Determining the optimal diagnostic and risk stratification approaches for people with hypertension in two rural populations in Kenya and The Gambia: a study protocol for IHCoR-Africa Work Package 2

Patient and Public Involvement

This study has a strong emphasis on the involvement of the patients and the public. Before protocol development, the study was discussed with community leaders in the study areas so they could suggest changes and discuss how to involve the community. The IHCoR-Africa project has been presented to local departments of health in Kilifi and Gambia, and their recommendations taken into consideration. Throughout the course of the we will meet with CHWs and patient groups, to understand their experiences as the project is implemented. The collection of data will be mainly performed by CHWs, who are members of the studied communities. After the conclusion of the study, we will involve CHWs and other members of the community in developing an intervention that should improve the detection, treatment, and control of hypertension in rural Gambia and Kenya.

Study setting

Activities will take place across two rural sites, one in East Africa (Figure 1). Both settings have a high burden of hypertension, as well as a well-developed research infrastructure to facilitate our work.

Figure 1. Map showing location of study sites.

West Kiang Health and Demographic Surveillance System (The Gambia) and Kilifi Health and Demographic Surveillance System (Kenya). This figure is an original figure produced by the authors for this article.

Kilifi County is one of the poorest regions in Kenya⁴⁰. The Kilifi Health and Demographic Surveillance System (KHDSS), covers an area of 900 km² and has a population of around 300,000 people. The prevalence of hypertension is 26% and only 3% of individuals have their blood pressure controlled⁴¹. Further evidence of the consequences of poor hypertension control is the high incidence of cardiovascular disease in the area, with more than one third of patients admitted with cardiovascular disease dying during their hospital stay³¹.

CHWs deliver health services at the community level and work within a Community Health Unit (CHU), a health service delivery structure within a defined geographical area covering approximately 5,000 people (500–1000 households). CHWs are selected from the community they serve and where possible, are required to have completed secondary education. Each CHU is required to have approximately 10 CHWs.

The Kiang West district in The Gambia covers 750km² and includes 36 villages with a population of over 14,000 individuals. It is one of the poorest regions in the country with all households in the lowest or low wealth groups. Hypertension prevalence in this region is estimated to be 40%, of whom at least 71% are undiagnosed, and only 4% controlled⁴². The health service delivery in the region is coordinated through the Regional Health Team of the Lower River Region (Mansakonko Administrative Area). The community health nurses (CHNs) deliver services in a defined village cluster within the district. There are 4 CHNs in Kiang West with each handling 7–9 villages. CHNs work with and supervise the village health workers (VHW) who are selected by their communities.

Study population and sample procedures

We will select an age-stratified random sample of participants aged ≥30 years from the population registers of the Kiang West Health and Demographic Surveillance System (KWHDSS, the Gambia) and Kilifi Health and Demographic Surveillance System (KHDSS, Kenya)^43,44. Half of the study population will be selected from The Gambia and half from Kenya. The following age strata will be used: ≥30 to <40, ≥40 to <50, ≥50 - <60, ≥60 - <70 and ≥70y.

Pregnant women will be excluded due to technical difficulties in conducting some of the study measurements such as the echocardiogram and while gestational hypertension remains important it is beyond the scope of our aims and objectives. The participant recruitment approach is shown in Figure 2.

Figure 2. Flow diagram showing participant recruitment, study procedures and objectives.

Inclusion criteria:

•   Registered in population registers of the KWHDSS (The Gambia) or KHDSS (Kenya)

•   Aged ≥30 years at the time of enrolment in the study

•   Able to provide written informed consent to participate in the study

Exclusion criteria

•   Pregnant women (self-reported)

Study aim and objectives

We aim to improve the evidence base for diagnosis and risk estimation for a CHW-led community-centred hypertension programme in two rural communities in SSA.

We have the following specific objectives:

Sample size

The overall sample size for the study is based on the assessment of hypertension prevalence. Assuming a prevalence of hypertension of 25–35%, as detected by the reference standard method (24-hr-ABPM), we will require 832–1003 participants to generate validity parameters for hypertension prevalence with a precision of ±3% overall and ±5% at each of the two sites^24,41. We have made a provision for loss of data in 25% of participants due to poor data quality, taking the total number of participants required to 1250, equally divided between Kilifi (N=625) and Kiang West (N=625). This overall sample size will provide sufficient statistical power to address the other aims stated above.

In our assessment of BP and risk factors, and assuming for illustrative purposes that 10% of individuals will have any of the putative factors and that the standard deviation of systolic blood pressure will be 15 mmHg, then 1,003 participants will provide at least 80% statistical power for an alpha of 0.05 to detect a 4mm Hg difference in systolic BP in those with and without the exposure of interest.

Assuming a prevalence of hypertension detected by the reference standard method (24-hr-ABPM) of 30%, we estimate that approximately 300 participants from those recruited in Objective 1 will be classified as hypertensive. With 300 hypertensive patients, we would be able to estimate the confidence interval for a conservative HMOD prevalence of 35% with lower and upper intervals of 29.7% and 40.4% respectively, and that we will have over 95% power for an alpha of 0·05 to test the null hypothesis that the prevalence of HMOD is below 25%. To further explore the difference in the prevalence of HMOD in the hypertensive and general population, we intend to assess all participants recruited into the study for HMOD.

When measured with the reference standard techniques, we expect a 35% prevalence of HMOD among the 300 estimated hypertensive participants and approximately 5% prevalence in the remaining non-hypertensive participants. This will provide us with a total available sample of 153 participants with HMOD. Assuming at least 85% sensitivity of detection of HMOD by the POC strategies we will have at least 80% power for an alpha of 0.05 to test the null hypothesis that the sensitivity is below 75%. As above, we will conduct POC HMOD assessments on all participants.

Study procedures

Baseline assessments. Baseline assessments will collect information including demographics, household assets, anthropometrics, diet, physical activity, history of hypertension and medication use for hypertension and other diseases (Table 3).

Blood pressure assessment. We will conduct a cross-sectional study over two weeks to determine the accuracy of three BP measurement methods compared to the reference standard of 24-hour Ambulatory BP Measurement (24-hr-ABPM). The alternative BP measurement methods are: 1) Attended Automated BP Measurements (aABPM); 2) Unattended Automated BP Measurements (uABPM); and 3) Home-Based BP Measurement (HBPM). Details for all BP measurement methods are summarised in Table 1 and Table 3.

All BP measurements will be conducted by CHWs trained by the study team to perform the blood pressure measurements based on study SOPs. The competencies of the CHWs will be assessed by the study team prior to data collection and at regular intervals during data collection. The CHWs will then collect blood pressure measurements in a random order at participants’ homes. The random order of blood pressure measurements for each of the participants will be set using a computer-generated randomisation list produced prior to measurements starting.

Each participant is expected to have completed all four BP measurements over a two-week period. A schematic overview of the procedure is shown in Figure 3. We will follow relevant international cardiological society guidelines for conducting all BP measurements^45–47.

Figure 3. Schematic overview of blood pressure measurements performed in each participant.

Hypertension mediated organ damage assessment. We will measure HMOD in the following organ systems: heart, eye, kidney, and vasculature. All measurements will be conducted during an in-clinic visit. The prevalence of HMOD will be assessed using reference-standard clinical assessments. In addition, we will validate the accuracy of POC assessments for HMOD compared to the reference-standard methods for the heart, eyes, and kidney. For all HMOD assessments, a detailed SOP will be used to ensure consistency between both study sites (Kenya and The Gambia) and staff will be trained in the relevant competencies. HMOD definitions in all categories will follow existing clinical guidelines²². These assessments are summarised here and described in Figure 4 and Table 2 and Table 3.

Figure 4. Schematic overview of hypertension-mediated organ damage measurements performed in each participant for Objectives 3 and 4.

Reference standard HMOD assessment

In the heart, left ventricular systolic dysfunction will be assessed with a transthoracic echocardiogram (TTE), and left ventricular hypertrophy will be assessed with a TTE and 12-lead electrocardiogram (ECG). In the eye, hypertensive retinopathy will be assessed from retinal images captured by a fixed retinal camera and graded for hypertensive retinopathy by trained staff. Kidney dysfunction will be assessed using clinical laboratory assessment of blood and urine to measure serum creatinine levels and estimate glomerular filtration rate (eGFR) and glomerular function using urine albumin:creatinine ratio and proteinuria. Finally, we will assess arterial stiffness using a specialised blood pressure cuff (Arteriograph, TensioMed, Hungary).

Point-of-care HMOD assessment

Point-of-care assessment of left ventricular systolic dysfunction and left ventricular hypertrophy will be assessed using a mobile phone based 6-lead ECG (KardiaMobile, AliveCor, United States of America) coupled with artificial intelligence (AI) tools developed for ECG analysis⁴⁸. In the eye, we will assess hypertensive retinopathy using a handheld, mobile phone based, non-mydriatic retinal camera (Remidio, India). Retinal image grading will be conducted by the same staff who graded the reference standard images. Finally, kidney dysfunction (defined as serum creatine levels, eGFR and glomerular function) will be measured using urine dipsticks, and POC clinical chemistry devices (Abbott iStat, Abbott, United States).

Genetic sampling. For the genetic analyses, DNA will be extracted from whole blood samples using Qiagen DNA Blood Mini kit. DNA extraction will make use of blood samples drawn for HMOD measurement and no new blood sample is required. We will then use PCR to determine presence of the genotypes of interest, including malaria and trypanosomiasis protective polymorphisms (Dantu, SCT, APOL1 or APOL2)^36,37. These analyses will be conducted in KWTRP.

Health economics assessment. We will use the human capital approach to collect primary data on patient costs including the opportunity cost of time spent administering blood pressure-monitoring methods, missing productive activity and any out-of-pocket costs incurred for undergoing study procedures. Health system cost data will be collected using a micro-costing, ingredient-based approach. Data will be collected from a structured questionnaires administered to three CHWs from each study site, structured observations of CHW and user time spent on BP and HMOD measurement, and a survey using a structured questionnaire to 200 individuals diagnosed with hypertension⁴⁹.

Statistical analysis plans

A detailed statistical analysis plan will be produced before data collection is completed and uploaded to a public repository. A summary of statistical analysis approaches that will be taken for this study is outlined here.

Hypertension prevalence. We will estimate sensitivity; specificity; positive and negative predictive values; and the positive and negative likelihood ratios of each BP measurement approach compared to the reference standard 24-hour BP measurement method.

Identifying hypertension risk factors. The primary outcome will be hypertension, defined as a BP of >140/90mmHG using 24-hour BP measurement, or ongoing therapy with anti-hypertensives. We will determine the relationship between the candidate risk factors and blood pressure and hypertension using multivariate linear and logistic regression methods as appropriate. We will include confounders in the regression models based on results of univariate models and existing literature on the subject.

Prevalence of organ damage. We will analyse the overall prevalence of HMOD and report age- and sex-standardised and stratified prevalence rates. We will also analyse prevalence rates stratified by type of HMOD (left ventricular hypertrophy, impaired left ventricular systolic function, renal dysfunction, hypertensive retinopathy, or arterial stiffness). Finally, we will construct logistic regression models to identify the participant parameters associated with presence of HMOD.

Accuracy of POC diagnostic tools. The diagnostic accuracy of the POC assessments will be estimated using sensitivity; specificity; positive and negative predictive values; and positive and negative likelihood ratios of each POC parameter compared to the conventional measurement. These comparisons will be:

Cost effectiveness analysis. We will model the comparative cost-effectiveness of different decision-making approaches for initiating treatment of hypertension. We will define cost-effectiveness by comparing the costs (to be obtained from primary data collection) and outcomes (to be obtained from secondary sources) of implementing the four different BP measurement approaches, as well as the reference and PoC assessments of HMOD, to guide initiation of treatment. This analysis will be used in developing the community-centred programme at a later stage of the project.

Ethical considerations

Approval for research involving humans. Ethical approval for this research has been given by the KEMRI Scientific & Ethics Review Unit (SERU4620; 6 February 2023), the MRCG Ethics Committee and the LSHTM Ethics Committee (28276; 6 March 2023).

Consent. We will obtain written informed consent from all the study participants. CHWs will be trained in the consenting procedures and will be responsible for obtaining consent. We will put in place all the necessary measures to remove potential barriers for informed consent, including language, literacy, and pressure to participate to access care.

All study participants will be provided with comprehensive information about the study in the language they are most comfortable with. This information will explain the study's aims, their specific role in it, and how data being collected will be used. Where a participant is illiterate an independent witness will be present to ensure the correct study information is presented to the participant. This witness will sign the consent form as well.

It will be made clear that participation in any part of the research will be entirely voluntary, will not influence future care and that participants will be free to withdraw at any stage, without giving a reason.

Withdrawals. Participants may withdraw from the study at any time, for any reason. The study team may withdraw participants if they are no longer contactable or if they develop conditions that would exclude them from the study, including a lack of capacity to consent.

Compensation. Participants will not be compensated for home blood pressure measurement visits as they will not disrupt daily activities and will take less than 15 minutes. Participants will be compensated for their time and expenses to attend hospital visits. For each visit, participants will be reimbursed 500 Kenyan shillings or 250 Gambian dalasi plus any travel expenses incurred traveling to and from the research visit.

Urgent care referral. During this research, it is possible that severe medical conditions could be identified. In such cases we will arrange for urgent care and further treatment will follow the locally available standard of care.