Observational Research Using Propensity Scores

Abstract

In most observational studies, treatments or other "exposures" (in an epidemiologic sense) do not occur at random. Instead, treatments or other such interventions depend on several patient-related and patient-independent characteristics. Such factors, associated with the receipt vs nonreceipt of treatment, may also be-independently-associated with outcomes. Thus, confounding exists making it difficult to ascertain the true association between treatments and outcomes. Propensity scores (PS) represent an intuitive set of approaches to reduce the influence of such "confounding" factors. PS is a computed probability of treatment, a value that is estimated for each patient in an observational study and then applied (in a variety of ways such as matching, stratification, weighting, etc.) to reduce distortion in the true nature of the association between treatment (or any similar exposure) and outcomes. Despite several advantages, PS-based methods cannot account for unmeasured confounding, ie, for factors that are not being included in the computation of PS. Published by Elsevier Inc. on behalf of the National Kidney Foundation, Inc.

Hyperkalemia and its Association With Mortality, Cardiovascular Events, Hospitalizations, and Intensive Care Unit Admissions in a Population-Based Retrospective Cohort

“…First is the use of propensity score matching, a statistical method that attempts to account for observable factors that may affect the likelihood of a patient having an outcome. 30 In our cohort, we were able to match 88,541 individuals of 93,667 initially identified hyperkalemic episodes, which resulted in groups that were comparable in baseline characteristics. Therefore, the exclusion of those who could not be suitably matched resulted in only a small reduction in overall sample.…”

Drug interactions in hospital prescriptions in Denmark: Prevalence and associations with adverse outcomes

“…The preference score is the probability that a patient be exposed whether this happened or not. Thus, if two patients have (almost) the same preference score but one was exposed and the other not, the exposure is a likely explanation for their difference in outcome 26–28 …”

“…Thus, if two patients have (almost) the same preference score but one was exposed and the other not, the exposure is a likely explanation for their difference in outcome. 26 , 27 , 28 …”

“…Thus, if two patients have (almost) the same preference score but one was exposed and the other not, the exposure is a likely explanation for their difference in outcome. [26][27][28] We used Cyclops 29 to compute high-dimensional propensity scores 27 with sparse lasso logistic regression models using up to 843 fea- version, continuous ! 1 feature).…”

Identification of technology spillover among airport alliance from the perspective of efficiency evaluation: The case of China

“…It is known that propensity score matching (PSM) method is popular in observational studies for uncovering the true association between treatments and outcomes (Raghunathan et al, 2016). However, due to of sample size constraints (especially the scale of allied airports), the PSM method is not suitable for small-sized samples.…”