Misspecification of confounder-exposure and confounder-outcome associations leads to bias in effect estimates.

BACKGROUND: Confounding is a common issue in epidemiological research. Commonly used confounder-adjustment methods include multivariable regression analysis and propensity score methods. Although it is common practice to assess the linearity assumption for the exposure-outcome effect, most researchers do not assess linearity of the relationship between the confounder and the exposure and between the confounder and the outcome before adjusting for the confounder in the analysis. Failing to take the true non-linear functional form of the confounder-exposure and confounder-outcome associations into account may result in an under- or overestimation of the true exposure effect. Therefore, this paper aims to demonstrate the importance of assessing the linearity assumption for confounder-exposure and confounder-outcome associations and the importance of correctly specifying these associations when the linearity assumption is violated. METHODS: A Monte Carlo simulation study was used to assess and compare the performance of confounder-adjustment methods when the functional form of the confounder-exposure and confounder-outcome associations were misspecified (i.e., linearity was wrongly assumed) and correctly specified (i.e., linearity was rightly assumed) under multiple sample sizes. An empirical data example was used to illustrate that the misspecification of confounder-exposure and confounder-outcome associations leads to bias. RESULTS: The simulation study illustrated that the exposure effect estimate will be biased when for propensity score (PS) methods the confounder-exposure association is misspecified. For methods in which the outcome is regressed on the confounder or the PS, the exposure effect estimate will be biased if the confounder-outcome association is misspecified. In the empirical data example, correct specification of the confounder-exposure and confounder-outcome associations resulted in smaller exposure effect estimates. CONCLUSION: When attempting to remove bias by adjusting for confounding, misspecification of the confounder-exposure and confounder-outcome associations might actually introduce bias. It is therefore important that researchers not only assess the linearity of the exposure-outcome effect, but also of the confounder-exposure or confounder-outcome associations depending on the confounder-adjustment method used.

Propensity score methods in observational research: brief review and guide for authors.

Causal inference in observational research requires a careful approach to adjustment for confounding. One such approach is the use of propensity score analyses. In this editorial, we focus on the role of propensity score-based methods in estimating causal effects from non-randomised observational data. We highlight the details, assumptions, and limitations of these methods and provide authors with guidelines for the conduct and reporting of propensity score analyses.

The effect of the stented iliac lesions TASC-II C, D on the femoropopliteal bypass patency: Propensity score-matched observational study.

INTRODUCTION: Concurrent stenting of complex iliac lesions during infrainguinal bypasses can increase the complexity of a case and impact outcomes. OBJECTIVE: Our aim was to evaluate the effect of inflow stenting of TASC-II C, D iliac lesions on femoropopliteal bypass patency. METHODS: A retrospective observational cohort study of patients who underwent femoropopliteal bypass with TASC-II C, D iliac artery stenting (hybrid group) or without inflow lesions (non-hybrid group) was conducted. After propensity score matching, 120 patients were included in the non-hybrid group and 60 patients in the hybrid one. The median follow-up was 432 (193; 1313) days in the hybrid group and 472 (196; 1376) days in the non-hybrid group (p = 0.94). RESULTS: No significant differences were found between the groups in 30-day morbidity and serious adverse events. At 3 years, primary and secondary bypass patency for the hybrid group and non-hybrid group were 62.2% versus 59.9% (p = 0.36) and 63.7% versus 64.3% (p = 0.077), respectively. The primary patency of the iliac stents in patients of the hybrid group was 95% at 3 years. The estimated hazard ratio for primary patency for hybrid versus non-hybrid was 0.77, with 90% CI: 0.50-1.21; the noninferiority upper bound being 1.31, which corresponds to a 10% additive noninferiority margin for probabilities. The 3 years of freedom from amputation in patients with chronic limb-threatening ischemia was 94.1% and 75.0% in the hybrid and non-hybrid groups, respectively (p = 0.09). CONCLUSION: The outcomes of the femoropopliteal bypass in hybrid surgery supplemented with stenting of TASC-II C, D iliac lesions was similar to femoropopliteal bypass with intact inflow arteries.

Factors associated with psychiatric admission and subsequent self-harm repetition: a cohort study of high-risk hospital-presenting self-harm.

BACKGROUND: Individuals presenting to hospital with self-harm of high lethality or high suicidal intent are at high risk of subsequent suicide. AIM: To examine factors associated with psychiatric admission and self-harm repetition following high-risk self-harm (HRSH). METHOD: A cohort study of 324 consecutive HRSH patients was conducted across three urban hospitals (December 2014-February 2018). Information on self-harm repetition was extracted from the National Self-harm Registry Ireland. Logistic regression models examined predictors of psychiatric admission and self-harm repetition. Propensity score (PS) methods were used to address confounding. RESULTS: Forty percent of the cohort were admitted to a psychiatric inpatient setting. Factors associated with admission were living alone, depression, previous psychiatric admission, suicide note and uncommon self-harm methods. History of emotional, physical or sexual abuse was associated with not being admitted. Twelve-month self-harm repetition occurred in 17.3% of cases. Following inverse probability weighting according to the PS, psychiatric admission following HRSH was not associated with repetition. Predictors of repetition were recent self-harm history, young age (18-24 years) and previous psychiatric admission. CONCLUSION(S): Findings indicate that psychiatric admission following HRSH is not associated with repeated self-harm and reaffirms the consistent finding that history of self-harm and psychiatric treatment are strong predictors of repetition.

Use of Electronic Cigarettes to Aid Long-Term Smoking Cessation in the United States: Prospective Evidence From the PATH Cohort Study.

Electronic cigarettes (e-cigarettes) are the preferred smoking-cessation aid in the United States; however, there is little evidence regarding long-term effectiveness among those who use them. We used the Population Assessment of Tobacco and Health Study to compare long-term abstinence between matched US smokers who tried to quit with and without use of e-cigarettes as a cessation aid. We identified a nationally representative cohort of 2,535 adult US smokers in 2014-2015 (baseline assessment), who, in 2015-2016 (exposure assessment), reported a past-year attempt to quit and the cessation aids used, and reported smoking status in 2016-2017 (outcome assessment; self-reported ≥12 months continuous abstinence). We used propensity-score methods to match each e-cigarette user with similar nonusers. Among US smokers who used e-cigarettes to help quit, 12.9% (95% confidence interval (CI): 9.1%, 16.7%) successfully attained long-term abstinence. However, there was no difference compared with matched non-e-cigarette users (cigarette abstinence difference: 2%; 95% CI: -3%, 7%). Furthermore, fewer e-cigarette users were abstinent from nicotine products in the long term (nicotine abstinence difference: -4%; 95% CI: -7%, -1%); approximately two-thirds of e-cigarette users who successfully quit smoking continued to use e-cigarettes. These results suggest e-cigarettes may not be an effective cessation aid for adult smokers and, instead, may contribute to continuing nicotine dependence.

Population-based long-term outcomes of open versus endovascular aortic repair of ruptured abdominal aortic aneurysms.

OBJECTIVE: Existing data regarding endovascular aortic repair (EVAR) of ruptured abdominal aortic aneurysm (rAAA) are conflicting in their findings. The purpose of this paper was to determine the long-term outcomes of EVAR vs open surgical repair (OSR) for treatment of rAAA. METHODS: A population-based retrospective cohort study of all patients 40 years or more that underwent OSR or EVAR of rAAA in Ontario, Canada, from 2003 to 2016 was conducted. Administrative data from the province of Ontario was used as the data source. The propensity for repair approach was calculated using a logistic regression model including all covariates and used for inverse probability of treatment weighting. Cox proportional hazards regression was conducted using the weighted cohort to determine the survival and major adverse cardiovascular event (MACE)-free survival of EVAR relative to OSR for rAAA up to 10 years after repair. RESULTS: A total of 2692 rAAA (261 EVAR [10%] and 2431 OSR [90%]) repairs were recorded from April 1, 2003, to March 31, 2016. Mean follow-up for the entire cohort was 3.4 years (standard deviation [SD], 3.9 years), with a maximum follow-up of 14.0 years. OSR patients were followed for a mean of 3.5 years (SD, 4.0 years) and maximum of 14.0 years, and EVAR patients were followed for a mean of 2.7 years (SD, 2.7 years) and a maximum of 11.4 years. Median survival was 2.7 years overall, and 2.5 and 3.7 years for OSR and EVAR patients, respectively. There were no significant baseline differences between EVAR and OSR patients after inverse probability of treatment weighting. EVAR patients were at lower hazard for all-cause mortality (hazard ratio, 0.49; 95% confidence interval, 0.37-0.65; P < .01), and MACE (hazard ratio, 0.51, 95% confidence interval, 0.40-0.66; P < .01) within 30 days of repair. There were no statistically significant differences between EVAR and OSR in the hazard for all-cause mortality or MACE from 30 days to 5 years, and 5 to 10 years. Despite this, the upfront mortality and MACE benefits of EVAR persisted for more than 4.5 years after repair. CONCLUSIONS: This population-based cohort study using administrative data from Ontario, Canada, demonstrated lower hazards for all-cause mortality and MACE within 30 days of operation in favor of EVAR, but no differences in the mid- or longer-term results. More work is needed to understand and improve the long-term outcomes of ruptured endovascular aortic aneurysm repair and ruptured open surgical repair.

A comparison of machine learning algorithms and covariate balance measures for propensity score matching and weighting.

Propensity score matching (PSM) and propensity score weighting (PSW) are popular tools to estimate causal effects in observational studies. We address two open issues: how to estimate propensity scores and assess covariate balance. Using simulations, we compare the performance of PSM and PSW based on logistic regression and machine learning algorithms (CART; Bagging; Boosting; Random Forest; Neural Networks; naive Bayes). Additionally, we consider several measures of covariate balance (Absolute Standardized Average Mean (ASAM) with and without interactions; measures based on the quantile-quantile plots; ratio between variances of propensity scores; area under the curve (AUC)) and assess their ability in predicting the bias of PSM and PSW estimators. We also investigate the importance of tuning of machine learning parameters in the context of propensity score methods. Two simulation designs are employed. In the first, the generating processes are inspired to birth register data used to assess the effect of labor induction on the occurrence of caesarean section. The second exploits more general generating mechanisms. Overall, among the different techniques, random forests performed the best, especially in PSW. Logistic regression and neural networks also showed an excellent performance similar to that of random forests. As for covariate balance, the simplest and commonly used metric, the ASAM, showed a strong correlation with the bias of causal effects estimators. Our findings suggest that researchers should aim at obtaining an ASAM lower than 10% for as many variables as possible. In the empirical study we found that labor induction had a small and not statistically significant impact on caesarean section.

Investigating the impact of the economic crisis on children's wellbeing in four European countries.

This paper analyses the impact of the economic crisis on children's wellbeing from a comparative European perspective using a multidimensional and fuzzy methodology. Comparisons of children's wellbeing based solely on monetary resources are likely to give a partial picture, because wellbeing can be related to different capabilities and the economic crisis impacted them differently. The use of several indicators captures the multidimensional and interrelated nature of wellbeing as experienced by children and can give a comprehensive view of the threats of future generations. A propensity score method is used to compare children's capabilities distributions and to assess the impact of the economic crisis on the net change between 2009 and 2014 using EU-SILC data for four European countries. Results show that the economic crisis had a wide range of consequences for children's wellbeing: a disparity is observed across the four countries in preserving the same level of wellbeing after the economic crisis.

How Do Propensity Score Methods Measure Up in the Presence of Measurement Error? A Monte Carlo Study.

Considering that the absence of measurement error in research is a rare phenomenon and its effects can be dramatic, we examine the impact of measurement error on propensity score (PS) analysis used to minimize selection bias in behavioral and social observational studies. A Monte Carlo study was conducted to explore the effects of measurement error on the treatment effect and balance estimates in PS analysis across seven different PS conditioning methods. In general, the results indicate that even low levels of measurement error in the covariates lead to substantial bias in estimates of treatment effects and concomitant reduction in confidence interval coverage across all methods of conditioning on the PS.

Changing patterns of use of osteoporosis medications in the years after launch: implications for comparative effectiveness research.

PURPOSE: Newly marketed medications may be used selectively in patients with more severe disease. Changes in patterns of use following a drug's introduction to the market can greatly influence results from non-experimental comparative effectiveness research. We sought to explore this issue by characterizing trends in oral and injectable prescription drug claims for the prevention and treatment of osteoporosis. METHODS: We examined a post-menopausal population of women age 55 years and older in the Truven Health Analytics MarketScan® Databases. We used propensity score (PS) methods to describe how predictors of new users of oral and injectable osteoporosis medications change over time. RESULTS: We found that injectable osteoporosis medications tended to be used more selectively in the higher risk patients shortly after launch. Over time, they appeared to be used increasingly in lower risk patients. CONCLUSION: If disease severity is incompletely captured in the data, comparative effectiveness of novel osteoporosis medications may be difficult to accurately estimate, particularly when medications are new to market.