curves for the parameters of two different MSMs and how they can be used to produce variance approximations for parameter estimates. Finally, the data analysis evaluating the impact of HCV on ESLD was undertaken using multiple imputations to account for the non-monotone missing data.

Super learning to hedge against incorrect inference from arbitrary parametric assumptions in marginal structural modeling

OBJECTIVE

Clinical trials are unlikely to ever be launched for many comparative effectiveness research (CER) questions. Inferences from hypothetical randomized trials may however be emulated with marginal structural modeling (MSM) using observational data, but success in adjusting for time-dependent confounding and selection bias typically relies on parametric modeling assumptions. If these assumptions are violated, inferences from MSM may be inaccurate. In this article, we motivate the application of a data-adaptive estimation approach called super learning (SL) to avoid reliance on arbitrary parametric assumptions in CER.

STUDY DESIGN AND SETTING

Using the electronic health records data from adults with new-onset type 2 diabetes, we implemented MSM with inverse probability weighting (IPW) estimation to evaluate the effect of three oral antidiabetic therapies on the worsening of glomerular filtration rate.

RESULTS

Inferences from IPW estimation were noticeably sensitive to the parametric assumptions about the associations between both the exposure and censoring processes and the main suspected source of confounding, that is, time-dependent measurements of hemoglobin A1c. SL was successfully implemented to harness flexible confounding and selection bias adjustment from existing machine learning algorithms.

CONCLUSION

Erroneous IPW inference about clinical effectiveness because of arbitrary and incorrect modeling decisions may be avoided with SL.

Making valid causal inferences from observational data

The ability to make strong causal inferences, based on data derived from outside of the laboratory, is largely restricted to data arising from well-designed randomized control trials. Nonetheless, a number of methods have been developed to improve our ability to make valid causal inferences from data arising from observational studies. In this paper, I review concepts of causation as a background to counterfactual causal ideas; the latter ideas are central to much of current causal theory. Confounding greatly constrains causal inferences in all observational studies. Confounding is a biased measure of effect that results when one or more variables, that are both antecedent to the exposure and associated with the outcome, are differentially distributed between the exposed and non-exposed groups. Historically, the most common approach to control confounding has been multivariable modeling; however, the limitations of this approach are discussed. My suggestions for improving causal inferences include asking better questions (relates to counterfactual ideas and "thought" trials); improving study design through the use of forward projection; and using propensity scores to identify potential confounders and enhance exchangeability, prior to seeing the outcome data. If time-dependent confounders are present (as they are in many longitudinal studies), more-advanced methods such as marginal structural models need to be implemented. Tutorials and examples are cited where possible.

Improved double-robust estimation in missing data and causal inference models

Recently proposed double-robust estimators for a population mean from incomplete data and for a finite number of counterfactual means can have much higher efficiency than the usual double-robust estimators under misspecification of the outcome model. In this paper, we derive a new class of double-robust estimators for the parameters of regression models with incomplete cross-sectional or longitudinal data, and of marginal structural mean models for cross-sectional data with similar efficiency properties. Unlike the recent proposals, our estimators solve outcome regression estimating equations. In a simulation study, the new estimator shows improvements in variance relative to the standard double-robust estimator that are in agreement with those suggested by asymptotic theory.

A propensity score approach to estimating child restraint effectiveness in preventing mortality

Confounding between the child's restraint use and driver behavior can bias restraint effectiveness estimates away from the null if survivable crashes are more common in certain restraint types. Analyzing only fatal crashes may introduce selection bias toward the null because any protective effects of a restraint type will underrepresent children in that restraint. A marginal-structural-model-type estimator suggests a 17% reduction in fatality risk for children aged 2 through 6 in child restraint systems relative to seat belts. This reduction is estimated at 22% when severe misuse of the restraint is excluded.