Obstetric Emergency Supply Chain Dynamics and Information Flow Among Obstetric Emergency Supply Chain Employees: Key Informant Interview Study

Introduction

Background

For the past several decades, the Ethiopian Ministry of Health (MOH) has worked to decrease the maternal mortality ratio (MMR)—the number of pregnant mothers dying per 100,000 live births [1]. However, with the most recently reported MMR of 267, Ethiopia still ranks high in the MMR globally and needs additional interventions to lower the MMR to the sustainable development goal of 70 [2]. Therefore, the Ethiopian MOH is focusing on improving “the health systems capacity to offer quality care that meets women’s needs (the supply side)” [3,4]. Toward this goal, the Amhara Regional Health Bureau in Amhara and the Ethiopian MOH have identified the need for a real-time obstetric emergency readiness system, which will assist in measuring and monitoring facility-level readiness to manage the 6 most common basic emergency obstetric care needs (BEmOC), including a set of core drugs; commodities; and resources to identify emergencies, treat them, and monitor-modify therapy as clinically indicated [5].

Formative work to articulate the specific needs, barriers, and facilitators for the development and implementation of a real-time obstetric emergency readiness system is foundational to meeting required needs and ensuring that the system matches the environment in which it is deployed [6].

Objective

This study aimed to describe the obstetric emergency supply chain (OESC) dynamics and information flow through semistructured qualitative interviews with key informants from different levels of Ethiopia’s health care system. This information would be used to guide the design and eventual implementation of electronic dashboards as a component of the real-time obstetric emergency readiness system.

Methods

Theoretical Model

The “Sociotechnical Model for Studying Health Information Technology in Complex Adaptive Healthcare Systems” guided this study (Figure 1) [7]. The Sociotechnical Model is a dynamic and interconnected model that provides a thorough picture of the process for designing, implementing, and evaluating health information technology (HIT) through 8 interconnected dimensions: hardware and software, clinical content, human-computer interface, people, workflow and communication, internal organization features, external rules and regulations, and measuring and monitoring (Textbox 1) [7].

‎

Guiding Methodology

Guided by qualitative descriptive methodology, we conducted semistructured interviews to further describe and contextualize the gaps in emergency supply availability discovered in an earlier quantitative analysis of facility-level obstetric emergency readiness in the region [8,9]. The interviews explored multifactorial causes of emergency commodity stockouts, the current health system approaches to restocking commodities (from the federal, regional, and facility level), and information flow gaps and dynamics that could contribute to emergency supply stockouts. Data collection and analysis occurred concurrently [10].

Recruitment and Sample

We recruited a purposive sample of OESC experts from across the federal, regional, and facility level of care in Ethiopia. This included federal MOH government officials working in supply chain management and maternal survival in Addis Ababa, regional health bureau officials with obstetric expertise, regional pharmaceutical supply system employees, and supply managers and pharmacists from individual health care facilities. To participate in this study, individuals had to be at least 18 years of age, speak English and Amharic, and be full-time employees in one aspect of Amhara’s OESC.

Ethical Considerations

The study was approved by institutional review boards at Columbia University (IRB-AAAU2006) and Emory University (MOD005-STUDY00005335) and received ethics approval from Amhara Public Health Institute (NoH/RfftT1D1o7144). All participants provided verbal consent before participation. All data were deidentified before data analysis. Participants received no compensation for their participation in this study.

Data Collection: Semistructured Interviews

We created an interview guide with open-ended questions to explore the flow of information through the OESC, including barriers and facilitators (factors that enhance or impede access and use) with paper-based systems and the computer-based integrated pharmaceutical logistics system (IPLS; Textbox 2 and Multimedia Appendix 1). We conducted interviews in English (federal level) and Amharic (regional and facility level). The interviews were audio recorded and lasted approximately 60 minutes. Amharic is the national language, and individuals working in Ethiopia’s health care field received their training and education in English. During the interviews, we performed member checks by summarizing the main concepts and ideas the participants voiced and asked for confirmation to ensure we accurately captured their perspectives [11].

Data Management and Preparation

We stored audio files in OneDrive (Microsoft Corporation), an encrypted, password-protected electronic site accessible only to the research team members [12]. All interviewees were identified by a participant number linked to the OESC level in which that individual worked (ie, federal, regional, or facility level). The participant numbers were linked to their names and stored behind an encrypted, password-protected site. We checked the transcripts for accuracy against the audio recording in English or Amharic and deidentified them before sharing them with other research team members. We used ATLAS.ti (ATLAS.ti GmbH version 9) to manage, code, and analyze all qualitative data [13].

Data Analysis

We performed directed content (ie, deductive) and inductive analysis of the interview transcripts. The directed content analysis was guided by a codebook developed by KD and SB (Multimedia Appendix 2) based on the a priori constructs of the Sociotechnical Model dimensions [7,14]. During data analysis, we added codes to the human-computer interface dimension of the codebook to reflect the emergence of additional themes. KD independently coded the transcripts, generated themes, and documented the coding. SB reviewed the documentation and refined codes with KD to establish the confirmability of the coding decisions. We created saturation tables using a priori constructs of the Sociotechnical Model, concurrently with data analysis to document if and when data saturation, a measure of data adequacy that signals that no further data collection is needed, had occurred (Multimedia Appendix 3) [15]. We specifically used a priori thematic saturation to guide the decision on when to stop data collection [16].

Results

Overview

We conducted 17 interviews (federal level: n=5, 29%; regional level: n=5, 29%; facility level: n=7, 41%) from February 17 to March 17, 2023. We relied on the qualitative research experience of the data collectors (KD and YA) and data saturation (Multimedia Appendix 3) to guide the decision on when to stop participant recruitment. The sample was predominantly male (14/17, 82%), with an age range of 30 to 60 years, and with 2 to 38 years of experience working within the OESC (Multimedia Appendix 4). While we analyzed all 8 dimensions of the Sociotechnical Model, in the main paper, we focused on the 5 (62%) dimensions most pertinent to the development of dashboards for monitoring facility-level readiness to manage obstetric emergencies and reported on the other 3 (38%) in Multimedia Appendix 5.

Hardware and Software

In 2014, Ethiopia launched IPLS which allows facilities to place orders for medical supplies either electronically or through paper forms. IPLS also enables the central Ethiopian Pharmaceutical Supply Service (EPSS) and regional hubs to see reports on supply availability at the regional and national levels.

Two software systems are used within the OESC to monitor the movement of medical supplies, namely Vitas and Dagu. Vitas is used by central EPSS, which supplies the country with medications and medical supplies. The director general of EPSS and input officers have access to Vitas dashboards, which allows them to check supply availability and track commodities at both the federal and regional hub levels. Dagu is the software used at individual health care facilities. Dagu is used at the facility level to monitor human resources and service availability, but it does not have dashboards for medical supply logistics and movement. Some software in the OESC have components that function offline while other features are fully dependent on live internet access. Barriers (eg, unstable internet access and frequent electric power outages) and facilitators (eg, some IPLS components available offline) and related quotes are summarized in Table 1.

Clinical Content

Respondents at all levels of the supply chain agreed that several pieces of information are critical to ensure facilities are ready to provide BEmOC and to prevent emergency supply stockouts. In many health care facilities, pharmacists hand-calculate the last month’s consumption, average monthly consumption, and they forecast future monthly and annual supply demands. In contrast, some health care facilities used components of the electronic IPLS to calculate expiration and current quantity of supplies automatically. These respondents stated that the automatic calculations assisted in their consumption calculations. Overall, participants reported a desire for an automated system at all health care facilities that will calculate consumption, waste, and medication expiration in real time.

Upon receiving orders from hubs, health care facilities must know and report the number of items they originally ordered, the number of items received, the number of items missing from the order, who sent and received the shipment, and the date the supplies arrived at the health care facility. These pieces of information are stored in various data sources, such as bin cards and stock record cards resulting in a fragmented, nontrackable, and nonintegrated reporting system.

Furthermore, when individual health care facilities are determining how much medicine and medical supplies to order, they must also know the disease burden in the population that they serve, and which items are available for ordering. For example, in obstetric emergencies, facilities should know the estimated size of their catchment population, the number of mothers who give birth at their facility for antenatal care service, the number of pregnant women who attend the facility, and the number of mothers who are typically referred to higher facility levels for more intensive treatment. These complex variables are implicitly factored into health care facility orders, but there is no system to formally account for those variables in creating orders from facilities. Respondent perceptions of barriers (eg, lack of reliable, high-quality, real-time consumption data) and facilitators (eg, integration of medical supplies into one place) as well as exemplar quotes are displayed in Table 2.

Workflow and Communication

In the Amhara region, there are 918 health centers, 103 hospitals, and 19 supply hubs. To determine the type and quantity of medical supplies to order, a facility must first determine its monthly consumption and average monthly consumption to forecast future needs. When requesting for supplies, facilities can request up to a maximum of 4 months of stock. The head of each department or unit at the health care facility plans monthly consumption every 2 weeks and submits an internal request and reporting forms to the central store or pharmacy in the facility. The pharmacist reviews and fulfills these biweekly requests from the facility’s stock and also completes and submits requesting and reporting forms (RRFs) to their designated regional supply hub to refill facility-level supplies. Pharmacists submit these forms every 1 to 2 months depending on the size of the population they serve and may also submit an emergency RRF if they have less than 2 weeks’ supply of an item or experience a stockout. The RRFs are either submitted in paper form or electronically through the IPLS. A health care facility–level respondent describes the critical role of IPLS saying, “It acts as the system’s brain and is crucial for having the right medication in the right quantity and of the right quality at the right time.”

After receiving the RRFs through the IPLS, regional hubs review the forms and if they have the stock on hand in the hub, they create a stock transfer voucher and fulfill the request by providing the commodity to the facility. In addition, the hub consolidates the consumption reports and RRFs for the individual health care facilities under their jurisdiction and estimates overall consumption and forecasting for the upcoming months, which they report in an RRF and send to the central or national office of the EPSS for fulfillment every 1 to 2 months.

Central EPSS reviews the RRFs from their hubs and fulfills the orders based on consumption reports and local and national forecasting numbers. The hubs receive the products from central EPSS and review the orders to see if they received the full or partial order. The hubs compare this information to the requests they have received from their facilities and send the supplies to the facilities every 1 to 2 months. After receiving the order from their designated hub, health care facilities then review what they have received and dispense it throughout the various departments or units at the facility. They track the new deliveries and existing inventory with bin cards and stock record cards that are paper-based. A summary of the various data sources and communication systems present in this supply chain is presented in Multimedia Appendix 6. Figure 2 illustrates the procurement process. The barriers and facilitators related to workflow and communication are presented in Table 3 with sample quotes from participants at various levels of the system.

‎

Human-Computer Interface

Individuals interact with the OESC at the federal, regional, and individual facility levels of the system through a variety of human-computer interfaces. Multimedia Appendix 7 describes the needs of the 3 types of users as well as how they interact with the interfaces available at their level of the health system. In addition to the current interfaces, some respondents wanted the electronic IPLS to be accessible via a mobile app or on mobile devices so that they have greater access to the data, particularly when desktop or laptop computers were not available at the health facilities and alternative mechanisms were needed to use the electronic components of IPLS. Multimedia Appendix 8 summarizes the barriers and facilitators identified for human-computer interfaces.

Discussion

Principal Findings

We used the Sociotechnical Model to explicate participants’ perceptions of barriers and facilitators to securing and managing supplies within the OESC in Amhara, Ethiopia, and to identify intervention targets to increase BEmOC supply availability at the facility level as a key step for reducing Ethiopia’s MMR [7]. Changing one component within the OESC can influence the success of the system, but current barriers may impede or interact with the system in unique and unintended ways. An example of this would be if the EPSS increased the number of computers and tablets available to health care facilities but failed to explore users’ usability and ease of use concerns. If the technology is still viewed as difficult to use and users continue to prefer paper-based forms, then EPSS as a whole will not see the full benefits of increasing access to the technology. This type of outcome is highlighted in a recent scoping review in low- and middle-income countries, which included Ethiopia, and it identified lack of training and user involvement as barriers to successful digital health interventions [17]. To ensure the success of the electronic components of IPLS the researchers and individuals working within this field can build on the preexisting strengths found in the IPLS to rectify barriers and concerns for HIT-based real-time monitoring of BEmOC supply provision and availability.

The lack of access to computers and tablets led to lower use of the electronic components of IPLS; this is often the case at small and rural health posts. If facilities do not have the equipment to access IPLS, then they will continue to rely on the paper-based method [18]. When the MOH provided tablets to individual facilities, they reported they were more likely to use the electronic components of IPLS. However, even if facilities had the hardware available, if they did not have consistent access to the internet, the electronic systems were still of little use because some features of the technology are unable to work offline. These findings, while novel, as they relate to technology specifically tailored for OESCs, also affirm what is found in the current literature surrounding digital health research in low- and middle-income countries, which found that increased access to technology and consistent reliable internet led to increased use of the digital health technology [17,18]. Creating dashboards and electronic components of the IPLS that work offline is crucial because internet access is not always consistent in all parts of Amhara [19]. To encourage users to transition from paper-based forms and ensure facilities can leverage the benefits of the electronic system, the central EPSS and MOH should ensure the hubs and individual facilities have the physical supplies necessary to access the government’s technology infrastructure.

Users report the traditional siloed process for ordering medical supplies for each health program or clinical condition was tedious and time-consuming. Having all the supplies needed to manage obstetric emergencies in one place, along with supplies for other medical situations, makes the ordering process more streamlined. Furthermore, having all this information available in one place is useful when pharmacists and supply managers are trying to view inventory levels, determine their consumption, and forecast future needs. Pharmacists stated they enjoy having this information available to them electronically. As a developmental next step, the Ethiopian MOH could take steps to automate the electronic components of IPLS (internet-based and hardware devices) at all health care facilities, which will help with the calculation of monthly consumption, waste, and expiration so that pharmacists can have these current, real-time data available when they are making supply requests to the regional hubs. This recommendation is based on participant reports of calculation errors and is supported by the preexisting literature which found forecasting errors to be a concern with Ethiopia’s medical supply chain [20]. In addition, knowing what is available at various hubs can assist in making supply distribution decisions and advancing supply accessibility across hubs and the facilities they service, which was reported in a study conducted in South Africa [21]. Without the availability of rich inventory data, facilities will continue to remain in the dark about their true readiness to manage basic obstetric emergencies, and the central EPSS and regional hubs will not have the necessary information to provide essential emergency supplies prospectively, before stockouts and maternal deaths occur [20,22].

Facilities reported using the same process for ordering and obtaining supplies was incredibly helpful. Using one system to order all basic emergency obstetric supplies created a uniform communication flow between individual facilities, regional hubs, and central EPSS. However, as the IPLS ages, it will benefit from ongoing updates and adaptation to the evolving needs of the system’s users at multiple levels, which aligns with best practices stating that technology should be routinely modified and updated to both align with user feedback and comply with recent regulatory standards [23]. For example, providing space to justify seasonal differences in supply request quantities would be a direct response to user requests, allowing facilities to explain the change in demand, helping hubs and central EPSS to both forecast future needs based on seasonal trends, and also helping central planners make informed, strategic decisions even in cases where they must distribute a smaller number of supplies to multiple hubs [24]. Furthermore, similar work conducted in Zambia identified seasonal changes to be a driver of malaria medication stockouts in the country; therefore, Ethiopia’s modification to include justification for seasonal differences would also align with the preexisting literature [25].

The greatest concern related to the workflow and communication of the OESC was an inability to accurately forecast future supply needs. This is a critical issue because inaccurate forecasting can lead to undersupplying and stockouts or oversupplying and waste of medication due to expiration. This same issue was identified in a similar study of hospital laboratories in Gambela, a region in the southwest of Ethiopia, which emphasizes that this may be a countrywide issue and not only seen Amhara [26]. When individuals know their true consumption, they can analyze the trend in their past use patterns and forecast their future needs. Increasing data visibility is not only a concern at facilities for preventing maternal deaths but also crucial at every stage of the supply chain to ensure consistent, predictable emergency supply availability for reducing facility-based mortality at the population level in Amhara. Previous research in the United States demonstrated the utility of using multilevel dashboards (ie, facility and national levels) to monitor the distribution and inventory status of COVID-19 vaccines [27]. Currently, the greatest communication breakdown occurs between regional hubs and individual facilities. Bridging the gap between these 2 areas will not only assist with forecasting but also provide an additional level of surveillance at the regional hubs which can monitor inventory levels at facilities and prospectively send out supplies if they notice a facility is reaching critically low levels to prevent stockouts in advance.

Technology updates and changes must be reviewed and monitored not only to measure their success but also to investigate if there are any positive or negative unintended consequences of the technology updates [23]. This is seen in the Dagu 1 to Dagu 2 updates which improved data transparency within the system but also introduced barriers to efficient use of the system at the facility level. On the basis of user complaints from this update, it is imperative to conduct a formative assessment with end users to ensure the technology updates cohere with the HIT’s intended purpose and existing user workflow [28]. Furthermore, the Dagu 1 to Dagu 2 updates underscore the notion that usefulness does not always equal usability, and updates to technology should be reviewed by targeted end users to ensure they will not have unintended negative consequences. This same problem occurred when electronic health records were widely integrated into hospitals in the United States in the 2010s [29]. Before implementing changes within the electronic components of IPLS, those individuals who are leading these updates should establish a process to monitor and measure the impact of the changes.

The findings from these interviews highlight the need to bridge the data visibility gap between facilities and regional hubs. Furthermore, facility-level participants stated that they would appreciate the ability to view inventory data electronically and having that information available would assist in forecasting decisions and supply requests. The discussions with individuals at all levels of the supply chain underscore the importance of having data views that are tailored to the different job types and are supported by the current literature. Previous health technology research has found curated technology displays for different end users or job tasks to be beneficial [30]. Providing health technology, such as dashboards with different views for the various stakeholders in the supply chain is helpful because it can provide a granular view for each stakeholder depending on their information needs and job functions [31]. Different types of dashboards that could be useful would include dashboards for individuals who work at facilities and regional hubs. The individual facility employees can use the dashboards to assist with performing accurate forecasting and supply requests, as well as monitoring their day-to-day inventory levels. In contrast, the dashboards for region hub employees could assist with the hub’s ability to monitor the inventory levels of the facilities under their jurisdiction and provide supplies in advance of stockouts.

One facilitator of the current human-computer interfaces is that there is already strong buy-in from the users because they believe the technology is both useful for their jobs and easy to use. While some reported ease of use issues, most respondents indicated the existing EPSS system and IPLS were usable. New dashboards must maintain these existing usability strengths. Future steps can be taken to curate additional individualized dashboards that provide the information that is most pertinent for each stakeholder across the health system from facilities to regional hubs and the central EPSS. In addition, if individuals continue to use the technology on tablets or mobile devices, then the EPSS could ensure the technology is tailored for use on those devices instead of only computers. This incremental adaptation of the IPLS and EPSS interfaces to mobile technology can ensure that the supply monitoring technology follows usability heuristics for mobile health, which is critical to ensure success of the technology on the new platform [32].

Limitations

One potential limitation of the study is that when transcripts were translated from Amharic to English, there may have been unintended translation errors. However, to mitigate this concern, a medical anthropologist, who lives in Amhara and has >14 years of experience performing qualitative data collection and translating data from Amharic to English, performed the translations. In addition, while this work was designed for health care facilities in the Amhara region of Ethiopia, all regional- and facility-level participants worked in Bahir Dar city, the capital of the region. This limitation on sampling outside the regional capital was due to civil unrest which prevented team members from safely visiting other locations [33]. Despite this potential limitation, the study sample represents different health care facility levels and regional positions and still provides a rich understanding of the current OESC dynamics in Amhara. Finally, the sample was predominately male, which may have unintentionally excluded the opinions and experiences of female individuals working within the OESC. However, this male-dominated sample matches preexisting HIT user staffing in the region.

Conclusions

This study captured insight into the information and communication flow within Amhara’s OESC. Participant responses identified several strengths and barriers related to the success of the current IPLS. A frequently noted strength was the high degree of integration of medications and supplies that can be purchased through the system. Common barriers include a lack of data transparency at the facilities and also between facilities and regional hubs. The findings offer several recommendations for how future technology can be designed and tailored to meet the needs of current OESC employees. In addition, the results of the interviews underscored the importance of conducting qualitative research early in the developmental process of HIT to ensure a rich understanding of the current environment and user tasks that the technology will need to accomplish. Although these data come from Amhara and national MOH informants, the EPSS system and IPLS are common across Ethiopia’s regions. Consequently, HIT-based innovations resulting from this research can be readily adapted to the hubs and facilities in multiple Ethiopian regions.