Dataset Information

Classification of drug use patterns.

ABSTRACT: Characterizing long-term prescription data is challenging due to the time-varying nature of drug use. Conventional approaches summarize time-varying data into categorical variables based on simple measures, such as cumulative dose, while ignoring patterns of use. The loss of information can lead to misclassification and biased estimates of the exposure-outcome association. We introduce a classification method to characterize longitudinal prescription data with an unsupervised machine learning algorithm. We used administrative databases covering virtually all 1.3 million residents of Manitoba and explicitly designed features to describe the average dose, proportion of days covered (PDC), dose change, and dose variability, and clustered the resulting feature space using K-means clustering. We applied this method to metformin use in diabetes patients. We identified 27,786 metformin users and showed that the feature distributions of their metformin use are stable for varying the lengths of follow-up and that these distributions have clear interpretations. We found six distinct metformin user groups: patients with intermittent use, decreasing dose, increasing dose, high dose, and two medium dose groups (one with stable dose and one with highly variable use). Patients in the varying and decreasing dose groups had a higher chance of progression of diabetes than other patients. The method presented in this paper allows for characterization of drug use into distinct and clinically relevant groups in a way that cannot be obtained from merely classifying use by quantiles of overall use.

SUBMITTER: Righolt CH

PROVIDER: S-EPMC7719192 | biostudies-literature | 2020 Dec

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Classification of drug use patterns.

Righolt Christiaan H CH Zhang Geng G Mahmud Salaheddin M SM

Pharmacology research & perspectives 20201201 6

Characterizing long-term prescription data is challenging due to the time-varying nature of drug use. Conventional approaches summarize time-varying data into categorical variables based on simple measures, such as cumulative dose, while ignoring patterns of use. The loss of information can lead to misclassification and biased estimates of the exposure-outcome association. We introduce a classification method to characterize longitudinal prescription data with an unsupervised machine learning al ...[more]

PMID: 33280248

Similar Datasets

Project description:ImportancePatients who have a fragility fracture are at high risk for subsequent fractures. Prescription drugs represent 1 factor that could be modified to reduce the risk of subsequent fracture.ObjectiveTo describe the use of prescription drugs associated with fracture risk before and after fragility fracture.Design, setting, and participantsRetrospective cohort study conducted between February 2015 and March 2016 using a 40% random sample of Medicare beneficiaries from 2007 through 2011 in general communities throughout the United States. A total of 168 133 community-dwelling Medicare beneficiaries who survived a fracture of the hip, shoulder, or wrist were included. Cohort members were required to be enrolled in fee-for-service Medicare with drug coverage (Parts A, B, and D) and to be community dwelling for at least 30 days in the immediate 4-month postfracture period.ExposuresPrescription drug use during the 4-month period before and after a fragility fracture.Main outcomes and measuresPrescription fills for drug classes associated with increased fracture risk were measured using Part D retail pharmacy claims. These were divided into 3 categories: drugs that increase fall risk; drugs that decrease bone density; and drugs with unclear fracture risk mechanism. Drugs that increase bone density were also tracked.ResultsA total of 168 133 patients with a fragility fracture (141 569 women; 84.2%) met the inclusion criteria for this study; 91.8% were white. Across all fracture types, the mean (SD) age was 80.0 (7.7) years, and 53.2% of the fracture cohort was hospitalized at the time of the index fracture, although this varied significantly depending on fracture type (100% of hip fractures, 8.2% of wrist fractures, and 15.0% of shoulder fractures). The frequency of discharge to an institution for rehabilitation following hospitalization also varied by fracture type, but the mean (SD) duration of acute rehabilitation did not: 28.1 (19.8) days. Most patients were exposed to at least 1 nonopiate drug associated with increased fracture risk in the 4 months before fracture (77.1% of hip, 74.1% of wrist, and 75.9% of shoulder fractures). Approximately 7% of these patients discontinued this drug exposure after the fracture, but this was offset by new users after fracture. Consequently, the proportion of the cohort exposed following fracture was unchanged (80.5%, 74.3%, and 76.9% for hip, wrist, and shoulder, respectively). There was no change in the average number of fracture-associated drugs used. This same pattern of use before and after fracture was observed across all 3 drug mechanism categories. Use of drugs to strengthen bone density was uncommon (≤25%) both before and after fracture.Conclusions and relevanceExposure to prescription drugs associated with fracture risk is infrequently reduced following fragility fracture occurrence. While some patients eliminate their exposure to drugs associated with fracture, an equal number initiate new high-risk drugs. This pattern suggests there is a missed opportunity to modify at least one factor contributing to secondary fractures.

Dataset Information

Classification of drug use patterns.

Publications

Classification of drug use patterns.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets