Using a pharmacoepidemiological approach to estimate diabetes type 2 prevalence in Portugal

• Published in: Pharmacoepidemiology and drug safety 2006-04, Vol.15 (4), p.269-274
• Main Authors: Duarte-Ramos, Filipa, Cabrita, José
• Format: Article
• Language: English
• Subjects: diabetes type 2; drug-consumption data; pharmacoepidemiology; prevalence
• ISSN: 1053-8569; 1099-1557
• DOI: 10.1002/pds.1186

Purpose To estimate the prevalence of drug‐treated diabetes type 2 in Portugal using drug consumption data and accounting for the proportion of patients treated with combinations of oral hypoglycaemic agents (OHAs). Methods The prevalence of diabetes treated by OHAs was calculated on the basis of: IMS‐Health data for Portugal (2003) and two measures of daily drug intake—defined daily dose (DDD) and prescribed daily dose (PDD), after correcting for the proportion of patients treated with two or more OHAs. The PDD and proportion of patients in combination therapies have been obtained in a descriptive, cross‐sectional national survey, conducted in 2003, by inquiry of 1046 type 2 diabetics countrywide distributed. Results Drug use study: We have studied 1046 type 2 diabetics (539 women, 501 men, 6 sex unkown), with a mean age of 64.5 (SD = 11) years. OHAs were prescribed as monotherapy in 46.8% (489/1046) of the patients and the remaining 557 (53.2%) received from 2 to 4 OHAs concomitantly. Diabetes prevalence: Using DDD as the mean daily intake consumption unit, the prevalence of drug‐treated type 2 diabetes was 4.15%, which dropped to 2.52% when the proportion of OHAs associations was taken into account. Using PDD these values were 4.48% and 2.72%, respectively. Conclusions Since a high proportion of patients are treated with combination of OHAs in clinical practice, it is imperative to account for that confounder in order to improve the accuracy of estimate from drug consumption data. The use of this methodology provided a slight under‐estimation of diabetes prevalence, compared with the National Health Authorities values (3%–5%). Nevertheless we consider this as an efficient tool to estimate drug‐treated diabetes prevalence that should be implemented in a regular way for longitudinal observations, in order to generate signals. Copyright © 2005 John Wiley & Sons, Ltd.