Outcomes and costs of implementing a community-based intervention for hypertension in an urban slum in Kenya.

OBJECTIVE: To describe the processes, outcomes and costs of implementing a multi-component, community-based intervention for hypertension among adults aged > 35 years in a large slum in Nairobi, Kenya.
METHODS: The intervention in 2012-2013 was based on four components: awareness-raising; improved access to screening; standardized clinical management of hypertension; and long-term retention in care. Using multiple sources of data, including administrative records and surveys, we described the inputs and outputs of each intervention activity and estimated the outcomes of each component and the impact of the intervention. We also estimated the costs associated with implementation, using a top-down costing approach.
FINDINGS: The intervention reached 60% of the target population (4049/6780 people), at a cost of 17 United States dollars (US$) per person screened and provided access to treatment for 68% (660/976) of people referred, at a cost of US$ 123 per person with hypertension who attended the clinic. Of the 660 people who attended the clinic, 27% (178) were retained in care, at a cost of US$ 194 per person retained; and of those patients, 33% (58/178) achieved blood pressure control. The total intervention cost per patient with blood pressure controlled was US$ 3205.
CONCLUSION: With moderate implementation costs, it was possible to achieve hypertension awareness and treatment levels comparable to those in high-income settings. However, retention in care and blood pressure control were challenges in this slum setting. For patients, the costs and lack of time or forgetfulness were barriers to retention in care.

Introduction

Cardiovascular diseases are the leading cause of death globally, killing 17.5 million people per year and 80% of deaths from these diseases occur in low- and middle-income countries., Evidence suggests that the main drivers of the global cardiovascular disease epidemic are urbanization and industrialization, which lead to an increase in sedentary lifestyles, unhealthy dietary patterns, tobacco consumption and increased alcohol consumption. Hypertension is a leading risk factor for cardiovascular diseases, and its prevalence is increasing worldwide – from 25% in 2000 to a projected 40% in 2025. The rising burden of hypertension in low- and middle-income countries is amplified by the public’s low levels of awareness, treatment and control of this condition, particularly among slum residents, who typically constitute a large portion of neglected urban populations in such settings., Studies in slum populations suggest that when people are made aware of having hypertension they do tend to seek care., However, the level of adherence to treatment for hypertension remains low for several reasons, including, but not limited to, the high costs of treatment and to patients’ perceptions of a low risk of cardiovascular diseases and belief in a one-time cure for disease rather than to lifelong preventive treatment and monitoring.–

In response to the rising burden of cardiovascular disease risk factors in slum populations in Kenya,, a community-based intervention was developed and implemented in the capital city, Nairobi. This intervention, known as SCALE UP (the sustainable model for cardiovascular health by adjusting lifestyle and treatment with economic perspective in settings of urban poverty), has been described in detail elsewhere. The intervention had multiple components with the overall aim of reducing cardiovascular diseases risk through awareness campaigns, improvements in access to screening and standardized clinical management of hypertension. The aim of this paper is to share experiences of implementing a comprehensive intervention for primary prevention of hypertension in a slum setting and to examine the processes, outcomes and costs of the intervention. The lessons learnt from this paper will be useful to policy-makers and other stakeholders looking to implement similar interventions in highly resource-constrained settings.

Methods

Context

Korogocho slum, located in Nairobi, is home to about 35 000 people resident across seven villages. Within this slum, two primary health-care centres were invited to participate: a private nonprofit facility and a community-owned facility. The intervention team set up cardiovascular diseases’ clinics at the facilities, provided basic screening equipment (such as blood pressure monitors), and trained a pair of nurses and clinical officers in each clinic to manage patients found to have hypertension, using a standardized treatment guideline developed by the study team in line with international practice. Although most patients made out-of-pocket payments for services received, the clinics offered services at highly subsidized prices, which were possible through donor funding. It was not practical in this setting to implement an intervention for all cardiovascular risk factors and therefore treatment focused on blood pressure screening and prescription of anti-hypertensive medication: hydrochlorothiazide, nifedipin or enalapril. If indicated, patients who also had diabetes were placed on metformin.

Study design

The intervention itself was part of a quasi-experimental study. For this analysis we aimed to measure the outcome and impact of each stage of the intervention in terms of hypertension control. The intervention study was approved by the Kenya Medical Research Institute’s national ethics review committee (NON-SSC protocol no. 339; current controlled trials no. ISRCTN84424579). All participants gave written informed consent to participate in the intervention both at the household screening and at the clinics.

Intervention

The intervention was developed based on a modelling exercise, described in more detail elsewhere., The intervention was implemented for 18 months from August 2012 and had four components: (i) raising awareness about cardiovascular diseases; (ii) improving access to screening; (iii) facilitating access to treatment; and (iv) promoting long-term retention in care.

A total of 50 community health workers were recruited and trained to conduct door-to-door household visits to raise awareness about the burden of cardiovascular diseases in the community and provide information about opportunities for screening, to conduct the screening and to provide brief counselling among the eligible population. The community health workers were trained for seven days at the implementing institution’s offices, with teaching facilitated by the project’s senior researchers in a classroom setting and including practice sessions. Health workers each received US$ 6 per day for transportation reimbursement.

Eligible people were all adults aged 35 years or older resident in Korogocho slum area who were listed in the database of the Nairobi urban health and demographic surveillance system, and who consented to participate. Anthropometric and clinical measurements were taken at participants’ homes, including height, weight, waist and hip circumference, blood pressure and blood glucose (early morning, dried blood-spot testing). Community sensitization about the household visits was conducted via local radio campaigns, community meetings and religious gatherings.

All persons with elevated blood pressure (≥ 140 mmHg systolic and/or ≥ 90 mmHg diastolic) were referred by the community health worker to either of the two participating clinics. As an incentive for patients to seek care, community health workers gave each referred person a paper voucher that entitled him or her to receive a free 1-month supply of medication – valued at about United States dollars (US$) 1.8 – on their first clinic visit. All subsequent monthly medication prescribed at the clinic, however, was to be paid for by the patients. The consultation and laboratory tests were provided free of charge at these clinics, as usual in public primary health-care facilities in Kenya. To motivate them to follow-up each patient, community health workers received an incentive of US$ 3.0 per appropriately referred patient who visited the health facility.

To promote retention in care, patients receiving treatment were organized into support groups by village. Each support group received an incentive – a group reduction in the price of medication by one-third (approximately US$ 0.6) – if they collectively achieved 80% or more attendance to follow-up visits for a consecutive period of six months. Financial incentives were also offered to community health workers to organize the support groups: US$ 1.8 per support group participant attending the clinics for at least six consecutive months as scheduled. Finally, we sent monthly mobile phone short message service (SMS) reminders to patients reminding them of scheduled clinic appointments.

Data collection and analysis

Data sources

The main sources of data were administrative records, activity reports, minutes of meetings and other relevant records. However, we supplemented these data with data sourced from population- and clinic-level surveys conducted at baseline and at the end of the intervention period. The population-level survey was conducted with randomly sampled participants in the study community at baseline (August to December 2012) and endline (February to April 2014). The clinic survey involved structured interviews with patients attending the clinics only. Data for the cost analysis were collected from financial records and time sheets and interviews with staff.

Processes and outcomes

To describe the processes we first listed the activities involved in the intervention and the inputs needed to implement each activity: for example, provision of facilities (input) required for training of community health workers (activity). We then described the result of each activity: for example, the number of community health workers trained (output).

To evaluate the outcomes of the components of the intervention we determined the number of people participating at each stage of the intervention and calculated the following measures: the proportion of the target population who were screened and referred to the clinic for treatment (awareness-raising and screening); the proportion of people with high blood pressure referred who attended the clinic for treatment (access to treatment); and the proportion of people who attended for treatment and made six or more clinic visits within a 12-month period (retention in care). We sought to identify possible explanations for the outcomes observed in each stage of the continuum of care. For screening and awareness we used field reports to document the reasons why not every prioritized adult was reached by the intervention. For treatment-seeking and retention in care, we conducted semi-structured interviews with a random sub-sample of referred patients who defaulted from scheduled visits or never attended the clinics.

Impact

To evaluate the overall impact of the intervention we collected data from anonymized routine medical records from the two clinics and calculated the levels of blood pressure control achieved among patients during the intervention. The main impact measure was the percentage of all patients retained in care (defined as patients with six or more clinic visits within a 12-month period) whose blood pressure was controlled to below 140/90 mmHg. We also calculated the percentage change in the mean systolic and diastolic blood pressures of these patients between their first and sixth clinic visits.

Costs

The costs of the intervention were estimated from a provider perspective and expressed in US$, based on the average conversion rate in 2013 of Kenyan shillings 85 to US$ 1. We included all service and above-service level costs for each itemized activity per intervention component for the 18-month duration of the intervention. We excluded evaluation and research-related activities as they are not a part of service delivery. We used a top-down costing approach by allocating costs to each component of the intervention, then to activities by input type (Box 1). This was done through a review of financial records and time sheets and interviews with staff. Part of the management staff costs were first allocated out proportionally to the time spent on other projects. This was ascertained through interviews. The intervention staff costs were then allocated to implementation activities based on the relative duration of research and implementation activities.

Personnel input:

Salaries of all categories of staff and consultants

Commodities and supplies input:

Costs of drugs, tests, consumables and all training and communication materials

Training input:

Costs of space, travel, food and accommodation for participants, excluding staff costs

Capital cost input:

Costs of equipment, furniture, buildings (tents, floors) and labour to set up

Building operating and maintenance input:

Costs of communication, security, cleaning and repairs

Transport input:

Mileage allowance for supervision visits

Intervention activities input:

All payments for incentives to community health workers and patients, short message service reminders, community mobilization and adherence support

Indirect expenses input:

Reported overhead expenses

For all costs, we first allocated those costs that could be clearly tracked to a particular component. For the remaining shared costs, we allocated them across components based on the level of effort in hours dedicated to the activities in each component. We accounted for additional costs such as security escorts for our staff due to the field conditions. We report economic costs including items for which there were no financial transactions, for example rental of clinic space (these were valued using market prices). Capital costs were converted to an annual rate using a discount rate of 3%.

The costs for all inputs by itemized activities within each intervention component were then totalled to determine the total amount spent on that component. We then divided the total cost per component by the quantifiable unit of outcome per component, resulting in the cost per unit of outcome per intervention component. Finally, we totalled the cost of all components and divided that by the number of people with blood pressure under control to obtain the cost per unit of health gain (patient with blood pressure controlled and retained in care).

Results

Processes and outcomes

Table 1 shows the details of all the activities and inputs and the resultant outputs for each component of the intervention.

Table 1

Results of process evaluation of the community-based intervention for hypertension management in Kenya, 2012–2013

Intervention component by input category	Inputs	Activities	Outputs
Awareness and screening
Community gatherings (baraazas)	Banners, public address system, facilitators (community leaders, expert patients)	7 baraazas held	Estimated between 50 and 80 people attended each meeting
Religious services	Facilitators (community health workers, religious leaders)	21 religious meetings held	Estimated between 30 and 50 people attended
Radio jingle	Jingle content developer, local radio station (Koch FM)	1 jingle lasting 50 seconds aired 3 times daily for 3 weeks	Koch FM radio listener numbers estimated at 250 000 people
Community health workers	Facilitators (medical/research officers), training facilities, allowances	1 training and 1 refresher training held	50 community health workers traineda
Door-to-door screening	Community health worker allowances, screening equipment and materials	39 community health workers conducted door-to-door screenings	4049 people screened
Referral	Free vouchers, confirmation of blood pressure by supervisor	39 community health workers conducted referrals	976 people referred
Treatment
Clinic staff	Facilitators (medical/research officers), training facilities, allowances	1 training and 1 refresher training held	2 nurses, 2 clinical officers and 1 medical records clerk trained
Standard treatment guidelines	Meetings and review by stakeholders	1 main meeting held with stakeholder. Guideline reviewed mostly by email correspondence	1 guideline document published
Upgrading and equipping of clinics	Construction of consultation area, equipment	2 clinics upgraded. Concrete floor constructed and tent erected in 1 clinic. Both clinics received 2 sets of screening equipment and light furniture for consulting areas	2 clinics upgraded
Management of referred patients at clinics	Clinic staff allowances, utilities and supplies (including medication)	Clinics held twice a week for 17 months	845 people attended clinic first time, of whom 660 were eligible for recruitment into care
Retention in care
Follow-up of defaulters	Community health workers’ allowances (including incentives) and resources	188 defaulters followed up and interviewed by community health workers	46 defaulters returned to clinic after follow-up
SMS reminders	Bulk SMS application	4519 SMS reminders sent	660 patients received SMS reminders
Support groups	Community health workers, facilitators, incentives	7 support groups formed and 28 support groups held	371 people attended support groups

SMS: short message service.

a Although 50 community health workers were trained, not all were deployed to conduct screening. Some dropped out of the study to pursue other interests.

Fig. 1 summarizes the outcomes of each stage of the intervention. Community health workers successfully screened and counselled 4049 out of 6780 (60%) of the target population. The principal reasons for exclusion from the study were because the person refused to participate in the study (164; 2%), was believed to be resident in the slum but could not be reached (1161/6780; 17%), was no longer resident in the slum (281; 4%) or had died (26; < 1%). Other reasons accounted for 68 (1%) of drop-outs. A further 1031 people were not reached during the screening campaign but were traced by community health workers during the intervention period and given a complementary blood pressure check. However, we did not collect any data from these people nor did we follow them up at the clinic. Out of the 4049 people screened, 976 (24%) people with raised blood pressure were identified and referred; 358 (9%) were known to have hypertension and 618 (15%) were newly diagnosed.

Fig. 1

Outcome and impact indicators for each stage in the cascade of hypertension diagnosis and management in the community-based intervention in Kenya, 2012–2013

Out of the 976 persons referred to the clinics, 845 (87%) attended the cardiovascular diseases’ clinic at least once. Of these, 185 were found to have normal blood pressure levels after confirmatory measurement by the nurses. Therefore 660 patients out of 976 referred (68%) started on treatment with a prescription from the clinic. Out of 131 patients who were referred to a clinic but did not attend, 61 answered the follow-up interviews conducted at their homes. The leading reported reasons for non-attendance were lack of time (17 patients; 28%), cost of treatment (14; 23%) and forgot clinic appointment (9; 15%).

By the end of the intervention period, a total of 4519 SMS messages had been sent to all patients, and seven support groups had been formed with a total of 371 persons attending the meetings at least once. The average number of support group meetings attended by each person was 3.5 over the entire intervention period.

A total of 178 out of 660 patients (27%) attending the clinics were retained in care. Out of the 482 patients not retained in care, the community health workers followed up 127 at their homes; the remainder could not be reached after up to three revisits. The main reasons for not being retained in care included cost of treatment (45 patients; 35%), lack of time (26; 21%) and forgot clinic appointment (22; 17%).

Impact

Out of 178 patients retained in care, 58 (33%) had their blood pressure controlled by the sixth visit. This amounts to 9% of all 660 patients recruited into the clinics (Fig. 1).

The mean systolic blood pressure of those retained in care and with complete data (n = 177) was 161.6 mmHg at the first visit and this was reduced by 19.8 mmHg (95% confidence interval, CI: 16.0–23.6) by the sixth visit. Their mean diastolic blood pressure was 100.5 mmHg at the first visit and this was reduced by 10.4 mmHg (95% CI: 8.2–12.7) by the sixth visit.

Costs

Table 2 shows the total costs of the intervention by input category and intervention component. The awareness and screening component of the intervention accounted for 38% (US$ 70 071) of the total cost of US$ 185 861, access to treatment for 44% (US$ 81 337) and retention in care for 19% (US$ 34 453). Personnel was the highest input cost at 53% (US$ 99 119) of the total cost.

Table 2

Total costs by input category and component of the 18-month community-based intervention for hypertension management in Kenya, 2012–2013

Input category	Cost, US$
	Awareness and screening	Treatment	Retention in care	All components (%)a
Personnel
CHWs facilitation fee (support groups)	–	–	329	329
Programme management	9 146	32 926	21 951	64 022
Field supervisor	3 747	3 747	3 747	11 241
Field team leaders	7 169	6 452	418	14 040
Clinical staff for cardiovascular diseases’ clinics	–	9 487	–	9 487
Total	–	–	–	99 119 (53)
Commodities and supplies
Medical consumables	2 042	1 602	–	3 645
Non-medical supplies	3 642	6 590	–	10 231
Medicationsb	0	0	0	0
Total	–	–	–	13 876 (8)
Training
Training sessions	4 498	1 519	748	6 765 (4)
Capital cost
Clinic upgrading	–	1 190	–	1 190
Equipment	19 126	1 460	–	20 586
Furniture	–	593	–	593
Total	–	–	–	22 369 (12)
Building operating and maintenance
Repairs	–	12	–	12
Field communication	302	272	18	591
Field security	590	–	–	590
Cleaning	–	104	–	104
Building rentc	–	141	–	141
Total	–	–	–	1 437 (1)
Transport
Transport for supervision visits	889	1 671	108	2 668 (1)
Intervention activities
CHWs for screening and referral	7 249	–	–	7 249
CHWs for retention in care	–	–	935	935
First free treatment voucher	–	1 165	–	1 165
Community gatherings (baraazas)	589	–	–	589
Religious services	271	–	–	271
Radio jingle	124	–	–	124
Running of support groups	–	–	376	376
Training the trainers sessions	–	–	515	515
SMS reminders	–	–	53	53
Total	–	–	–	11 275 (6)
Indirect expenses
Programme overheads (estimated at 18%)	10 689	12 407	5 256	28 352 (15)
All categories	70 071	81 337	34 453	185 861 (100)

CHWs: community health workers; SMS: short message service; US$: United States dollars.

a Total cost of input category as a percentage of total intervention cost (US$ 185 861).

b Drug costs were paid by patients.

c Donated; value was estimated and included utilities (electricity, water, medical waste disposal etc.).

Notes: The average conversion rate during 2013 was 85 Kenyan shillings to US$ 1. Dashes indicate data not applicable.

Table 3 summarizes the unit costs per patient with blood pressure controlled for the three components of the intervention. The unit cost per person reached via screening and awareness was US$ 17. For access to treatment, the cost was estimated at US$ 123 per person seeking treatment. It cost US$ 194 per person to retain a patient in care. The overall cost of getting a person screened, treated, retained in care and to have their blood pressure under control was US$ 3205.

Table 3

Summary of costs per unit of outcome at each stage of the community-based intervention for hypertension management in Kenya, 2012–2013

Intervention component	Cost, US$	No. of people reached	Cost per person reached, US$
Awareness and screening	70 071	4 049	17
Treatment	81 337	660	123
Retention in care	34 453	178	194
Blood pressure control	185 861	58	3 205

US$: United States dollars.

Note: The average conversion rate during 2013 was 85 Kenyan shillings to US$ 1.

Discussion

In summary, the intervention reached 60% of the target population, provided access to treatment to 68% of eligible patients with hypertension, retained 27% in care and achieved blood pressure control among 33% of patients retained in care.

These results show that, despite the intervention, the so-called rule of halves – “half the hypertensive population is undetected, half of those detected are untreated, and in half of those treated hypertension is not controlled” – applied in our setting. Studies in developing countries have shown that the levels of awareness, treatment and control of hypertension are still quite low, with control rates ranging from 4% to 47% among patients aged 35–49 years. Even worldwide, only 13% of people with hypertension have adequate blood pressure control. Our study showed that with a comprehensive community-based intervention it is possible to achieve awareness and initial treatment rates above 50%. Achieving awareness and access to treatment levels that are comparable to high-income countries is commendable, especially in unstable populations such as those in slum areas, where annual migration rates alone could reach 30%. However, retention in care and blood pressure control rates in our population remains suboptimal despite the intervention components designed to address them.

We also found that our financial incentives were not strong enough to keep the majority of patients retained in care. We believe that this finding is specific to our setting: over 90% of slum residents in Kenya make out-of-pocket payments for health and more than 50% report being food insecure. Moreover, many of those who defaulted from the clinic cited cost as the main reason. In other words, although treatment costs were subsidized, it was still a barrier to care. Cost is an issue in other settings too. A study in 36 mostly low- and middle-income countries found that 1 month of daily treatment with one hypertensive drug cost on average 1.8 days’ wages. The World Health Organization has set a global target of a 25% reduction in the prevalence of hypertension by 2025. If this target is to be achieved, then mechanisms need to be found to make drugs more affordable, as has been achieved with tuberculosis treatment and antiretroviral therapy (ART) to supress human immunodeficiency virus (HIV) infection.

The cost of the entire intervention per person with blood pressure controlled was US$ 3205. This compares favourably to other multifaceted public health interventions in Kenya. For example, the implementation of the option B+ approach, in which HIV-positive pregnant women are started immediately on ART and continued for life – a comparable intervention as it includes screening, diagnosis and chronic treatment – was recently estimated to be US$ 6015 per infection averted. However, when we place the cost of each component of the hypertension intervention in the context of other public health interventions for cardiovascular disease prevention we find that our costs are high. For example, a recent systematic review found that the costs of using general medical practitioners in community hypertension programmes ranged from only US$ 0.81 to US$ 8.67 per patient per year, versus our unit cost of US$ 123 per person seeking treatment. Furthermore, the per capita cost of the intervention was almost three times higher than the gross domestic product per capita of Kenya in 2014 (US$ 1290). Yet compared with the treatment costs of other chronic diseases our costs are favourable. For example, from a provider perspective it costs US$ 273 and US$ 258 to treat drug-susceptible tuberculosis in lower middle-income and low-income countries, respectively. The median cost of antiretroviral therapy per patient per year is estimated to range from US$ 682 to US$ 1089 in low-income countries and from US$ 156 to US$ 3904 in lower middle-income countries.

The study has several limitations. First, due to budgetary limitations we were unable to collect all the data to compute the cost–effectiveness of our intervention, as originally intended in our study protocol for a quasi-experimental community-based trial. This meant that we did not obtain fasting blood glucose levels for all study participants to determine their 10-year cardiovascular diseases risk. Second, our study was conducted in a slum setting, which limits the generalizability of our findings to other settings. Nonetheless, as slums are said to constitute up to 60% of urban areas in low- and middle-income countries we believe our findings will be useful to urban health practitioners in other settings.

In conclusion, the intervention achieved reasonable success in terms of raising awareness and hypertension treatment levels in a challenging resource-constrained setting at a cost that in principle would be regarded as affordable when compared with public health interventions such as ART. We recommend, however, that further research be conducted to address low levels of retention in care and of blood pressure control in such settings. In terms of scalability and sustainability, we believe that the strength of the intervention is that it simplifies the process of identifying persons with high blood pressure at community level and linking them into care. This characteristic has the potential to make it applicable to other contexts. Indeed, certain elements of the intervention have been implemented successfully in other contexts. For example, an observational study in Bangladesh, Guatemala, Mexico and South Africa demonstrated that community health workers could do community-based screenings to predict cardiovascular disease risk as effectively as physicians or nurses. The clinics that we set up as part of this study were handed over to the local government in Kenya and continue to be operational in 2016.