LAUSR

: LAN.005 Getting High Quality Data to Drive Programs: How is the Quality of the Data Collection System Associated with the Quality of Routine Health Data in Malawi? R. O’Hagan, M. Marx, K. Finnegan, P. Naphini, K. Ng’ambi, K. Laija, E. Wilson, L. Park, S. Wachepa, J. Smith, L. Gombwa, A. Misomali, T. Mleme, S. Yosefe; Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA, Johns Hopkins Bloomberg School of Public Health, Baltimore, USA, Johns Hopkins University, Baltimore, Maryland, USA, Ministry of Health, Malawi, Lilongwe, Malawi, Ministry of Economic Planning and Development, Lilongwe, Malawi, National Statistical Office, Zomba, Malawi, National Statistical Office, Zomba, Malawi, Johns Hopkins Bloomberg School of Public Health, Lilongwe, Malawi Background: Routine data can be a rich source of information for health systems. However, the perceived and actual quality of routine health data in lowand middle-income countries hinders its use for policy and programming. We conducted a data quality assessment (DQA) with the aim of characterizing the quality of routine data in Malawi’s health system and identifying associated systems-level factors. Methods: The DQA was led by the Central Monitoring and Evaluation Division of the Ministry of Health of Malawi. It was conducted in 15 randomly selected districts, stratified by zone. The sample included 16 hospitals, 90 randomly selected health centres, and 16 district health offices (DHOs), including one district with two DHOs. Registers, monthly reports, and computerized records were reviewed for five service areas: antenatal care (ANC4), family planning, HIV testing and counseling (HTC), and acute respiratory infection (ARI) and pneumonia diagnosis. Interviews were conducted with facility and district personnel to assess current Health Management Information System (HMIS) functioning. Data quality was characterized within four domains: availability; completeness; consistency; and validity. Analysis of variance and multiple linear regression were used to measure the association between data quality and facility and DHO performance in HMIS functional areas. Findings: Data quality varied across service areas; median verification ratios, comparing register and report totals, ranged from 0.78 [IQR 0.25 e 1.07] for ARI to 1.00 [IQR 0.96-1.00] for HTC. Procedures required by Malawi’s HMIS policy are not implemented at many facilities: only 60% of facilities report receiving a documented supervisory visit for HMIS in the six months preceding the assessment. Adherence to data quality practices is low, with a mean score of 0.51 out of 1.00 [SD 0.30]. Half of facilities have a full-time statistical clerk; however, employment of statistical clerks at facilities is not significantly associated with the availability or completeness of data. Interpretation: These findings can guide improvements in Malawi’s HMIS, including increased awareness of and adherence to existing policies. The associations between systems-level factors and data quality can inform efforts to strengthen HMIS in other LMICs. Source of Funding: Funding was provided by Global Affairs Canada, the World Health Organization, Save the Children, and the Supporting Service Delivery Integration (SSDI) project.