Instrumental variables (IV) are used to draw causal conclusions about the effect of exposure E on outcome Y in the presence of unmeasured confounders. IV assumptions have been well described: 1) IV affects E; 2) IV affects Y only through E; 3) IV shares no common cause with Y. Even when these assumptions are met, biased effect estimates can result if selection bias allows a non-causal path from E to Y. We demonstrate the presence of bias in IV analyses on a sample from a simulated dataset, where selection into the sample was a collider on a non-causal path from E to Y. By applying inverse probability of selection weights we were able to eliminate the selection bias. IV approaches may protect against unmeasured confounding but are not immune from selection bias. Inverse probability of selection weights used with IV approaches can minimize bias.

Automated algorithms may facilitate identification of patients and/or providers most likely to need more intensive screening and/or intervention for nonmedical opioid use. Additional implementation research in real-world settings would clarify their utility.

BackgroundRestricted mean survival time (RMST) is an underutilized estimand in time-to-event analyses. Herein, we highlight its strengths by comparing time to (1) all-cause mortality and (2) initiation of antiretroviral therapy (ART) for HIV-infected persons who inject drugs (PWID) and persons who do not inject drugs.MethodsRMST to death was determined by integrating the Kaplan-Meier survival curve to 5 years of follow-up. To account for the competing risks of death and loss-to-clinic when estimating time to ART, we calculated RMST to ART initiation by estimating the area between the survival curve for ART initiation and the cumulative incidence curve for death or loss-to-clinic. We standardized all curves using inverse probability of exposure weights.ResultsWe followed 3044 HIV-positive, ART-naive persons from enrollment into the Johns Hopkins HIV Clinical Cohort from 1996 to 2014. PWID had a − 0.19 year (95% confidence interval (CI): − 0.29, − 0.10) difference in survival over 5 years of follow-up compared to persons who did not inject drugs. There was no difference between the two groups in time not on ART while alive and in clinic (RMST difference = 0.08, 95% CI: -0.10, 0.36).ConclusionsPWID have similar expected time to ART initiation after properly accounting for their greater risk of death and loss-to-clinic.Electronic supplementary materialThe online version of this article (10.1186/s12874-018-0484-z) contains supplementary material, which is available to authorized users.

The MCID values were similar for the revision and primary lumbar fusion groups, even when subgroup analysis was done for different diagnostic etiologies, simplifying interpretation of clinical improvement. The results of this study further validate the use of patient-reported HRQOLs to measure clinical effectiveness, as a patient's previous experience with care does not seem to substantially alter an individual's perception of clinical improvement.