“Quality of prenatal and maternal care: bridging the know-do gap” (QUALMAT study): an electronic clinical decision support system for rural Sub-Saharan Africa

The CDSS is part of a complex intervention plan within the QUALMAT study. The development of the CDSS preceded the interventions at the study sites. Study activities of the project are carried out at rural primary health care centres in the three countries in SSA (Figure 1). In each country both intervention and control districts were chosen according to several criteria. The study districts were particularly disadvantaged based on comparisons within the countries. Study sites are public facilities at least 10 km away from a town and with a minimum of infrastructure. Availability of electricity was not a prerequisite and was organized through the project where necessary. All the health centres involved in the study have maternity facilities equipped to accommodate uncomplicated deliveries including a 24-hour observation period after delivery. However, none of the sites is equipped or staffed to provide full emergency obstetric care, for example with assisted vaginal delivery methods. Staff members at the sites are health professionals with 1-3 years of training, but no physicians are present. All health centres included in the study are no more than 2 hours drive from a district hospital where patients can be referred. Ambulances from the district health authorities are available upon request (Ghana and Burkina Faso, partly Tanzania) or ambulances are part of the facilities (most facilities in Tanzania). The local district hospitals provide full emergency obstetric care including the possibility for caesarean sections. One district with 6 study sites in each country will be subject to the intervention package, whereas the control district will have no interventions. In Ghana the intervention district is Kassena – Nankana and the control sites are located in Builsa. In Burkina Faso the intervention district is Nouna and the control sites are in Solenzo. In Tanzania, the intervention district is Lindi Rural and the control sites are situated in Mtwara Rural.

The QUALMAT project consortium, in close collaboration with local health authorities at national and sub-national level, has planned to introduce a CDSS and a performance-based incentive scheme in the research sites of the intervention districts. The development and introduction of these interventions is accompanied by research projects, whereby the quality of maternal care, the competence and motivation of those providing such care, and the cost of the interventions is assessed before, during, and after their introduction. In addition, specific studies will be conducted to evaluate the CDSS to analyse adoption and usability of the CDSS, attitudes towards computer use, and the influence of the decision support on medical care and on the workflow at the healthcare centres

Ethical clearance for the QUALMAT study was provided by the Ethics Committee of the Medical Faculty, University of Heidelberg, Germany (ref. S-173/2008), the Institutional Review Board of the Navrongo Health Research Centre, Ghana (ref. NHRCIRB 085), the Muhimbili University of Health and Allied Sciences Ethical Review Committee, Tanzania (ref. MU/RP/AEC/Vol.XIII/96) and the Ethics Committee for Health Research, Burkina Faso (ref. 2010.05/CLE/CRSN).

The development of the QUALMAT CDSS tried to anticipate critical factors for success of clinical decision support as identified by Kawamoto and colleagues [14]. Future users, medical experts, and IT specialists jointly developed the CDSS using the following steps:

During initial visits to study sites, the local situation of the rural health care facilities in all three countries and the workflow of patient care at the sites were assessed. Future users - mostly, as expected, computer illiterates or individuals with minimal computer experience - and local representatives of the health system from both national and sub-national levels described their environment and outlined their expectations.

There had been no previous HIT projects in the targeted districts for interventions with the exception of Ghana, where the expansion of the national health insurance scheme included the introduction of computers for administrative work at the faculties. In general, HIT projects previously known to the project partners in SSA were mainly focused on facilitating administrative work. Therefore the concept of a CDSS initially required detailed discussions to reach a common understanding regarding the conceptual possibilities of decision support at the point of care. In a second step, about 15 months into the project, a pilot software showing possible functionalities and options was programmed and representatives from all levels of the health system, including future users as stakeholders of the project were asked for their feedback in personal meetings. This was done after future users and individuals from district health authorities had had the opportunity to experiment with using the pilot software. In addition, local research partners investigated the prototype in detail. Feedback on functionality, style, and administrative necessities of the system were received and directed the development phase of the software. At that stage the future users and district health authorities were by now familiar with the concept of decision support, and were very active in expressing their expectations and in proposing changes and additional functionality. The development team was well briefed for the final programming and a first release candidate of the final version was rolled out after an additional 6 months. This release candidate underwent testing and was translated into local languages before being forwarded to the study sites 3 months later.

The test phase for the release candidates included a first part, where the system was tested with use cases and mock patients and a second iteration, where real patient data were entered retrospectively, without using the system for patient care. Care was taken, that a thorough test phase with all pre-planned steps was completed even when it became obvious that timelines for the actual start of the implementation were delayed. A repeated test phase with the second release candidate was added for a final feedback on all changes prior to implementing the final version. The detection of possible problems with the software before its use for patient care was given highest priority. Translation of the system proved to be challenging, whilst conducting the needed training sessions and feedback rounds took a great deal of effort on the part of the local research team. This too was considered essential to ensure that the software could be successfully introduced in the routine care at the study sites. The 3 iterations of testing release candidate versions with iterative programming to accommodate comments received from future users and the task of translations took an unexpected 12 months. Taking account of all the steps - as Figure 2 shows - the initial planning and assessment, programming and testing the pilot versions and programming and testing the release candidate versions - the final version was ready to be implemented for patient care about 3 years after the QUALMAT study consortium had met for the first time.

The software consists of 4 parts:

The user interface allows for the system to be operated with only minimal computer knowledge. The interface supports and proposes complete actions for antenatal care and facilitates the supervision of patients under delivery. Users are able to enter and review patient data, which will be stored in an xml database. The xml data may be used for data analysis or other project needs. Algorithms derived from the above mentioned guidelines screen values in the xml database.

In addition, documents may be added to the software at various locations, for example for training purposes. A set of international documents (mostly WHO) has been centrally included and local research teams have added documents from their national or district health authorities.

The decision support has three different approaches, identified as likely useful in rural health care environments:

The first approach is to provide decision support by electronic checklists, aimed to ensure that comprehensive information is available for decision making and all necessary actions for safe maternal care are observed and executed. This approach therefore supports thorough history-taking, physical examination, basic laboratory tests, as well as provision of counselling and preventive measures (Figures 3 and 4 give an examples). The results of these actions will be entered into the system and allow for review of all data from previous visits in the same way as in a patient’s chart.

The second approach is based on algorithms running in the background of the system that screen entered data and suggest diagnoses or alert the user about dangerous situations that require consideration during the visit (Figure 5). These algorithms consider information for example from the medical history, examination, basic laboratory test results. There are various options for displaying results of the analysis using the different algorithms. Algorithms may produce (i) different kinds of instant warnings (“watch dogs”, as shown in Figure 6), or they may (ii) integrate a broader amount of data to detect medical situations of concern and appear as more detailed hints at the end of a section, where diagnoses and actions (exams, treatment, or referral) are suggested. As an example, Figure 7 shows a situation, where the medical history revealed symptoms of blurred vision, the physical examination showed elevated blood pressure and the laboratory results showed proteinuria. At the end of the visit the caregiver was therefore asked to consider the diagnosis of preeclampsia. Within each recommendation the user is referred to the standard source (basic documents such as the WHO guideline itself or local guidelines if provided), which may be issued on demand (Figure 7, arrow).

The third approach for decision support is the provision of an electronic partograph, which provides continuous monitoring of the delivery process on the screen and provides the above mentioned features of watch dog or detailed recommendations in addition. Figure 8 shows a partograph, where the entry for labour progress remained in the yellow alert area at 4 hours after the active phase of labour has started. At this time, the caregiver will receive detailed instructions (Figure 9). At 6 hours the action line has been crossed. Then there would be a red prompt which explains, that actions would be urgently necessary (Figure 9). A similar graph with decision support follows the postpartum phase until 24 h post delivery or until discharge.

A CDSS for a resource-poor setting needs to strike a balance between the best technical possibilities and affordable, reliable, and robust technical solutions that can be used in rough rural environments. For the hardware environmental factors like high temperature, dust levels, and high humidity were taken into account; however the most suitable hardware, so called ruggedized laptops, was outside the tight research budget and may always be outside a health system’s budget in resource-poor countries. Therefore, affordable equipment was chosen that had been used previously in such settings and for which local maintenance could be ensured.

Study sites are equipped with one laptop per site with personal password protected accounts for each staff member involved in maternal care, as well as for supervisory staff. The infrastructure provides the possibility of having a laptop running 24 hours a day, seven days a week using mostly solar power and appropriate batteries. All staff received training to acquire general computer skills and specific knowledge required for safe handling of hard- and software and the collected data. As faulty hardware may be a powerful negative motivator and may thus interrupt local workflow and adversely affect patient care, round-the-clock technical support will be available provided by the local research team. A dedicated IT administrator visits each site regularly (at least bi-weekly) to identify and solve problems arising from soft- and hardware. This person is also responsible for the administration of local user accounts and passwords and for collecting data during the lifetime of the QUALMAT study. Having completed the appropriate training of future users in workshops, and after the iteration of test phases at the sites, the final version of the software has now been implemented and authorized for use in patient care at the intervention study sites in all three countries.

The effect of the CDSS on the quality of maternal care, on the competence and motivation of the providers, and the cost-effectiveness of introducing such software will ultimately decide upon the success of the system. Therefore our project will closely monitor the introduction of the CDSS to facilitate timely recognition if problems of acceptance, hardware, software, or workflow hinder the implementation. Early adaptations, corrections and trouble-shooting as well as continuous discussions with and support of users may catalyze the adoption of the QUALMAT CDSS. This approach was successfully taken in a Kenyan project, where the early detection of problems through evaluation and the provision of corrective measures for problems helped to overcome the initial frustration of users [22]. Therefore a study program has already started to closely evaluate the adoption, usability as well as the attitudes towards the CDSS and its effects on the workflow at the different sites. Quantitative and qualitative assessments before, at certain time points during, and after the intervention will be carried out using questionnaires, interviews, observations and focus group discussions and will be supplemented with additional computer log analyses.