Methodological comparison of marginal structural model, time-varying Cox regression, and propensity score methods: the example of antidepressant use and the risk of hip fracture

Long-term PM1 exposure and hypertension hospitalization: A causal inference study on a large community-based cohort in South China

Limited evidence exists on the effect of submicronic particulate matter (PM1) on hypertension hospitalization. Evidence based on causal inference and large cohorts is even more scarce. In 2015, 36,271 participants were enrolled in South China and followed up through 2020. Each participant was assigned single-year, lag0-1, and lag0-2 moving average concentration of PM1 and fine inhalable particulate matter (PM2.5) simulated based on satellite data at a 1-km resolution. We used an inverse probability weighting approach to balance confounders and utilized a marginal structural Cox model to evaluate the underlying causal links between PM1 exposure and hypertension hospitalization, with PM2.5-hypertension association for comparison. Several sensitivity studies and the analyses of effect modification were also conducted. We found that a higher hospitalization risk from both overall (HR: 1.13, 95% CI: 1.05-1.22) and essential hypertension (HR: 1.15, 95% CI: 1.06-1.25) was linked to each 1 µg/m3 increase in the yearly average PM1 concentration. At lag0-1 and lag0-2, we observed a 17%-21% higher risk of hypertension associated with PM1. The effect of PM1 was 6%-11% higher compared with PM2.5. Linear concentration-exposure associations between PM1 exposure and hypertension were identified, without safety thresholds. Women and participants that engaged in physical exercise exhibited higher susceptibility, with 4%-22% greater risk than their counterparts. This large cohort study identified a detrimental relationship between chronic PM1 exposure and hypertension hospitalization, which was more pronounced compared with PM2.5 and among certain groups.

Handling time-varying treatments in observational studies: A scoping review and recommendations

OBJECTIVE

Time-varying treatments are common in observational studies. However, when assessing treatment effects, the methodological framework has not been systematically established for handling time-varying treatments. This study aimed to examine the current methods for dealing with time-varying treatments in observational studies and developed practical recommendations.

METHODS

We searched PubMed from 2000 to 2021 for methodological articles about time-varying treatments, and qualitatively summarized the current methods for handling time-varying treatments. Subsequently, we developed practical recommendations through interactive internal group discussions and consensus by a panel of external experts.

RESULTS

Of the 36 eligible reports (22 methodological reviews, 10 original studies, 2 tutorials and 2 commentaries), most examined statistical methods for time-varying treatments, and only a few discussed the overarching methodological process. Generally, there were three methodological components to handle time-varying treatments. These included the specification of treatment which may be categorized as three scenarios (i.e., time-independent treatment, static treatment regime, or dynamic treatment regime); definition of treatment status which could involve three approaches (i.e., intention-to-treat, per-protocol, or as-treated approach); and selection of analytic methods. Based on the review results, a methodological workflow and a set of practical recommendations were proposed through two consensus meetings.

CONCLUSIONS

There is no consensus process for assessing treatment effects in observational studies with time-varying treatments. Previous efforts were dedicated to developing statistical methods. Our study proposed a stepwise workflow with practical recommendations to assist the practice.

Metabolically healthy obese individuals are still at high risk for diabetes: Application of the marginal structural model

AIM

To assess the effect of obesity phenotype on the incidence of diabetes, considering phenotype as a time-varying exposure.

METHODS

We used community-based cohort data from the Korean Genome and Epidemiology Study, with a 16-year follow-up period. Obesity phenotype was determined using body mass index and metabolic syndrome criteria. The influence of obesity phenotype on the occurrence of diabetes was evaluated using a Cox proportional hazard model and a marginal structural model (MSM).

RESULTS

Obesity phenotypes were defined in 6265 individuals, with diabetes identified in 903 (14.4%) during the follow-up period. Individuals with metabolically healthy obesity (MHO) exhibited a higher risk of diabetes compared to those with metabolically healthy normal weight (MHNW), with a hazard ratio (HR) of 1.48 (95% confidence interval [CI] 1.15-1.90). This association remained significant after applying the MSM (HR 1.49, 95% CI 1.01-2.20). Moreover, various sensitivity analyses consistently demonstrated a higher risk of diabetes in individuals with MHO compared to those with MHNW.

CONCLUSIONS

Even when obesity phenotype was treated as a time-varying exposure, individuals with MHO were still at higher risk for developing diabetes than those with MHNW. Consequently, such individuals should aim to avoid transitioning to a metabolically unfavourable state and strive to reduce their body weight to a normal range.

Rheumatoid Arthritis Disease Activity and Hospitalized Infection in a Large US Registry

OBJECTIVES

The association between disease activity and infection risk among patients with rheumatoid arthritis (RA) is not clear, and it is challenging to determine because of confounding due to the effects of RA treatments and comorbidities.

METHODS

Using patients with RA in the CorEvitas registry with Medicare coverage in 2006-2019, we identified eligible patients who had at least 1 visit with moderate disease activity based on the Clinical Disease Activity Index (CDAI; CDAI >10 and ≤22). Follow-up started at the subsequent CorEvitas visit. Hospitalized infection during follow-up was assessed in linked Medicare data. We calculated the incidence rate of hospitalized infection for patients in remission, and low and moderate disease activity, and estimated the effect of time-varying CDAI on hospitalized infection by controlling for baseline and time-dependent confounders using marginal structural models (MSMs).

RESULTS

A total of 3,254 patients with RA were eligible for analysis, among which 529 hospitalized infections were identified during follow-up. The crude incidence of hospitalized infection was 3.8 per 100 person-years for patients in remission, 6.6 for low disease activity, and 8.0 for moderate disease activity. Using MSMs and compared with being in remission, the hazard ratio of hospitalized infection associated with low disease activity was 1.60 (95% confidence interval [95% CI] 1.13-2.28) and with moderate disease activity was 1.83 (95% CI 1.30-2.64).

CONCLUSION

The risk of hospitalized infection was higher for patients with RA in low or moderate disease activity than for those in remission after accounting for the interplay of disease activity, RA treatments, treatment switching, and other potential confounders.

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.

Antidepressants and Vertebral and Hip Risk Fracture: An Updated Systematic Review and Meta-Analysis

Although antidepressant drugs appear to play an active role in increasing fracture risk, their weight is still unclear. We conducted a PRISMA compliant systematic review and meta-analysis through PubMed/Scopus/Cochrane libraries and registered with PROSPERO (registration number CRD42021254006) to investigate the relationship between antidepressant drugs categories, including SSRIs, SNRIs, and TCAs, and the risk of hip and vertebral fractures. After screening 3122 items, we finally found 26 papers for qualitative analysis and 11 for quantitative synthesis. A total of 15,209,542 adult and elderly patients were identified, with a mean follow-up of 51 months and a major prevalence of women. We identified results largely for SSRIs, with only a small amount of data for SNRIs, TCAs, and NaSSA. No data were found among the most recent categories of antidepressants, such as vortioxetine and esketamine. All included studies reported hip fractures, while three of them also included vertebral fractures. Overall, we observed a significant effect of SSRIs on fracture risk with a mean effect of 0.98 (95% CI = 0.75–1.20). This meta-analysis reveals that the use of SSRIs increases the risk of fractures. Clinicians’ awareness in antidepressant prescription should optimize their potential while reducing this risk.

Sulfonylurea and Cancer Risk Among Patients With Type 2 Diabetes: A Population-Based Cohort Study

BACKGROUND

Current evidence of the association between the use of sulfonylurea and cancer risk is highly conflicting and little evidence of this association is from the mainland Chinese population. This study aimed to evaluate the potential effects of sulfonylurea use on cancer risk among patients with type 2 diabetes mellitus (T2DM).

METHODS

A retrospective cohort study of T2DM patients who were new users of sulfonylurea or metformin was conducted using the Yinzhou Regional Health Care Database. A marginal structural Cox model was used to estimate the hazard ratio (HR) of cancer associated with the use of sulfonylurea compared with metformin, with time-varying confounders controlled by inverse probability weighting. Secondary analyses using different glucose-lowering drugs (GLDs) as comparator and sensitivity analyses for potential bias due to latency period, model misspecification, missing data, analyses strategy (intention-to-treat and per-protocol), and diagnosis validation were performed to examine the robustness of the results.

RESULTS

After fully controlling for time-varying confounding, baseline confounding, and competing risk, the use of sulfonylurea was not associated with the risk of any cancer (HR 1.09; 95% CI, 0.93-1.27), compared with the use of metformin. In the secondary analyses, compared with α - glucosidase inhibitors, thiazolidinediones, glinides, other GLDs except sulfonylure and insulin, and T2DM patients not treated with sulfonylureas, the HRs of the association between sulfonylurea use and cancer risk were 0.92 (95% CI; 0.78-1.08), 0.89 (95% CI; 0.66-1.19), 0.85 (95% CI; 0.71-1.02), 1.04 (95% CI; 0.89-1.22), and 1.07 (95% CI; 0.99-1.16), respectively. The results of analyses for various subgroups, risk of site-specific cancers, cumulative duration, dose-response relationship, and sensitivity analyses of different latency periods and missing data were generally consistent with the findings of the primary analyses.

CONCLUSION

No association between sulfonylurea use and cancer risk was found in this study after properly controlling biases due to time-varying confounders and other sources. Further studies on the association between sulfonylurea use and the risk of cancer by using data from a Chinese population with higher representativeness are needed.

Effects of statins and steroids on coronary artery disease and stroke in patients with interstitial lung disease and pulmonary fibrosis: A general population study

Purpose

To determine the effects of statins and steroids on the risk of coronary artery disease (CAD) and stroke in patients with interstitial lung disease and pulmonary fibrosis (ILD-PF).

Methods

We retrospectively enrolled patients with ILD-PF who were using statins (statin cohort, N = 11,567) and not using statins (nonstatin cohort, N = 26,159). Cox proportional regression was performed to analyze the cumulative incidence of CAD and stroke. Adjusted hazard ratios (aHRs) and 95% confidence intervals (CIs) of CAD and stroke were determined after sex, age, and comorbidities, as well as the use of inhaler corticosteroids (ICSs), oral steroids (OSs), and statins, were controlled for.

Results

Compared with those of patients without statin use, the aHRs (95% CIs) of patients with statin use for CAD and ischemic stroke were 0.72 (0.65–0.79) and 0.52 (0.38–0.72), respectively. For patients taking single-use statins but not ICSs/OSs, the aHRs (95% CIs) for CAD and ischemic stroke were 0.72 (0.65–0.79)/0.69 (0.61–0.79) and 0.54 (0.39–0.74)/0.50 (0.32–0.79), respectively. For patients using ICSs/OSs, the aHRs (95% CIs) for CAD and ischemic stroke were 0.71 (0.42–1.18)/0.74 (0.64–0.85) and 0.23 (0.03–1.59)/0.54 (0.35–0.85), respectively.

Conclusions

The findings demonstrate that statin use, either alone or in combination with OS use, plays an auxiliary role in the management of CAD and ischemic stroke in patients with ILD-PF.

Bias of time-varying exposure effects due to time-varying covariate measurement strategies

Purpose

In studies of effects of time‐varying drug exposures, adequate adjustment for time‐varying covariates is often necessary to properly control for confounding. However, the granularity of the available covariate data may not be sufficiently fine, for example when covariates are measured for participants only when their exposure levels change.

Methods

To illustrate the impact of choices regarding the frequency of measuring time‐varying covariates, we simulated data for a large target trial and for large observational studies, varying in covariate measurement design. Covariates were measured never, on a fixed‐interval basis, or each time the exposure level switched. For the analysis, it was assumed that covariates remain constant in periods of no measurement. Cumulative survival probabilities for continuous exposure and non‐exposure were estimated using inverse probability weighting to adjust for time‐varying confounding, with special emphasis on the difference between 5‐year event risks.

Results

With monthly covariate measurements, estimates based on observational data coincided with trial‐based estimates, with 5‐year risk differences being zero. Without measurement of baseline or post‐baseline covariates, this risk difference was estimated to be 49% based on the available observational data. With measurements on a fixed‐interval basis only, 5‐year risk differences deviated from the null, to 29% for 6‐monthly measurements, and with magnitude increasing up to 35% as the interval length increased. Risk difference estimates diverged from the null to as low as −18% when covariates were measured depending on exposure level switching.

Conclusion

Our simulations highlight the need for careful consideration of time‐varying covariates in designing studies on time‐varying exposures. We caution against implementing designs with long intervals between measurements. The maximum length required will depend on the rates at which treatments and covariates change, with higher rates requiring shorter measurement intervals.

Associations of Endometriosis and Hormone Therapy With Risk of Hyperlipidemia

Using claims data from the universal health insurance program of Taiwan, we conducted a retrospective cohort study to investigate whether endometriosis and hormone therapy are associated with the risk of developing hyperlipidemia. We selected 9,155 women aged 20-55 years with endometriosis diagnosed during the period 2000-2013 and 212,641 women without endometriosis with a median follow-up time of 7 years. Among patients with endometriosis, 86% of cases were identified on the basis of diagnosis codes with an ultrasound claim, and 14% were defined by diagnostic laparoscopy or surgical treatments. In a Cox proportional hazards model, the adjusted hazard ratio was 1.30 (95% confidence interval (CI): 1.19, 1.41) for all women, 1.04 (95% CI: 0.81, 1.32) for women under 35 years of age, 1.17 (95% CI: 1.03, 1.32) for women aged 35-44 years, and 1.34 (95% CI: 1.18, 1.52) for women aged 45-54 years. Hysterectomy and/or bilateral oophorectomy accounted for 46.9% of the association between endometriosis and hyperlipidemia, and hormone therapy accounted for 21.6%. Among women with endometriosis, the marginal structural model approach adjusting for time-varying hysterectomy/bilateral oophorectomy showed no association between use of hormone medications and risk of hyperlipidemia. We concluded that women with endometriosis are at increased risk of hyperlipidemia; use of hormone therapy by these women was not independently associated with the development of hyperlipidemia.

Quantitative Bias Analysis for a Misclassified Confounder: A Comparison Between Marginal Structural Models and Conditional Models for Point Treatments

Observational data are increasingly used with the aim of estimating causal effects of treatments, through careful control for confounding. Marginal structural models estimated using inverse probability weighting (MSMs-IPW), like other methods to control for confounding, assume that confounding variables are measured without error. The average treatment effect in an MSM-IPW may however be biased when a confounding variable is error prone. Using the potential outcome framework, we derive expressions for the bias due to confounder misclassification in analyses that aim to estimate the average treatment effect using an marginal structural model estimated using inverse probability weighting (MSM-IPW). We compare this bias with the bias due to confounder misclassification in analyses based on a conditional regression model. Focus is on a point-treatment study with a continuous outcome. Compared with bias in the average treatment effect in a conditional model, the bias in an MSM-IPW can be different in magnitude but is equal in sign. Also, we use a simulation study to investigate the finite sample performance of MSM-IPW and conditional models when a confounding variable is misclassified. Simulation results indicate that confidence intervals of the treatment effect obtained from MSM-IPW are generally wider, and coverage of the true treatment effect is higher compared with a conditional model, ranging from overcoverage if there is no confounder misclassification to undercoverage when there is confounder misclassification. Further, we illustrate in a study of blood pressure-lowering therapy, how the bias expressions can be used to inform a quantitative bias analysis to study the impact of confounder misclassification, supported by an online tool.

Dietary patterns related to cardiovascular disease based on reduced rank regression analysis of healthy middle-aged Koreans: data from the community-based Korean Genome and Epidemiology Study (KoGES) cohort

BACKGROUND

Dietary patterns (DPs) provide a comprehensive picture of the foods consumed by an individual.

OBJECTIVES

Using 12-y follow-up data from the Korean Genome Epidemiology Study (KoGES), we determined the associations of DPs with incident cardiovascular disease (CVD) using reduced rank regression (RRR).

METHODS

This study analyzed the data of 7354 CVD-free subjects aged 40-69 y drawn from the community-based KoGES cohort. Based on the daily intake of 26 food groups at baseline, we identified DPs based on retinol, vitamin B-2 (riboflavin), and vitamin B-3 (niacin) intakes using RRR. The effects of the DPs on incident CVD were assessed using HRs with 95% CIs. Furthermore, using a marginal structural model, the association between DPs and incident CVD was evaluated after adjusting for time-varying confounders.

RESULTS

The incidence of CVD during the follow-up period was 3.7 per 1000 person-years (n = 274). The identified DP accounted for 28.99% of the variation in the response variables (i.e., the intake amounts of all 3 nutrients) and was characterized by high intakes of eggs, fish, milk, and dairy products. The effect of DP quintile on incident CVD differed by sex (Pinteraction = 0.03); the highest DP quintile was associated with a protective effect against the development of CVD in women (HR: 0.44; 95% CI: 0.22, 0.89), but not in men (HR: 1.57; 95% CI: 0.82, 3.00), compared with the lowest quintile. Even after adjusting for time-dependent variables, the effect of DP on incident CVD was significant in women (HR: 0.43; 95% CI: 0.22, 0.84), but not in men (HR: 1.49; 95% CI: 0.71, 3.10).

CONCLUSIONS

In this study, we identified DPs related to CVD, and a DP characterized by high intakes of eggs, fish, milk, and dairy products protected against incident CVD in women.

Fitting Marginal Structural and G-Estimation Models Under Complex Treatment Patterns: Investigating the Association Between De Novo Vitamin D Supplement Use After Breast Cancer Diagnosis and All-Cause Mortality Using Linked Pharmacy Claim and Registry Data

Studies have shown that accounting for time-varying confounding through time-dependent Cox proportional hazards models may provide biased estimates of the causal effect of treatment when the confounder is also a mediator. We explore 2 alternative approaches to addressing this problem while examining the association between vitamin D supplementation initiated after breast cancer diagnosis and all-cause mortality. Women aged 50-80 years were identified in the National Cancer Registry Ireland (n = 5,417) between 2001 and 2011. Vitamin D use was identified from linked prescription data (n = 2,570). We sought to account for the time-varying nature of vitamin D use and time-varying confounding by bisphosphonate use using 1) marginal structural models (MSMs) and 2) G-estimation of structural nested accelerated failure-time models (SNAFTMs). Using standard adjusted Cox proportional hazards models, we found a reduction in all-cause mortality in de novo vitamin D users compared with nonusers (hazard ratio (HR) = 0.84, 95% confidence interval (CI): 0.73, 0.99). Additional adjustment for vitamin D and bisphosphonate use in the previous month reduced the hazard ratio (HR = 0.45, 95% CI: 0.33, 0.63). Results derived from MSMs (HR = 0.44, 95% CI: 0.32, 0.61) and SNAFTMs (HR = 0.45, 95% CI: 0.34, 0.52) were similar. Utilizing MSMs and SNAFTMs to account for time-varying bisphosphonate use did not alter conclusions in this example.

The impact of high frequency oscillatory ventilation on mortality in paediatric acute respiratory distress syndrome

BACKGROUND

High-frequency oscillatory ventilation (HFOV) use was associated with greater mortality in adult acute respiratory distress syndrome (ARDS). Nevertheless, HFOV is still frequently used as rescue therapy in paediatric acute respiratory distress syndrome (PARDS). In view of the limited evidence for HFOV in PARDS and evidence demonstrating harm in adult patients with ARDS, we hypothesized that HFOV use compared to other modes of mechanical ventilation is associated with increased mortality in PARDS.

METHODS

Patients with PARDS from 10 paediatric intensive care units across Asia from 2009 to 2015 were identified. Data on epidemiology and clinical outcomes were collected. Patients on HFOV were compared to patients on other modes of ventilation. The primary outcome was 28-day mortality and secondary outcomes were 28-day ventilator- (VFD) and intensive care unit- (IFD) free days. Genetic matching (GM) method was used to analyse the association between HFOV treatment with the primary outcome. Additionally, we performed a sensitivity analysis, including propensity score (PS) matching, inverse probability of treatment weighting (IPTW) and marginal structural modelling (MSM) to estimate the treatment effect.

RESULTS

A total of 328 patients were included. In the first 7 days of PARDS, 122/328 (37.2%) patients were supported with HFOV. There were significant differences in baseline oxygenation index (OI) between the HFOV and non-HFOV groups (18.8 [12.0, 30.2] vs. 7.7 [5.1, 13.1] respectively; p < 0.001). A total of 118 pairs were matched in the GM method which found a significant association between HFOV with 28-day mortality in PARDS [odds ratio 2.3, 95% confidence interval (CI) 1.3, 4.4, p value 0.01]. VFD was indifferent between the HFOV and non-HFOV group [mean difference - 1.3 (95%CI - 3.4, 0.9); p = 0.29] but IFD was significantly lower in the HFOV group [- 2.5 (95%CI - 4.9, - 0.5); p = 0.03]. From the sensitivity analysis, PS matching, IPTW and MSM all showed consistent direction of HFOV treatment effect in PARDS.

CONCLUSION

The use of HFOV was associated with increased 28-day mortality in PARDS. This study suggests caution but does not eliminate equivocality and a randomized controlled trial is justified to examine the true association.

Long-Term Physical Activity and Subsequent Risk for Rheumatoid Arthritis Among Women: A Prospective Cohort Study

OBJECTIVE

To evaluate the effects of long-term physical activity on subsequent risk of rheumatoid arthritis (RA) in a prospective cohort study.

METHODS

This study investigated physical activity and RA risk among women from the Nurses' Health Study II (1989-2015). Physical activity exposures and covariates were prospectively obtained using biennial questionnaires. Two rheumatologists independently reviewed the medical records of women who self-reported a new diagnosis of RA on biennial questionnaires and who screened positive for RA based on a supplemental survey. All incident RA cases met the 1987 American College of Rheumatology (ACR) or 2010 ACR/European League Against Rheumatism (EULAR) classification criteria for RA. The primary analysis investigated the long-term cumulative average number of hours spent in recreational physical activity 2-8 years prior to the RA diagnosis, a time span chosen to reduce the potential for reverse causation bias, since early RA affects physical activity prior to diagnosis. Estimated Cox regression hazard ratios (HRs) with 95% confidence intervals (95% CIs) were used to assess the risk of RA serologic phenotypes (all, seropositive, or seronegative) in relation to physical activity categories. The analyses were adjusted for body mass index (BMI) at age 18 years and time-varying potential confounders, and the mediating effect of updated BMI on the interaction between physical activity and RA risk was quantified.

RESULTS

Among the 113,366 women analyzed, 506 incident RA cases (67.0% with seropositive RA) were identified during 2,428,573 person-years of follow-up. After adjustment for confounders, including smoking, dietary quality, and BMI at age 18 years, increasing cumulative average total hours of recreational physical activity was associated with a reduced risk of RA, as follows: HR 1.00 for <1 hour/week (reference), HR 1.00 (95% CI 0.78-1.29) for 1 to <2 hours/week, HR 0.92 (95% CI 0.72-1.17) for 2 to <4 hours/week, HR 0.84 (95% CI 0.63-1.12) for 4 to <7 hours/week, and HR 0.67 (95% CI 0.47-0.98) for ≥7 hours/week (P for trend = 0.02). The proportion of the effect between physical activity and RA mediated by updated BMI was 14.0% (P = 0.002) for all RA and 20.0% (P = 0.001) for seropositive RA.

CONCLUSION

Higher levels of physical activity were associated with reduced RA risk. These results add to the literature implicating metabolic factors in the pathogenesis of RA.

Combined effect of posttraumatic stress disorder and prescription opioid use on risk of cardiovascular disease

AIM

Prescription opioid analgesic use (OAU) is associated with increased risk of cardiovascular disease (CVD). OAU is more common in patients with than without posttraumatic stress disorder (PTSD), and PTSD is associated with higher CVD risk. We determined whether PTSD and OAU have an additive or multiplicative association with incident CVD.

METHODS AND RESULTS

Veterans Health Affairs patient medical record data from 2008 to 2015 was used to identify 2861 patients 30-70 years of age, free of cancer, CVD and OAU for 12 months before index date. We defined a four-level exposure variable: 1) no PTSD/no OAU, 2) OAU alone, 3) PTSD alone and 4) PTSD+OAU. Cox proportional hazard models estimated the association between the exposure variable and incident CVD. The mean age was 49.0 (±11.0), 85.7% were male and 58.3% were White, 34.4% had no PTSD/no OAU, 32.9% had PTSD alone, 10.6% had OAU alone, and 22.1% had PTSD+OAU. Compared with patients with no PTSD/no OAU, those with PTSD alone were not at increased risk of incident CVD (hazard ratio = 0.82; 95% confidence interval (CI): 0.63-1.17); however, OAU alone and PTSD+OAU were both significantly associated with incident CVD (hazard ratio = 1.99; 95% CI:1.36-2.92 and hazard ratio = 2.20; 95% CI: 1.61-3.02). There was no significant additive or multiplicative PTSD and OAU association with incident CVD.

CONCLUSION

OAU is associated with nearly a two-fold increased risk of CVD in patients with and without PTSD. Despite no additive or multiplicative interaction effects, the high prevalence of OAU in PTSD may represent a novel contributor to the elevated CVD burden among patients with PTSD.

A marginal structural model approach to analyse work-related injuries: an example using data from the health and retirement study

BACKGROUND

Biases may exist in the limited longitudinal data focusing on work-related injuries among the ageing workforce. Standard statistical techniques may not provide valid estimates when the data are time-varying and when prior exposures and outcomes may influence future outcomes. This research effort uses marginal structural models (MSMs), a class of causal models rarely applied for injury epidemiology research to analyse work-related injuries.

METHODS

7212 working US adults aged ≥50 years, obtained from the Health and Retirement Study sample in the year 2004 formed the study cohort that was followed until 2014. The analyses compared estimates measuring the associations between physical work requirements and work-related injuries using MSMs and a traditional regression model. The weights used in the MSMs, besides accounting for time-varying exposures, also accounted for the recurrent nature of injuries.

RESULTS

The results were consistent with regard to directionality between the two models. However, the effect estimate was greater when the same data were analysed using MSMs, built without the restriction for complete case analyses.

CONCLUSIONS

MSMs can be particularly useful for observational data, especially with the inclusion of recurrent outcomes as these can be incorporated in the weights themselves.

Selective serotonin reuptake inhibitor and selective serotonin and norepinephrine reuptake inhibitor use and risk of fractures in adults: A systematic review and meta-analysis

OBJECTIVE

To evaluate the association between selective serotonin reuptake inhibitor (SSRI) and selective serotonin and norepinephrine reuptake inhibitor (SNRI) use and risk of fractures in older adults.

METHODS

We systematically identified and analyzed observational studies comparing SSRI/SNRI use for depression with non-SSRI/SNRI use with a primary outcome of risk of fractures in older adults. We searched for studies in MEDLINE, PsycINFO, Embase, DARE (Database of Abstracts or Reviews of Effects), the Cochrane Library, and Web of Science clinical trial research registers from 2011 for SSRIs and 1990 for SNRIs to November 29, 2016.

RESULTS

Thirty-three studies met our inclusion criteria; 23 studies were included in meta-analysis: 9 case-control studies and 14 cohort studies. A 1.67-fold increase in the risk of fracture for SSRI users compared with nonusers was observed (relative risk 1.67, 95% CI 1.56-1.79, P = .000). The risk of fracture increases with their long-term use: within 1 year, the risk is 2.9% or 1 additional fracture in every 85 users; within 5 years, the risk is 13.4% or 1 additional fracture in every 19 users. In meta-regression, we found that the increase in risk did not differ across age groups (odds ratio = 1.006; P = .173). A limited number of studies on SNRI use and the risk of fractures prevented us from conducting a meta-analysis.

CONCLUSIONS

Our systematic review showed an association between risk of fracture and the use of SSRIs, especially with increasing use. Age does not increase this risk. No such conclusions can be drawn about the effect of SNRIs on the risk of fracture because of a lack of studies.

Methods for time-varying exposure related problems in pharmacoepidemiology: An overview

Purpose

Lack of control for time‐varying exposures can lead to substantial bias in estimates of treatment effects. The aim of this study is to provide an overview and guidance on some of the available methodologies used to address problems related to time‐varying exposure and confounding in pharmacoepidemiology and other observational studies. The methods are explored from a conceptual rather than an analytical perspective.

Methods

The methods described in this study have been identified exploring the literature concerning to the time‐varying exposure concept and basing the search on four fundamental pharmacoepidemiological problems, construction of treatment episodes, time‐varying confounders, cumulative exposure and latency, and treatment switching.

Results

A correct treatment episodes construction is fundamental to avoid bias in treatment effect estimates. Several methods exist to address time‐varying covariates, but the complexity of the most advanced approaches—eg, marginal structural models or structural nested failure time models—and the lack of user‐friendly statistical packages have prevented broader adoption of these methods. Consequently, simpler methods are most commonly used, including, for example, methods without any adjustment strategy and models with time‐varying covariates. The magnitude of exposure needs to be considered and properly modelled.

Conclusions

Further research on the application and implementation of the most complex methods is needed. Because different methods can lead to substantial differences in the treatment effect estimates, the application of several methods and comparison of the results is recommended. Treatment episodes estimation and exposure quantification are key parts in the estimation of treatment effects or associations of interest.

Pre- and post-diagnostic beta-blocker use and prognosis after colorectal cancer: Results from a population-based study

Recent experimental and epidemiological studies have suggested that beta-blocker use might be associated with better cancer prognosis, but results were inconclusive and only few studies have investigated the association specifically for colorectal cancer (CRC) patients. We investigated this hypothesis using a linked dataset of the Eindhoven area of the Netherlands Cancer Registry and the PHARMO record linkage, including patients diagnosed with CRC between 1998 and 2011. Eligible patients were matched on propensity scores to control for potential confounders such as socio-demographic factors, comorbidity, cancer treatment and use of other medications. Controls were subsequently restricted to active comparators. The association between pre-diagnostic and time-dependent post-diagnostic beta-blocker use and overall survival was estimated using Cox proportional hazard regression models. Subgroup analyses by cancer site and stage and by beta-blocker type were conducted. Of 8,100 CRC patients with a median follow-up of 6.6 years, 1,813 (22%) used beta-blockers prior to diagnosis. In multivariate analysis, we observed no significant association in overall mortality for pre-diagnostic [hazard ratio 1.07, 95% confidence interval (0.96-1.19)] and post-diagnostic [1.10 (0.98-1.23)] beta-blocker use, respectively. Analyses by beta-blocker type, by cancer site, cancer stage and by cumulative dose showed no significant survival improvements for beta-blocker users. However, there was a significant association between cumulative duration of use of 1-12 months and increased overall mortality [1.20 (1.03-1.39)]. Thus, our results do not support the hypothesis of a beneficial effect of pre- or post-diagnostic beta-blocker intake on CRC prognosis, neither for specific patient subgroups nor for specific types of beta-blockers.

Tips and tricks of the propensity score methods in the thoracic surgery research

Without randomization, the differences in the distribution of baseline covariates can confound the evaluation of outcomes between the groups. To estimate the causal effects of treatment on the results randomized clinical trials (RCT) are routinely used. When RCT are not feasible for cost, time, and ethical issues, the effect of treatment on an inevitable outcome could be investigated by using a non-experimental study design. The propensity score (PS) refers to the individual probability, for a subject involved in a study, of receiving a new treatment rather than the control treatment. PS matching is a valuable and increasingly popular tool for dealing with observational data and non-random treatment assignment. Nevertheless, surgeons need to proceed with caution and apply PS methods appropriately.

A population-based study of the comparative effectiveness of second-generation antipsychotics vs older antimanic agents in bipolar disorder

OBJECTIVES

Numerous antimanic treatments have been introduced over the past two decades, particularly second-generation antipsychotics (SGAs). However, it is not clear whether such newer agents provide any advantage over older treatments.

METHODS

A historical cohort design investigated the nationwide population of outpatients with bipolar disorder treated in the Department of Veterans Affairs who were newly initiated on an antimanic agent between 2003 and 2010 (N=27 727). The primary outcome was likelihood of all-cause hospitalization during the year after initiation, controlling for numerous demographic, clinical, and treatment characteristics. Potential correlates of effect were explored by investigating time to initiation of a second antimanic agent or antidepressant.

RESULTS

After control for covariates, those initiated on lithium or valproate monotherapy, compared to those beginning SGA monotherapy, were significantly less likely to be hospitalized, had a longer time to hospitalization, and had fewer hospitalizations in the subsequent year. Those on combination treatment had a significantly higher likelihood of hospitalization, although they also had a longer time to addition of an additional antimanic agent or antidepressant.

CONCLUSIONS

The present analysis of a large and unselected nationwide population provides important complementary data to that from controlled trials. Although various mechanisms may be responsible for the results, the data support the utilization of lithium or valproate, rather than SGAs, as the initial antimanic treatment in bipolar disorder. A large-scale, prospective, randomized, pragmatic clinical trial comparing the initiation of SGA monotherapy to that of lithium or valproate monotherapy is a logical next step.