Strengthening health systems through eHealth: two mixed-methods case studies at 10 facilities in Malawi
Item statusRestricted Access
Embargo end date30/06/2019
Background International agencies such as the World Health Organisation have highlighted the potential of digital information and communications technologies to strengthen health systems, which are underpinned by the ‘building blocks’ of information, human resources, finances, commodities, leadership and governance, and service delivery. In high income countries, evidence of the positive impacts of ‘eHealth’ innovations on the cost-effectiveness of healthcare is growing and many governments are now providing incentives for their adoption. In contrast, the use of eHealth in developing countries has remained low and efforts to introduce these new approaches have experienced high failure rates. There is even scepticism regarding the feasibility of eHealth in low-resource settings, which may be hindered by high costs, indeterminate returns on investment, technical problems and socio-organisational barriers. More research is needed to document both the value of eHealth for strengthening resource-limited health systems and the challenges involved in their implementation and adoption, so that insights from such research may be used to inform future initiatives. While many studies of eHealth for patient care in low- and middle-income countries (LMIC) are taking place, evidence of its role in improving administrative processes such as financial management is lacking, despite the importance of ‘good governance’ (transparency and accountability) for ensuring strong and resilient health systems. The overall objective of this PhD was to elucidate the enablers, inhibitors and outcomes characterising the implementation and adoption of a modular eHealth system in a group of healthcare facilities in rural Malawi. The system included both clinical and billing modules. The specific objectives were (i) to understand the socio-technical, organisational and change management factors facilitating or hindering the implementation and adoption of the eHealth system, (ii) to assess the quality of data captured by the eHealth system compared with conventional paper-based records, and (iii) to understand how information within the eHealth system was used for service delivery, reporting and financial management. A further aim was to contribute to the corpus of mixed-methods case studies exploring eHealth system implementation processes and outcomes (including data quality) in LMIC. As described in the following chapters, the research also gave rise to unanticipated and serendipitous findings, which led to new lines of enquiry and influenced the theoretical perspectives from which the analysis drew. Methods Mixed-methods case study was used for the research, taking a ‘soft-positivist’ approach to analysis, which encompasses both inductive and deductive forms of enquiry. Two case studies were undertaken in rural Malawi: one at a 300-bed fee-for-service hospital, and the other at nine primary care health centres that surround the hospital. At the outset of the research, the ‘logic model’ underpinning the eHealth system implementation programme was mapped, based on formative scoping to articulate the goals and intentions of those commissioning and supplying the eHealth system, along with literature-informed theory. This provided a framework against which to evaluate the processes and outcomes of eHealth system implementation at the ten facilities. For the hospital case study (Case Study 1), a retrospective single-case embedded design was employed, with outpatient and inpatient departments being the two units of analysis. Qualitative data included document review and in-depth key informant interviews, while quantitative data was obtained from the web-based District Health Information System (DHIS2), patient files and the hospital’s finance records. For the study of primary health centres (Case Study 2), a single-case embedded design was also used, with the rollout project as the case and the three units of analysis being 3 Early Adopter Facilities, 4 Late Majority facilities and 2 Laggard facilities. This case study used a prospective design, with data being collected 7 months and 24 months after implementation of the eHealth system due to a mismatch between the independent eHealth implementation project and the PhD research. Data sources included documentation screened against the criteria listed in the Performance of Routine Information System Management (PRISM) tools, information extracted from the eHealth system, health indicators drawn from DHIS2 and qualitative data from focus group discussions. In both case studies, framework analysis was used for qualitative data, while quantitative data was analysed by calculating data completeness, accuracy and agreement. Descriptive statistics and the Mann-Whitney U-test were used for analysing finance data in Case Study 1. Content analysis was also used to gain insights from Case Study 2. Results Based on the initial logic model, staff-, service delivery- and management-level outcomes were moderated through the organisational change management and socio-technical factors described below. Key organisational and process factors influencing system implementation Change management processes: Organisational strategies aimed at facilitating the introduction of the eHealth system included training clinical and clerical staff in the computer skills required to use it (see below) and adapting work processes to accommodate and optimise adoption. At the three health facilities where the billing module was implemented, the latter included introducing new procedures for providing electronic receipts to clients and service providers. At Madalo Hospital this also involved the creation of a new category of administrative staff with responsibility for managing the appropriate capture, entry and exchange of data using the system. However, such data clerks were only introduced within the inpatient department, whilst already overburdened clinical staff in the outpatient department were expected to integrate the eHealth system into their existing work routines. Outpatient departments at the health centres resorted to task-shifting patient data entry roles from clinicians to lower-educated allied staff such as janitors and security guards. Infrastructure and security issues: Organisational enablers were infrastructural and policy interventions aimed at securing equipment and patient data. These included installations of locks and burglar-proof bars, enhanced engagement of security guards and frequent backup of data. An organisational intervention undertaken at the health centres was the introduction of backup batteries and solar power, aimed at providing a continuous electricity supply. However, problems with battery depletion, frequent connectivity interruptions between the client computers and the server and electricity fluctuations and outages, affected both the efficiency of the batteries and the practical utility of the eHealth system. Highly efficient nano-computing units were later introduced, to reduce electricity demands and improve the consistency of available power for the purposes of using the system. Socio-technical issues arising during the implementation process Technical/software problems: There were 24 problems identified with the eHealth system, encompassing its design flaws, security protocols, and hardware and database limitations. For instance, entry of patient data was in multiple windows needing to be minimised, passwords expired with no one at the facilities with rights to issue new passwords, there were frequent disconnections between the client computers and the server, and lists of drugs and indicators for reporting in its database were limited. Although health centre staff used the system for backup storage and retrieval of data, only Early Adopters reported use of the eHealth system’s search function. Socio-technical issues: The technical problems outlined above resulted in a heavy reliance on paper records by the health centres, although centres varied in their attitude towards and persistence with eHealth system implementation, with Early Adopter sites overcoming most challenges. At the hospital, the eHealth system was subjected to such inappropriate use by staff that even establishing rules and an IT centre to regulate usage were ineffective, leading to a system crash in 2012 due to viruses and other malware. Such inappropriate use included staff depleting hospital server space by storing personal files (videos, music, pictures, games), being on Facebook instead of attending to patients, sharing of login credentials and not always logging off their account after use, and removal of cables from the computers. Leadership: At the hospital, there was strong management support for the eHealth system. In contrast, there were strong opinions from staff at Late Majority and Laggard facilities about the ineffective engagement of health facility “in-charges”. Further, many system champions were senior staff and thus busier and more mobile, most often leaving the junior staff at the health centres, who were not formally trained, to be using the eHealth system. Training: Limitations in the scope and number of staff formally trained was perceived to be a barrier to eHealth system adoption at the health centres, particularly lack of training in basic troubleshooting and maintenance. Even peer training lacked follow-up formal training. At the hospital, developing an appropriately skilled cadre of system users was hindered by high staff turnover and departmental rotations, which required frequent rounds of basic training. Staff at the hospital and health centres were nevertheless happy about the computer knowledge they had gained as a result of the implementation programme, although most expressed a lack of confidence in using the eHealth system. Technical support: For reasons including those already outlined, staff requested support for a range of hardware and software problems, not all of which it was possible to fulfil in a timely way, due to lack of sufficient IT personnel. Lack of in-country technical support for the software was also a considerable barrier to progress, particularly for the IT team based at the hospital, requiring requests for changes to be passed to the parent company. In one attempt to address this, the rights to a partial version of the software was passed to a local foundation for onward management, however the software developers were unwilling to release the source code so that further enhancements and customisation could be made. Efforts to recruit more hospital IT workers and reorganising responsibilities were frustrated by high staff turnover among the IT team. As a result, response to calls from health centres for technical support by the IT team was said to be slow and ineffective (except at Late Majority Facilities), and there was no transfer of basic troubleshooting and minor repair skills from the IT team to the health facility staff. Perceived outcomes: Despite the challenges described above, some tracer outcomes of the eHealth system were detectable from the qualitative and numerical results, relating to data quality, service delivery, reporting and decision-making, and financial management. Perceived and measured outcomes of eHealth system implementation Documentation and associated workload: In both case studies, implementation of the eHealth system illuminated the dysfunctional paper-based system, particularly loss of documents. At the health centres (Case Study 2), only Early Adopters reported reduced administrative and patient care workload following eHealth implementation, while the other adopter groups reported increased workload due to dual use of paper and electronic systems, as well as staff shortage and high patient load. Data quality: Both case studies reported poor data quality in the eHealth system, mainly due to the dual use of the paper-based and electronic systems, and staff defaulting to using the paper-based system only. This was aggravated by infrastructure and leadership problems at the health centres. Across the health centres, completeness of outpatient registration data in the eHealth system was 82.4%, as compared to DHIS2 (100.0% for Early Adopters, 73.9% for Late Majority), equivalent to an average monthly omission of 1,271 clients. When compared to DHIS2 data at Madalo Hospital, outpatient registration data in the eHealth system was 76.0% complete, under-reporting by an average 577 clients per month. Compared with the hospital’s paper-based records, inpatient registration and diagnosis data in the eHealth system, as entered by ward clerks, was 93.6% complete and 68.9% accurate. Service delivery (efficiency and patient experience): At Madalo Hospital, the eHealth system was reported to have made retrieval of patients’ paper files faster, as the implementation project had also led to changes in the hospital’s filing system. This new filing system also facilitated retrieval of data for patients with lost paper records, and allowed linking of patients’ outpatient and inpatient records. Reported service delivery improvements at the health centres included enhanced ability for tracing patients, treatment continuity, identifying the correct patient, ensuring patient confidentiality, keeping health workers alert and available, following clinical protocols, identifying the need to change prescription for (or refer) a recurrent patient, and reportedly showing the patient that the provider was paying attention. Improvements in patient experience were perceived to be through avoiding the need for patient details to be re-entered at subsequent visits, better management of queues, and patients feeling more understood by the service provider and having more confidence in the services. Perceived negative patient experiences were associated with staff members’ slow typing skills and unfamiliarity with the eHealth system, dual entry of patient information into both the electronic and paper systems, extra steps added to the patient journey through the care process, and disrupted patient-provider interaction. Efficiency of reporting: After its implementation at the hospital site, the eHealth system had become routinely used to generate data for measuring quality of care, and partly for national reporting purposes (HMIS). Customised reports for the hospital were created and used for decisions such as allocation of wards, advocacy and funding applications. In contrast, all the primary healthcare facilities were still using paper registers to compile HMIS reports, a few in combination with the eHealth system, because of lack of knowledge of the reporting module, poor design of the system’s reports, and disruptions in electricity and network connections to the server. Management of finances: Financial management was reported to have improved at Madalo Hospital due to better-quality data capture and tracking of service charges, separation of billing and receiving roles by recruiting ward clerks, enhanced oversight by management, and fraud prevention through greater transparency and accountability. Although median monthly revenue was significantly higher after eHealth system implementation (P=0.024), micro- and macro-contextual factors confounded this effect, and the descriptive and qualitative data revealed that genuine improvement only came about after recruitment of ward clerks towards the end of the study period. At the health centres, the eHealth system reportedly helped staff in the accounts department with billing, the facility in-charges with financial oversight, and clients with more trust in printed receipts. Conclusion Converging the results of these two case studies illustrates the potential of eHealth to strengthen LMIC health systems through developing human resource capacity (skills, staff roles), facilitating service delivery, and improving financial management and governance. However, realising such improvements is dependent upon understanding the socio-technical interactions mediating the integration of new systems into organisational processes and work practices, and implementing appropriate change management interventions. The results of this study suggest that, for effective implementation and adoption of eHealth systems, healthcare leaders should (1) recruit data entry clerks to relieve clinical staff, improve workflow and avoid data fraud, (2) facilitate appropriate data use among system users and an information culture at the facilities, and (3) strengthen knowledge and skills transfer from eHealth system developers to local implementers and system champions, to optimise responsiveness and ensure sustainability. Further interdisciplinary research is needed to obtain additional insights into factors affecting the quality of eHealth data and its use in the management of LMIC health systems, including the role of social, professional and technological influences on financial good-governance.