Factors associated with data quality in the routine health information system of Benin

  • Archives of Public Health, Volume 72, No. 1, Year 2014

listen audio

Study Justification:
The aim of this study was to identify the factors associated with poor data quality in the routine health information system (RHIS) in Benin. The study was conducted because the insufficient quality of data produced by these systems limits their usefulness in decision-making for health sector planning.
Highlights:
– The study found a significant link between data quality and level of responsibility, sector of employment, RHIS training, level of work engagement, and perceived self-efficacy.
– Focus groups confirmed a positive relationship with organizational factors such as availability of resources, supervision, and perceived complexity of technical factors.
– The study provides strategic decision support in improving the performance of the RHIS in Benin.
Recommendations:
– Increase RHIS training and retraining opportunities for health workers.
– Improve supervision and support for health workers responsible for data collection.
– Enhance availability of material resources for RHIS activities.
– Address organizational factors such as resource availability and perceived complexity of technical factors.
Key Role Players:
– Ministry of Health: Responsible for policy development and implementation.
– National Health Information System Unit: Oversees the RHIS and coordinates data collection.
– District Health Management Teams: Provide supervision and support to health workers.
– Health Facility Managers: Responsible for completing reporting forms and ensuring data quality.
Cost Items for Planning Recommendations:
– Training and retraining programs for health workers on RHIS.
– Supervision and support systems for health workers.
– Procurement of material resources for RHIS activities.
– Capacity building for organizational factors such as resource availability and technical complexity.
Please note that the cost items provided are general categories and not actual cost estimates.

The strength of evidence for this abstract is 7 out of 10.
The evidence in the abstract is based on a cross-sectional descriptive and analytical study conducted in Benin. The study used a combination of techniques and tools, including interviews, questionnaires, and focus groups, to assess the factors associated with poor data quality in the routine health information system (RHIS) in Benin. The study found significant links between data quality and factors such as level of responsibility, sector of employment, RHIS training, work engagement, and perceived self-efficacy. The study also identified organizational factors that influence data quality, such as the availability of resources and supervision. The study provides valuable insights into the factors affecting data quality in the RHIS in Benin and could inform strategic decision-making to improve the system’s performance. However, the evidence is based on a single study conducted in a specific context, and further research is needed to validate the findings and assess their generalizability to other settings. To improve the evidence, future studies could consider using a longitudinal design to assess the long-term impact of interventions aimed at improving data quality in the RHIS. Additionally, incorporating qualitative methods, such as in-depth interviews or case studies, could provide a deeper understanding of the factors influencing data quality and potential strategies for improvement.

Background: Routine health information systems (RHIS) are crucial to the acquisition of data for health sector planning. In developing countries, the insufficient quality of the data produced by these systems limits their usefulness in regards to decision-making. The aim of this study was to identify the factors associated with poor data quality in the RHIS in Benin. Methods: This cross-sectional descriptive and analytical study included health workers who were responsible for data collection in public and private health centers. The technique and tools used were an interview with a selfadministered questionnaire. The dependent variable was the quality of the data. The independent variables were socio-demographic and work-related characteristics, personal and work-related resources, and the perception of the technical factors. The quality of the data was assessed using the Lot Quality Assurance Sampling method. We used survival analysis with univariate proportional hazards (PH) Cox models to derive hazards ratios (HR) and their 95% confidence intervals (95% CI). Focus group data were evaluated with a content analysis. Results: A significant link was found between data quality and level of responsibility (p = 0.011), sector of employment (p = 0.007), RHIS training (p = 0.026), level of work engagement (p < 0.001), and the level of perceived self-efficacy (p = 0.03). The focus groups confirmed a positive relationship with organizational factors such as the availability of resources, supervision, and the perceived complexity of the technical factors. Conclusion: This exploratory study identified several factors associated with the quality of the data in the RHIS in Benin. The results could provide strategic decision support in improving the system's performance.

This was a cross-sectional, descriptive and analytical study conducted between October and November 2012. In Benin, the RHIS is organized according to the pyramid structure of the healthcare system and includes the public and private sectors. The illegal facilities of the private sector are not included. Furthermore, the resources available for training and supervision are mostly dedicated to the public sector. As in most developing countries, the RHIS is composed of data collected from patient’s information records. These data are assembled in periodic summary reports produced by health staff in outlying areas. These reports are activity summary tables that are filled in using the data from the records. The reporting forms are linked to outpatient care, maternal care, immunization, financial management, laboratory, and anti-malarial activities. The number of data items is variable depending on the form. The shortest is the anti-malarial activities reporting form with 45 items; the longest is the outpatient care reporting form with 815 items. There is no computer in the health facilities; thus, the reporting forms are completed by hand. One person, often the facility manager, is charged with completing the reporting forms. When there are enough human resources in the same health facility, the head of the health district can assign one person to each type of reporting form. Accordingly, in the same facility, one or more persons are designated to complete the different reports, but the same person completes one type of reporting form. No data are collected about the time taken to complete the different reporting forms. The reporting forms are filled in monthly and are periodically sent, typically monthly, to the district health management team. The facility manager must analyze the report and use the information for decision making; however, information use is still insufficient. Computerization (still rudimentary) is usually performed at the health district level. A Microsoft Access database is created and sent to the intermediate level, and then to the central level. This study was conducted in first-line public and private health centers of the RHIS in four municipalities of the department of Atlantique-Littoral in the south of Benin. The municipality of Cotonou was selected deliberately because it is the only urban municipality in the department. Three rural municipalities were selected randomly from among the department’s eight rural municipalities. All of the public and private health centers in the three rural municipalities were selected. In the city, where the workforce in the private health centers was very large and unmanaged, random sampling was performed from the list available in order to select equal numbers of private and public health centers. The population targeted by the study were health workers responsible for RHIS data in the health centers. We also collected all of the data generated by each health worker in the twelve months preceding the survey. Sixty-nine public and private health centers including one hundred and twenty health workers were included in the study. The research protocol for this study was approved by the Benin National Ethics Committee for Research in Health (Comité National d’Ethique pour la Recherche en Santé) and also by the regional authorities and the local health authorities. The participants were informed about the objectives and anonymity of the survey. They were invited to give their informed consent before receiving the questionnaire. The dependent variable was the quality of the data generated by the health worker (good versus poor). The independent variables were: •socio-demographic characteristics: age, gender, general level of education, basic vocational training; •work-related characteristics: work location (urban/rural), sector of employment (public/private), type of contract (open-ended/fixed-term), responsibility of the health center (Yes/No), RHIS training or retraining in the last 12 months (Yes/No), supervision concerning RHIS data quality received in the last six months, receipt of financial incentives and availability of material resources for RHIS activities; •the perceived complexity of the technical factors (Yes/No), •the personal resources of the health worker such as their work engagement and their perceived self-efficacy concerning RHIS activities. The techniques and tools used included a document review with a processing form to assess the quality of the data and an interview with a self-administered questionnaire to collect other information. We also held two focus groups with health workers to determine the reasons for the poor quality of the data they produced. Health workers were selected for focus groups among all the health workers in charge of data collection in the municipalities involved in the study. They were randomly selected among the volunteers who wanted to participate in the focus group. Each group was composed of six people. The quality of the data batch was assessed using the Lot Quality Assurance sampling (LQAS) method with n = 32 and d* + 1 = 3 (n: size of the sample and d* + 1: maximum number of defective units expected per sample). Other parameters included in the equation were: N large, P0 = 20%, Pa = 5%, N being the size of the batch, P0 proportion of defective units, and Pa the maximum proportion of defective units expected to consider the lot of good quality [8-10]. The batch was defined as all of the data generated by a single health worker in the twelve months preceding the survey. The data were all numerical values that had to be entered in the periodic report produced by the health worker. The quality of the data randomly sampled, was assessed for comprehensiveness, reliability and accuracy. In this study, data comprehensiveness was defined as the “availability of the data across all of the documents in which it must be provided” for the twelve months. If the data were missing or if the document had not been produced by the health worker, the data were considered to be incomplete. In this study, data reliability was defined as “case correspondence to the case definition in the national guidelines”. The data were judged as being unreliable if the cases reported did not correspond with the case definition. Verification of reliability in this study was based on the clinical information entered in the records and did not involve verification with the actual clinic concerning the patient. For example, if the reporting form mentioned 5 cases for simple malaria, the surveyor checked every case reported in the register source: Does each case correspond to “fever + positive rapid diagnostic test” as recommended in the national guidelines? If one reported case did not match the national definition used in the guidelines, we considered the data to be unreliable. Data accuracy was defined by “the numerical correspondence between the data recorded in the document and that in the record”. A relative difference of 5% was permitted. Data were judged as lacking if they did not meet all three criteria. The batch was rejected if three defective items of data were found in 32 random samples. In each batch, 32 data items are expected to be randomly sampled with tables of random numbers. The sampling was stopped as soon as the maximum number of 3 defective items of data was reached. The number of samples prior to the batch’s rejection was, therefore, variable. The quality of the data generated by a health worker was judged as poor if his/her data batch was rejected. Work engagement for RHIS activities was measured using the French version of the Utrecht Work Engagement Scale (UWES) with 9 items [11-14] and was adapted to the field of health information system activities. The UWES comprised 3 items for each dimension: absorption, dedication and vigor. Each item was rated on the Likert scale from 0 (situation never found) to 6 (situation found every day). An average score was obtained for each dimension and for the overall scale (ranging between 0 and 6). A high score reflected a high level of engagement by the health worker in RHIS activities. In the analyses, the overall score for work engagement was divided into two categories obtained by applying the median threshold (high level of engagement versus low level of engagement). Perceived self-efficacy was measured with the HMIS tasks self-efficacy questionnaire (confidence level in their own abilities) taken from the Organizational and Behavioral Assessment Tool (OBAT) used to assess the Performance of Routine Health Information Systems (PRISM) in contexts similar to that of Benin [15,16]. In seven questions, the subject was invited to rate his/her perceived self-efficacy in performing various RHIS tasks on a scale of 0 to 100%. The average score obtained for the seven questions expressed as a percentage was used. A high score reflected a high level of perceived self-efficacy and a high level of confidence in the health worker’s own abilities. This score was recoded based on a median threshold in two categories of perceived self-efficacy: High and Low. The descriptive statistics used were the average and its standard deviation for the scores and percentages for the qualitative variables. The percentages were compared using the chi2 test. For the analytical component, we conducted survival analysis using the rejection of the data batch as the event and the number of samples before the rejection as the time variable. Univariate proportional hazards (PH) Cox models were applied to derive hazard ratios (HR) and their 95% confidence intervals (95% CI) and p-values using the Wald test. The proportional hazard assumption was checked using the test and plots based on Schoenfeld residuals and examining the plots ln (-ln (S (t)); S (t) is the survival curve. We presented Kaplan Meier survival curves to compare the probability of batch rejection (probability that the data batch would be of poor quality) according to the independent variables. The median number of samples (and interquartile range) was presented. We also performed a comparison of the average scores for work engagement and perceived self-efficacy based on the quality of the batches using the Student t-test. The significance threshold used was 5%. For the qualitative component concerning the focus group, information was categorized by topic, and a content analysis was completed.

N/A

The study “Factors associated with data quality in the routine health information system of Benin” aimed to identify the factors associated with poor data quality in the routine health information system (RHIS) in Benin. The study found several factors that were linked to data quality, including the level of responsibility, sector of employment, RHIS training, level of work engagement, and perceived self-efficacy. The study also conducted focus groups to determine the reasons for the poor quality of the data. The study used the Lot Quality Assurance Sampling method to assess the quality of the data.

Based on this study, some potential innovations to improve access to maternal health could include:

1. Training and capacity building: Providing comprehensive training and capacity building programs for health workers responsible for data collection in public and private health centers. This could include training on data collection, data management, and data quality assurance.

2. Supervision and support: Implementing regular supervision and support mechanisms to ensure that health workers receive ongoing guidance and feedback on their data collection practices. This could include regular site visits, mentoring, and supportive supervision.

3. Improved resources: Ensuring that health centers have the necessary resources, such as computers and internet access, to facilitate data collection, management, and reporting. This could involve providing funding and technical support to upgrade infrastructure and equipment.

4. Streamlined data collection processes: Simplifying and standardizing data collection processes to reduce the burden on health workers and improve data quality. This could involve developing standardized reporting forms, implementing electronic data capture systems, and using mobile technology for data collection.

5. Data validation and quality assurance: Implementing robust data validation and quality assurance mechanisms to ensure the accuracy and reliability of the data collected. This could involve conducting regular data audits, implementing data quality checks, and providing feedback to health workers on their data quality performance.

6. Use of data for decision-making: Promoting the use of data for decision-making at all levels of the health system. This could involve training health workers on data analysis and interpretation, establishing data review and analysis forums, and integrating data into health planning and decision-making processes.

These innovations could help improve access to maternal health by ensuring that accurate and reliable data is available for decision-making, resource allocation, and monitoring of maternal health programs and interventions.
AI Innovations Description
The recommendation that can be developed into an innovation to improve access to maternal health based on the study is to implement targeted interventions aimed at improving the quality of data in the routine health information system (RHIS) in Benin. This can be achieved by addressing the factors identified in the study that are associated with poor data quality.

Some specific recommendations include:

1. Enhancing training and retraining programs for health workers responsible for data collection in public and private health centers. This can help improve their knowledge and skills in data collection, entry, and management.

2. Strengthening supervision and support for health workers in maintaining data quality. Regular supervision visits can provide guidance and feedback to ensure accurate and reliable data reporting.

3. Improving the availability of resources, such as computers and other necessary equipment, in health facilities. This can facilitate the transition from manual data entry to computerized systems, which can improve data accuracy and efficiency.

4. Addressing organizational factors, such as the complexity of technical factors and the availability of resources. This can be done by streamlining data collection processes, providing clear guidelines and standard operating procedures, and ensuring adequate resource allocation.

5. Promoting work engagement and self-efficacy among health workers involved in RHIS activities. This can be achieved through supportive work environments, recognition of their contributions, and opportunities for professional development.

By implementing these recommendations, the quality of data in the RHIS can be improved, leading to more accurate and reliable information for decision-making in maternal health. This, in turn, can contribute to better access to maternal health services and improved health outcomes for women and their babies.
AI Innovations Methodology
Based on the provided description, here are some potential recommendations to improve access to maternal health:

1. Strengthen training programs: Provide comprehensive training programs for health workers responsible for data collection in public and private health centers. This should include training on data quality assurance, data management, and reporting procedures.

2. Increase supervision and support: Implement regular supervision visits to health centers to ensure data quality and provide support to health workers. This can help identify and address any challenges or gaps in data collection and reporting.

3. Improve availability of resources: Ensure that health centers have the necessary resources, such as reporting forms, registers, and basic equipment, to facilitate accurate and timely data collection. This may involve providing additional resources to health centers, especially in rural areas.

4. Enhance data management systems: Explore the possibility of introducing computerization or digital solutions to streamline data collection, reporting, and analysis processes. This can help reduce errors and improve the efficiency of the routine health information system.

To simulate the impact of these recommendations on improving access to maternal health, a methodology could be developed as follows:

1. Define indicators: Identify key indicators related to access to maternal health, such as the number of antenatal care visits, skilled birth attendance, and postnatal care coverage. These indicators should be measurable and reflect the desired outcomes of improved access to maternal health.

2. Baseline data collection: Collect baseline data on the identified indicators before implementing the recommendations. This can be done through surveys, interviews, or analysis of existing data sources.

3. Implement recommendations: Implement the recommended interventions, such as training programs, supervision visits, resource provision, and data management system improvements. Ensure that these interventions are implemented consistently across the selected health centers.

4. Monitor and evaluate: Continuously monitor and evaluate the impact of the interventions on the identified indicators. This can be done through regular data collection and analysis. Compare the post-intervention data with the baseline data to assess any changes or improvements in access to maternal health.

5. Analyze and interpret results: Analyze the data collected to determine the impact of the recommendations on improving access to maternal health. This can involve statistical analysis, such as comparing means or proportions, and interpreting the findings in the context of the study population.

6. Adjust and refine: Based on the results and findings, make any necessary adjustments or refinements to the interventions. This may involve scaling up successful interventions, addressing any challenges or barriers identified, and continuously improving the implementation strategies.

7. Repeat the process: Repeat the data collection and evaluation process periodically to assess the sustained impact of the recommendations on improving access to maternal health. This will help ensure ongoing monitoring and improvement of the routine health information system.

By following this methodology, it will be possible to simulate the impact of the recommendations on improving access to maternal health and inform decision-making for further interventions and improvements in the routine health information system.


Statistics:
Citations: 28
Identifiers:
DOI: 10.1186/2049-3258-72-25
ISSN: 07787367
Research Areas:
Environmental, Health System and Policy, Infectious Diseases, Maternal Access, Maternal and Child Health, Quality of Care
Study Countries:
Benin
Study Design:
Cross Sectional Study, Exploratory Study, Grounded Theory
Study Approach:
Mixed-methods, Qualitative, Quantitative
Share this:
Facebook
Twitter
LinkedIn
WhatsApp
Email