An algorithm to identify patients with type 2 diabetes among undocumented migrants using data on drug dispensation by charities.

Abstract

Background. Electronic databases of chronic diseases are available in many countries. Health data of residents, natives and documented migrants, are thus easily accessible. This does not happen for the growing population of undocumented migrants.

Methods. We analysed the databases of two Italian non-governmental organisations (NGOs) containing the records of all drug dispensations to 12,386 undocumented migrants from January 1^st, 2013 to December 31^st, 2016, with the aim to identify treated for type 2 diabetes (T2D) on the basis of demographic data and dispensed medicines.

Medications were identified according to the Anatomical Chemical Therapeutic (ATC) classification. All the patients with at least one dispensation per year of any A10 (antidiabetic) drug were selected. An algorithm to match this observation with the diagnosis of type 2 diabetes mellitus (T2D), on the basis of demographic data and use or not of insulin, was developed. The algorithm was validated in 400 diabetic and 400 non-diabetic patients randomly selected.

Results. The algorithm correctly identified all patients (N=660) with T2D. When our patients were grouped according to ethnicity, we found that all ethnic groups contributed a comparable percentage of patients with T2D. Also, no difference was seen between the group of EU citizens living in poverty cared for by the NGOs and any of the ethnic groups.

Conclusions. This algorithm can be used to identify patients treated for T2D when no diagnostic codes are available, as is frequently the case with undocumented migrants. Therefore it can be useful for many aspects of public health.