Menu
Menu
Menu
Menu
Introduction Health Management Information Systems (HMIS) are vital to ensure accountability and for making decisions including for tracking the Sustainable Development Goals. The Ethiopia Health Sector Transformation Plan II includes preventing data falsification as a major strategic initiative and our study aimed to explore the reasons why healthcare providers intentionally falsify maternal and newborn health (MNH) data in two regions of Ethiopia. Methods We conducted a qualitative study in two hospitals, four health centres and their associated health posts in Oromia and Amhara regions. We conducted 45 in-depth interviews with health facility managers, quality improvement (QI) focal persons, health information technicians, MNH care providers, Health Extension Workers and QI mentors. Data were collected in local languages and transcribed in English. During analysis we repeatedly read the transcripts, coded them inductively using NVivo V.12, and categorised the codes into themes. Results Participants were hesitant to report personal data falsification but many reported that falsification is common and that they had experienced it in other facilities or had been told about it by other health workers. Falsification was mostly inflating the number of services provided (such as deliveries). Decreasing the number of deaths or reclassifying neonatal death into stillbirths was also reported. An overarching theme was that the health system focuses on, and rewards, the number of services provided over any other metric. This focus led to both system and individual level incentives for falsification and disincentives for accurate reporting. Conclusion Our finding suggests that to reduce facility level data falsification policy makers might consider disentangling reward and punishments from the performance reports based on the routine HMIS data. Further studies examining the high-level drivers of falsification at regional, national and global levels and effective interventions to address the drivers of data falsification are needed.
We conducted a qualitative study between July and August 2018 in Amhara and Oromia regions of Ethiopia. Oromia and Amhara regions are the most populous in Ethiopia with an estimated 2020 population of 38 and 22 million, respectively. Together the two regions constitute around 60% of the total population of Ethiopia.19 They contribute the largest number of maternal and neonatal deaths of Ethiopia’s 11 regions and rank 3rd and 4th in terms of their neonatal mortality rate. Their neonatal mortality rate is 46 and 39 deaths per 1000 live births for Oromia and Amhara, respectively,20 and their estimated maternal mortality ratio 520 and 369 deaths per 100 000 live births.21 This study was part of a larger qualitative study to explore the functioning of the prototype phase of an MNH quality improvement (QI) initiatives being implemented by the Ministry of Health supported by the Institute of Healthcare Improvement.22 The QI intervention included the formation of learning collaboratives at woreda (district) level, the formation of facility MNH QI teams to plan, implement and monitor QI projects and change ideas and collaborative level learning sessions to share experiences, build clinical skill and develop and review change ideas. QI teams were supported through visits by QI mentors who, among providing other support, validated HMIS data and worked with the team to improve accuracy through trust building, training and feedback.23 From each region we selected one woreda, within which we selected the hospital, a less accessible health centre located at a remote village with access by a rough road and a more accessible health centre located near a main road. Both health centres were ‘typical’ in that they had no unusual additional interventions or staff in place and were in a typical geographic area for the woreda. The two hospitals were better equipped and staffed than the health centres, and two of the health centres had intermittent electricity and one had no running water. One was an upgraded health post and had poor building quality and lacked equipment. Within each facility we conducted in-depth interviews (IDI) with 5–8 participants including the health centre/department manager, the MCH focal person, the health information technicians (HITs), MNH clinical care providers (midwives, nurses and health officer) and HEWs. To be eligible for interviews participants needed to have some experience of the QI project either through attending learning sessions or being part of the facility QI team. We also interviewed three QI mentors. In the majority of cases, participants were identified by the data collectors as facilities were small. When needed, the facility head and the QI mentors helped identify the MCH focal person, the HIT and other who were involved in QI, but they were not directly involved in approaching or recruiting participants. Identified participants were approached at the facility by one of the qualitative interviewers who checked that they were not busy with clinical work and found a private place to take consent and conduct the interview. IDIs were conducted in Amharic or Afaan Oromo languages, and digitally recorded. In total, we interviewed 45 participants. The majority of the participants were females who had worked in the health facilities for 1–4 years. Table 1 summarises the study participants. Background characteristics of the study participants (n=45) *Three missed data. †Excludes woreda/Institute of Healthcare Improvement mentors. ‡Two missed data. QI, quality improvement. Six trained and experienced qualitative interviewers conducted the IDIs using pretested semi-structured interview guides. The guides collected data on the QI intervention but also contained questions related to data quality with participants asked explicitly about falsification: ‘There can be many reasons why health facilities exaggerate their performance. Can you share any experiences of this?’. During the interviews the data collectors took field notes. Daily debriefing meetings with study investigators were organised to discuss the adequacy of the topic guides, review and give feedback on transcripts, increase reflexivity and assess saturation. Analysis began during the daily de-briefs where the importance of data falsification first emerged and was discussed within data collection team. Using the grounded theory approach,24 the research team developed hypotheses as we reviewed all transcripts and extracted all data related to data falsification. Indeed, extracts centred on particular incidents or behaviours were coded inductively through in depth and repeated reading to identify a first set of themes and codes followed by a thorough coding using NVivo software V.12. The research team met regularly during the analysis to discuss the emerging codes, whether any codes should be merged and to discuss patterns, links and contradictions in the data. This study was conducted to assess the perspectives of health workers involved in the planning and implementation maternal and newborn healthcare services and did not involve patients or the public in the design, conduct, reporting, or dissemination of the study findings.
N/A
