Quality of age data in patients from developing countries.

Abstract

Age misreporting is common in demographic studies but the prevalence and magnitude of age misreporting in clinical cohorts is unknown. We analysed single-year age distribution and terminal digit preference in cancer patients from developing countries. Age distribution was analysed by plotting a single-year age of 3874 cancer patients from 72 different countries, mainly from the Indian subcontinent and the Middle East, who resided in the UAE at the time of cancer diagnosis. Preference for age ending with digits '0' and '5' was evaluated using Whipple's index (WI), which has value 100 in cohorts without preference. Preference for all 10 terminal digits was expressed as the difference between the found and expected frequencies using Myers blended method and was graphed. Age data quality was low in cancer patients from the Indian subcontinent (WI = 177) and Middle Eastern countries (WI = 113-204). Females of all nationalities supplied better quality of age data (lower WI) than males. Preference for age ending with digits '0' and '5' was found in all populations except the UAE male citizens who did not show preference for terminal digit '0'. Age data quality in this cohort of patients from developing countries was low. Preference for age ending with numbers '0' and '5' is common. In studies conducted in developing countries, age data quality should be analysed as it may bias results and weaken the power of the study.