Evaluating the accessibility and utility of HIV-related point-of-care diagnostics for maternal health in rural South Africa: A study protocol

Introduction: Poor healthcare access is a major barrier to receiving antenatal care and a cause of high maternal mortality in South Africa (SA). ‘Point-of-care’ (POC) diagnostics is a powerful emerging healthcare approach to improve healthcare access. This study focuses on evaluating the accessibility and utility of POC diagnostics for maternal health in rural SA primary healthcare (PHC) clinics in order to generate a model framework of implementation of POC diagnostics in rural South African clinics. Method and analyses: We will use several research methods, including a systematic review, quasi-experiments, survey, key informant interviews and audits. We will conduct a systematic review and experimental study to determine the impact of POC diagnostics on maternal health. We will perform a cross-sectional case study of 100 randomly selected rural primary healthcare clinics in KwaZulu-Natal to measure the context and patterns of POC diagnostics access and usage by maternal health providers and patients. We will conduct interviews with relevant key stakeholders to determine the reasons for POC deficiencies regarding accessibility and utility of HIV-related POC diagnostics for maternal health. We will also conduct a vertical audit to investigate all the quality aspects of POC diagnostic services including diagnostic accuracy in a select number of clinics. On the basis of information gathered, we will propose a model framework for improved implementation of POC diagnostics in rural South African public healthcare clinics. Statistical (Stata-13) and thematic (NVIVO) data analysis will be used in this study. Ethics and dissemination: The study protocol was approved by the Ethics Committee of the University of KwaZulu-Natal (BE 484/14) and the KwaZulu-Natal Department of Health based on the Helsinki Declaration (HRKM 40/15). Findings of this study will be disseminated electronically and in print. They will be presented to conferences related to HIV/AIDS, diagnostics, maternal health and strengthening of health systems.

A summary of the methodology used by this multiphase/component study can be found in table 1. In this study, we will use several research methods to determine the impact of POC diagnostics on maternal health in rural KZN. We define rural as sparsely populated areas in which people farm or depend on natural resources, including the villages and small towns that are dispersed through these areas. In addition, they include the large settlements in the former homelands, created by the apartheid removals, which depend, for their survival, on migratory labour and remittances. Summary of the methodology CIs, confidence intervals; DHIS, District Health Information System; KZN, KwaZulu-Natal; MMR, maternal mortality rate; NA, not applicable; NHI, National Health Insurance; NPV, negative predictive value; PHC, primary healthcare; POC, point-of-care; PPS, probability proportional to size; PPV, positive predictive value; QMS, quality management systems; SA, South Africa; SANAS, South African National Accreditation System. The programme evaluation theory is a promising approach to explain how a programme produces the desired effects.17–19 This theory argues that effective implementation of a programme requires gathering of evidence from relevant stakeholders.17 20 The programme evaluation theory involves three distinctive approaches: postpositivism, interpretivism and critical normative science paradigm.21 Adopting the postpostivism paradigm will enable us to rationally deduce research experience and interpret into concepts and knowledge.22 The interpretivism paradigm will enable us to contextualise subjective realities of study participants in terms of their experiences of POC diagnostics’ challenges and barriers and enable us to attach this to meaning and qualitative evidence.23 The critical normative science paradigm will enable a critical analysis of POC diagnostic services in order to determine its merit. Combining the programme evaluation theory paradigms, we develop a theoretical framework to guide this study (figure 1). We will use data triangulation to increase confidence and diversity regarding the research data.16 24 Data for this study will be collected from seven sources: surveys, interviews, record review, District Health Information Software (DHIS) routine data, peer-reviewed literature and audits. A theoretical framework underpinning this study, programme evaluation theory adapted to the local context. POC, point-of-care; STDs, sexually transmitted diseases; TB, tuberculosis. The evaluation study will be carried out in KZN province, SA. KZN is located in the southeast of the country on the coast of the Indian Ocean, shares borders with three other provinces and the countries of Mozambique, Swaziland and Lesotho. KZN is the largest province in SA consisting of 11 districts and 52 municipalities and consists of a mix of urban, semiurban and rural areas. The province has a total population of ∼10 694 400, of which 86.8% are black Africans and Zulu speakers.25 A representative subset of maternal health primary healthcare clinics in rural KZN will be sampled for this study. The full data analysis plan for each objective/component can also be found in table 1. In this study, various methods of data collection and interpretation will be used as an integrated form to carry out methodological triangulation of sources for collecting quantitative and qualitative samples.22 23 Objective 1: To investigate the typology, supply chain POC diagnostics in rural PHC clinics in SA. Data sources: Survey (survey tool has been provided as online supplementary material file). The cross sectional survey study protocol was developed a priori and was registered in the Clinical Trial.gov website. ClinicalTrials.gov Identifier: {“type”:”clinical-trial”,”attrs”:{“text”:”NCT02692274″,”term_id”:”NCT02692274″}}NCT02692274, available at: https://clinicaltrials.gov/ct2/show/{“type”:”clinical-trial”,”attrs”:{“text”:”NCT02692274″,”term_id”:”NCT02692274″}}NCT02692274. bmjopen-2016-011155supp.pdf Sampling: We will conduct a stratified random sampling of PHC clinics to ensure generalisability (external validity). PHC clinic healthcare professionals responsible for the POC diagnostic services in the clinic will be requested to participate in the survey. A sample size of 100 primary units has been demonstrated to be an appropriate sample size for this type of facility-based survey.26 The most recent (2014) data on PHC headcount, professional nurse clinic workdays, annual nurse’s estimate and average headcount per week was requested from the South African DHIS in order to assign sample strata. Four strata were created based on the above sets. A total of 25 facilities were sampled within each of these strata using probability proportional to size (PPS). Proportionate stratification was implemented to ensure that the sample size of each stratum is proportionate to the population size of the stratum among all 11 KZN districts. The sample size of each stratum is proportionate to the population size of the stratum. Strata sample sizes were determined by the following equation: where nh is the sample size for stratum h, Nh the population size for stratum h, N the total population size and n is the total sample size. Table 2 shows the sampling frame per district. Project sampling frame Objective 2: To investigate the deficiencies and their causes for POC diagnostics in rural PHC clinics in SA. Data sources: In-depth interviews. Sampling: Clinics with low POC diagnostic availability and usage based on the overall average level of availability and usage of POC diagnostics from the sampled clinics. Variables: The interviews will be aimed at gaining rich data on patients-centred and staff-centred advantages, barriers, challenges of current POC diagnostic services and future service needs. Objective 3: To investigate the quality management systems emplaced to ensure reliability of the HIV-related POC diagnostics for maternal health in their current setting. Data sources: Audit and validation test. Sampling: To determine the reliability of POC diagnostic services in clinics with high availability, accessibility and usability of POC diagnostics, the quality management systems implemented in the clinics will be assessed against relevant quality indicators as prescribed by the most recent WHO guidelines for POC diagnostics in resource-limited settings, through an audit.27 The audit will include evaluation of the performance, operational characteristics of the test and linkage to healthcare. To determine reliability of the results produced by PHC clinics WHO standards, a POC diagnostic validation test will be carried out. Full blood samples will be requested from consenting women who will be attending the clinic and receiving an HIV POC diagnostic test for laboratory testing. A validation test will be conducted against the gold standard test for HIV, the ELISA test. The sample size at the required absolute precision level for sensitivity and specificity will be dependent on the survey, vertical audit results and clinic size. It will be calculated by Buderer’s formula16 which is demonstrated below: and Here, n is the required sample size, SN the anticipated sensitivity, Sp the anticipated specificity, α the size of the critical region (1-α is the confidence level), Z1−α/2 the standard normal deviate corresponding to the specified size of the critical region (α) and L is the absolute precision desired on either side (half-width of the CI) of sensitivity or specificity. Owing to lack of laboratory infrastructure in rural and resource-limited KZN and to ensure reliability of the POC performance validation test, only PHC clinics that are within 60 km to the testing laboratory will be included in the POC test evaluation. It is anticipated that sensitivity (or specificity) of a given POC test is 80% for detecting a given outcome against the laboratory gold standard, assuming an absolute precision of ±10% and the prevalence of outcome in the study population is 27%28 and based on the average patient headcount per week, then it will be necessary to sample and test 207 study subjects using both the POCT and the laboratory gold standard. Objective 4: To determine the impact of HIV-related POC diagnostics on maternal mortality using an interrupted time series study. Data sources: South African DHIS. Sampling: Retrospective data on rural KZN maternal mortality rate from all KZN districts. The time of POC test implementation in KZN rural clinics will be obtained from the Department of Health archives. Variables: The following is an explanation of the method of Wagner et al,29 applied to our practical analysis. Time series of maternal mortality rate will be assessed using segmented negative binomial regression analysis, which is a method of estimating changes in levels and trends in an outcome associated with an intervention (POC diagnostics). The time series regression equation for this model is as follows: is the outcome (mean number of deaths per quarter), time indicates the number of quarters) from the start of the series, intervention is the dummy variable taking the values 0 in the preintervention segment and 1 in the postintervention segment, time_after_intervention is 0 in the preintervention segment and counts the quarters in the postintervention segment at time t. The coefficient β0 estimates the base level of the outcome (number of deaths) at the beginning of the series, β1 estimates the base trend, that is, the change in outcome per quarter in the preintervention segment, β2 estimates the change in level of deaths on the postintervention segment, β3 estimates the change in trend in deaths in the postintervention segment, et estimates the error. This model will be used to estimate the impact of HIV-related and MH-related POC diagnostics on maternal mortality in rural KZN. Objective 5: To evaluate whether the introduction of POC diagnostics into algorithms for diagnosing maternal patients improves maternal health for HIV-infected women using a systematic review. The systematic review protocol was developed a priori and was registered in the PROSPERO international prospective register of systematic reviews and in publication.30 PROSPERO record: CRD42014015439, available at: http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42014015439#.VSfoV-_GPug. The systematic review will follow recommendations described in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement31 and the Cochrane Handbook for Intervention Reviews.32 The findings of the systematic review will be disseminated through publication in a peer-reviewed journal and will be formatted according to the specific journal publication guidelines. Data source: The studies will be selected by evaluation of the inclusion and exclusion criteria. This will be carried out in duplicate and independently by two authors with agreement assessed using κ-statistics. Objective 6: To develop a model framework and recommendations for improved implementation of POC diagnostics on SA rural PHCs. Guided by the information gathered from the above objectives, we will propose a model framework for improved implementation of POC diagnostics in rural SA PHCs. Interviews will be conducted with consenting participants in English. Field notes from observations will be written down about each observation session. The observations will be used to contextualise the interview findings and confirm the validity of interpretations. We will perform a verbatim transcription of all interviews and check transcripts with study participants to seek points of clarification in relation to issues arising from interviews. We will also perform an audit trial for assessing the entire research process. Thematic content analyses will be performed to identify the patterns of POC diagnostic key utility areas and deficiencies from respondent’s interviews, using NVIVO software. First, participants’ responses will be coded into categories which were then grouped into themes. The codes will be grouped into similar concepts that reflect the context about local factors that determine healthcare workers and patients’ engagement with POC diagnostics. Finally, the identified themes will be validated by the study supervisor. Anticipated themes include: management, human resources, infrastructure, staff knowledge, skill, attitude, believes, relevant key stakeholders’ perceptions on the quality of POC diagnostics and relevant key stakeholders’ perception on POC diagnostics scale-up. Quantitative data will be entered into a project-specific Microsoft Access database and extracted manually onto a categorised table. Data will be grouped into two levels: facility level and individual level. Facility-level data will include data from clinic audits and PHC clinic nurses. Inferential statistics will be used to determine significant differences in the availability of POC diagnostics from the sampled clinics. Standard Student’s t-test and analysis of variance (ANOVA) will be used to compare means across groups while the Pearson χ2 test will be used for contingency tables. Factors such as distance to the nearest emergency hospital and clinic size in terms of patient volumes and staff numbers, frequency of use for the diagnostics and level of need for the diagnostics (list of POCT requested by the clinic staff during survey) will be taken into account. The reliability and accuracy of the POCT test results versus the laboratory gold standard will be estimated along with 95% CIs. Table 1 show variables and analysis for the data obtained from the audit. Systematic review analysis will follow the relevant PRISMA guidelines as stipulated in the published protocol.30 All quantitative data will be processed and analysed using Stata V.13.0 (StataCorp. 2013. Stata Statistical Software: Release 13. College Station, Texas, USA: StataCorp L). Descriptive statistics that include frequency distribution, percentages and percentiles, means and SDs and cross-tabulations will be used to describe the characteristics of POC diagnostic service in rural KZN. A 95% CI will be constructed around point estimates given the sampling design. We will address any missing data using appropriate statistical methods.33 Data can be missed for many reasons, such as on occasions when a participant did not show up to participate in a study; or one group had more participants than another; or a device did not record the data correctly. The nature of missing data will determine the statistical analysis methods to be used. We will undertake a careful and prolonged planning of the study to reduce or eliminate the potential sources of bias including sampling bias, recall bias and reporting bias. Recall bias can be introduced into the qualitative data collection stage of investigation, presenting a major threat to the internal validity and credibility of the study. To overcome this limitation in this study, participants will be provided with enough time before answering the question, to reflect and think through a sequence of events in their professional history. The risk of bias for (eg, internal validity) the studies included in a systematic review will be evaluated using the Cochrane Risk of Bias Tool.32

N/A

• mashamba-thompson, tivani p.
• drain, paul k.
• sartorius, benn