Exploring the potential of introducing an electronic death registration system in South Africa

Abstract

Background

Despite South Africa’s well-established Civil Registration and Vital Statistics system (CRVS) and good completeness of death registration, challenges persist in terms of the quality of cause of death information and the delayed availability of mortality statistics. The introduction of an electronic medical certification of cause of death (eMCCD) system may offer opportunities to improve both the quality and timeliness of this information.

Methods

This study used an exploratory mixed methods design to investigate perceptions surrounding an electronic solution for registering deaths in South Africa. We conducted 14 key informant interviews from 23 key informants invited, surveyed 208 out of 250 targeted health workers and engaged with more than 500 South African health professionals in a participatory workshop about the acceptability of introducing an eMCCD. Mentimeter was used to obtain feedback from the participants. Rapid qualitative analysis methods were used to analyse the key informant interviews and descriptive statistics for the survey and workshop data. During the interpretation phase, qualitative and quantitative data were integrated according to key themes that emerged from the data.

Results

During the qualitative interviews, the underlying factors mentioned as contributing to inadequate mortality data quality included insufficient MCCD training, diseases subjected to stigma, limited access to and quality of patient health information, and a significant proportion of deaths occurring outside medical facilities. More than 80.8% of the surveyed health professionals (168/208) rated the importance of mortality statistics as high, but the current quality was rated as low or very low by 29.3% (61/208). An eMCCD appeared to be acceptable to most workshop participants who registered on Mentimeter and was perceived as a means to strengthen the timeliness of mortality information. However, 43.0% (68/158) of the workshop participants who responded to this question on Mentimeter saw poor internet use as the main barrier to introducing an online system.

Conclusions

Our results shed light on critical aspects surrounding cause of death information in South Africa, as well as the challenges faced in maintaining the quality of such data. The findings suggest that the implementation of an eMCCD system could provide opportunities to strengthen the cause of death information in South Africa if it is designed to integrate into the current system and provide supplementary functionalities.

Background

The completeness of death registration in South Africa has improved significantly over recent decades. For adults, completeness increased from 50.0% in 1994 to over 90.0% by 2019, while registration for children under 5 years rose from 57.7% in 2000 to around 72.0% since 2012 [1]. Despite these gains, challenges persist regarding the quality of cause of death (COD) information and delays in the availability of official mortality statistics [2]. Currently the latest available data is for 2019, a time lag of 5 years [3]. These issues pose serious implications for public health planning and policymaking in South Africa. Without accurate, timely mortality data, it is difficult to identify the leading underlying causes of death or design effective interventions to reduce the burden of disease. Furthermore, the absence of timely mortality information limits public health authorities’ ability to respond rapidly to emerging health threats, such as infectious disease outbreaks or new drug-resistant infections [4–6].

The coronavirus disease 2019 (SARS-CoV-2) pandemic underscored the critical gap in COD data availability for public health purposes. In South Africa, a key barrier is the lack of access to COD information at the time of death registration, which hinders real-time decision-making. Additionally, the long delays in processing mortality data and the release of summary statistics, compounded by disagreements among key stakeholders over the use of provisional COD data for surveillance, further impede effective response [5]. The quality of COD data from the Civil Registration and Vital Statistics (CRVS) system is also concerning, with issues such as a high proportion of unusable codes [7], underreporting of human immunodeficiency virus (HIV)/acquired immunodeficiency syndrome (AIDS) deaths [8], and inaccurate profiles of injury-related deaths [9]. The current paper-based death registration system involves multiple manual steps, from the time of death until the data reach the national death statistics office [10].

Several countries have adopted electronic death registration systems (EDRS) to enhance the completeness, quality and timeliness of mortality data. For example, Denmark [11] and Peru [12] have made significant strides by implementing EDRS that streamline data collection processes and improve data quality. These electronic systems enable real-time data entry with built-in quality checks, centralized data management, and efficient data storage, analysis and sharing. The benefits of EDRS include more accurate data, quicker access to mortality information, and the ability for health authorities to promptly identify and respond to mortality trends, thereby improving public health surveillance [13, 14].

However, the implementation of an EDRS is complex and requires careful consideration of several factors. These include the country’s legislative framework, coordination between departments, information technology infrastructure, and stakeholder buy-in [15]. In South Africa, one key consideration is the need for regulatory changes to facilitate the timely use of provisional COD data for public health purposes, with the understanding that official statistics will be released later. Additionally, confidentiality and data security concerns must be addressed, given the sensitive nature of mortality data.

In response to these challenges, the South African Medical Research Council (SAMRC) initiated a scoping study to explore the feasibility of an EDRS in South Africa. This study involved a review of international EDRS best practices and global initiatives [16], including a participatory webinar to consolidate lessons learned from these initiatives. A technical landscape analysis was also conducted to assess the current CRVS process and evaluate the health information and demographic systems that could potentially be integrated into an electronic medical certificate of cause of death (eMCCD) or EDRS system in future [17]. This analysis led to the development of a proposed model for a South African system. Further, a detailed technical review of international eMCCD processes was undertaken to identify potential requirements for a local system [18].

As part of the scoping study, we also explored healthcare workers’ perceptions of the potential introduction of an eMCCD system in South Africa, aiming to gauge the acceptability and practical considerations of such a transition, which is the focus of this paper.

Methods

Study design

This was an exploratory mixed methods study aimed at investigating the potential of introducing an eMCCD in South Africa. The study commenced with key informant interviews to gain insights from stakeholders regarding the death registration system. Building on the findings from these interviews, we designed a survey targeting health workers to quantitatively assess their perspectives on mortality statistics, including their perceptions of the challenges and opportunities associated with implementing an eMCCD in South Africa. The study culminated in a final participatory workshop designed to explore strategies for strengthening the death registration system.

Data collection

Key informant interviews

Key informant interviews (KII) were conducted with stakeholders, including medical practitioners and funeral undertakers. Participants were recruited purposively using a snowball approach to obtain a diverse range of individualized experiences. Identified stakeholders were sent a letter informing them about the project and requesting permission for the project staff to contact them for an online interview.

All interviews were conducted by phone using Microsoft (MS) Teams. Key informants were given an information sheet at the beginning of their interview. If consent was provided, verbal consent and the the full interview was audio-recorded. The interviews followed the interview guide that was developed and tested for this study. Two of the researchers (N.K. and S.G.) received training in qualitative methods. Probes were used to elicit additional information on the basis of the course of conversations and the interviewer’s observations.

Survey with health workers

After a rapid analysis of the key informant interviews, we designed the survey to assess the perceptions of an EDRS and to identify major challenges with the current death registration system. The survey questions were pretested by the project team prior to conducting the online survey.

A convenience sample of 250 medical and allied health practitioners was sought. The survey was distributed through existing professional associations, including the Health Professionals Council of South Africa (HPCSA) and the South African Medical Association (SAMA), as well as via South Africa Medical Research Council's (SAMRC) social media accounts (Facebook and Twitter). To ensure that input from rural doctors was obtained, we attended the Rural Doctor’s Association (RUDASA) annual conference and distributed a flyer asking participants to participate in the online survey. To ensure wide distribution across South Africa, the online survey was also disseminated to all health workers registered with the Knowledge Hub, a professional development platform for the National Department of Health (https://www.knowledgehub.org.za/).

A digital consent form was included in the survey. Participants were asked to provide online consent before being able to access the questionnaire. No personal identifiers were captured, but demographic and occupational information was requested from the participants.

Participatory workshop

This online participatory workshop was conducted through a knowledge hub webinar hosted by the National Department of Health. These webinars are presented regularly as part of continuing medical education and have a wide audience. These included health workers (medical doctors, health managers, nurses, paramedics, and pharmacists) from across South Africa. We familiarized participants with the CRVS in South Africa, emphasising the current collection and processing of cause of death information, and during the discussion, we used Mentimeter to question the adequacy of South Africa’s current cause of death data, potential enhancements, and the role an eMCCD would play, as well as the barriers and opportunities for such a digital system. All questions were presented in a multiple-choice format.

Data analysis

Rapid qualitative analysis

The key informant interviews were analysed using rapid qualitative methods [19]. The information collected during the interviews was compiled and summarized using a rapid assessment procedure (RAP) sheet. The RAP sheet was created using the domains of acceptability and feasibility, which allowed the findings to be categorized using a deductive approach [20]. This approach allowed the researchers to immerse themselves in the data and permitted early identification of findings as the data were being collected. Thereafter, inductive coding was used to identify emerging themes. Representative quotes were extracted from the audio recordings and categorized in the RAP sheet.

Rigour and validity were established by independently coding and assigning the data to themes by a second researcher using the recording, followed by discussions between the investigators. After initial analysis using the RAP sheets, selective transcribing allowed the extraction of quotes for the different themes [19]. Only selected sections of the recording were transcribed.

Descriptive analysis of the online survey and participatory workshop data

The survey data consisted of categorical variables that were described as proportions. A stratified analysis was also performed that compared the responses of doctors and nurses.

A descriptive analysis of the responses to the closed questions captured on Mentimeter was conducted. Similarly, after coding the responses to the open-ended questions, a descriptive analysis of these data was performed. Survey responses were stored in a password-protected computer, and descriptive analysis of the survey results was conducted using the statistical program R Studio v3.3 [21].

Ethical review

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee (SAMRC human research ethics committee, EC009-4/2022) and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. This project was reviewed in accordance with Centres for Disease Control (CDC) human research protection procedures and was determined to be non-research public health surveillance.

All participants in the study provided informed verbal consent, which was recorded, after being given a written study information sheet and a verbal explanation of the consenting process.

Results

Key informant interviews were conducted with 14 of the 23 stakeholders we contacted, including medical practitioners (8/14), health managers (3/14) and funeral undertakers (3/14), from across four provinces: Western Cape, Eastern Cape, Gauteng, and KwaZulu-Natal. The participants included seven males and seven females. The online descriptive survey was completed by 208 health workers or 83% of the target of 250. Further demographic information is provided in Table 1.

Table 1.

Descriptive statistics of survey sample stratified by health setting. The potential of an electronic death registration system in South Africa: a feasibility and acceptability study (July—November 2022)

	Urban N = 136 n (%)	Rural N = 58 n (%)	Other N = 14 n (%)	Overall N = 208 n (%)
Age group (years)
< 20	1 (0.7)	0 (0.0)	0 (0.0)	1 (0.5)
20–29	8 (5.8)	1 (1.7)	0 (0.0)	9 (4.3)
30–39	30 (22.1)	20 (34.5)	3 (21.4)	53 (25.4)
40–49	32 (23.5)	16 (27.6)	5 (35.7)	53 (25.4)
50–59	42 (30.8)	13 (22.4)	3 (21.4)	58 (27.9)
60–65	17 (12.5)	4 (6.9)	3 (21.4)	24 (11.5)
> 65	6 (4.4)	4 (6.9)	0 (0.0)	10 (4.8)
Gender
Woman	78 (57.7)	35 (60.3)	8 (57.1)	121 (58.6)
Man	55 (40.7)	23 (39.6)	6 (42.9)	84 (40.6)
Prefer not to say	2 (1.5)	0 (0.0)	0 (0.0)	2 (1.0)
Province
Eastern Cape	6 (4.8)	4 (8.7)	3 (21.4)	13 (7.0)
Free State	6 (4.8)	2 (4.4)	1 (7.1)	9 (4.9)
Gauteng	49 (39.2)	8 (17.4)	4 (28.6)	61 (33.0)
KwaZulu Natal	14 (11.2)	8 (17.39)	0 (0.0)	22 (11.9)
Limpopo	2 (1.6)	6 (13.04)	0 (0.0)	8 (4.3)
Mpumalanga	6 (4.8)	6 (13.04)	0 (0.0)	12 (6.5)
North West	5 (4.0)	1 (2.2)	2 (14.3)	8 (4.3)
Northern Cape	4 (3.2)	4 (8.7)	1 (7.1)	9 (4.9)
Western Cape	33 (26.4)	7 (15.2)	3 (21.4)	43 (23.2)
Role
Clinician	60 (44.1)	20 (34.5)	4 (28.6)	84 (40.3)
Nurse	23 (16.9)	12 (20.7)	2 (14.3)	37 (17.8)
Allied health professionals	21 (15.4)	10 (17.2)	2 (14.3)	33 (15.9)
Public health	7 (5.2)	3 (5.2)	2 (14.3)	12 (5.8)
Management	17 (12.5)	11 (19.0)	2 (14.3)	30 (14.4)
Other	8 (5.9)	2 (3.5)	2 (14.3)	12 (5.8)

We also conducted an online participatory workshop with approximately 500 health workers, including medical doctors, nurses, health managers, and other allied health professionals. Mentimeter was used to interact with workshop participants; however, not all participants downloaded mentimeter and responded to the questions. We thus used the total number of responses received for each question as the denominator for this analysis. These denominators are indicated in the Results section below.

The findings of each data collection method (interviews, online descriptive survey and participatory workshop) are presented by the main themes that emerged from the data in Table 2. To provide a more complete and nuanced description of the acceptability, we merged the qualitative and quantitative data, including statistics with key themes.

Table 2.

Overview of the quantitative and qualitative findings from the study, The potential of an electronic death registration system in South Africa: a feasibility and acceptability study (July—November 2022)

Emerging themes	Questionnaire answers			Interview perceptions and quotes
	Survey (N = 208)	Online workshop questionnaire (N varies by question)
Importance of cause of death data	80.8% (168/208) indicated mortality statistics were very important, giving a rating of 5/5, 16.8% (35/208) gave a 4/5 rating for the importance of mortality statistics and < 3.0% gave a rating of 3/5 or less 82.1% (69/84) of doctors and 75.7% (28/37) of nurses gave a rating of 5/5 for importance of mortality statistics, 16.7% (14/84) of doctors and 18.9% (7/37) of nurses gave a rating of 4/5, 1.2% (1/84) of doctors and 5.4% (2/37) of nurses gave a rating of 3/5 or less	When asked “How could these data (mortality data) help with public health planning and monitoring?”   (N = 171) Respondents reported: - 38.6% (66/171) stated for health planning, prevention, and interventions - 24.0% (41/171) stated to determine resource allocation - 14.6% (25/171) stated to identify disease priorities - 8.2% (14/171) stated for health policy development -8.2% (14/171) stated for public health surveillance 6.4% (11/171) stated to identify  areas of training		“Death statistics are one of the only pure sources of data in terms of understanding what is happening in the community… to identify (health) needs.” ID6, female, health facility manager “Critically important … if we had better and more complete recording on cause of death data, I think it would be really useful to really understand the burden of disease in South Africa and mortality and to do more nuanced analysis around population groups.” ID2, female, public health specialist “Important, especially when it comes to funeral insurance. Most funeral directors have funeral insurance packages, so the mortality statistics are used there….it also helps control and manage your business better.” ID12, male, funeral director
Perceived quality of cause of death data	46.2% (96/208) rated the quality 3/5 29.3% (61/208) rated it  2/5 or less 12.5% (26/208) rated it 5/5 (good quality) Among providers, 52.4% (44/84) of doctors and 48.6% (18/37) of nurses gave a rating of 3/5 for the quality of current COD data A much higher percentage of nurses rated the quality 5/5 compared with doctors [32.4% (12/37) of nurses versus 2.4% (2/84) of doctors]	When asked “Why would you be concerned with the high proportion of ill-defined causes in a mortality data set?”   (N = 85) Respondents reported: - 72.8% (62/85) stated “it would bias the data and affect its usefulness for planning.” -15.3% (13/85) stated “it shows the lack of training of medical certifiers and a lack of understanding of the importance of the data.” - 11.8% (10/85) had other concerns		“Probably quite bad, particularly for people that die at home, you do not necessarily know what they die of. For people who die in-hospital … that is fairly reliable information, so it depends on where you are.” ID9, female, medical doctor “High percentage of unknown causes, which leads to poor quality data” ID10, female, medical officer “Not accurate, take an example, during COVID, there were a lot of COVID deaths, … we had families that insisted that (COVID) should not go on the death notification. It distorted the figures.” ID12, male, funeral director
Drivers behind the quality of death data	Not applicable	When asked “How could cause of death data be made fit for purpose (N = 115) Respondents reported: - 43.5% (50/115) stated “improve training in MCCD and ICD-10 coding” - 20.9% (24/115) stated “introducing an online, centralized system” - 12.3%(14/115) stated “improve data capture accuracy” - 7.0% (8/115) stated “timely release of the data” - 7.0% (8/115 stated “improve access to the data” - 4.3% (5/115) stated “integrate mortality with other systems for example, Dept of Justice.” - 2.6% (3/115) stated “Implement audit of MCCD/M&M meetings” - 1.7% (2/115) stated “Review/update MCCD form” - 0.9% (1/115) stated that system design should be simple		“Junior doctors don’t understand that the form is for statistical purposes, they think the form is just for legal purposes….when they fill in the form….they want to tell the whole story…so they do not get sued.” ID6, female, medical director “There are a lot of deaths at home … so there are a lot of deaths notified where it is mostly guesswork and I do not think there is a lot we can do about that” ID10, female, medical officer “Resources, obviously, most of our people do not have a regular physician that they attend … stigma as well. Certain diseases people do not want to be associated with.” ID12, male, funeral director “The focus must be on the quality of the data that is submitted and that really does require having the right system, the right tool, and the right training of healthcare providers to submit their data.” ID2, female, public health specialist
Acceptability of an EDRS	“Would you use an electronic system?” - 144/208 (69.2%) responded “yes” - 45/208 (21.6%) responded “maybe” - 19/208 (9.1%) responded “no”	When asked to identify the barriers to implementing an online system (N = 158) Respondents reported: - 43.0% (68/158) said “poor internet access.” - 19.6% (31/158) said “poor connectivity owing to power outages.” - 9.5% (15/158) said that poor computer literacy would be an issue - 8.9% (14/158) felt that poor funding and limited staff would be a problem -6.3% (10/158) said Corruption and theft of cables, electronic devices, and funding -5.1% (8/158) said “Integration of data from different sources/organizations” - 7.6% (12/158) identified other barriers		“It will stop papers from getting lost” ID10, female, medical officer “It will get the data to Stats SA5 quicker than the paper copy and maybe that will decrease the time lag of the publication of data “ ID10, female, medical officer “It would make our lives so much easier…removing the person (at Home affairs) behind the computer…” ID11, male, funeral director “I think it is possible to have an online system … that will give you immediate analytics and summaries which would be very powerful data, for action” ID2, female, public health specialist

COD Cause of Death, MCCD medical certificate of cause of death, ICD-10 International Classification of Diseases 10th revision, EDRS electronic death registration system, Stats SA Statistics South Africa

Perceptions about the importance, uses and quality of mortality statistics. (2388)

The overall perception of stakeholders was that mortality data are very important but that the current quality of cause of death information in South Africa is suboptimal.

Key respondents mentioned that COVID-19 had highlighted the importance of mortality statistics and that the quality and timeliness of SA cause of death data were suboptimal, especially in rural settings. Among survey respondents (N = 208), most rated the quality of statistics as average (46.2%), and 29.3% rated the quality as poor (score of 2 or less out of 5). Most workshop participants thought that the current mortality statistics are not fit for purpose.

Some doctors expressed a desire to use mortality statistics at the local hospital level to monitor trends within a facility. Currently, mortality statistics are reported at the national, provincial, and district levels, with mortality data available at only the national or provincial level and no feedback returned to the district level or health facilities.

• “When one is asked to fill in forms and collect data, it is much more motivating when you see your data are getting used and you have access to that data, but when it goes into a central repository that we never have access to…it should be easy to have access to it.” (ID13, male, senior medical doctor and researcher)

The workshop respondents (n = 171) indicated that mortality data are useful for health planning (38.6%), resource allocation (24.0%), identification of disease priorities (14.6%), health policy development (8.2%), public health surveillance (8.2%), and identification of areas of training (6.4%).

Several drivers of the poor quality of mortality statistics were identified. The interviews highlighted the lack of medical information available to the certifying medical doctor for the decedent. Almost half of all deaths in South Africa occur outside of a health facility[22], and medical records for the decedent are often unavailable, thus the certifier has to rely on information provided by the emergency medical services (EMS) staff or from family or friends accompanying the decedent. For in-facility deaths, after hours staff, who may not have been the treating physician and thus unfamiliar with the patients medical history, are expected to complete MCCDs. Many doctors are not aware of the public health importance of the MCCD and thus do not prioritize completing the MCCD. It is seen as an administrative task that is left to the end of a shift or for the next shift.

• “Initially, I viewed filling in the form as more of a chore, something you would get the intern to do, but now thinking about public health and the bigger picture….., I realise it is very important.” (ID8, male, medical doctor)

• “There is not that internal incentive to strengthen the quality of the information because (for the staff) it just seems pointless…” (ID3, male, public health specialist)

The responsibility for completing the MCCD typically falls on junior doctors, most of whom lack clinical experience and MCCD training, which in turn can lead to incorrect diagnoses in the medical records and errors on the MCCD. Other reasons contributing to poor quality data highlighted by interviewees included the design of the MCCD form and a reluctance by certifiers to report certain stigmatized conditions (for example, HIV) on the death certificate.

MCCD training and form

Overall, 8 of the 14 KIIs reported having received no formal training in the MCCD. In total, four reported formal training later in their careers, and two received an undergraduate lecture at university but felt that it was insufficient. ID3 reported receiving formal training, but this had occurred many years earlier. Similar results were found among the survey and workshop participants. More than half of the survey participants reported no training in the MCCD. For individuals who did receive training, the majority stated that it was informal training or on-the-job training.

• “Formal training is definitely needed; most things are quite self-explanatory but particularly around the cause of death. I think people struggle with primary and secondary causes of death, and I think training to standardize these understandings would be useful.” (ID2, female, public health specialist)

Several interviewees expressed the need for access to real-time advice on MCCD at their workplace and suggested that facility MCCD audits would be useful for training purposes. When probed specifically about audits, some interviewees reported that their facility has had audits. In one of the interviews, the interviewee reported that she reviews every completed MCCD in her health facility and provides feedback directly to the doctors who completed the MCCD.

• “They make a copy of the last cause of death page (of the MCCD) and bring them to me every few days….if I see the doctors are making errors, especially the senior doctors, I provide feedback to them immediately.” (ID6, female, health facility manager)

Although doctors are ultimately responsible for what is reported on the South African death notification form issued by the Department of Home Affairs (DHA-1663), nurses often complete the demographics section of the form before the doctor completes and signs the MCCD form. This is time consuming for the nursing staff. “For me as a doctor, the process is pretty easy—the nurses do the main bulk of the work and I fill in one or two pages, so the process takes me 5 min. However, for nurses, it is labour intensive; there is only one book of the MCCD in the hospital, so they first need to locate the book, find it, then bring it to the ward where the doctor is, fill it in, wait for the doctor, and then take it back to the funeral undertaker. For the nurses, it is labour intensive” (ID8, male, medical officer).

Funeral undertakers complete a section of the MCCD as part of an administrative process, which includes the collection of next-of-kin (NOK) information. They must approach a doctor to complete the MCCD and then register for death at the Department of Home Affairs (DHA). Where no family doctor is available, the funeral home appoints an external doctor to certify the patient’s death and confirm that it is owing to natural causes. Funeral undertaker interviewees highlighted the challenges of registering a death at the DHA. They mentioned how each DHA office works differently, making it challenging for families to navigate the processes.

• “The process of registering a death is a big burden for families—unfortunately, there is no uniformity within Home Affairs, every office applies their own rules and regulations.” (ID12, male, funeral director)

Potential for an electronic death registration and MCCD

The majority of interviewees, survey respondents, and workshop participants agreed that an online MCCD system would be acceptable in South Africa.

The doctors interviewed expressed concern that the current paper forms could be lost and that the information was not secure. They felt that an electronic system would overcome these issues, be less time-consuming, and increase the accuracy of the data collected with built-in checks.

• “It could help the doctor fill in the right form, for example, was it a live birth? They then get directed to the correct form, depending on the answer provided.” (ID14, male, head of the clinical department)

The funeral directors were also positive about an electronic system, stating that it would save time by eliminating the need to travel to the DHA office and would be more flexible by enabling registration of a death at any time of day. They explained that they would need a hybrid system (electronic and paper combined), as citizens still require a printed hardcopy of the death certificate to access services. The death could be registered online, and then the printed hardcopy of the certificate could be immediately provided to the waiting relatives.

Despite the generally positive opinions, workshop participants (N = 158) identified several potential barriers to the implementation of an eMCCD including poor internet access (43.0%), power outages (19.6%), poor computer literacy (9.5%), lack of funding and staff (8.9%), corruption and theft of hardware (6.3%), integration of data from the three key agencies (DHA, NDoH and Stats SA) in the CRVS system (5.1%), resistance to change (2.5%), inter-agency co-operation (1.9%), and issues around confidentiality (1.9%). The interviewees mentioned similar challenges, including access to computers/internet in some facilities, as well as staff who are not computer literate and could have difficulty adapting to an electronic system.

• “I am very wary of electronic things because they depend on all things being perfect, everyone having access…. availability of the internet, everyone being tech savvy, which is not always the case. Having a system with someone to drive it would be important.” (ID7, female, hospital consultant doctor).

For the potential success of an electronic solution, several recommendations were given (see details in Table 2). Most survey participants and interviewees indicated that in-house data quality checks to ensure a valid underlying cause of death should be included in the tool. The interviewees also suggested that automatic International Classification of Diseases (ICD)-10 coding would be very helpful. They also suggested an internal quality review function, allowing access for senior doctors only, to enable review of the form before uploading. The respondents stressed the importance of a simple, user-friendly interface, with the collection of only the most critical information.

The respondents also considered the integration of electronic solutions with other departmental information systems crucial [for example, Department of Home Affairs, Stats SA, Department of Justice and the District Health Information System (DHIS)]. Further integration with other systems would improve efficiency by reducing the data capture workload, as well as potentially allowing access to decedents’ laboratory results, which could aid in determining the cause of death and could allow for cancelling of chronic prescriptions and prevent wasteful dispensing of medication.

• “I think it would be great if it was integrated with DHA. If I put in an ID number, then everything else (demographic information) would self-populate.” (ID13, male, senior medical doctor and researcher)

Several ideas were proposed to ensure digital security and prevent fraud, such as allowing access to the EDRS via a fingerprint or face ID or using the staff number or ID number of the doctor.

All respondents agreed that once the certificate is uploaded, it should not be changed. If editing is needed, only authorized personnel should have access to a code, and only a few people should have such a code. Separate levels of access could be granted, depending on a person’s authorized level of access to the system.

At funeral homes, death certificates use a very particular type of printing paper, so one interviewee mentioned that they would need controlled access to this paper and a process of allocating and tracking the paper to prevent the possibility of fraudulent certificates being used.

The interviewees were asked specifically if an online system would be preferred to an offline system. There were mixed responses. Many answered that an online system would be better, with benefits being immediate access to analytics that could be used for decision making and access to integrated systems (for example, DHA). Others were hesitant, citing rampant connectivity problems, and felt that an offline system that allows for the automatic upload of data when online would be a better solution (Table 3).

• “There has to be an option to fill in the MCCD, even if there is no internet…it would be truly frustrating if a patient comes from far away and we could not do anything because there is no signal or there is no power…” (ID8, male, medical officer)

Table 3.

Summary of the recommendations suggested by the key informants. The potential of an electronic death registration system in South Africa: a feasibility and acceptability study (July—November 2022)

Challenge/concern	Recommendations
Poor quality of cause of death (CoD) data	· Digital tool, with features, such as skip-logic and quality checks, to ensure accurate capturing of cause of death · MCCD training of certifiers with emphasis on the importance of accurate CoD data, and CoD causal sequence
Reservations about a digital system	· Use of an offline system which allows information to be automatically uploaded when the tool is connected to the internet · Computer literacy training for staff
Choice of device/platform	· Web-based platform, with software that can be used across devices, so individuals with limited access to a computer can use their mobile device · Accessible on mobile devices, as this is the most easily accessible device. Issue of privacy and confidentiality can be addressed by ensuring individual access codes for each user, with additional protection, such as biometrics or face recognition (featured on most smartphones)
Desired features of the tool	· In-built quality check features for data verification, for example · ID number and date of death to confirm age · Pop-ups that can prompt doctor to check ill-defined conditions and causal sequence · Drop-down boxes and tick-boxes where possible · Integration with DHA death registration system which could improve efficiency of operations · Integration with Department of Justice to enable update with inquest findings
Incorporating tool into routine practise	· Adequate staff training, with access to IT personnel who would be able to assist with any real-time issues · Making the new digital tool compulsory, which requires buy in from decision-makers in management positions
Design of the electronic solution	· In the design of the solution, several stakeholders should be involved: o Users (junior/senior doctors, as well as urban/rural, funeral homes) o Stakeholders (Department of Health, Statistics South Africa, forensic pathologists, Department of Home Affairs) o Data analysts and IT designers to advise on feasibility o International partners [for example, WHO or United Nations Children’s Fund (UNICEF)] o Traditional leaders and community representatives to engage with the community

Discussion

Our findings suggest that an electronic system for medical certification of cause of death would be a welcome intervention by health professionals to strengthen mortality statistics in South Africa. The current challenges with South Africa’s mortality statistics have been well described in literature and include delays in the release of mortality statistics [4], poor access to mortality data for public health purposes, and poor quality of cause of death information [5], as reflected in the high proportion of unusable codes (32.0%) in the national mortality data [23], misclassification of HIV/AIDS deaths [8], and inaccurate injury mortality profiles [23].

This study highlights several factors contributing to the poor quality of cause-of-death (COD) data, many of which are consistent with findings from other countries. A common issue across countries, regardless of income level, is inadequate MCCD training for medical certifiers. Training often emphasizes legal or forensic aspects, leading to a limited understanding of the importance of COD data for public health purposes [24]. Lack of access to medical information is a particular problem in lower income countries where health information systems might be poor and a high proportion of deaths occur in the community, resulting in physicians who certify a death often not knowing the medical history of the deceased [25]. The misclassification and underreporting of deaths that are associated with stigma, such as HIV/AIDS or suicide, is also a recurring challenge [26]. Furthermore, evidence from Ecuador and other settings shows that the quality of COD data may also be influenced by factors such as, the sex, age, and gender of the deceased, potentially introducing biases that further undermine the reliability of mortality statistics [27].

Whilst health workers in our study generally agreed that a well-designed EDRS has the potential to improve the quality and timeliness of South African mortality statistics, they also emphasized the need for complementary interventions, such as training certifiers, improving access to medical records and revising the MCCD form, which does not comply fully with the WHO internationally recommended form, so health professionals cannot easily capture all the relevant information, resulting in higher unusable COD codes for public health purposes.

Experiences from countries, such as Portugal [28], Ecuador [29], Peru [12] and the United Kingdom, which have implemented eMCCD/EDRS systems, highlight the importance of holistic interventions when transitioning to digital systems as critical for success. Lessons in these countries highlight that key factors for successful implementation include strong multisectoral leadership, legal alignment with the CRVS framework, early stakeholder engagement, and addressing barriers such as legislative constraints, user acceptance and infrastructure limitations [16].

In these countries, the adoption of electronic systems has resulted in significant improvements in the quality of data and the efficiency of the registration process [12, 28, 29]. The implementation of an EDRS system in South Africa aligns with the current global trend towards the digitalization of information systems [30]. However, consideration needs to be given to the digital health readiness of South Africa [31, 32]. The overall readiness of Portugal is high, with South Africa having a moderate level of readiness with challenges of digital literacy and access in rural areas. Peru and Ecuador have similar low-moderate levels of readiness. Portugal with its high level of readiness managed to transition from paper to 100% electronic MCCD within a year. Peru and Ecuador with low-moderate level of readiness managed to reach 86% and 37%, respectively, within 3 years. Whilst South Africa has challenges to address in the digital sphere it has a higher level of readiness than Peru and Ecuador, suggesting that the overall digital health phase is unlikely to be an insurmountable barrier to implementation of an EDRS.

South Africa is currently undergoing major reforms to digitize its health information systems. The National Digital Health Strategy 2019–2024 of South Africa’s health systems outlines the vision and path towards “Better health for all South Africans enabled by person-centred digital health” [33]. Therefore, it seems very timely for the country to consider this transition towards an EDRS within the CRVS. This study highlighted the importance of integrating a future EDRS with existing information systems in South Africa (within and outside the health sector) for its successful implementation. A future EDRS would replace the current paper based CRVS and could be linked to the National Population Register which is part of the current CRVS and the National Health Patient Registration System (patient demographic information systems), which would then provide the backbone for an EDRS, as suggested by several interviewees in our study. The design and scale-up of an integrated EDRS should involve a multisectoral process that considers the complexity of different systems. The importance of a coordinated and multisectoral effort that overcomes institutional barriers during the design of an EDRS cannot be overstated. The World Health Organisation and African Union both recognize the importance of strengthening coordination between the health sector and other national CRVS partners by setting up a formal national inter-agency collaboration to oversee CRVS strengthening [34, 35]. South Africa can capitalize on existing experiences and innovations that can inform the design of an EDRS [36].

Given the limited resources available in South Africa, it will be important to leverage existing infrastructure to provide access to the internet and necessary hardware to facilitate the rollout of an EDRS, including contingency plans to deal with breaks in internet connectivity and power. Several studies have demonstrated the importance of assessing the existing technology and ensuring that it is available and adequate for the needs of the new digital information system [37].

In addition, digital literacy was identified as a potential challenge that needs to be addressed. Socioeconomic inequalities in South Africa have resulted in a disproportionate level of digital “knowledge”, which could be addressed through mandatory training programs for users of the system [38]. To improve the quality of medical certification of causes of death, a free, online training tool has been developed (www.deathcertification.org). However, wider dissemination of this tool is needed to achieve its intended impact. To increase uptake, discussions are underway to include the tool in the curriculum of final-year medical students and during the orientation of new interns.

Furthermore, the implementation of a digital system can be used to ensure that the data that have been entered are immediately accessible to health facilities and districts for provisional use for public health purposes, while the data processing and production of official statistics continue at the relevant stakeholders. Several respondents highlighted how this feedback loop to medical certifiers, health facilities, and district managers would enable direct health actions and support the continuous improvement in the quality of the MCCD. When the data submitted are translated into effective information used for action, the completion of reports will not be seen as a bureaucratic activity.

Finally, it is crucial to ensure that any digital solution is designed with robust data security and privacy laws to prevent fraud [39].

Limitations

We encountered challenges in obtaining the desired samples for each of the modalities, which could have introduced bias in our results. Securing participation from funeral undertakers for key stakeholder interviews proved challenging, given our first-time outreach. However, doctors showed greater receptiveness, likely owing to their existing connections within our networks. Administrative hurdles in certain provinces hindered our ability to interview doctors in those regions. Additionally, the online workshop saw participants entering and leaving at different intervals, resulting in varying response rates across questions. Promoting our online survey proved difficult, ultimately falling short of our target sample size. Nevertheless, we sought to triangulate our results across methods and mitigate the limitations inherent in each approach.

Conclusions

This study showed that an EDRS would be acceptable to the majority of stakeholders and would be feasible if thoughtfully designed to specifically address the South African landscape. The eMCCD has the potential to improve and strengthen the quality of South African cause of death data and make it suitable for public health surveillance and health planning purposes. A well-designed EDRS system that incorporates features to aid in correct medical certification could improve the accuracy of COD data and reduce the data capture time. In addition, transitioning to an electronic system offers an opportunity for the integration of the new EDRS with other existing information systems, improving efficiency and benefiting both families and the CRVS system.

As a first step, the three key agencies responsible for civil registration in South Africa, namely, the Department of Home Affairs, Department of Health, and Statistics South Africa, would need to engage and set up governance mechanisms to provide leadership for the CRVS that could develop a harmonized plan of action to set up an EDRS. This would involve an update of the technical landscape analysis to include recent digital health developments and an update of the readiness assessment of the digital health ecosystem in South Africa prior to the development of system requirements.

Author contributions

C.S.F., N.K. and P.G. designed the study, conducted data collection and data analysis and wrote the initial and final drafts of the manuscript; S.G. contributed to data collection, data analysis and the initial draft of the manuscript; D.B., D.M. and D.C. participated in the study design and offered technical input throughout. All authors reviewed and approved the final manuscript.

Funding

This project was supported by the President’s Emergency Plan for AIDS Relief (PEPFAR) through the US Centers for Disease Control and Prevention (CDC) under the terms of cooperative agreement GH002193.

Availability of data and materials

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study was approved by the South African Medical Research Council ethics committee (EC009-4/2022) and an exemption from an ethics review by the Swiss Tropical and Public Health Institute. This project was reviewed in accordance with CDC human research protection procedures and was determined to be non-research public health surveillance. All participants in the study provided informed verbal consent, which was recorded, after being given a written study information sheet and a verbal explanation of the consenting process. All interviewed study participants provided recorded consent to participate, and their responses were recorded.

Consent for publication

All authors provided consent for publication of the manuscript.

Competing interests

The authors declare no competing interests.