Refining estimates of prescription durations by using observed covariates in pharmacoepidemiological databases: an application of the reverse waiting time distribution

Abstract

The algorithm allows estimation of prescription durations based on the reverse WTD, which can depend upon observed covariates. Statistical uncertainty intervals and tests allow statistical inference on the influence of observed patient and prescription characteristics. The method may replace ad hoc decision rules. Copyright © 2017 John Wiley & Sons, Ltd.

Use of vitamin K antagonists and risk of prostate cancer: Meta–analysis and nationwide case–control study

“…Furthermore, we modeled exposure as ever–use (at least 1 filled VKA prescription) and, to evaluate any dose–response relationship, as an ordinal variable according to cumulative duration of use (< 1, 1–3, 3–5, 5–10 and > 10 years). To define the duration assigned to each prescription fill, we fitted a reverse waiting time distribution (rWTD) model for warfarin and phenprocoumon prescriptions filled in 2005 adjusting for age and number of pills redeemed (100, 200, 300+) . If the next prescription for VKAs occurred within the duration defined by the rWTD model, we assumed that the treatment episode had continued.…”

Using probability of drug use as independent variable in a register‐based pharmacoepidemiological cause‐effect study—An application of the reverse waiting time distribution

“…Based on theory for renewal processes, we have recently proposed a method whereby the probability of (still) being treated can be modeled as a function of the distance from the latest prescription . We have extended the method to also include multivariable modeling such that the probability function may depend on covariates, such as age, sex, package size, coprescribed medication, and comorbidity …”

Refining estimates of prescription durations by using observed covariates in pharmacoepidemiologic databases: Necessary refinements to stimulate alternative approaches

“…This is, however, rather important whenever quantitative and temporal drug exposure is of relevance for pharmacoepidemiological research. We read the paper by Støvring and colleagues with interest, addressing the timely topic of accurate drug exposure estimation. Methodologically, the authors previously described the waiting time distribution (WTD) to derive prescription durations from a distributional pattern in the total sample (i.e., aggregated data) .…”

“…Methodologically, the authors previously described the waiting time distribution (WTD) to derive prescription durations from a distributional pattern in the total sample (i.e., aggregated data) . Now, the authors introduce the reverse WTD as the distribution of times from each patient's last prescription within a time window to the end of the time window . Distribution percentiles at a certain cutoff can be used to derive the typical prescription duration.…”

“…To explore the potential of Støvring's idea, we adapted our recent modeling and simulation framework of our comparative method evaluation . Although the distribution of administered daily doses and their ratio to the dispensed dose unit are still based on primary data from an observational cohort, we introduce random variables to be related to this measure and can therefore be used to predict “true” doses in situations of only single redemptions.…”