Assessment of Data Quality on Expanded Programme on Immunization in Ghana: The Case of New Juaben Municipality

Organizations of all kinds make decisions based on the data they have at their disposal. In healthcare reporting, data quality and consistency are critical to ensuring patient safety and communicating health service delivery. Quality data provides accurate and timely information to manage services. It also provides good information to manage service effectiveness as well as aids to prioritize and ensure the best use of resources. The main objective of the study was to improve on data quality on expanded programme on immunization (EPI) in the New Juaben Municipality. The work aimed at improving the accuracy of reported number of vaccinations at vaccination delivery sites as well as assesses the completeness of data that was provided. A descriptive cross-sectional study was employed. The study involved structured observation of tallied data from EPI tally books from the eight health sub-districts in the Municipality. Purposive sampling was used for this study. Data Quality Self-Assessment Tool (DQS) was the main instrument used in presenting and analyzing the accuracy and discrepancy ratios of the data. The result demonstrated discrepancies in tallied data at the vaccination delivery sites, facility summary report and report submitted to the Municipal Health Directorate. In 2011, there was 2674 over reported data to the district level while 2824 over reported data was recorded in 2012 from the eight (8) health facilities used for the study. It was observed that less importance was attached to data capture at some health facilities in the sub-districts. It was also ascertained that data storage and retrieval was very poor in some facilities visited. There is therefore the need for regular monitoring in the sub-districts’ RCHs and Health Centers in the New Juaben Municipality of Ghana to correct the mistakes.


Introduction
Immunization is among the very important public health interventions for preventing and reducing global child morbidity and mortality. The global effort to use vaccination as a public health intervention began in 1974 when the World Health Organization (WHO) launched the Expanded Programme on Immunization (EPI). There are often discrepancies between tallied data at the vaccination delivery sites and reported data from the same vaccination sites to the Municipal Health Directorate. This has necessitated the need for the research since high coverage of EPI and MDG 4 cannot be achieved without quality data. EPI coverage has to be improved but this cannot also be achieved without quality data or strengthening of the data management system. The New Juaben Municipality is the smallest of the 26 districts in the region and houses the Regional Capital (Table 1).

Problem statement
The weak health information system support within the subdistricts is one of the obstacles to effective and efficient management in the health service delivery. At the level of operation where service delivery is the main task, there are problems with data capture, analysis, storage and utilization of health service information for decision making. In EPI, tally data (recounted data) is usually different from summary data at the facility level and also that submitted to the District Health Directorate. This research would assess the quality aspect of EPI data generated in the municipality.

Main objective
To improve on data quality on Expanded Programme on Immunization (EPI) in the New Juaben Municipality.

Specific objectives
To determine the accuracy of reported number of vaccinations at vaccination delivery sites.
To assess the completeness of data.
To assess data storage and retrieval of data for action.

Conceptual framework
The framework above ( Figure 1) explains most likely factors that lead to good EPI data quality. These include complete recording and documentation in EPI tally books, availability of EPI tally books and monthly reporting formats, availability of computers for data capture and analysis, availability of proper filing system and availability of patient registers. EPI tally books are very important since services provided in health facilities are to be documented for analysis as well as for future reference. The registers serve as source documents where summary reports are generated. The non existence of these registers implies that services were not provided. The correlation between the existences of these registers indicates whether or not data is manufactured at fake health facilities. Records keeping are one of the critical components of quality data. Good records keeping make data

Introduction
Immunization is a vital mechanism for controlling as well as eradicating infectious disease and in order to achieve this aim there should be quality data to aid in knowing the increasing coverage and decreasing drop-out rates. Quality data is one of the major problems facing the New Juaben municipality. In relation to EPI, children vaccinated are not tallied correctly and summary data do not always tally with reported data. The old EPI tally books are mishandled and misplaced and others. These have necessitated the need to research into this area to diagnose all associated problems and address them.
Data: Data is defined as abstract representations of selected characteristics of real world objects, events and concepts expressed and understood through explicitly definable conventions related to their meaning, collection and storage. The key to making decision about fitness for use is, knowing what the data is intended to represent and

Introduction
This chapter outlines the various methods and materials that were used to achieve the study objectives. Both primary and secondary data were used as sources of information from both the vaccination delivery sites and the New Juaben Municipal Health Directorate, Koforidua. It also focuses on the strategies used in sampling and collecting data and how the data will be analyzed. In addition, data was entered into the DQS toolbox for analysis. Tables, frequencies and proportions were used to present the data.

Study sites:
The study was carried out in the Reproductive and Child Health units (RCHs compound) and Health Centers in the New Juaben Municipality in the Eastern Region of Ghana. The Municipality has eight sub-districts RCHs and Health Centers represented in Table  1 above.

Research design:
The study was descriptive cross sectional in content.
Study population: All the eight (8) sub-districts were visited for the research.
Sample size: Eight (8) health facilities were selected for the research. These facilities are the main eight sub-districts in the municipality. The study compared tallied data in the EPI tally registers at the vaccination delivery sites to facility summary report as well as to reported data at the Municipal Health Directorate to check for discrepancies.
Sampling procedure: Purposive sampling was used to assess data from all the eight sub-districts health facilities. The reason for adopting this sampling procedure was to identify any discrepancies in EPI tally books and reasons accounting for the unreliable data.

Data collection technique:
The data quality self-assessment tool (DQS) was used to compare facility tallied data with reported data to check for discrepancies. In addition, an observational checklist was used to check for how data was captured, stored, retrieved, analyzed and utilize in the facility. Moreover, the study relied on vials using existing data from the district health directorate information unit. Data was obtained primarily from the EPI register books which specified number of vials given. This was collated by the Midwives in the various sub-districts at the end of every month and submitted to the Municipal Health Directorate (MHD) in a form of hardcopies since they lack computers.
During the process, tallies from EPI tally books were compared with summary data at the facility and the municipal health directorate using the observation technique. Tallies in EPI books and summaries collated were observed and verified to check for discrepancies using the DQS data collection tool ( Table 2).

Instrument for data collection:
The data quality self-assessment tool (DQS), Table 3 below was used to review EPI records from 2011 to 2012. Tallied data (recounted data) was compared to summary report at the facility and later cross checked with the District summary report. The daily vaccinations administered in the EPI tally books were recorded into an EPI data collection tool. The aggregates for each month per antigen were then entered into the DQS tool which generated accuracy ratio (verification factor) and Discrepancy rate per health unit. This exercise is vital because it also provides an opportunity to evaluate coverage data accuracy and correct it.
An observational check list was also employed to assess data quality issues in the various health facilities. The observational check list covered areas such as data capture, storage and retrieval and data utilization.
Data processing and analysis: The DQS Tool was used in the quantitative data entry, processing and analysis. Data was entered into the DQS tool which generated accuracy ratio (verification factor) and Discrepancy rate per health unit and antigen. The DQS has the period of which data is collected by month thus January to December with its corresponding antigen that is assessed. When these indicators on the DQS are completely filled, the tool will automatically generate accuracy ratio represented on the tool as 'VF' and discrepancy levels which is also indicated as 'discrepancy'.
Epi-Info software was also used to design a template of which the observational check list was entered. Statistical analysis was run using analysis command of the Epi-Info software.  to check for their previous EPI vaccination registers, copies of previous summary reports, data completeness in the EPI registers, graphs showing trends of EPI coverage, EPI monitoring charts, reports from data validation team and finally how data was stored and retrieved in the facility.

Delimitation of the study:
The study was limited to only New Juaben Municipality. The target for the study was only children below 1 year and pregnant women. Only Penta 1, Penta 3, Measles and TT2+ for pregnant women were assessed due to time constraints. Some health facilities could not produce primary source data (EPI tally book) for some of the periods under study. This will some way affect the analysis of data collected. Some health facilities were visited more than twice to complete data collection due to the volume of data to be assessed and with this, time and transport was a limitation.
Ethical issues: Permission was sought from the Municipal Health Directorate before the study was carried out. Upon reaching the health facilities, permission was again sought from the in-charges (Heads) before embarking on the exercise. The study also acknowledged all other sources of information used in reviewing literature thereby making it worth reading and source of information for future studies.

Analysis and Findings of Data Introduction
The analysis focused on the accuracy and discrepancies of data in the eight (8) health facilities. Tallied data in EPI tally books at the vaccination delivery sites and summary report were considered in this analysis. All tallies in the EPI tally books for PENTA 1, PENTA 3, Measles and TT2+ for the period of 2011 to 2012 were counted and compared with the facility summary report and report submitted to the MHD to check for discrepancies. Again, a comparative analysis was drawn between the periods from 2011 to 2012 to ascertain whether there was improvement in the quality of EPI data [15,16]. With this, the data gathered from the EPI tally books were entered into the DQS toolbox and analyzed with the aid of graphical and statistical tabulation. The use of observational checklist was also employed. In all, eight (8) observational checklists were used in the eight health facilities in the eight sub-municipals. Figure 2 above shows that, Koforidua Poly-Clinic had a high discrepancy rate of 48% of PENTA 1 antigen, followed by Medical Village which recorded 26%. Akwadum RCH, Zongo health centre, and Jumapo health centre recorded discrepancy of 9%, 3% and 1% respectively. Effiduase RCH and Densuano RCH recorded 100% accuracy ratio with no discrepancy in PENTA 1 antigen [17][18][19][20]. It must be noted that the higher the discrepancy rate, the less quality the data is. These discrepancies occurred as a result of wrong tabulation and additions of tallied data in the vaccination sheet. It was also observed that some health workers do not tally direct into tally books during immunization services but rather use piece of papers to document and later transcribe into the tally book. Errors are made during the transcription of data from pieces of papers used during the immunization sections into the tally books. Totals of 19% discrepancy and 81% accuracy were recorded in 2011 for PENTA 1. It should be noted that if the discrepancy recorded is more than 10% then the data is not reliable for decision making and planning of immunization programme. It is worth mentioned that structures are put in place to validate captured data from these vaccination centers before submitting it to the next level. Figure 3 depicts that Jumapo Health Center had a high discrepancy rate of 77% of PENTA 1 antigen followed by Koforidua Poly-Clinic (44%) and Densuano RCH (11%). Effiduase RCH, Zongo Health Center, and Oyoko Health Center had a discrepancy level of 4%, 2%, and 3% respectively. Akwadum RCH and Medical Village RCH recorded 1% discrepancy level each. It must be noted that the higher the discrepancy rates the less quality the data is. Although Medical Village RCH had 101% accuracy ratio, there was 1% discrepancy rate recorded for PENTA 1 antigen which indicates under reporting. None of the health facilities visited had 100% accuracy ratio (AR) with no discrepancy in PENTA 1 antigen [21].

Data quality validation on PENTA 1, PENTA 3, Measles and TT2+
Making comparative analysis between accuracy ratio (AR) and the discrepancies of data in the reporting period of 2011 and 2012 of PENTA 1 antigen, it was noted that the accuracy ratio of PENTA 1 antigen in 2011 was 81% with 19% discrepancy level and still remains the same in 2012 although some health facilities recorded improvement of data quality with PENTA 1 antigen over the period of January to December, 2012. It was also observed that three (3) health facilities Akwadum Health Centre, Medical Village RCH, Zongo Health Centre recorded an improvement in data accuracy of the 2012 EPI data over 2011. Figure 4 above depicts that Koforidua Poly-clinic had a high discrepancy rate of 55% of PENTA 3 antigen followed by Medical Village RCH (25%) and Zongo Health Centre (5% Recounted Data(Tally Sheet)  Figure 4.5 above. It was also ascertained in Figure 4.5 that 81% and 19% was recorded as accuracy ratio and discrepancy level for measles antigen from January to December, 2011. It can also be said that with the discrepancy ratio of 19%, measles data for the year 2011 was of poor quality and not useful for any planning and decision making towards the immunization programme [23].
From Figure 7, none of the health facilities recorded 100% of accuracy ratio in measles antigen. One (1) Facility, Medical Village RCH recorded 101% of accuracy ratio with 1% of discrepancy which indicated under reporting while all the other facilities recorded less than 100% accuracy ratio showing over reporting. Two (2) health facilities, Jumapo Health Center and Koforidua Poly-Clinic recorded the highest discrepancy rate of 81% and 45% respectively A comparative analysis of Figure 6 and Figure 7 depicts that, there was a 3% improvement in measles data accuracy in 2012 over 2011. It was also ascertained that three (3) health facilities Medical Village RCH, Oyoko Health Centre and Koforidua Poly-Clinic recorded improvement of data accuracy in 2012 over 2011. It was also noted that Koforidua Poly-Clinic recorded the highest discrepancy level of 50% and 45% in 2011 and 2012 of measles data respectively. From Figure 8 above, Koforidua Poly-Clinic had the highest discrepancy rate of 53%, followed by Effiduase RCH (48%), Oyoko Health Center recorded 38%, Medical village recorded 31%, Zongo and Jumapo Health Centers both recorded 1% discrepancy ratio each. Jumapo recorded 101% accuracy ratio (AR) with 1% of discrepancy ratio which indicated under reporting while all the other facilities recorded less than 100% accuracy ratio thus over reporting.
It was also noted that TT2+ recorded 59% and 41% of accuracy ratio and discrepancy level respectively from January to December, 2011. It must be noted that it is only the TT2+ antigen that recorded high discrepancy against accuracy. It can therefore be emphatically stated that, data quality for TT2+ was very low in this study. This is mainly issues with poor documentation on the part of health workers.
From Figure 9, none of the health facilities recorded 100% of accuracy ratio in TT2+ antigen. Two (2) facilities, Medical Village RCH and Zongo Health Center recorded 113% accuracy ratio with 13% discrepancy and 104% accuracy ratio with 4% of discrepancies respectively which indicated under reporting. Jumapo Health Center had the highest discrepancy ratio of 77% with 23% accuracy. Effiduase RCH recorded 0% of accuracy ratio with 100% discrepancy and this indicates the highest over reported data for TT2+ antigen. It was also noted from figure 4.8 that TT2+ antigen recorded 60% and 40% accuracy ratio and discrepancy level respectively. It is noted that TT2+ had the poorest data quality as compared with the other antigens under study.
A comparative analysis of Figure 8 and Figure 9 depicts that, there was 1% improvement in TT2+ data accuracy in 2012 over 2011. It was also ascertained that the general performance on the data capture or documentation on TT2+ antigen for 2012 was appalling in the Municipality although there was slight improvement over the 2011 data assessed. discrepancy level of 4%, 3%, 2% and 1% respectively. Although Effiduase RCH had 104% accuracy ratio, there was 4% discrepancy rate recorded for PENTA 3 antigen which indicates under reporting. Only one health facility, Jumapo Health Center had 100% accuracy ratio with no discrepancy in PENTA 3 antigen. It must be noted that the higher the discrepancy rate, the less quality the data is. Figure 4 show that PENTA 3 recorded 80% accuracy with 20% discrepancy for the period from January to December, 2011. As stated earlier, the accepted range of discrepancy level should be less than or equal to 10%. This indicates that PENTA 3 data quality for 2011 was appalling and managers should address this challenge [22].
In Figure 5 above, only Oyoko Health Center recorded 100% accuracy ratio. Jumapo Health Center recorded a very high discrepancy rate of 76% followed by Koforidua Poly-Clinic (41%). These two facilities were not able to provide a complete EPI tally book for verification and this resulted in the high discrepancy. Moreover, apart from Oyoko Health Center, all the other facilities visited recorded less than 100% Accuracy (AR) for PENTA 3 Antigen which means there was over reporting.
A comparative analysis of figures 5 and 6 shows that data accuracy in 2011(80%) of PENTA 3 antigen still remains the same in 2012. It was also noted that three (3) health facilities Medical Village RCH, Koforidua Poly-Clinic and Oyoko Health Center recorded a slight improvement of data accuracy of PENTA 3 antigen in 2012.
In Figure 6 above, three (3) health facilities, Zongo Health Center, Jumapo Health Center and Effiduase RCH recorded 100% accuracy ratio. Koforidua Poly-Clinic recorded the highest discrepancy rate of 50% followed by Medical Village RCH (25%). Generally the performance of data accuracy on the measles antigen was good with  the average 668 data was over reported per antigen to the Municipal Health Directorate. This depicts the low quality of data with respect to Expanded Programme on Immunization (EPI) where 2674 figures were over reported to the District, Regional and National Level from the eight health facilities assessed. Figure 11 above shows the tallies of all antigens against the summary reported to the District level. Penta 1, Penta 3, Measles, and TT2+ recorded 590, 615, 479 and 1140 over reported data respectively to the municipal health directorate in 2012. It is observed that TT2+ had the highest discrepancy followed by PENTA 3, PENTA 1 and Measles Antigen. This depicts low quality of data with respect to Expanded Programme on Immunization (EPI) where 2824 figures were over reported to the District, Regional and National Level from the eight health facilities assessed. If AR is <100%, data is over reported . If AR is >100%, data is under reported. Table 4 illustrates the accuracy ratio (AR) and discrepancy ratio of the various antigen data. It is shown in Table 4 above that PENTA 3 antigen recorded the highest data accuracy of 87% with 13% discrepancy level while both PENTA 1 and Measles recorded 81% with 19% discrepancy. TT2+ antigen also recorded 59% of data accuracy with 41% discrepancy. The total data accuracy reported by these antigens from the eight (8) health facilities was 77% and 23% discrepancy rate. Table 5 illustrates the accuracy ratio of the various antigen data. It is shown in Table 5 that Measles antigen recorded the highest data accuracy of 83% with 17% of discrepancy while PENTA 1, PENTA 3, and TT2+ recorded 81% accuracy with 19% discrepancy, 80% accuracy ratio with 20% discrepancy , and 60% accuracy ratio with 40% discrepancy respectively. The total data accuracy ratio reported by these antigens from the eight (8) health facilities was 76% with 24% discrepancy.   reports are generated. The non-existence of these registers implies that services were not provided. The correlation between the existences of these registers indicates whether or not data is manufactured at fake health facilities. Table 7 above illustrates that all health facilities assessed had copies of summary sheet (100%) at their level which was encouraging. Good records keeping make data accessible at any point in time for decision making at all levels. It is worth mentioned that all patients' registers, tally books and all source data are kept well for continuity of care and for future references. If the source data (registers, tally books, patient folders) are not kept well, it becomes very difficult to make accurate decision as and when it is needed.

Analysis of the observational checklist
Data from Table 8 shows that 50% of health facilities assessed did not have complete tally books while 50% had their tally books complete.
From the review of the literature, data quality refers to data that is accurate, valid, reliable, relevant, legible, complete and available in a timely manner to decision makers for healthcare delivery and planning purposes. Complete data gives the right detail of information hence making analysis and interpretation easier and arriving at practical solutions to address challenges. Data from Table 9 shows that 25% of health facilities assessed had excellent data storage/ retrieval (filing system) while 12.5%, 25% and 37.5% had very good, good and poor filing system respectively.
There are many ways by which EPI data can be stored in the health facilities for easy retrieval in decision making. Good records keeping make data accessible at any point in time for decision making at all levels. It is worth mentioned that all patients' registers, tally books and all source data are kept well for continuity of care and for future references.
Data from Table 10 shows clearly that all health facilities assessed had EPI monitoring chart for the current year pasted. Table 11 illustrates that 50% of health facilities assessed have their EPI monitoring chart of the previous month updated while 50% of health facilities have not updated their EPI monitoring chart.

Summary of Findings, Conclusions and Recommendation Introduction
The summary of findings of the research, conclusions drawn based on the findings and recommendations made are discussed in this chapter.

Data completeness
Data completeness deals with whether or not the right detail of information is provided. It was observed in table 4.5 that 50% of health facilities had incomplete tally books. This is likely to affect data quality in the sense that the right level of details of information that should be provided is not available for good decision making and planning purposes. For example estimates for vaccines to be used for vaccinations are base on the documentations in the tally books hence vaccines would be over budgeted or under budgeted for EPI programmes.

Data storage and retrieval (filing system)
Stored data are of use only if information can be retrieved quickly in an understandable form. Having a large quantity of information put together in common file might cause difficult access to any particular piece of information. That was the situation in our health facilities. It was being noted that a large quantity of different data is kept together in health facilities and this makes it difficult for data retrieval when decisions are to be made.
It was also observed that all health facilities assessed did not have electronic data storage system but relies on the manual system of data storage and retrieval. It was again ascertained that data storage and retrieval was very poor in some facilities especially Koforidua Poly-Clinic, Jumapo Health Center and Medical village RCH.

Conclusion
Judging from the findings of the research, the conclusion drawn was that the quality of data with respect to the Expanded Programme on Immunization (EPI) is generally poor.
As indicated earlier on, source data (tally books) from three (3) health facilities could not be produced for verification and this resulted in wide margins of over reported cases. The poor data quality can also be attributed to poor handling of data in these health facilities.

Recommendations
In analyzing the data and findings of the research work, some problems had been identified and as a result the following recommendations and suggestions have been provided; 1. EPI tallies must be done as soon as vaccines are administered to avoid distortion of data in the vaccination delivery sites.
2. Data audit teams should be instituted at all facilities to validate monthly and quarterly data before submission to the next level.

The Municipal Health Directorate and Sub-municipal heads should
intensify Facilitative supervisory visits at all levels and orient health staff involved in data collection on the importance of quality data.

The Municipal
Health Directorate should provide cabinets, files, computers and accessories for health facilities to improve on the data storage and retrieval system in the Municipality.

5.
Health facility heads or sub-municipal heads should make sure that daily summaries of children vaccinated and logistics used are documented in tally books to enhance accurate data recordings at the end of the month.
6. Data quality assessment should be ritualized in the routine monitoring and support visits to all health facilities.

7.
Further research should be conducted into factors affecting poor data quality on expanded programme on immunization by the Municipal Health Directorate.