Validity of chronic drug exposure presumed from repeated patient interviews varied according to drug class Pernelle Noize a,b,c, * , Fabienne Bazin a,b , Antoine Pariente a,b,c , Carole Dufouil d,e,f , Marie-Laure Ancelin g,h , Catherine Helmer i,j , Nicholas Moore a,b,c , Annie Fourrier-R eglat a,b,c a Univ. de Bordeaux, U657, F-33000 Bordeaux, France b INSERM, U657, F-33000 Bordeaux, France c CHU de Bordeaux, Service de Pharmacologie, F-33000 Bordeaux, France d Univ. de Bordeaux, U708, F-33000 Bordeaux, France e INSERM, U708, F-33000 Bordeaux, France f INSERM, CIC-EC7, F-33000 Bordeaux, France g Univ. Montpellier 1, U1061, F-34093 Montpellier, France h INSERM, U1061, F-34093 Montpellier, France i Univ. de Bordeaux, U897, F-33000 Bordeaux, France j INSERM, U897, F-33000 Bordeaux, France Accepted 18 April 2012; Published online 21 July 2012 Abstract Objective: To evaluate the validity of chronic drug exposure presumed from cross-sectional interviews taking reimbursement data as reference. Study Design and Setting: The study concerned 2,985 elderly persons of the French Three-City cohort (1) who were interviewed on current drug use 2 and 4 years after inclusion and (2) whose reimbursement data were obtained from the main health care insurance system. Validity (sensitivity, Se; specificity, Sp; positive predictivevalue, PPV; negative predictivevalue, NPV) of chronic exposure presumed from follow-up interviews was investigated taking two exposure definitions from reimbursements as reference for the period between interviews: at least 80% coverage with and without a maximal time between reimbursements of 60 days. Results: Using 80% coverage as reference, validity of interview data was substantial for cardiovascular and antithrombotic drugs (Se, 85.3e95.4%; Sp, 67.1e97.6%; PPV, 65.9e86.6%; NPV, 93.3e99.3%). For benzodiazepines, nonsteroidal anti-inflammatory drugs, or analgesics, validity was low especially owing to PPVs (15.8e51.4%). Conclusion: Using reported use at cross-sectional interviews as a proxy for chronic exposure between interviews was valid for drugs used regularly but not so for drugs used more irregularly. Ó 2012 Elsevier Inc. All rights reserved. Keywords: Pharmacoepidemiology; Pharmaceutical preparations; Aged; Health insurance reimbursements; Interview; Validity (epidemiology) 1. Introduction In pharmacoepidemiology, determining drug exposure is fundamental, particularly in the framework of etiological studies designed to evaluate the association between a drug exposure and an outcome occurrence [1]. In such studies, errors on exposure measurement may affect point estimates of associations by introducing misclassification bias that may potentially lead to erroneous conclusions [2]. There are several sources of data available for drug exposure assess- ment in pharmacoepidemiological research including patient interviews and computerized databases recording reimburse- ment claims but all are limited in a certain way in their ability to reflect true patient exposure. Whereas reimbursement claims databases remain attractive because they totally avoid memory bias and allow an easier accessibility to large data sets [3], they suffer from a lack of data regarding over-the- counter (OTC) drugs and also that reimbursement does not necessarily mean use. Information obtained through well- conducted patient interviews may be exhaustive with regard to the use of OTC drugs or self-medication [4], yet when Conflict of interest statement: The authors have no conflict of interest to declare. * Corresponding author. Service de Pharmacologie, Universit e Bordeaux Segalen, CHU de Bordeaux, Zone Nord Carreire, B^ atiment 1A, Case 36, 146 rue L eo Saignat, Bordeaux Cedex 33076, France. Tel.: þ33-5-57571560; fax: þ33-5-57574671. E-mail address: pernelle.noize@u-bordeaux2.fr (P. Noize). 0895-4356/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. doi: 10.1016/j.jclinepi.2012.04.009 Journal of Clinical Epidemiology 65 (2012) 1061e1068