This study identified a set of interrelated concepts across the study informants that influenced the grounded phenomenon, “the overall use of DHIS2”. The study findings were organized under four main key themes:
Key theme 1: Perceptions of DHIS2. This came from capturing users’ knowledge, experience, expectations, and their overall acceptability towards an electronic HMIS versus a paper-based HMIS.
Key theme 2: Perceived barriers to implementing DHIS2. These were drawn from the study informants’ discussion of obstacles to DHIS2 use related to individual capacity, institutional support, or technical issues related to DHIS2 software.
Key theme 3: Perceived facilitators to implementing DHIS2. These were drawn from the study participants’ discussion of positive determinants of DHIS2 use related to individual capacity, institutional support, or technical issues related to DHIS2 software.
Key theme 4: Recommendations to improve DHIS2 functionality. The study participants shared their suggestions for overcoming challenges or how to make the system more functional.
Key theme 1: Perceptions of DHIS2
The majority of study participants expressed a strong, positive preference toward using DHIS2 for RMNCAH data collection. They described DHIS2 as a dynamic system that has improved overall medical record keeping and accountability of data reporting from community clinics at the periphery to district-level hospitals.
Online is a perfect system. Previously I used to collect data in papers, and at the end of the year my office gets full of papers. It was also very difficult to retrieve data from thousands of piled up paper forms. Now, in online, by clicking the date or by name or phone number of the patients, I can easily check the data. I am getting the data collection form even in my mobile, by which I can fill up the form, from any place and at any time! So, it is easier. — Community health care provider, IDI
The supervisory team perceived that initiating such technology has contributed to instant monitoring, cross-checking of collected data, setting priorities, and making decisions, which was time-consuming with the previous paper-based system. With DHIS2, statisticians are assigned to tabulate the data and share the generated summary reports with district and divisional health managers. Managers observe and flag the gaps in service delivery and note achievements. Findings are discussed at monthly review meetings in the presence of field staff. At these meetings, which are held in each sub-district, district, and divisional health manager’s office, comparisons are made with the previous month, present month, and yearly national targets to track improvements in performance and identify any hindrances to achieving targets. Most respondents, from the community to the national level, identified this review meeting as a platform for RMNCAH-related data observation, monitoring, and instant planning for the coming weeks.
From DHIS2, along with [the] national scenario, we can see the status of districts and sub-districts, even unions and wards. All the field staff are forwarding data on rate of using contraceptives, maternal death, amount of IUD [intrauterine device] delivered, and number of oral contraceptives supplied. — Information communication technology focal person, IDI
Additional factors also strongly influenced users’ perception of DHIS2. These were at both the individual and institutional levels. DHIS2 users who were more frequent users and had sufficient training perceived the true need for it. Availability of sufficient technical equipment at the field level, like laptops, desktops, and tablets, made the users more enthusiastic.
The demand for using DHIS2 goes beyond RMNCAH. Key informants who had been involved with DHIS2 since its inception explained that the software is continually maturing. In 2009, when DHIS2 was launched, it was not used for data visualization and decision making because accessing the system was challenging. As soon as DHIS2 introduced the dashboard concept in 2012, it drew the attention of directorates working at the national level, who demanded the platform be used for their own reporting. As a result, the online data entry forms increased from 12 in 2012 to 32 in 2013. The perceived need for DHIS2 is explained in the below quote:
In 2013, the DHIS2 log-in dashboard became much [more] popular, all users could access it. At that time, 5,000 to 6,000 graphs were made using DHIS2, which eventually increased to 15,000 to 16,000. It means people were trying to use it. To justify my argument, I must say, these graphs were prepared by users from 64 districts, not by a single user. That means people are using it! — HMIS expert, KII
A few health managers expressed a contrasting view, arguing that staff orientation and adaptation to technology sometimes works as a major obstacle to electronic HMIS implementation. One health manager shared his concern saying that, “In some places . . . a complex device, [like a] computer has been handed over to the hand of an old community health worker, hence she cannot use it.”
While aggregated data are being reported monthly, automated data reporting is not possible within the current DHIS2 system. This makes the data entry process time-consuming and complicated. The field-level workers (i.e., CHCPs) have been maintaining both paper and electronic forms so they can cross-check data from missing reports. Moreover, insufficient understanding of data entry and how to report the RMNCAH indicators leads to unintentional errors in data entry. This ultimately results in more misreporting and less data use.
Key theme 2: Perceived barriers to implementing DHIS2
Several technical challenges with the DHIS2 platform were highlighted during the KIIs. Absence of an automated data aggregation process increases the possibility of data disparity and errors.
DHIS2 has a problem. . . . There are [boxes] for entering aggregated data. But, now, it is needed to use the formula. Many of the staff do not understand these formulas. In training sessions, I provide them the formula, explain this using multimedia presentation. Many [field staffs] do not understand it. In several cases, they put the value of one indicator in boxes designated for other indicator. — IT expert for MIS, KII
Respondents pointed out some technical issues with the data collection forms that should be checked to decrease misreporting and improve efficiency. DHIS2 has the provision to “SKIP” for all indicators, which serves as a source of data incompleteness. Respondents involved in data analysis identified minor issues with the data collection forms that should be checked to decrease misreporting.
In [the] individual server, first, I put mother's name, her EDD [estimated date of delivery], date of enrolment, and then a box will pop up for gender. There is male, female and transgender. The data is meant to be for the pregnant mother. I don’t understand what the need of gender then? There should be a system that [the] computer would recognize the gender automatically when pregnant woman has been marked. We should not put it manually. Here our field workers are making mistake[s]. — District statistician, FGD
Instead of using unique health identification numbers to track patients, patients’ cell phone numbers are used. However, it is difficult and time-consuming to search the database with a cell phone number. To get around this, CHCPs prefer to enroll follow-up patients as new ones. This raises a data quality issue since repeat clients are identified in the system as new clients. According to the key informants, this has created a gap in the system, as it is not possible to track the health status of a single patient in the existing system during data analysis and visualization. It was suggested that the system could be linked to Bangladesh’s National Identification Database to get a unique identifier
Updating the data entry forms to facilitate comparisons among data variables is challenging. DHIS2 started with version 2.6 on 2009, which was upgraded, version-by-version, to 2.13 at the end of 2013 to make the system faster. Each time the data entry forms are changed it becomes more difficult to compare the old and new data because the software cannot match the data variables, resulting in invalid findings.
For those, who are computer literate, for them, a version change is an “attraction.” “Let us explore, what are the new features?” But our CHCPs do not perceive it in this way. They think, there was a box here in the older version, where did the box goes now with the newer version? They don’t understand, we are trying to make their work easier! It will take some time, to change the culture. — HMIS expert, KII
Several informants reported that in the existing system, searching for sub-districts is a time-consuming process.
At the supervisory level, district and sub-district health managers could not find the time to use DHIS2 on a daily basis because they were involved in other activities. Sometimes they avoided using DHIS2 altogether.
A health manager knows clearly about his district's targets on immunization coverage, or ANC coverage, or even for facility births from their years of experience. So they do not need to open the computer and get into the DHIS2. The mechanism is such; you cannot trap him for this reason. — Senior programmer, KII
Reporting to DHIS2 is an additional task for the statisticians with other regular administrative duties (e.g., preparing salary sheets, drafting letters). They need to commit an extra hour of work for that. National-level key personnel acknowledged the shortage of statisticians or other staff trained in data analysis. They admitted that in many areas, qualified statisticians had not been recruited. Even so, many statisticians are not proficient in using computer software and do not understand health indicators and data compilation. In many areas, statisticians do not even attend trainings.
The job description and responsibilities of statistician should be separate. But in many districts there is no designated statistician…. In area “YY,” a ward boy does all the work of a statistician; you cannot expect anything better from him! There should be an assigned person, who will do research [with data]. — District health manager, KII
The RMNCAH data collected by the MIS Division of DGHS is also used by the RMNCAH line directorate of DGHS. However, data retrieval from the DHIS2 platform is not the regular practice for the RMNCAH line directorates; like all other directorates they rely on their own reporting format.
Statisticians reported not receiving any specific training on DHIS2, rather it was a part of computer literacy training. Participants received DHIS2 training manuals, though these were not updated to reflect changes in newer versions of the software and forms. Since DHIS2 was introduced, all the line directorates want to incorporate their relevant indicators to be collected and analyzed through DHIS2 using the same workforce.
Now everybody wants their data from DHIS2. [The] non-communicable disease division add some indicator[s], RMNCAH add some too. In some cases, the reporting format is also different than the one used by DHIS2. For example, if [the] EmNOC [emergency newborn and obstetric care] reporting format for [the] MIS division and RMNCAH would be [the] same, I can get the report by clicking on DHIS2 data. But [the] EmNOC report for RMNCAH directorates have 27 indicators while it is 25 in [the] DHIS2 database. — Sub-district statistician, IDI
Although the participants said the number of electronics provided for data collection is sufficient, slow Internet connectivity makes real-time data entry difficult. As one CHCP described:
At dawn, sometimes the Internet speed is better. In most cases, I enter the data at this time. It happened, I could not report for one week, two weeks, as the speed was slow. With a weak connection, I cannot even log in into the system. —CHCP, IDI
Providing offline data entry could make things easier. The process of sending broken tablets to the capital city for repairs and transporting them back to the community takes a long time. The majority of respondents reported internet modem shortages as well. In many areas, sub-district and district health managers personally obtained a modem and Wi-Fi router.
Key Theme 3: Facilitators of data collection and analysis with DHIS2
Mandatory quality checks at different tiers have played a significant role in improving data quality. At the data entry level, the system does not allow incorrect data. The system ensures self-validation of data by adopting three approaches: input validation, adding appropriate ranges, and validation rules. When a data operator adds any value that is out of the expected range, they get an error message. Moreover, DHIS2 allows local-level data access and correction before it is reported at the national warehouse. An IT focal person with a medical background was assigned at a sub-district hospital and at a district-level civil surgeon office to check data errors. Consequently, regular monthly feedback meetings are organized at the sub-district level in the presence of field workers from both the health and family planning wings to minimize duplication of RMCAH data. Similarly, monthly feedback meetings are also organized at the district and national levels. The national core MIS committee, chaired by the MIS directorates, meet monthly to get feedback on technical issues and to monitor data coming from all the districts. The key informants greatly appreciated this national-level meeting where government officials, donors, and technical people participate.
A national-level expert shared his experience with checking data validity:
For example, when we check MMR [maternal mortality ratio], we locate where the ratio is high. Then we review the ratio of that particular district for consecutive months to explore the consistency of data and reporting status, either it was low or high for the previous months. We check all these. Then we send an e-mail, to respective authority, to look into the matter. — HMIS expert, KII
So far, DHIS2’s performance has been measured from the perspective of timeliness and completeness. The sub-district and districts are evaluated according to their overall reporting rate. A positive competition for service improvement has been nurtured. The best-performing district or division receives recognition from the national level.
In our monthly meeting we discuss our shortfall; we plan how to improve the reporting rate. We always analyze the data, hence our performance is better!! We have a silent competition with other districts of this division and we do better always and got national award as model district. — District health manager, KII
International donors strongly support strengthening Bangladesh’s HMIS. They share financial costs with the government for national- and international-level staff training, IT equipment purchases, and other needs. In collaboration with other nongovernmental organizations, like icddr,b, they are providing technical support to the IT programmer for improving the online platform and organizing a training on the DHIS2 manual for staff working at different tiers of the health system. Donor organizations have demonstrated a strong commitment to the successful implementation of DHIS2 by deploying their staff as monitoring officers at each administrative division and ensuring their physical presence and participation during monthly coordination meetings at the divisional and central levels.
The government has limited capacity and could not develop that capability till now. From the side of development partners, we are giving them that support. If development partners withdraw their support, how will the system run? But the DHIS2 dashboard is already sustainable, and its automatic; staff have training and they can handle it. The government is cordial, and they have sufficient resources, training arrangements, and hardware. In this context, strict monitoring and defined role of staffs are important. In addition, ownership of data is a major concern, many health managers do not own the data. — HMIS expert from donor, KII
A monitoring and evaluation framework is used to identify DHIS2-related facilitating factors at all steps ranging from input, process, and outcome. (Figure 2)
Key theme 4. Recommendations for strengthening the HMIS to improve RMNCAH outcomes
Based on the study findings, the participants’ major recommendations for strengthening the HMIS to ameliorate RMNCAH outcomes in Bangladesh are elaborated in this section. The DHIS2 platform should be programmed to generate automated data for specific RMNCAH indicators. A pop-up box with the indicator definition, calculation (if applicable), and any possible disaggregation should be included. This will provide instant help to the CHCPs and standardize data collection. The software should be translated into Bangla (the local language) to help create a clear understanding of instructions and RMNCAH indicators. An online dashboard should be installed in the platform where instant RMNCAH-related reporting and performance status updates can be exhibited automatically at the sub-district and district levels. Statisticians should be informed in advance about software updates and notified of specific changes so they can prepare the CHCPs.
Data collection forms should be simplified to ease the data collection process and data reporting. Creating unique health identification numbers for patients and issuing individual health cards will decrease the time spent on data entry and help mitigate data duplication. Since the system will contain clients’ contact information, statisticians can verify the collected data through random phone calls. A geographic information system should be installed in CHCPs’ electronic devices used for data collection to track providers’ movement. Users should be able to enter data into DHIS2 daily, as aggregated data increases the risk of errors and compromises data quality.
Since DHIS2 is used at different levels of the health system, the DHIS2 training curriculum should be tailored to the needs of health professionals working at different levels. The IDIs and FGDs revealed a need for separate training sessions on medical terminology for community- and sub-district-level staff. After every update to the software or data collection forms, refresher trainings should be organized to improve staff knowledge and efficiency. A standardized training curriculum and tools are also needed. Furthermore, soft copies of training manuals should be shared with staff via e-mail so they can be easily updated and disseminated.
Along with a statistician, another staff member should be trained in data compilation and analysis to complement the statistician’s work and support the statistician in his/her absence. A separate MIS unit can be formed comprising, at a minimum, a statistician and a supporting staff member who will be assigned to perform all MIS-related tasks only. Sub-district and district health managers should be more involved in data reporting and analysis to develop ownership and a regular practice of using DHIS2.
Computers and other electronic devices for data collection should be repaired at the local level to save money and time. Providing CHCPs with an Internet data subscription can ensure timely reporting. The number of modems at the sub-district and district levels should be increased, and each municipality should have its own dedicated laptop for the statisticians to use to ensure timely reporting.
The country would benefit from a national e-health strategy and implementation framework to facilitate a culture of DHIS2 use for planning, setting priorities, and decision making among different stakeholder groups. This strategy should include how the country intends to provide the resources to fund DHIS2’s long-term sustainability when donor support is no longer available.