Data quality of reported age-sex structure from a community diagnosis program of a hilly region of Nepal

Abstract

Introduction: Indicators of health are extensively based on the age-sex data and thus warrants a detailed assessment of data. The main purpose of this study is to assess and then adjust the various types of bias and error that can occur during any census or survey. 
 Method: This study is based on a cross-sectional de jure census of Ugrachandi Nala VDC, Kavre, Nepal, which was conducted during August 30 – September 5, 2005. Digit and gender preference is analysed using Whipple Index Myers Index and, Age-Sex Accuracy Indices. The data is adjusted using Beer’s Method and Hill Technique. Microsoft Excel 2003 and SPSS 11.5 programs were used for data entry and analysis respectively. 
 Result: The Whipple Indices revealed that the single year data as rough and Myers Index confirmed that 0 and 5 are the most preferred digits. Age-Sex Accuracy Indices disclosed the quniquennial age data as highly inaccurate. Single-year age data adjusted using Beer’s Method and quinquennial age data adjusted using Hills Technique are found to be an ideal age-sex structure. 
 Conclusion: Although the reported age shows serious problems of digit preference and age displacement in Ugrachandi Nala VDC, its data quality is better than the Nepal and Kavre district. This study shows that bias and errors related to the reported age and gender in a census and survey can be identified and adjusted using various demographic tools. Thus, it is suggested to use adjusted age-sex structure for calculating demographic, epidemiological and other health indicators.