Dora Dadzie1, Richard Okyere Boadu2, Cyril Mark Engmann3,4,5, Nana Amma Yeboaa Twum-Danso6,7. 1. Cape Coast Teaching Hospital, Cape Coast, Ghana. 2. Department of Health Information Management, University of Cape Coast, Cape Coast, Ghana. 3. Maternal, Newborn and Child Health and Nutrition, PATH, Seattle, WA, United States of America. 4. Department of Paediatrics, University of Washington School of Medicine, Seattle, WA, United States of America. 5. Department of Global Health, University of Washington School of Public Health, Seattle, WA, United States of America. 6. TD Health, Accra, Ghana. 7. Gillings School of Global Public Health, University of North Carolina, Chapel Hill, NC, United States of America.
Abstract
BACKGROUND: Cause-specific mortality data are required to set interventions to reduce neonatal mortality. However, in many developing countries, these data are either lacking or of low quality. We assessed the completeness and accuracy of cause of death (COD) data for neonates in Ghana to assess their usability for monitoring the effectiveness of health system interventions aimed at improving neonatal survival. METHODS: A lot quality assurance sampling survey was conducted in 20 hospitals in the public sector across four regions of Ghana. Institutional neonatal deaths (IND) occurring from 2014 through 2017 were divided into lots, defined as neonatal deaths occurring in a selected facility in a calendar year. A total of 52 eligible lots were selected: 10 from Ashanti region, and 14 each from Brong Ahafo, Eastern and Volta region. Nine lots were from 2014, 11 from 2015 and 16 each were from 2016 and 2017. The cause of death (COD) of 20 IND per lot were abstracted from admission and discharge (A&D) registers and validated against the COD recorded in death certificates, clinician's notes or neonatal death audit reports for consistency. With the error threshold set at 5%, ≥ 17 correctly matched diagnoses in a sample of 20 deaths would make the lot accurate for COD diagnosis. Completeness of COD data was measured by calculating the proportion of IND that had death certificates completed. RESULTS: Nineteen out of 52 eligible (36.5%) lots had accurate COD diagnoses recorded in their A&D registers. The regional distribution of lots with accurate COD data is as follows: Ashanti (4, 21.2%), Brong Ahafo (7, 36.8%), Eastern (4, 21.1%) and Volta (4, 21.1%). Majority (9, 47.4%) of lots with accurate data were from 2016, followed by 2015 and 2017 with four (21.1%) lots. Two (10.5%) lots had accurate COD data in 2014. Only 22% (239/1040) of sampled IND had completed death certificates. CONCLUSION: Death certificates were not reliably completed for IND in a sample of health facilities in Ghana from 2014 through 2017. The accuracy of cause-specific mortality data recorded in A&D registers was also below the desired target. Thus, recorded IND data in public sector health facilities in Ghana are not valid enough for decision-making or planning. Periodic data quality assessments can determine the magnitude of the data quality concerns and guide site-specific improvements in mortality data management.
BACKGROUND: Cause-specific mortality data are required to set interventions to reduce neonatal mortality. However, in many developing countries, these data are either lacking or of low quality. We assessed the completeness and accuracy of cause of death (COD) data for neonates in Ghana to assess their usability for monitoring the effectiveness of health system interventions aimed at improving neonatal survival. METHODS: A lot quality assurance sampling survey was conducted in 20 hospitals in the public sector across four regions of Ghana. Institutional neonatal deaths (IND) occurring from 2014 through 2017 were divided into lots, defined as neonatal deaths occurring in a selected facility in a calendar year. A total of 52 eligible lots were selected: 10 from Ashanti region, and 14 each from Brong Ahafo, Eastern and Volta region. Nine lots were from 2014, 11 from 2015 and 16 each were from 2016 and 2017. The cause of death (COD) of 20 IND per lot were abstracted from admission and discharge (A&D) registers and validated against the COD recorded in death certificates, clinician's notes or neonatal death audit reports for consistency. With the error threshold set at 5%, ≥ 17 correctly matched diagnoses in a sample of 20 deaths would make the lot accurate for COD diagnosis. Completeness of COD data was measured by calculating the proportion of IND that had death certificates completed. RESULTS: Nineteen out of 52 eligible (36.5%) lots had accurate COD diagnoses recorded in their A&D registers. The regional distribution of lots with accurate COD data is as follows: Ashanti (4, 21.2%), Brong Ahafo (7, 36.8%), Eastern (4, 21.1%) and Volta (4, 21.1%). Majority (9, 47.4%) of lots with accurate data were from 2016, followed by 2015 and 2017 with four (21.1%) lots. Two (10.5%) lots had accurate COD data in 2014. Only 22% (239/1040) of sampled IND had completed death certificates. CONCLUSION:Death certificates were not reliably completed for IND in a sample of health facilities in Ghana from 2014 through 2017. The accuracy of cause-specific mortality data recorded in A&D registers was also below the desired target. Thus, recorded IND data in public sector health facilities in Ghana are not valid enough for decision-making or planning. Periodic data quality assessments can determine the magnitude of the data quality concerns and guide site-specific improvements in mortality data management.
Authors: Bethany L Hedt-Gauthier; Lyson Tenthani; Shira Mitchell; Frank M Chimbwandira; Simon Makombe; Zengani Chirwa; Erik J Schouten; Marcello Pagano; Andreas Jahn Journal: BMC Health Serv Res Date: 2012-07-09 Impact factor: 2.655