Summary. Randomized controlled trials (RCTs) can provide unbiased estimates of sample average treatment effects. However, a common concern is that RCTs may fail to provide unbiased estimates of population average treatment effects. We derive the assumptions that are required to identify population average treatment effects from RCTs. We provide placebo tests, which formally follow from the identifying assumptions and can assess whether they hold. We offer new research designs for estimating population effects that use non-randomized studies to adjust the RCT data. This approach is considered in a cost-effectiveness analysis of a clinical intervention: pulmonary artery catheterization.

Recommended Citation:Radice, Rosalba; Ramsahai, Roland; Grieve, Richard; Kreif, Noemi; Sadique, Zia; and Sekhon, Jasjeet S. (2012) AbstractPropensity score (Pscore) matching and inverse probability of treatment weighting (IPTW) can remove bias due to observed confounders, if the Pscore is correctly specified. Genetic Matching (GenMatch) matches on the Pscore and individual covariates using an automated search algorithm to balance covariates. This paper compares common ways of implementing Pscore matching and IPTW, with Genmatch for balancing time-constant baseline covariates}. The methods are considered when estimates of treatment effectiveness are required for patient subgroups, and the treatment allocation process differs by subgroup. We apply these methods in a prospective cohort study that estimates the effectiveness of Drotrecogin alfa activated, for subgroups of patients with severe sepsis. In a simulation study we compare the methods when the Pscore is correctly specified, and then misspecified by ignoring the subgroup-specific treatment allocation. The simulations also consider poor overlap in baseline covariates, and different sample sizes. In the case study, GenMatch reports better covariate balance than IPTW or Pscore matching. In the simulations with correctly specified Pscores, good overlap and reasonable sample sizes, all methods report minimal bias. When the Pscore is misspecified, GenMatch reports the least imbalance and bias. With small sample sizes, IPTW is the most efficient approach, but all methods report relatively high bias of treatment effects. This study shows that overall GenMatch achieves the best covariate balance for each subgroup, and is more robust to Pscore misspecification than common alternative Pscore approaches.

Instrumental variables can be used to make inferences about causal effects in the presence of unmeasured confounding. For a model in which the instrument, intermediate/treatment, and outcome variables are all binary, Balke and Pearl (1997, Journal of the American Statistical Association 92: 1172-1176) derived nonparametric bounds for the intervention probabilities and the average causal effect. We have implemented these bounds in two commands: bpbounds and bpboundsi. We have also implemented several extensions to these bounds. One of these extensions applies when the instrument and outcome are measured in one sample and the instrument and intermediate are measured in another sample. We have also implemented the bounds for an instrument with three categories, as is common in Mendelian randomization analyses in epidemiology and for the case where a monotonic effect of the instrument on the intermediate can be assumed. In each case, we calculate the instrumental-variable inequality constraints as a check for gross violations of the instrumental-variable conditions. The use of the commands is illustrated with a recreation of the original Balke and Pearl analysis and with a Mendelian randomization analysis. We also give a simulated example to demonstrate that the instrumental-variable inequality constraints can both detect and fail to detect violations of the instrumental-variable conditions.

The instrumental variables (IV) method is a method for making causal inferences about the effect of a treatment based on an observational study in which there are unmeasured confounding variables. The method requires a valid IV, a variable that is independent of the unmeasured confounding variables and is associated with the treatment but which has no effect on the outcome beyond its effect on the treatment. An additional assumption that is often made for the IV method is deterministic monotonicity, which is an assumption that for each subject, the level of the treatment that a subject would take if given a level of the IV is a monotonic increasing function of the level of the IV. Under deterministic monotonicity, the IV method identifies the average treatment effect for the compliers (those subject who would take the treatment if encouraged to do so by the IV and not take the treatment if not encouraged). However, deterministic monotonicity is sometimes not realistic. We introduce a stochastic monotonicity condition which relaxes deterministic monotonicity in that it does not require that a monotonic increasing relationship hold within subjects between the levels of the IV and the level of the treatment that the subject would take if given a level of the IV, but only that a monotonic increasing relationship hold across subjects between the IV and the treatment in a certain manner. We show that under stochastic monotonicity, the IV method identifies a weighted average of treatment effects with greater weight on subgroups of subjects on whom the IV has a stronger effect. We provide bounds on the global average treatment effect under stochastic monotonicity and a sensitivity analysis for violations of the stochastic monotonicity assumption. We apply the methods of IV inference under stochastic monotonicity that we develop to a study of the effect of premature babies being delivered in a high volume, high technology neonatal intensive care unit (NICU) vs. a lower level unit.

Decision makers require cost-effectiveness estimates for patient subgroups. In nonrandomized studies, propensity score (PS) matching and inverse probability of treatment weighting (IPTW) can address overt selection bias, but only if they balance observed covariates between treatment groups. Genetic matching (GM) matches on the PS and individual covariates using an automated search algorithm to directly balance baseline covariates. This article compares these methods for estimating subgroup effects in cost-effectiveness analyses (CEA). The motivating case study is a CEA of a pharmaceutical intervention, drotrecogin alfa (DrotAA), for patient subgroups with severe sepsis (n = 2726). Here, GM reported better covariate balance than PS matching and IPTW. For the subgroup at a high level of baseline risk, the probability that DrotAA was cost-effective ranged from 30% (IPTW) to 90% (PS matching and GM), at a threshold of £20 000 per quality-adjusted life-year. We then compared the methods in a simulation study, in which initially the PS was correctly specified and then misspecified, for example, by ignoring the subgroup-specific treatment assignment. Relative performance was assessed as bias and root mean squared error (RMSE) in the estimated incremental net benefits. When the PS was correctly specified and inverse probability weights were stable, each method performed well; IPTW reported the lowest RMSE. When the subgroup-specific treatment assignment was ignored, PS matching and IPTW reported covariate imbalance and bias; GM reported better balance, less bias, and more precise estimates. We conclude that if the PS is correctly specified and the weights for IPTW are stable, each method can provide unbiased cost-effectiveness estimates. However, unlike IPTW and PS matching, GM is relatively robust to PS misspecification.