Skip to main content
NCBI home page
Health Policy and Planning logoLink to Health Policy and Planning
. 2020 Dec 25;36(4):397–406. doi: 10.1093/heapol/czaa176

An innovative model for management of cardiovascular disease risk factors in the low resource setting of Cambodia

Nazaneen Nikpour Hernandez 1,✉, Samiha Ismail 2,3, Hen Heang 4, Maurits van Pelt 5, Miles D Witham 6,†, Justine I Davies 7,8,9,†
PMCID: PMC8128014  PMID: 33367513

Abstract

Non-communicable diseases are increasing in developing countries and control of diabetes and hypertension is needed to reduce rates of the leading causes of morbidity and mortality, stroke and ischaemic heart disease. We evaluated a programme in Cambodia, financed by a revolving drug fund, which utilizes Peer Educators to manage diabetes and hypertension in the community. We assessed clinical outcomes and retention in the programme. For all people enrolled in the programme between 2007 and 2016, the average change in blood pressure (BP) and percentage with controlled hypertension (BP < 140/<90 mmHg) or diabetes (fasting blood glucose (BG) < 7mg/dl, post-prandial BG < 130 mg/dl, or HBA1C < 7%) was calculated every 6 months from enrolment.  Attrition rate in the nth year of enrolment was calculated; associations with loss to follow-up were explored using cox regression. A total of 9139 patients enrolled between January 2007 and March 2016. For all people with hypertension, mean change in systolic and diastolic BP within the first year was −15.1 mmHg (SD 23.6, P < 0.0001) and −8.6 mmHg (SD 14.0, P < 0.0001), respectively. BP control was 50.5% at year 1, peaking at 70.6% at 5.5 years. 41.3% of people with diabetes achieved blood sugar control at 6 months and 44.4% at 6.5 years.  An average of 2.3 years [SD 1.9] was spent in programme. Attrition rate within year 1 of enrolment ranged from 29.8% to 61.5% with average of 44.1% [SD 10.3] across 2008–15. Patients with hypertension were more likely to leave the program compared to those with diabetes and males more likely than females. The programme shows a substantial and sustained rate of diabetes and hypertension control for those who remain in the program and could be a model for implementation in other low middle-income settings, however, further work is needed to improve patient retention.

Keywords: Health systems, diabetes, hypertension

Key Messages.

  • The author’s institute in Cambodia utilizes Peer Educators (PEs) to screen and manage diabetes and hypertension in the community and a revolving drug fund (RDF) is used to finance the programme.

  • Use of peer educators promotes sustained rates of both hypertension and diabetes control. Attrition rate is variable but substantial.

  • The programme shows a significant and sustained rate of diabetes and hypertension control for those who remain in the program. There is a utility for rolling this model out across Cambodia and in other low- and middle-income country settings; however, further work is needed to improve patient retention.

Introduction

The global aim of achieving Universal Health Coverage (UHC) requires innovative, wide-reaching interventions that improve access to care for vulnerable populations. Ageing populations, globalization, reduction in physical activity and changing diets have led to a rapid rise in non-communicable diseases (NCDs) against a backdrop of prevalent infectious diseases in many lower- and middle-income countries (LMICs). Cambodia, a LMIC in South East Asia, has a rapidly growing economy with aims to reach upper-middle income country status by 2030. Although poverty rates are falling, 4.5 million of the population of over 16 million remain in near poverty and are vulnerable (World Bank, 2020). Life expectancy at birth was 72.7 years and 66.8 years for females and males, respectively, in 2017; an increase of 10 years in the last two decades (IHME, 2017). As expected with rapid economic development, NCDs are becoming prevalent. NCDs accounted for 64% of all deaths in 2018, and in 2016 stroke was the leading cause of death with ischaemic heart disease ranked as fourth (WHO, 2018). Type-2 diabetes mellitus and hypertension—modifiable risk factors for these conditions—were found in 5.9% and 11.2% of the population, respectively, in 2016 (Sophal Oum, 2010; WHO, 2016), making them high-priority conditions to start tackling NCD morbidity and mortality. Despite this, in 2018, fewer than 25% of primary health centres reported to offer cardiovascular disease (CVD) risk factor stratification, guidelines for CVD prevention were not used in >50% of facilities, and only 30% of facilities reported medicines for treating NCDs were generally available (WHO, 2018). Data from countries in South East Asia have shown that fewer than 40% of the population with hypertension have been diagnosed and fewer than 10% have this controlled to target (Geldsetzer et al., 2019). The figures are only slightly better for people with diabetes (Manne-Goehler et al., 2019). In countries with underdeveloped health systems, such as Cambodia, community-centred models of care may provide solutions to ensure all healthcare needs of individuals are met (Schneider and Lehmann, 2016).

The delivery of chronic disease healthcare models has evolved in different ways across different settings. For example, in the UK, primary care facilities are used in screening, initiating treatment and monitoring. There is a strong focus on encouraging self-management for achieving disease control, and General Practitioners are incentivized with the Quality and Outcomes Framework (QOF) to screen patients (NHS Digital, 2020). However, in the UK, contact with the health services is frequent and people are offered health checks after the age of 40. In contrast, the USA provides the majority of management through secondary care facilities using national and private health insurance programs (Commonwealth Fund, 2016). In LMICs, strategies for dealing with NCDs are in their infancy, given the focus of healthcare provision has largely been on providing care for acute conditions. For screening, suggested approaches involve Community Healthcare Workers or a similar cadre. However, these have not been widely rolled out in real-world settings (Ndou et al., 2013; Tang et al., 2014; Munshi et al., 2019). Many suggested approaches for treatment in LMICs developed in recent years include the diagonal approaches of integrating existing infectious disease services to manage long-term conditions. These services are largely run by non-governmental organizations and financed with out-of-pocket expenditure (Janssens et al., 2007; Datta et al., 2019). Approaches specific to managing chronic diseases, via strengthened primary care infrastructures, have also begun to gain popularity, but are still often dependent on private funding grants rather than national revenue (Collins et al., 2019; Laar et al., 2019). Regardless of approach or setting, analyses have shown that large proportions of patients with diabetes or hypertension are lost at all stages of the cascade of care, from testing and diagnosis, to treatment and ultimately control. In LMICs, a large proportion is lost in the first stage of the cascade (testing) (Geldsetzer, 2019; Manne-Goehler et al., 2019). Therefore, there is a need for innovative new healthcare models that address the common barriers to accessing healthcare throughout the cascade of care, from disease recognition to adequate treatment, including high out-of-pocket expenses or reliable medication supply (Joshi et al., 2008). Robust assessment of new programmes that aim to address the entire cascade are needed to inform the development of effective chronic disease management in LMICs.

MoPoTsyo applies a unique healthcare model that utilizes Peer Educators (PEs), patients with diabetes and/or hypertension themselves, to return to their local villages to screen and initiate management of fellow community members. Elsewhere, the role of PEs in diabetes and self-management has also been recognized in the last two decades (Brownson and Heisler, 2009) with increasingly robust data that shows impact (Debussche et al., 2018). The Programme structure also uses a Revolving Drug Fund (RDF), which stabilizes essential medicine provision and creates a sustainable income source (Cross et al., 1986; Murakami et al., 2001; van Olmen et al., 2016). The PE roles have since been formalized into the existing national health infrastructure of two Operational Districts (ODs) in Cambodia, with a view to expand, whereas the RDF management remains under the organization of The Programme. One previous study (Taniguchi et al., 2017) has examined the clinical effectiveness of The Programme between 2007 and 2013 in Takeo province, where it first started a rural network. Our article expands on these clinical outcomes to include data until 2016 (a total of 8 years), and additionally to describe the long-term retention of patients, which is an important element of its long-term effectiveness.

Methods

Target population

The study population includes people aged over 18 years living in the largely rural Takeo province in Cambodia, who were screened and diagnosed with diabetes or hypertension. Takeo is the sixth largest of the 25 provinces in Cambodia (population 899 485 [5.9%]) (National Institute of Statistics, 2019).

Programme description

MoPoTsyo uses a community-centred approach to long-term care for diabetes and hypertension management. PEs, patients selected for their motivation, are trained to screen and manage these conditions in their local village and health centre catchment area. PEs carry out screening of people aged 40 years and over and are given small payments for travel (US$0.04 per km), registering new patients (US$0.12 per patient), and more recently measuring blood glucose (BG) (US$0.62 per patient) and blood pressure (BP) (US$0.12 per patient). Local guidelines define the presence of diabetes as fasting blood glucose (FBG) ≥ 126 mg/dl or post-prandial blood glucose (PPBG) ≥ 180 mg/dl and hypertension as systolic ≥ 140 mmHg and diastolic ≥ 90 mmHg. Only patients with BG or pressure results above these thresholds are enrolled in the program, all other adults in the household are advised to contact the peer educator if they notice symptoms of diabetes. The role of PEs also includes arranging village health meetings to discuss lifestyle measures and diet. PEs are active in 21 of 100 operational districts (OD) of Cambodia and each OD is served by a referral hospital where physician-led prescriptions are made. The programme itself provides laboratory services and operates an RDF for vetted, reliable and lower-priced drugs supplied to contracted pharmacies. PEs act as an intermediary between the physician and hard to reach patients to deliver and explain blood results, pick-up prescriptions and monitor blood sugars and pressures. Early in its inception, in 2007, the program engaged at a national level to have PEs recognized as national healthcare workers; they were recognized in the National Strategic Plan for Prevention and Control of NCD 2013–20 (Ministry of Health, 2012). The role of the PE has also been described in the ‘National Standard Operating Procedures for Diabetes and Hypertension in Primary Care’ (Ministry of Health, 2019). However, training of PEs is still undertaken by MoPoTsyo with a view to transitioning this to the care of national services in the coming 5 years.

Data

Data from all patients enrolled in Takeo province from 1st February 2007 to 31st March 2016 were included in the study. Multiple electronic datasets containing enrolment data, laboratory results, doctors’ follow-up and PE follow-up were merged.

Variables captured at enrolment were: sex, age, whether prescription medication or traditional medicines were already being taken, smoking status, height and weight (using generic scales and a height measurement stick and from which body mass index (BMI) was derived in kg/m2), FBG or PPBG (mg/dl), and systolic and diastolic BP (mmHg). Capillary blood glucose variables were measured using 2007 Roche Accu-Chek, which was changed to ACON On Call Plus machine from 1st December 2009 onwards. HbA1c was measured in the laboratory using SIEMENS DCA. BP was measured using Automatic Citizen CH-432B machine in 2009, AS-35B in 2010 and in 2015 both ALP-K2 and Nissei SL 1902.

PEs classified patients with a fasting capillary blood glucose of ≥ 126 mg/dl or PPBG of ≥ 180 mg/dl as having diabetes. Patients with systolic BP ≥ 140 mmHg and diastolic BP ≥ 90 mmHg were classified as having hypertension. The clinical variables extracted from the follow-up databases were systolic and diastolic BP (mmHg), capillary glucose reading of FBG (mg/dl), PPBG (mg/dl) and/or HBA1C (%).

Study variables

Outcome measures

Our main outcome was change in BP over time for patients enrolled into the program with hypertension.

Other outcomes were (1) The population in the program enrolled with hypertension or diabetes who were controlled over time and (2) loss to follow-up of participants enrolled in the program.

Patients were defined as having either hypertension or diabetes based on the PE diagnosis.

For hypertension, three disease groupings explored in the analyses were all people with hypertension (with or without diabetes), diabetes and hypertension (concurrent diabetes and hypertension), or hypertension alone (without diabetes). BP control at follow-up was defined as systolic BP < 140 mmHg and diastolic BP < 90 mmHg (%). A separate analysis was done for patients with concurrent diabetes and hypertension, using a BP target of <130 mmHg systolic and <80 mmHg diastolic as per clinical guidelines for patients with diabetes.

For diabetes, all people with diabetes (with or without hypertension) were included. Blood sugar controlled to target at follow-up was defined, as per local guidelines, as FBG of ˂ 130 mg/dl, PPBG ˂ 180 mg/dl or HBA1C ˂ 7%. In patients who had more than one form of blood sugar measurement at a given follow-up time, HbA1c was chosen as the preferred measure, PPBG was the second preferred, and FBG was the least preferred, based on our latest understanding of the correlation of point of care (POC) blood sugar testing with HbA1c levels (Ketema and Kibret, 2015).

Time to follow-up was calculated as the date of follow-up visit minus enrolment date, in years to the nearest 6 months. Patients were categorized as lost to follow-up (‘left the programme’) if they had not reattended by one year after their last follow-up date. Timepoint 0 represents the date of enrolment.

Explanatory variables

Overweight was classified as having a BMI of ≥ 23 kg/m2 to reflect lower BMI average in Asia (WHO expert consultation, 2004), BMI of ≥ 25 kg/m2 was also used to show comparison with global standards. BMI ≥ 30 kg/m2 was defined as obese. Smoking was captured by self-report and taken to mean current smoker. Participants with no response to the question on smoking were classed as non-smokers.

Age-groups were categorized as aged under 60 or aged 60 and over. Whether on prescription or traditional medicine at baseline was captured as yes or no. In addition, the number of prescription medicines prescribed by the program to control BP or diabetes was captured as continuous variables at 1-year post enrolment, as well as the maximum number of medications taken by each individual at any time point in the program.

Analyses

All data were analysed using open-source statistical program RStudio Inc – Version 1.0136 2016 under public license. For all analyses, a two-sided P value of <0.05 was taken as significant.

Entries with incomplete data—mostly missing follow-up dates—or those with follow-up dates that occurred before their enrolment day (incorrect data entry) were excluded from analyses. Outlier entries with systolic or diastolic BP ˂ 20 mmHg were also excluded. Data are described as n (%) for categorical and mean (SD) for normally distributed variables (after testing for normality of distribution).

For patients with hypertension, the systolic and diastolic BP (mmHg) change from baseline were computed using the follow-up BP closest in time to each 6 monthly follow-up time point. Where multiple BP recordings existed for a patient that were equally close to the follow-up year, the mean reading was used for analysis. The mean BP change from enrolment for all patients at each timepoint is displayed for each hypertension group and disaggregated by sex (male and female) and age category (under 60 years and age 60 and over).

A paired samples t-test was carried out to test the significance of the mean change from baseline in systolic and diastolic BP at each timepoint and an unpaired t-test was carried out to test significance between male and female sex.

Percentage of patients with systolic and diastolic BP within the target range at each follow-up time was displayed for each hypertension group. For all patients with hypertension, binary logistic regression was done using complete cases to assess the effects of explanatory variables; the presence of diabetes, age group, sex, smoking status, BMI and number of medications for hypertension taken at the 1-year time point, on whether hypertension was controlled to target.

For all participants with diabetes, the percentage of patients with controlled diabetes at each follow-up time was calculated. To explore associations with diabetes controlled to target, binary logistic regression was done using complete cases, using the explanatory variables; the presence of hypertension, age group, sex, smoking status, BMI and number of diabetes medications taken at the 1-year time point.

Given attrition rates decline with increased duration in a program, attrition rates were calculated dependent on year of follow-up (n, the ‘n’th year after enrolment) and the year (Y). Patients were defined as lost to follow-up if they had not been seen in the year after their last visit, therefore, for any year, attrition rate, is the number of people enrolled in year ‘Y − n’, seen in year ‘Y − 1’ and not seen in year ‘Y’, divided by the total number of people enrolled in year ‘Y − n’ who were seen in year ‘Y − 1’.

The formula used was as follows:

graphic file with name M0001-Latex.gif

To explore the probability of any patients leaving the program at n years after enrolment, Cox regression analysis was done using complete cases with assumption of proportional hazards. Data were censored for those who remained in the program for the full duration of the data period. Explanatory variables included in the model were disease type (diabetes alone, hypertension alone, diabetes and hypertension), sex, age group, BMI, smoking status and total number of medications taken at 1 year. Kaplan–Meier survival plots were plotted for disease groups.

This study is a retrospective exploratory analysis of real-world data and therefore no sample size calculation was carried out.

Ethics

Ethical approval was granted from the Cambodian National Ethics Committee for Health Research (NECHR) and the London School of Hygiene and Tropical Medicine (LSHTM).

Results

A total 9139 patients were enrolled in the programme between January 2007 and March 2016, of which 66.4% were female, mean age was 56.4 [SD 13.5], 80.3% had a diagnosis of hypertension, and 47.6% had a diagnosis of diabetes (see Supplementary Appendix Table 1 for characteristics of all patients enrolled). 6677 patients had data available for inclusion in analyses (see Supplementary Appendix Table 2 for characteristics of excluded patients). Depending on analysis, not all 6677 patients had information available, 57 patients were excluded from analyses of blood pressure due to missing blood pressure measurements and 10 patients were excluded from analyses of diabetes due to missing blood sugar measurements (Figure 1). Of the total 6677 patients included, 59.2% (3956) had diabetes, 74.7% (4987) had hypertension, 34.5% (2302) had diabetes and hypertension and 40.2% (2685) had hypertension alone (Table 1). 67.1% patients included were female and the mean patient age was 55.7 years (SD 12.9). 15.7% patients were taking traditional medicines at enrolment and 5.4% reported smoking. 46.2% of all patients (48.5% of patients with diabetes) were overweight (BMI ≥ 23 kg/m2). The mean (SD) number of medications taken for hypertension or diabetes at the 1-year time point was 1.1 (1.2) and 0.8 (1.0), respectively. Mean (SD) maximum number of medications used to treat hypertension or diabetes in the program was 1.4 (1.3) and 1.0 (1.0), respectively (Table 1).

Figure 1.

Figure 1

Open in a new tab

Study inclusion and exclusion chart.

Table 1.

Characteristics of patients enrolled between 2007 and 2016

Pts included in analysis (n = 6677)
 All diabetes (n = 3956, [59.2%])a
 All hypertension (n = 4987, [74.7%])a
 Diabetes and Hypertension (n = 2302, [34.5%])a
 Hypertension alone (n = 2685, [40.2%])a
n % n % n % n % n %
Male 2198 32.9 1276 32.3 1634 32.8 727 31.6 907 33.8
Female 4479 67.1 2680 67.7 3353 67.2 1575 68.4 1778 66.2
Age mean (SD) 55.7 (12.8) 54.5 [12.5] 56.6 (12.1) 55.9 (11.8) 57.3 (12.4)
Under 60 years 4127 61.8 2639 66.7 2905 58.3 1435 62.3 1470 54.7
60 years and over 2550 38.2 1317 33.3 2082 41.7 867 37.7 1215 45.3
Taking anti-HTN med at enrolment 514 7.7 187 4.7 497 10 170 7.4 327 12.2
Taking anti-DM med at enrolment 451 6.8 449 11.3 235 4.7 234 10.2 N/A N/A
Taking traditional med at enrolment 1051 15.7 474 12 910 18.2 335 14.6 575 21.4
Smoking at enrolment 358 5.4 210 5.3 248 5 100 4.3 148 5.5
Mean number of HTN drugs at 1 year (SD) 1.1 (1.2) N/A N/A 1.3 (1.2) 1.2 (1.2) 1.3 (1.3)
Mean number of DM meds at 1 year (SD) 0.8 (1.0) 1.4 (0.9) N/A N/A 1.4 (0.9) N/A N/A
Mean maximum number of HTN meds at any time (SD) 1.4 (1.3) N/A N/A 1.5 (1.0) 1.6 (0.9) 1.5 (0.4)
Mean maximum number of DM meds at any time (SD) 1.0 (1.0) 1.6 (0.9) N/A N/A 1.6 (0.9) N/A N/A
Mean BMIb (SD) 22.8 (3.8) 23.0 (3.8) N/A 23.0 (3.7) 23.5 (3.7) 22.6 (3.7)
BMI > 23 kg/m2 3085 46.2 1917 48.5 2420 48.5 1258 54.6 1162 43.3
BMI >25 kg/m2 1722 25.8 1060 26.8 1372 27.5 714 31 658 24.5
BMI >30 kg/m2 211 3.2 126 3.2 176 3.5 93 4 83 3.1
Open in a new tab

Pts, patients; n, number; %, percentage; SD, standard deviation; BMI, body mass index; HTN, hypertension; DM, diabetes.

a

Denominator 6677 used to calculate percentage of all patients included.

b

Patients with no height or weight or BMI < 8 excluded.

From the time of enrolment to their last review date, patients had between 1 and 84 follow-up visits, with an average of 10.9 (SD 11.1) visits.

Blood pressure control

For all patients with hypertension, change in BP (mmHg) from enrolment is shown in Figure 2 (Supplementary Appendix Table 3 shows the maximum and mean BPs for the 1174 patients who had measurements that were equidistant from the 6-month analysis timepoint). The mean change in systolic and diastolic BP from baseline within the first year was −15.1 mmHg (SD 23.6, P < 0.0001) and −8.6 mmHg (SD 14.0, P < 0.0001), respectively. The greatest fall in systolic blood pressure occurred at 2.5 years after enrolment, (−18.9 mmHg [SD 37.4, P < 0.0001]) and at 3.5 years for diastolic BP (−11.4 mmHg [SD 24.5, P < 0.0001], Figure 2). For patients who had diabetes and hypertension, mean change in BP from baseline 1 year after enrolment was −10.9 mmHg (SD 22.9, P < 0.0001) systolic and −7.1 mmHg (SD 12.9, P < 0.0001) diastolic (Supplementary Appendix Figure 1). For patients with hypertension alone, mean change in BP from baseline at 1.0 year was −20.0 mmHg (SD 24.5, P < 0.0001) systolic and −8.5 mmHg (SD 14.2, P < 0.0001) diastolic (Supplementary Appendix Figure 1.1).

Figure 2.

Figure 2

Open in a new tab

Average change in blood pressure from baseline for all patients with hypertension. Error bars show standard deviation. Number of results included at each time from enrolment are shown along the x-axis. BP at each time point is statistically significant compared to baseline (P < 0.001). Year 7.5 to 8.5 are excluded as fewer than 30 patients’ data are available.

Considering all patients with hypertension, mean systolic or diastolic BP change from baseline was similar for males and females and for those older or younger than 60 years. (Supplementary Appendix Figures 1.2–1.5).

Percentages of all people with hypertension with systolic and diastolic BP controlled to target at each follow-up time are shown in Figure 3. The proportion of patients with controlled BP climbs steadily to a peak of 70.6% at 5.5 years. Supplementary Appendix Figure 2 shows control for patients with concurrent diabetes and hypertension within the target of <130/<80 mmHg as per clinical guidelines. Binary logistic regression showed the presence of diabetes, being younger than 60 years old and smoking at enrolment were all significantly associated with having hypertension controlled to target at 1 year after enrolment. Whereas the number of antihypertensive medications taken, sex, and BMI were not significantly associated (Supplementary Appendix Table 4).

Figure 3.

Figure 3

Open in a new tab

Percentage of patients who have hypertension (H) or diabetes (D) controlled to target at each time point after enrolment. Numbers of results included at each time point from enrolment are shown on the x-axis.

Blood sugar control

The proportion of patients within blood sugar levels controlled remained stable throughout the study period (Figure 3) at close to 40%. Binary logistic regression showed number of anti-glycaemic medications and BMI were significantly associated with blood sugar control at 1 year. Presence of hypertension, age group, sex and smoking status were not significantly associated (Supplementary Appendix Table 5).

Attrition rate

The average time spent in the program for all patients was 2.3 years [SD 1.9]. Average rate of attrition within the first year after enrolment, across the 7-year period (2008–15) was 44.1% (SD 10.3, range 29.8% to 61.5%). Attrition in the 2nd year after enrolment ranged between 32.4% and 50.0% with a mean of 38.3% [SD 5.8] (Figure 4). Cox Regression analyses results are shown in Table 2. Those with hypertension alone were more likely to leave compared to those with diabetes alone. Age under 60 years was protective against leaving the programme as was those with higher BMI compared to those with BMI under 23 kg/m2. Those who reported smoking were more likely to leave the programme as were those who took between 1 and 3, compared to 0, medications for diabetes or hypertension at 1 year. Figure 5 shows Kaplan–Meier plots of attrition for patients with hypertension, diabetes, or both hypertension and diabetes.

Figure 4.

Figure 4

Open in a new tab

Attrition rate for people leaving within 1st and 2nd year (yr) of enrolment.

Table 2.

Cox regression for explanatory variables on probability of leaving the programme

Explanatory variable Group HR (95% CI) P value
Disease Diabetes alone Referent
Hypertension alone  1.91 (1.75, 2.08) <0.001
Diabetes and hypertension 1.04 (0.95, 1.13) 0.394
Sex Female Referent
Male 1.15 (1.08, 1.22) <0. 001
Age group 60 years and over Referent
Under 60 years 0.87 (0.82, 0.93) <0.001
BMI BMI under 23 Referent
BMI 23 and over 0.93 (0.86, 1.0) 0.047
BMI 25 and over 0.84 (0.78, 0.91) <0.001
BMI 30 and over 0.81 (0.68, 0.97) 0.025
Smoking status NO Referent
YES 1.49 (1.30, 1.71) <0.001
Total number of medications taken at 1 year 0 Referent
1 1.16 (1.02, 1.33) 0.027
2 1.14 (1.05, 1.23) 0.001
3 1.11 (1.02, 1.22) 0.022
4 0.96 (0.86, 1.09) 0.547
5 0.90 (0.76, 1.07) 0.269
6 0.75 (0.53, 1.06) 0.107
7 1.09 (0.15, 7.78) 0.929
Open in a new tab

HR, hazard ratio; 95% CI, 95% confidence intervals.

Figure 5.

Figure 5

Open in a new tab

Kaplan–Meier survival graph for remaining in program between 2007 and 2016 by diagnosis.

Discussion

The analyses show that a program using a revolving drug fund and peers to screen and follow-up patients with hypertension and diabetes is effective at reducing BP and improving glycaemic control. However, retention in the program is low, with attrition being similar in each year after enrolment. The focus on healthcare provision in many LMICs has been on improving care for infectious diseases and reducing maternal and neonatal mortality and morbidity. However, with the Sustainable Development Goals has come a focus also on reducing premature mortality from non-communicable diseases and provision of universal health coverage (UHC). Unfortunately, the reality is that coverage of diseases under UHC, even in countries where this exists, may not extend to non-communicable diseases (The Lancet, 2014), and looking towards other models of ensuring access to medicines for people with these conditions remains a valuable lesson. Thus, even though The Programme uses an Out-of-Pocket payment system for ensuring medicines availability, our findings have important policy implications for other countries without UHC looking to scale up care for these diseases and suggest that although the model works in the short term, more needs to be done to encourage retention.

The most rapid reduction in systolic and diastolic BP was within the first year of enrolling with MoPoTsyo; the comparatively lower reduction in diastolic compared to systolic BP is in keeping with our understanding of the effect of common antihypertensive agents (Wu et al., 2005). However, sustained BP reduction was observed throughout the study period. Although we did not formally assess changes in cardiovascular risk over time, the increasing proportion of patients with BP and diabetes controlled over time suggests that patients enrolled within MoPoTsyo will have lowered their cardiovascular risk (NICE National Institute for Health and Care Excellence, 2019).

Our findings, with more patients and longer duration of follow-up were similar to those previously found (Taniguchi et al., 2017). Whilst it is disappointing that results haven’t improved over time, MoPoTsyo appears to achieve much higher levels of control compared to data analysed from the World Health Organizations (WHO) Study on Global Ageing and Adult Health (SAGE) where BP control ranged from 4.1% in Ghana to 14.1% in India (Lloyd-Sherlock et al., 2014). BP and diabetes control was also higher in MoPoTsyo than has been shown in analyses of the cascade of care in other South East Asian or LMIC settings (Chow et al., 2013; Geldsetzer et al., 2019; Manne-Goehler et al., 2019). Our relatively positive findings may be a particularity of the societal context or relatively developed setting of Cambodia. However, assessment of a Cambodian Médecins San Frontier (MSF) clinic for Diabetes from 2002 to 2008 (Raguenaud et al., 2009) working in the same region as our programme, found only 24% patients achieved blood sugar control within 1 year. This suggests that the program itself had positive benefits, although whether these are due to the use of peer educators or a revolving drug fund is not possible to tell from our study. The percentage control amongst MoPoTsyo is more comparable to those seen in a chronic care model implemented in Philippines called First-Line Diabetes Care (FiLDCare), which saw rates of control of 50.6% within six months. FiLDCare initiated self-management education in local health facilities using existing community health workers to encourage self-management at the community level. in contrast to the author’s institute, The Philippines programme had the family physician being the primary contact for patients rather than the community health worker. The payments for medications in FiLDCare were also out of pocket although with a plan to become subsidized (Ku and Kegels, 2015).

Experiences from settings where healthcare is free at the point of access suggest that cost of medications is not the only limiting factor to achieving diabetes control. For example, in the UK, where medications are either free or available for a small prescription charge, rates of control were historically low; only 24% of people with diabetes had blood glucose measurements within target levels between 1998 and 2002 (Fox et al., 2006). However, more recent data show control rates for type 2 diabetes as high as 66.8% in 2016–17 (NHS Digital, 2017). Potentially, multiple factors lie behind this increase, including health education—with an increase in awareness of diabetes and hypertension in the general public over time (Li et al., 2010; Falaschetti et al., 2014; Nazar et al., 2015) and presence of pay for performance incentives for healthcare practitioners (Vaghela et al., 2009). Although health literacy in general and specifically relating to cardiovascular diseases and its risk factors has not been formally assessed in MoPoTsyo, the use of PEs and the regular screening involved is likely to increase the knowledge of the general population. PEs are also paid on a per patient basis and this is likely to have a similar motivating effect to the UK pay for performance initiative.

That MoPoTsyo’s outcomes compare favourably with those from other LMIC settings is encouraging. The focus of this paper was to illicit whether sustained reduction in clinical outcome measures were achieved in this real-world assessment of the programme and we are unable to comment on what aspects of it, e.g. peer educator or revolving drug fund, contribute to the favourable outcomes. Other analyses in LMIC provide evidence that PEs may contribute to improving glycaemic control; for example, a randomized control trial carried out in Mali where PEs delivered structured education programmes to promote self-management showed higher levels of diabetes control (Debussche et al., 2018). The Programme’s own PEs were implemented to specifically address the screening and diagnosis of patients as well as aiming to reduce barriers in accessing care and medications (van Olmen et al., 2016). However, to what extent this influences clinical outcomes requires further research to ascertain, but this knowledge would enhance the applicability of elements of the chronic disease programme in other LMICs. Existing research supports the use of non-physician health workers, such as PEs, as a way of achieving key aspects of the WHO health system building blocks needed to deliver NCD care (Heller et al., 2019). Peer groups have also been trialled in women’s health programmes to improve maternal and child health (Lewycka et al., 2013) and use of peers may improve care by reducing distances to accessing care, increasing patient knowledge and empowerment for improving behaviours to combat cardiovascular risk factors (Khetan et al., 2017; Brownson and Heisler, 2009; Tang et al., 2014).

Despite the short-term success of The Programme, the attrition rates are unacceptably high given that diabetes and hypertension are usually lifelong conditions. In some LMIC settings, the lack of recognition of diabetes or hypertension as chronic conditions by patients, combined with the costs of medications leads to high attrition rates (Rachlis et al., 2016; Amira and Okubadejo, 2007; Buabeng, Matowe and Plange-Rhule, 2004). However, this isn’t always the case; one hypertension management programme in Vietnam found drop-out was as low as 14.3% over two years (Nguyen et al., 2011) and the results from a diabetes management programme run in the same province of Takeo, by MSF, showed attrition over a 5-year period of 34% (Raguenaud et al., 2009). The reasons for the high attrition rate in MoPoTsyo need to be explored. Given that we didn’t have information about death of patients, and that a large proportion of those enrolled were over 60 years of age, it may be that the attrition rate could partly be explained by patient death, especially as the life expectancy in Cambodia is currently around 70 years. However, mortality is unlikely to be the only explanation and further qualitative research needs to be done to explore the reasons for leaving the programme and offer solutions. It is encouraging that over time we saw an overall decline in first-and second-year attrition for the programme. This may be related to MoPoTsyo’s evolution and improved practise, with additional increased funding as it scaled up in 2010.

Study limitations

This real-world study used secondary data, and data capture was designed for clinical service rather than research studies. There were changes to the programme’s data entry methodology over the period 2007–16 linked with the set-up and expansion of MoPoTsyo, which led to gaps in data and some discrepancies with dating, which meant that large numbers of data entries had to be excluded between 2007 and 2009. The gaps in data and lack of information regarding reasons for leaving the program limits our ability to comment on causal factors for the results. The covariates included in our analyses were limited by the data that were available for the majority of patients. This limited, for example, our ability to reliably ascertain the influence of smoking status on attrition from the programme. In addition, we have no information regarding the patients’ use of alternative health services alongside MoPoTsyo, although our assumption is that it provided the majority of care for patients enrolled. There are low numbers of patients who were enrolled in the programme in the first few years of inception, which leads to low numbers of patients with data available for longer than six years. Some patients were already diagnosed with diabetes and/or hypertension, therefore our study cannot assess the first two parts of the cascade (population prevalence to screened and screened to given diagnosis). However, assessing the latter parts of the cascade gives vital information about the ability of the health service to deliver good quality clinical care. In addition, for the diabetes group, there were multiple methods for analysing blood sugar used by MoPoTsyo, therefore we were only able to look at diabetes outcomes as a binary variable (controlled or not ) compared to blood pressure analyses which allowed for change in BP analyses.

Death was not captured as an outcome, thus in our attrition analyses we were not able to distinguish between this and voluntarily leaving the program. Some of the improvements in outcomes seen may have been due to regression to the mean, however, given that this is an analysis of a real-world program and treatments were given as they would be in clinical practice, we have not accounted for blood pressure at baseline in our analyses, as this would not be done in clinical practice to determine treatment success. Finally, it is also important to consider that in assessing the clinical effectiveness of novel chronic disease programmes, the possible adverse effects of overtreatment of both BP and blood sugar, which leads to hypotension and hypoglycaemia, need to be considered, especially amongst frail older persons (Monami et al., 2013).

Conclusion

This study shows that the innovative use of peer educators and a revolving drug fund as part of a community-centred chronic disease programme results in substantial and sustained rates of diabetes and hypertension control for patients who remain in the program. The role of PEs at a village level reduces many barriers to accessing healthcare for the majority rural population in Cambodia. PEs bring point of care testing to village homes, act as a medium for long-term monitoring and drug prescription collection for less mobile patients, and share peer to peer health education at a local level. MoPoTsyo could prove to be a model for implementation in other low middle-income settings, however, further research is needed to determine reasons for the variable attrition rates.

Supplementary data

Supplementary data are available at Health Policy and Planning online.

Conflict of interest statement. Heang Hen and Maurits van Pelt work for The Programme. All other authors declare no conflict of interest.

Ethical approval. Ethical approval was granted from the Cambodian National Ethics Committee for Health Research (NECHR) and the London School of Hygiene and Tropical Medicine (LSHTM).

Supplementary Material

czaa176_Supp
Click here for additional data file. (636.2KB, zip)

Contributor Information

Nazaneen Nikpour Hernandez, Centre for Tropical Medicine and Global Health, University of Oxford, Oxford, OX3 7LG, UK.

Samiha Ismail, Institute for Applied Research, Birmingham University, Birmingham, B15 2TT, UK; Centre for Medical Education, Health Sciences Education, Queen Mary University of London, London, E1 4NS, UK.

Hen Heang, MoPoTsyo Patient Information Center, Stung Meanchey Commune, 12352 Phnom Penh, Cambodia.

Maurits van Pelt, MoPoTsyo Patient Information Center, Stung Meanchey Commune, 12352 Phnom Penh, Cambodia.

Miles D Witham, AGE Research Group, NIHR Newcastle Biomedical Research Centre, Translational and Clinical Research Institute, Faculty of Medical Sciences, Newcastle University and Newcastle Upon Tyne Hospitals NHS Trust.

Justine I Davies, Institute for Applied Research, Birmingham University, Birmingham, B15 2TT, UK; Centre for Global Surgery, Department of Global Health, Stellenbosch University, Cape Town,South Africa; Medical Research Council/Wits University Rural Public Health and Health Transitions Research Unit, Faculty of Health Sciences, School of Public Health, University of the Witwatersrand, Johannesburg, South Africa.

References

  1. Amira CO, Okubadejo NU.. 2007. Factors influencing non-compliance with anti-hypertensive drug therapy in Nigerians. The Nigerian Postgraduate Medical Journal 14: 325–9. [PubMed] [Google Scholar]
  2. Brownson CA, Heisler M.. 2009. The role of peer support in diabetes care and self-management. The Patient: Patient-Centered Outcomes Research 2: 5–17. [DOI] [PubMed] [Google Scholar]
  3. Buabeng KO, Matowe L, Plange-Rhule J.. 2004. Unaffordable drug prices: the major cause of non-compliance with hypertension medication in Ghana. Journal of Pharmacy Pharmaceutical Science 7: 350–2. [PubMed] [Google Scholar]
  4. Chow CK, Teo KK, Rangarajan S. et al. 2013. Prevalence, awareness, treatment, and control of hypertension in rural and urban communities in high-, middle-, and low-income countries. JAMA 310: 959–68. [DOI] [PubMed] [Google Scholar]
  5. Collins TE, Nugent R, Webb D. et al. 2019. Time to align: development cooperation for the prevention and control of non-communicable diseases. BMJ 366: l4499. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Cross PN, Huff MA, Quick JD, Bates JA.. 1986. Revolving drug funds: conducting business in the public sector. Social Science & Medicine 22: 335–43. [DOI] [PubMed] [Google Scholar]
  7. Datta BK, Husain MJ, Asma S.. 2019. Assessing the relationship between out-of-pocket spending on blood pressure and diabetes medication and household catastrophic health expenditure: evidence from Pakistan. International Journal for Equity in Health 18: 9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Debussche X, Besanc O S, Balcou-Debussche M. et al. 2018. Structured peer-led diabetes self-management and support in low-income country: the ST2EP randomised control trial in Mali. PLoS One 13: e0191262. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Falaschetti E, Mindell J, Knott C, Poulter N.. 2014. 'Hypertension management in England: a serial cross-sectional study from 1994 to 2011. The Lancet 383: 1912–9. [DOI] [PubMed] [Google Scholar]
  10. Fox KM, Gerber PharmD RA, Bolinder B, Chen J, Kumar S.. 2006. Prevalence of inadequate glycemic control among patients with type 2 diabetes in the United Kingdom general practice research database: a series of retrospective analyses of data from 1998 through 2002. Clinical Therapeutics 28: 388–95. [DOI] [PubMed] [Google Scholar]
  11. Geldsetzer P, Manne-Goehler J, Marcus M-E. et al. 2019. The state of hypertension care in 44 low-income and middle-income countries: a cross-sectional study of nationally representative individual-level data from 1.1 million adults. The Lancet 394: 652–62. [DOI] [PubMed] [Google Scholar]
  12. Heller DJ, Kumar A, Kishore SP. et al. 2019. Assessment of barriers and facilitators to the delivery of care for noncommunicable diseases by nonphysician health workers in low- and middle-income countries: a systematic review and qualitative analysis. JAMA Network Open 2: e1916545. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. IHME 2017. Cambodia. Institute of Health Metrics and Evaluation. http://www.healthdata.org/cambodia, Last accessed 07 December 2020.
  14. Janssens B, Van Damme W, Raleigh B. et al. 2007. Offering integrated care for HIV/AIDS, diabetes and hypertension within chronic disease clinics in Cambodia. Bulletin of the World Health Organization 85: 880–5. p [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Joshi R, Jan S, Wu Y, MacMahon S.. 2008. Global inequalities in access to cardiovascular health care: our greatest challenge. Journal of the American College of Cardiology 52: 1817–25. [DOI] [PubMed] [Google Scholar]
  16. Ketema EB, Kibret KT.. 2015. Correlation of fasting and postprandial plasma glucose with HbA1c in assessing glycemic control; systematic review and meta-analysis. Archives of Public Health 73: 43. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Khetan AK, Purushothaman R, Chami T. et al. 2017. The effectiveness of community health workers for CVD prevention in LMIC. Global Heart 12: 233–43.e6. [DOI] [PubMed] [Google Scholar]
  18. Ku GMV, Kegels G.. 2015. Implementing elements of a context-adapted chronic care model to improve first-line diabetes care: effects on assessment of chronic illness care and glycaemic control among people with diabetes enrolled to the First-Line Diabetes Care (FiLDCare) Project in the Northern Philippines'. Primary Health Care Research & Development 16: 481–91. [DOI] [PubMed] [Google Scholar]
  19. Laar AK, Adler AJ, Kotoh AM. et al. 2019. Health system challenges to hypertension and related non-communicable diseases prevention and treatment: perspectives from Ghanaian stakeholders. BMC Health Services Research 19: 693. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Lewycka S, Mwansambo C, Rosato M. et al. 2013. Effect of women's groups and volunteer peer counselling on rates of mortality, morbidity, and health behaviours in mothers and children in rural Malawi (MaiMwana): a factorial, cluster-randomised controlled trial. The Lancet 381: 1721–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Li R, Zhang P, Barker LE, Chowdhury FM, Zhang X.. 2010. Cost-effectiveness of interventions to prevent and control diabetes mellitus: a systematic review. Diabetes Care 33: 1872–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Lloyd-Sherlock P, Beard J, Minicuci N, Ebrahim S, Chatterji S.. 2014. Hypertension among older adults in low- and middle-income countries: prevalence, awareness and control. International Journal of Epidemiology 43: 116–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Manne-Goehler J, Geldsetzer P, Agoudavi K. et al. 2019. Health system performance for people with diabetes in 28 low- and middle-income countries: a cross-sectional study of nationally representative surveys. PLOS Medicine 16: e1002751. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Ministry of Health, Kingdom of Cambodia, Ministry of Health. 2012. National Strategic Plan for the Prevention and Control of Non-Communicable Diseases 2013-2020. Kingdom of Cambodia.
  25. Ministry of Health Kingdom of Cambodia, Department of Preventive Medicine. 2019. National Standard Operating Procedure for Diabetes and Hypertension Management in Primary Care 2019.
  26. Monami M, Vitale V, Lamanna C. et al. 2013. HbA1c levels and all-cause mortality in type 2 diabetic patients: epidemiological evidence of the need for personalised therapeutic targets. Nutrition, Metabolism and Cardiovascular Diseases 23: 300–6. [DOI] [PubMed] [Google Scholar]
  27. Munshi S, Christofides NJ, Eyles J.. 2019. Sub-national perspectives on the implementation of a national community health worker programme in Gauteng Province, South Africa. BMJ Global Health 4: e001564. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Murakami H, Phommasack B, Oula R, Sinxomphou S.. 2001. Revolving drug funds at front-line health facilities in Vientiane, Lao PDR. Health Policy and Planning 16: 98–106. [DOI] [PubMed] [Google Scholar]
  29. National Institute of Statistics. 2019. General Population Census of the Kingdom of Cambodia 2019. Ministry of Planning, Kingdom of Cambodia, National Institute of Statistics. https://www.nis.gov.kh/nis/Census2019/Provisional%20Population%20Census%202019_English_FINAL.pdf. Last accessed 07 December 2020. [Google Scholar]
  30. Nazar CMJ, Bojerenu MM, Safdar M, Marwat J.. 2015. Effectiveness of diabetes education and awareness of diabetes mellitus in combating diabetes in the United Kigdom; a literature review. Journal of Nephropharmacology 5: 110–5. [PMC free article] [PubMed] [Google Scholar]
  31. Ndou T, van Zyl G, Hlahane S, Goudge J.. 2013. A rapid assessment of a community health worker pilot programme to improve the management of hypertension and diabetes in Emfuleni sub-district of Gauteng Province, South Africa. Global Health Action 6: 19228. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Nguyen QN, Pham ST, Nguyen VL. et al. 2011. Implementing a hypertension management programme in a rural area: local approaches and experiences from Ba-Vi district, Vietnam. BMC Public Health 11: 325. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. NHS Digital. 2017. National Diabetes Audit, 2016-17 Report 1: Care Processes and Treatment Targets short report, NHS Digital. https://files.digital.nhs.uk/publication/g/3/national_diabetes_audit_2016-17_short_report__care_processes_and_treatment_targets.pdf Last accessed 07 December 2020.
  34. NHS Digital. 2020. QOF 2018/19 Results. Quality and Outcomes Framework. NHS Digital. https://qof.digital.nhs.uk/. Last accessed 07 December 2020
  35. NICE National Institute for Health and Care Excellence. 2019. Hypertension in Adults: Diagnosis and Management. NICE guideline NG136. https://www.nice.org.uk/guidance/ng136/evidence/d-targets-pdf-6896748209, Last accessed 07 December 2020.
  36. Rachlis B, Naanyu V, Wachira J. et al. 2016. ‘Identifying common barriers and facilitators to linkage and retention in chronic disease care in western Kenya’. BMC Public Health 16: 741. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Raguenaud M-E, Isaakidis P, Reid T. et al. 2009. ‘Treating 4,000 diabetic patients in Cambodia, a high-prevalence but resource-limited setting: a 5-year study’. BMC Medicine 7: 33. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Schneider H, Lehmann U.. 2016. From community health workers to community health systems: time to widen the horizon? Health Systems & Reform 2: 112–8. [DOI] [PubMed] [Google Scholar]
  39. Sophal Oum PRP, Eng M, Khuon V, Mey et al. 2010. Prevalence of Non-Communicable Disease Risk Factors in Cambodia. University of Health Sciences, Preventive Medicine Department of the Ministry of Health, Kingdom of Cambodia. https://www.who.int/ncds/surveillance/steps/2010_STEPS_Report_Cambodia.pdf, Last accessed 07 December 2020. [Google Scholar]
  40. Tang TS, Funnell M, Sinco B. et al. 2014. Comparative effectiveness of peer leaders and community health workers in diabetes self-management support: results of a randomized controlled trial. Diabetes Care 37: 1525–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Taniguchi D, LoGerfo J, van Pelt M. et al. 2017. Evaluation of a multi-faceted diabetes care program including community-based peer educators in Takeo province, Cambodia, 2007-2013. PLoS One 12: e0181582. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. The Commonwealth Fund. 2016. The U.S Health Care System. International Health Care System Profiles: The Commonwealth Fund. https://international.commonwealthfund.org/countries/united_states/, Last accessed 07 December 2020.  
  43. The Lancet. 2014. Grand convergence: a future sustainable development goal? The Lancet 383: 187. [DOI] [PubMed] [Google Scholar]
  44. Vaghela P, Ashworth M, Schofield P, Gulliford MC.. 2009. Population intermediate outcomes of diabetes under pay-for-performance incentives in England From 2004 to 2008. Diabetes Care 32: 427–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. van Olmen J, Eggermont N, van Pelt M. et al. 2016. Patient-centred innovation to ensure access to diabetes care in Cambodia: the case of MoPoTsyo. Journal of Pharmaceutical Policy and Practice 9: 1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. WHO. 2016. Cambodia. World Health Organization. https://www.who.int/diabetes/country-profiles/khm_en.pdf, Last accessed 07 December 2020. [Google Scholar]
  47. WHO. 2018. Noncommunicable Diseases Country Profiles 2018. Geneva: World Health Organization, 223. [Google Scholar]
  48. WHO expert consultation. 2004. Appropriate body-mass index for Asian populations and its implications for policy and intervention strategies. The Lancet 363: 157–63. [DOI] [PubMed] [Google Scholar]
  49. World Bank. 2020. Cambodia Home, Overview. The World Bank In Cambodia. Worldbank.org. https://www.worldbank.org/en/country/cambodia/overview, Last accessed 07 December 2020. [Google Scholar]
  50. Wu J, Kraja AT, Oberman A. et al. 2005. A summary of the effects of antihypertensive medications on measured blood pressure. American Journal of Hypertension 18: 935–42. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

czaa176_Supp
Click here for additional data file. (636.2KB, zip)

Articles from Health Policy and Planning are provided here courtesy of Oxford University Press

RESOURCES