Comparison of methods for estimating the attributable risk in the context of survival analysis

Abstract

BackgroundThe attributable risk (AR) measures the proportion of disease cases that can be attributed to an exposure in the population. Several definitions and estimation methods have been proposed for survival data.MethodsUsing simulations, we compared four methods for estimating AR defined in terms of survival functions: two nonparametric methods based on Kaplan-Meier’s estimator, one semiparametric based on Cox’s model, and one parametric based on the piecewise constant hazards model, as well as one simpler method based on estimated exposure prevalence at baseline and Cox’s model hazard ratio. We considered a fixed binary exposure with varying exposure probabilities and strengths of association, and generated event times from a proportional hazards model with constant or monotonic (decreasing or increasing) Weibull baseline hazard, as well as from a nonproportional hazards model. We simulated 1,000 independent samples of size 1,000 or 10,000. The methods were compared in terms of mean bias, mean estimated standard error, empirical standard deviation and 95% confidence interval coverage probability at four equally spaced time points.ResultsUnder proportional hazards, all five methods yielded unbiased results regardless of sample size. Nonparametric methods displayed greater variability than other approaches. All methods showed satisfactory coverage except for nonparametric methods at the end of follow-up for a sample size of 1,000 especially. With nonproportional hazards, nonparametric methods yielded similar results to those under proportional hazards, whereas semiparametric and parametric approaches that both relied on the proportional hazards assumption performed poorly. These methods were applied to estimate the AR of breast cancer due to menopausal hormone therapy in 38,359 women of the E3N cohort.ConclusionIn practice, our study suggests to use the semiparametric or parametric approaches to estimate AR as a function of time in cohort studies if the proportional hazards assumption appears appropriate.Electronic supplementary materialThe online version of this article (doi:10.1186/s12874-016-0285-1) contains supplementary material, which is available to authorized users.

Prevalence and prognostic impact of non‐cardiac co‐morbidities in heart failure outpatients with preserved and reduced ejection fraction: a community‐based study

“…absence of the co‐morbidity). Using the crude survival function S( t ) = P(T > t) and the standardized one S(t) adj = S(T > t|Z = z*), the PAF for time‐to‐event outcomes can be written as follows: …”

Risk Factors Associated With Atrioventricular Block

“…The population-attributable risk of incident AV block was calculated for modifiable risk factors found to have a statistically significant association with incident AV block using a semiparametric approach. 20,21 Specifically, the ratio of the mean excess risk associated with the exposure of interest to the mean observed risk was calculated. Reference levels for systolic blood pressure (120 mm Hg) and fasting glucose level (100 mg/dL) were chosen according to recent guidelines, 22,23 and 95% CIs for these population-attributable risk estimates were obtained using bootstrap resampling with 500 repetitions.…”

Population attributable fractions of the main type 2 diabetes mellitus risk factors in women: Findings from the French E3N cohort

“…The prevalence of the exposition and hazard ratios (HRs) of T2DM risk associated with the exposition and adjusted for potential confounding factors were taken into account in estimating the PAFs. The CIs were estimated using the multivariable delta method . The percentage of T2DM cases that could have been prevented if all women were in the lowest risk group was indicated by a positive PAF; conversely, a negative PAF is the percentage of T2DM cases that could have additionally occurred.…”

“…The CIs were estimated using the multivariable delta method. 24,25 The percentage of T2DM cases that could have been prevented if all women were in the lowest risk group was indicated by a positive PAF; conversely, a negative PAF is the percentage of T2DM cases that could have additionally occurred.…”