Assessing a computerized routine health information system in Mali using LQAS

Background: Between 1987 and 1998 Save the Children conducted a child survi val programme in Mali with the goal of reducing maternal and child morbidit y and mortality. An integral part of this programme was a computerized demo graphic surveillance and health information system (HIS) that gathered data on individuals on an on-going basis. Objective: To assess the overall coverage and quality of the data in the HI S, to identify specific health districts that needed improvements in data c ollection methods, and to determine particular areas of weakness in data co llection. Methods: Random samples of 20 mothers with children <5 years were selected in each of 14 health districts. Mothers were interviewed about pregnancies, live births, deaths of children <5, and children's growth monitoring and i mmunization status. The Lot Quality Assurance Method (LQAS) was used to ide ntify districts in which records and interview results did not meet predete rmined levels of acceptability. Data collected in the interviews were combi ned to estimate overall coverage and quality. Results: When all variables were analyzed, all 14 lots were rejected, and i t was estimated that 52% of all events occurring in the community were regi stered in ProMIS. Much of this poor performance was due to immunization and growth monitoring data, which were not updated due to printer problems. Co verage of events increased (92%) when immunizations and growth monitoring w ere excluded, and no lots were rejected. When all variables were analyzed f or quality of data recorded, six lots were rejected and the overall estimat ion was 83%. With immunizations and growth monitoring excluded, overall qua lity was 86% and no lots were rejected. Conclusions: The comprehensive computerized HIS did not meet expectations. This may be due, in part, to the ambitious objective of complete and intens ive monitoring of a large population without adequate staff and equipment. Future efforts should consider employing a more targeted and streamlined HI S so that data can be more complete and useful.