Study design, setting, population, and data source

This study utilized the longitudinal data from the Hospital Information System (HIS) for a government hospital providing services at primary care level located approximately 250 km North of the largest city in KZN. The hospital oversees 17 fixed clinics, and services a large and predominantly rural sub-district population of over 200,000 mainly Zulu speaking individuals within an area of 1400 km². The general population of the district as well as details about the Health Information System have been described in detail elsewhere^13,14.

The dataset contains demographic information and International Classification of Disease version 10 (ICD-10) codes assigned by healthcare professionals with a clinical and coding background. Information on 24,262 inpatient individuals of all ages between January 2011 and February 19, 2015 were available. Record selection criteria were adults 15 years and above discharged from the hospital prior to February 19, 2014 (to allow minimum one-year observation), with surgical adult patients being excluded from the study. The dataset is de-identified and publicly available, and the study was approved by the Biomedical Research Ethics Committee of the University of KwaZulu-Natal, South Africa.