Inverse Probability Weighting Enhances Absolute Risk Estimation in Three Common Study Designs of Nosocomial Infections

Abstract

Purpose When studying nosocomial infections, resource-efficient sampling designs such as nested case-control, case-cohort, and point prevalence studies are preferred. However, standard analyses of these study designs can introduce selection bias, especially when interested in absolute rates and risks. Moreover, nosocomial infection studies are often subject to competing risks. We aim to demonstrate in this tutorial how to address these challenges for all three study designs using simple weighting techniques. Patients and Methods We discuss the study designs and explain how inverse probability weights (IPW) are applied to obtain unbiased hazard ratios (HR), odds ratios and cumulative incidences. We illustrate these methods in a multi-state framework using a dataset from a nosocomial infections study (n = 2286) in Moscow, Russia. Results Including IPW in the analysis corrects the unweighted naïve analyses and enables the estimation of absolute risks. Resulting estimates are close to the full cohort estimates using substantially smaller numbers of patients. Conclusion IPW is a powerful tool to account for the unequal selection of controls in case-cohort, nested case-control and point prevalence studies. Findings can be generalized to the full population and absolute risks can be estimated. When applied to a multi-state model, competing risks are also taken into account.

Validation of the prevalence to incidence conversion method for healthcare associated infections in long-term care facilities

“…Our primary hypothesis for this study was that in the LTCF setting, length-biased sampling seen in acute-care may occur in the opposite direction: residents developing HAIs may be under-represented in PPS, due to a greater chance of being transferred to acute-care facilities or of dying, leading to underestimated HAI prevalence estimates [ 12 ]. Comparing HAI incidence obtained from H4LS with the estimate calculated from HAI prevalence from the pilot PPS, no significant difference was found, supporting the validity of the ECDC method.…”

“…When applying the corrected formula, we still found HAI incidence remained underestimated by 50%. Another solution could be applying inverse probability weighting, as proposed for correcting length-biased sampling in PPS of HAIs in acute-care hospitals [ 12 ]. Further, Doerken et al .…”

“…When applying the corrected formula, we still found HAI incidence remained underestimated by 50%. Another solution could be applying inverse probability weighting, as proposed for correcting length-biased sampling in PPS of HAIs in acute-care hospitals [12]. Further, Doerken et al have previously suggested that PPS of HAIs in acute-care hospitals should also collect some follow-up data from included patients, namely any incident HAIs and discharge dates [17].…”

“…This method requires a certain degree of approximation due to the cross-sectional nature of PPS: in particular, residents are assumed to permanently reside in LTCFs, and HAI duration is assessed by multiplying by 2 the median duration from day of onset to the day of the PPS for each specific HAI type [ 5 ]. Further, in the assessment of HAIs in acute-care hospitals, selection bias can occur due to cases being over-represented: in fact, patients developing HAIs usually have longer lengths of stays compared to controls, and therefore have an increased probability of being sampled in PPS, leading to an overestimation of HAI prevalence [ 12 , 13 ]. Our hypothesis was that in LTCFs, the opposite may occur: residents developing HAIs either may have a greater chance of being transferred to acute-care facilities or of dying, and therefore could be under-represented in PPS, leading to an underestimation of HAI prevalence ( Fig 1 ).…”