1.1. Health Management Information System

An HMIS is designed to manage healthcare data by collecting, storing, manipulating, and transmitting patient data, by helping with hospital operational management and by supporting healthcare policy decisions at community, district, provincial, and national levels (Ariani et al., 2017). HMISs are critical in unifying communications through fully integrated audio-visual data communications and electronic systems. The architecture of an HMIS comprises three key levels of use and aggregation, namely, the patient-carrying level, the territory level, and the central government level. The patient level carries information about a patient’s medical history, whilst the territory and government levels carry aggregated information about the medical staff, departments, operations, and management policies, as well as other decision-making related information (Asemahagn, 2017).

A good HMIS should provide relevant information that shows where service is lacking and should highlight the intervention needed in such areas. It should also allow access to reliable, timely, and authorised data by all relevant people, including partners, for implementation of different programs that will improve the healthcare (Chatterjee, 2015). Information is considered an indispensable operational asset necessary for effective management and development of health services; hence the need for an HMIS to support management at all levels (Alwan et al., 2015). The result of using an HMIS will be quality, cost effective, and efficient healthcare services for all (Asemahagn, 2017).

The adoption of HMISs in Zimbabwe would provide horizontal and vertical linkages of health systems such that patient data can be shared across all health systems with the aim of providing improved healthcare to patients. It should encompass all the aspects of healthcare, such as patient data storage, management of healthcare, and allocation of resources. Typically, an HMIS comprises various subsystems such as billing, intensive care unit systems, outpatients, maternity, drug inventory, duty allocations, and many others and should be accessible throughout the whole healthcare system.

1.2. Public Health System in Zimbabwe

In Zimbabwe, the largest proportion of healthcare services is accessed through the public health system and supplemented by mission hospitals and sometimes by non-governmental organizations. The health delivery system in Zimbabwe is divided into a four-level structure: primary, secondary, tertiary, and quaternary. The primary level, known as primary healthcare, drives most of the healthcare programmes such as the essential drugs programme, expanded immunisation, communicable disease control, curative and essential preventive services, food production and nutrition education, health education, and maternal and child services (Ministry of Health and Child Welfare, 2009).

According to official studies such as the Zimbabwe Maternal and Perinatal Mortality study of 2007, the Zimbabwe Demographic and Health Survey of 2005/2006, and the Multiple Indicator Monitoring Survey of 2009, the Zimbabwean populace is succumbing to diseases that can easily be prevented or treated, such as diarrhoea, Human immunodeficiency virus (HIV) and tuberculosis (TB)-related illnesses, malaria, malnutrition, and injuries, and in most cases, deaths are recorded in unusually high numbers. The management and treatment of patients suffering from such illnesses and other chronic diseases takes place over a lengthy period and is carried out by different healthcare providers in different settings. To achieve higher quality and safety in the delivery of healthcare, information should be presented in a structured manner, ready for use in clinical decision making, which can be achieved with the implementation of HMISs that support online access to health records, secure transmission of medical information between systems, and shared longitudinal electronic health records (RTI International, 2014).

The current system from facility to district health information is paper based, becoming computer based on provincial health information to national health information (United Nations Development Programme [UNDP], 2014b), and as such there is little support for accessibility to health consumer information. Attempts at computerisation in Zimbabwe have seen the development of various systems targeted at specific programmes, diseases, or areas. The systems are not integrated, nor do they cover the entire healthcare continuum of patients, thus allowing for a national picture of healthcare provision. Starting in 1985, a Microsoft Access based system was commissioned by the Ministry of Health and Child Care (MoHCC) to collect and report on routine data, but it soon became slower and less efficient. From 2011 to 2012, the country implemented the District Health Information Software version 1.4 (DHIS1). In 2012 the MoHCC got help from The United States President’s Emergency Plan for AIDS Relief (PEPFAR) to pilot a DHIS-2 in Manicaland, and was scaled to other provinces in 2013 with the help of funding from UNDP, PEPFAR, and other partners (UNDP, 2020). Another system of note is the FrontlineSMS system launched by the MoHCC, a mobile-phone-based system used for transmitting early infant diagnosis of HIV from selected health facilities to the Zimbabwe’s National Microbiology Reference Laboratory.

In 2012 the government, through the MoHCC and in partnership with various donor agencies such as World Health Organization (WHO), UNDP, Joint United Nations Programme on HIV/AIDS, the National AIDS Council, Centre for Disease Control, and the Research Triangle Institute, started piloting an electronic patient management system for HIV and TB. The first phase began in 2013, with the system installed at 85 sites and covering 65% of all national patients; 161 more sites were added in 2014, taking the total to 246. The plan was to implement Phase 3 of the project between 2015 and 2016, adding the remaining 184 facilities and bringing the number of patients covered to 97% (UNDP, 2014a). At Chitungwiza Hospital, a system using the System Applications and Products in Data Processing (SAP) Health Software application had approximately 100 desktop computers in all departments and wards, and these were used as access points. The system was linked to a central server at MoHCC headquarters with a fast fibre-optic link. Members were given login details to access the system. The system was used for tasks such as clinical information storage, capture of patient demographic data, and handling accounts of patients (Techzim, 2014).

1.3. Problem Statement

Challenges with regard to the old manual and current systems include difficulties in maintaining manual records and in the preparing and sharing of reports, a lack of standard names and codes to be used in service delivery, difficulty in monitoring and evaluation of hospital performance without real-time data, and a lack of adequate data to perform evidence-based program management (Asemahagn, 2017; Tan, 2009). The need to have HMISs in hospitals is driven by the growing need for connectivity among healthcare providers, for patients to participate in their personal care, for the managing of market pressures associated with healthcare provision, and for exploiting the great improvements in IT that have improved systems interoperability (International Center for AIDS Care and Treatment Programs at Columbia University, 2015; Tan, 2009). The challenges being experienced with old technologies such as tapes and folders is now pushing executives and healthcare givers to consider the implementation of innovative HMISs to improve healthcare delivery (Herrick et al., 2010; Ross & Venkatesh, 2016).

Though the benefits of using HMISs are known, few studies have dealt with the critical success factors (CSFs) for their adoption in Zimbabwe. The ICT infrastructure has developed to an extent where it is possible to computerize some or all of the repetitive information processing in hospitals and to meet the required information standards in minimal time with less effort on the part of staff, patients, and healthcare professionals. The opportunity to access health care information, including patient hospital records, which is fundamental to the delivery of healthcare services, has been hampered by a lack of suitable information systems. Studies documenting the issues around the adoption of HMIS in healthcare institutions in Zimbabwe are not available. Proper evaluations and proof regarding the perceived benefits and challenges to adoption in Zimbabwe are generally lacking.

The goal of this research study is to ascertain the CSFs that drive the adoption of an HMIS in Zimbabwe. The factors from the technology, organization, and environment (TOE) theory are used with additional constructs from other organizational and individual theories, such as the diffusion of innovation (DOI) theory, theory of reasoned behavior (TRB), theory of reasoned action (TRA), and the technology acceptance model (TAM), in order to construct a modified theoretical framework for explaining the CSFs influencing the adoption of HMISs at the organizational level.

1.4. Theoretical Framework

The main theory underlying this study is the TOE model, modified with constructs from other adoption models. The TOE framework is the work of Tornatzky and Fleischer (1990) and depicts the process a firm goes through during the adoption and implementation of technological innovations in the context of technological, organizational, and environmental factors. The internal and external technologies available and relevant to a firm constitute the technological context. The resources available and the internal characteristics of the organization constitute the organizational context. The industry’s structure and size, the existing competition to the firm, the macroeconomic environment, and the regulatory framework and environment is what constitutes the environmental context (Tornatzky & Fleischer, 1990).

The TRA posits that an individual’s behavior is a function of their intention, which in turn is a function of their attitude and subjective norms (Fishbein & Ajzen, 1975).

The theory of planned behavior (TPB) by Ajzen (1985) is similar to the TRA as it shows the link between an individual’s behavior as influenced by intention, which in turn is a function of attitude, subjective norms, and perceived behavioral control (PCB).

The TAM by Davis (1989) has been widely used and accepted in adoption studies and it explains how perceived ease of use (PEOU) and perceived usefulness (PU) are regarded as the primary factors influencing an individual’s intention to use a new innovation.

The TAM2 developed by Venkatesh and Davis (2000) is an extension of the TAM model and incorporates additional constructs, namely social influence and cognitive instrumental processes. TAM2 uses subjective norm, voluntariness, and image as the three interrelated social forms in determining whether an individual will accept or reject a new innovation.

The DOI theory from the work of Rogers (1962) states that different people in a society will not adopt an innovation uniformly but will do so at different times and stages. Rogers (1962) identifies the five main factors responsible for affecting the adoption of an innovation as relative advantage, compatibility, complexity, trialability, and observability, which have a different extent of influence in the five adopter categories. These are the five factors used from DOI to assess the technological category in the next section on conceptual framework and constructs.

The relationships between the models used are shown in Fig. 1.

Fig. 1

Relationships between adoption models. TRA, theory of reasoned action; DOI, diffusion of innovation; TOE, technology, organization, and environment; TAM, technology acceptance model; TPB, theory of planned behavior.

1.5. Conceptual Framework and Constructs

The main research aim of this study was to find the technological, organizational, environmental, and individual factors that are critical for the successful adoption of HMISs in Zimbabwean healthcare institutions. Appropriate research constructs validated in prior studies were borrowed from the literature and used for questionnaire development.

The TOE model by Tornatzky and Fleischer (1990) was adapted to determine the factors influencing adoption of IT innovations at the organizational level. While numerous studies have used the model with varying modifications to explain how organizations adopt technology, not all the modifications have been related to the health industry or have included individual behavioral models such as TPB and TRA. To bridge this knowledge gap, the TOE model was integrated with individual behavioral models, and additional constructs, such as attitude (TRA), intention to use, PEOU (TAM), and PU (TAM), to get a better understanding of the dynamics of technological adoption in healthcare institutions. The constructs to be operationalized and measured are provided in Table 1.

The focus of this research is therefore to test the relationships among the constructs of HMIS adoption in the context of technology, organization, environment, and individual characteristics. In the investigation of technological, organizational, environmental, and individual factors affecting the adoption of HMIS by health institutions, 21 constructs were used in order to extend the existing knowledge base of technology adoption.

1.6. Research Questions

The specific research questions that are addressed in this article are as follows:

1.7. Research Model and Hypotheses

The model was derived from the combination of TOE, DOI, TRA, TPB, TAM, and other adoption models and was tested for relevance and confirmation in public health provision settings. The model and hypotheses are shown in Fig. 2.

Fig. 2

Research model and hypotheses: H1, Environmental factors influence technological decision making on adoption; H2, Technological factors influence technological decision making on adoption; H3, Organizational factors influence technological decision making on adoption; H4, Individual factors influence technological decision making on adoption; H5, Technological decision making on adoption influences adoption. ICT, information and communication technology.

Hypotheses 1-3 are direct derivatives of the original TOE model with additional constructs from DOI; hypothesis 4 is an addition, adapted from the mixture of TOE and technology acceptance theories (TRA, TPB, TAM, TAM2), and H5 is from the TOE model.

3.1. Research Approach

The mixed methods (pragmatic) researchapproach was used for the empirical study. It is a methodology that uses both quantitative and qualitative research techniques in one study (Creswell, 2014). Quantitative and qualitative methods were used concurrently for collecting both forms of data, labelled as the parallel-databases variant. The data from the two methods were kept distinct, that is, the data collection and analysis for the two approaches were done independently, and only combined at the interpretation stage, where the results were given equal priority in drawing conclusions. The collection and analysis of both sets of data occurred at approximately the same time.

The researcher compared the results for convergence during the interpretation stage. Where divergent results were obtained in quantitative and qualitative findings, the researcher considered the potential complementarity and completeness of the findings. Confirmation of findings was addressed in a later phase of the study by use of a third set of data collected to evaluate the model developed after the first two sets of data collection and analysis. The approach is depicted in Fig. 3.

Fig. 3

Convergence model of mixed method data analysis for the study. QUAN, quantitative; QUAL, qualitative. Adapted from Creswell, 2014. Research design: Qualitative, quantitative, and mixed methods approaches (4th ed.). Sage Publications.

3.2. Phase 1: Quantitative Data Collection and Analysis

This phase discusses the sampling, data collection, and data analysis of the quantitative data collected using the CSF questionnaire.

3.3. Phase 2: Qualitative Data Collection and Analysis

This phase comprised sampling, data collection, and data analysis for the qualitative data collected, that is, for the CSF interviews.

3.4. Phase 3: Model Confirmation

In this phase, another set of data was collected in order to confirm the results that had been collected in the first two phases and to clarify all discrepancies.

3.5. Reliability and Validity

Reliability and validity were ensured by monitoring the parameters of the instruments to maintain the same properties. A pilot test was used to ensure reliability by giving the instrument to different respondents in different health institutions to check the consistency of the results prior to the main data collection exercise. Validity can be ensured in two ways, namely internal and external. Internal validity ensures the obtained results are legitimate, and for this study it was ensured by the way that the respondents were selected (health workers), and the way that the data was handled, i.e., recorded and analysed.

The sampling criteria used provided the ability to externally validate or generalise the results. Although random sampling was used, the information can be deemed to be reliable to localisation of the research to a particular area where the respondents were typical or representative of the whole population. A high level of validity (internal and external) was achieved through strict protocol execution and study design.

4.1. CSF Questionnaire Results

The 122 completed questionnaires were captured into IBM SPSS statistics 20 for analysis and the results are presented below:

4.2. CSF Interview Results

This section presents the results of the interviews for answering the research questions and hypotheses.

After the coding of the responses from the interviewees, the themes and codes were identified and used to answer the research questions, as shown in Table 7. All the identified themes were mapped to one or more identified codes.

The results showed that environmental (external support, support infrastructure, government support, market scope, industry and competitive pressure), technological (relative advantage, trialability, technology availability, observability, compatibility, complexity), organizational (management support, size, communication processes, organizational readiness, ICT expertise) and individual factors (PEOU, PU, attitude, intention to use) do have an influence on the adoption of HMIS in Zimbabwe, as shown in Table 7.

More information on themes, sub-themes, and codes is provided in Table 4. All the codes and themes answer all four research questions as well as the fifth research question (“How can the CSFs be integrated into an adoption model for HMIS in public health institutions in Zimbabwe?”), helping us to come up with a model that explains the factors at play in the organization’s adoption of technology, as shown in the Fig. 4.

Fig. 4

Final health management information system adoption model. ICT, information and communication technology.

4.3 Model Confirmation Results

Because of the discrepancy between the CSF questionnaire and CSF interview results and in order to achieve completeness, complementarity, and triangulation, another instrument was designed and administered. The results of the two methods of correlation and thematic analyses are presented in the two sections below.

4.4. Summary of the Results

The results of the CSF questionnaire proved only three factors as significant for the adoption of HMIS, namely, the environmental factor construct of industry and competitive pressure, the technological factor construct of relative advantage, and the organizational factor construct of management support and available skills. The interview data showed that the following factors influenced HMIS adoption in public hospitals in Zimbabwe: the organizational factor constructs of communication processes, management support, organizational readiness, and ICT expertise; the technological factor constructs of technology availability, relative advantage, compatibility, complexity, trialability, and observability; the environmental factor constructs of external support, technology infrastructure support and government regulation; and the individual factor constructs PEOU, PU, attitude, and intention. Because of the discrepancy between the results of the interviews and the questionnaires, a third instrument was designed, including constructs that had been validated in either the interviews or the questionnaire. The set of the constructs were validated as shown in the HMIS model in Fig. 4.