NEW UPDATE: Completed trials may now upload and register supplementary documents (e.g. null results reports, populated pre-analysis plans, or post-trial results reports) in the Post Trial section under Reports, Papers, & Other Materials.
The Nueva Ecija cardiovascular risk experiment: An evaluation of the impact of risk information and screening on primary prevention of cardiovascular disease
Initial registration date
April 09, 2018
April 11, 2018 2:48 AM EDT
Erasmus University Rotterdam
Other Primary Investigator(s)
University of the Philippines Diliman
Additional Trial Information
This study seeks to assess how beliefs about health risks, specifically the risk of cardiovascular disease (CVD), affect health lifestyles and the demand for preventive care in a low-income setting. It also aims to establish the effectiveness of the Package of Essential Noncommunicable Disease Interventions in the Philippines (PhilPEN) in delivering primary prevention of CVD.
To realize the first objective, we will measure the accuracy of beliefs about exposure to CVD risk and, subsequently, randomly provide information on personal CVD risk based on measured risk factors. This will allow assessment of the extent to which biased beliefs constrain demand for primary prevention and sustain unhealthy lifestyles. In addition, we will test whether beliefs about susceptibility to CVD are responsive to the receipt of information on personal risk, and whether health behaviors and the demand for CVD screening and medication are affected by any revision of beliefs. To meet the second objective we will randomly encourage uptake of the PhilPEN program’s risk screening by offering entry to a money prize lottery conditional on attending a health clinic where the program operates. The induced random variation in clinic attendance will be used to estimate the program's impact on exposure to risk factors, medication of hypertension, the predicted risk of CVD and awareness of this risk.
Meeting both objectives will allow us to distinguish between scenarios. One is that PhilPEN is effective in preventing CVD of patients who access the program but its impact on population health is muted because poor information on susceptibility to CVD reduces the demand for primary prevention. Another is that even if improved information is effective in raising this demand, this will have little impact on population health through PhilPEN because of deficiencies in the operation of the program in health facilities. Within the Nueva Ecija province, we will randomly sample barangays (N=304), subsequently households (n=5019) and, finally, one person aged 40-70 within each household. At the barangay level, we will randomly allocate to a treatment group receiving the lottery incentive to attend a health clinic (n=2261), another treatment group receiving information on personal CVD risk (n=497) and a control group (n=2261). A baseline survey (January-April 2018) will record data on initial health, health behavior, health knowledge, risk perceptions, risk attitudes, time preferences, health care utilization and expenditure and socioeconomic characteristics, and deliver the treatments. A follow-up survey 9-12 months later will record outcomes.
Capuno, Joseph and Owen O'Donnell. 2018. "The Nueva Ecija cardiovascular risk experiment: An evaluation of the impact of risk information and screening on primary prevention of cardiovascular disease." AEA RCT Registry. April 11.
Two interventions will be randomly allocated: 1) the provision of information on CVD risk, and 2) the offer of a lottery ticket conditional on going for a checkup at a health clinic.
1) Information on CVD risk
Respondents will receive information on the predicted probability of having a heart attack or stroke within 10 years. The predictions will be obtained from the Globorisk tool (www.globorisk.org). All information will be provided within a risk perceptions module of the baseline survey. Only this module will differ across the two treatment groups (information and lottery) and the control group. Information obtained from earlier modules will be retrieved automatically and used to make predictions of CVD risk consistent with the risk factor profile of the respondent.
Three types of information on CVD risks will be provided: a CVD base rate, a personalized CVD risk and an optimal CVD risk. The CVD base rate will be predicted from the respondent’s age and sex only. This information will be provided immediately after the respondents are asked to report the chance of someone of their age and gender having a heart attack or a stroke within ten years, and immediately before being asked to report the chance that they will have a heart attack or stroke within that period. The respondents will be presented with the information both as a chance of X in 100 and as the number of people (X) out of 100 the same age and sex as the respondent that doctors would expect to have a heart attack or stroke within ten years. In addition, a visual aid (sliding ruler) will be used to deliver the information.
After reporting their own chance of having a heart attack or stroke within ten years, the respondents in the treatment group will be told the risk for someone with the same age, sex, smoking status, body mass index (BMI) and blood pressure as them. Each respondent will not be told that this is their own chance of having a heart attack or stroke. Rather, if the respondent is female, she will be told that out of 100 women her age, with the same weight and blood pressure as her, and who (don’t) smoke like her, doctors would expect X to have a heart attack or stroke within 10 years. Analogous information will be provided to male respondents. After being given this information, the respondent is asked again to report their 10-year CVD risk. This is interpreted as the posterior belief (after the provision of personalized information). Finally, a treatment group respondent will receive information on what the 10-year CVD risk would be for someone of the same age and gender who did not smoke, and had normal blood pressure and BMI. This will be presented as the optimal risk that could be achieved through primary prevention. If the respondent’s risk is above the optimal, then they will be told the reason(s) for this - smoking, above a healthy weight or blood pressure reading is above normal. If the respondent’s risk is not above the optimal, then they will be congratulated and advised how to keep the risk low.
2) Lottery incentive to attend health clinic
Respondents will be offered a ticket for a lottery with a money prize on condition that they visit a specific public health clinic for a checkup. There will be one prize per barangay giving each respondent a one in ten chance of winning P5000 (US$100). The prize is equivalent to approximately 14 days earnings at the regional minimum wage. There is no supply-side component to the intervention. Respondents will simply be told that they can enter a lottery if they go to the specified clinic for a checkup. The health facilities will be told to conduct an assessment deemed appropriate for any particular patient that requests to be issued with a lottery ticket. No instructions will be given that the facilities should follow the PhilPEN protocol. We will evaluate whether they do implement the protocol for patients who qualify (by age if nothing else) for full risk screening. Prior approval of the study and intervention will be obtained from Mayors and Medical Health Officers. The study team will identify a point person in each health clinic and liaise with them to identify respondents who appear at each clinic. Respondents will be given a voucher at the baseline interview informing them of their entitlement to enter the lottery on attending a clinic. At the clinic, they will exchange the voucher for a lottery ticket. They will be instructed to send an SMS to a number accessed by the study team after obtaining a lottery ticket. The detailed logistics for the operation of the lottery are providing in the study protocol available that can be downloaded from this website.
Intervention Start Date
Intervention End Date
Primary Outcomes (end points)
Mean 10-year risk of CVD event (heart attack/stroke)
Primary Outcomes (explanation)
Predicted probability of having a heart attack or stroke within 10 years obtained from office version of Globorisk (www.globorisk.org) based on age, sex, systolic blood pressure, body mass index (BMI) and smoking status recorded in end-point survey. Group mean of predictions will be calculated.
Secondary Outcomes (end points)
1) Proportion with 10-year CVD risk ≥ 10%.
Risk Factors used in calculation of predicted 10-year CVD risk
2) Mean systolic blood pressure.
3) Proportion with elevated blood pressure (systolic ≥140) .
4) Mean BMI.
5) Proportion overweight/obese (BMI>25).
6) Proportion currently smoking.
7) Mean waist circumference
8) Proportion with waist circumference ≥ 90cm (men) / 80cm (women). Diagnosis and medication of hypertension
10) Proportion with undiagnosed hypertension 10) Proportion taking antihypertensive medication in the last 2 weeks.
11) Alcohol consumption.
12) Diet (intake of fruit, vegetables and salty foods).
13) Exercise 14) Knowledge of CVD and diabetes risk factors.
Outcomes specific to lottery intervention
15) The proportion of smokers/ex-smokers who have been advised by a doctor or health worker to quit smoking.
16) The proportion of smokers/ex-smokers who have received counselling on smoking cessation.
17) The proportion who have been advised by a doctor or other health worker to drink less alcohol (out of all who have ever consumed alcohol).
18) The proportion who have been advised by a doctor or other health worker to eat less salty and/or fatty food.
19) The proportion who have been advised by a doctor or other health worker to eat more fruit & vegetables and/or grains & pulses.
20) The proportion who have been advised by a doctor or other health worker be more physically active.
21) The proportion of individuals overweight or obese (at baseline) who have been encouraged by a health professional to lose weight.
Outcomes specific to information intervention recorded in baseline survey
22) Mean perceived 10-year risk of heart attack or stroke for someone of same age and sex as respondent.
23) Mean perceived own 10-year risk of heart attack or stroke.
24) Mean perceived own 10-year risk of heart attack or stroke if were to adopt healthy lifestyle.
Outcomes not specific to CVD risk
25) General health measured by SF-36v.1
26) Labour market employment, hours and earnings.
27) Health care utilization and expenditures.
Secondary Outcomes (explanation)
1) Predicted risk obtained from Globorisk as for primary outcome. If power permits, will also estimate effects on proportion with CVD risk>20% and >30%.
2)-6) Predicted CVD risk is function of blood pressure, BMI and smoking. We will also estimate effects on these risk factors separately.
2) & 3) Mean of last two SBP measures on single visit. BP measured using electronic (OMRON) wrap cuff monitor.
4) & 5) Height and weight measured using standardized instruments.
7) & 8) - Globorisk predicted 10-year CVD risk is not a function of central adiposity, but this is measured as part of PhilPEN risk assessment. Weight circumference will be measured followed a standardized procedure.
10) Numerator = systolic/diastolic BP ≥ 140/90 + not diagnosed with hypertension. Denominator = all respondents.
11)-13) Measures consistent with those of WHO STEPS.
14) Knowledge of CVD and diabetes risk factors assessed using questions adapted from previously fielded instruments.
15)-21) For the lottery intervention, we will examine whether attending a health clinic for a checkup increases the probability of receiving medical advice on the adoption of healthier habits, as is prescribed by the PhilPEN protocol.
22)-24) These outcomes will be measured in the baseline survey in response to information provided during the interview.
The study is a randomized parallel experiment with two separate, non-overlapping treatment groups and one control group. Randomized assignment of treatment will be done at the level of the barangay, which is the smallest administrative unit in the Philippines roughly equivalent to an electoral ward. The 847 barangays in the Nueva Ecija province will be stratified by urban/rural classification. Within each stratum, barangays will be drawn by simple random sampling and randomly allocated to the lotttery treatment (n=137), information treatment (n=30), control (n=137) and those not surveyed (n=543).
Within a barangay, interval sampling will be used: starting from a random location, a procedure will be followed to randomly select households moving out from that point. Selected households will be screened to establish whether each includes anyone aged from 40 to 70 inclusive. If so, one such person will be randomly selected as the main survey respondent and the potential subject of the intervention in the treatment group barangays. Further screening will be undertaken to establish whether the selected person is eligible for the study, mainly on the basis of CVD history. In addition to the main respondent, we will ask the head of household, spouse or other adult to answer questions about the household in general. We will aim to have 10 persons per barangay completing both the baseline and endpoint surveys and eligible for the treatments. To achieve this, allowing for attrition and non-eligibility, we will survey 17 households per barangay.
The baseline survey (January-April 2018) will record data on initial health, health behavior, health knowledge, risk perceptions, risk attitudes, time preferences, health care utilization and expenditure and socioeconomic characteristics, and deliver the treatments. The end-point survey 9-12 months later will record outcomes.
Experimental Design Details
Random assignment of barangays to treatment/control groups done by study team in office by computer. Random selection of households within each barangay done in field by enumerators using interval sampling following predefined procedure. Random selection of person aged 40-70 within selected households done by enumerator following predefined procedure.
Was the treatment clustered?
Sample size: planned number of clusters
Sample size: planned number of observations
5019 individuals aged 40-70.
Sample size (or number of clusters) by treatment arms
137 barangays control, 137 barangays given lottery incentive to go for checkup at health clinic, 30 barangays given information on personal CVD risk.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
0.2 standard deviation units.
INSTITUTIONAL REVIEW BOARDS (IRBs)
Commission cantonale (VD) d'ethique de la recherche sur l'etre humain (CER-VD)
IRB Approval Date
IRB Approval Number
IRB Approval Date
IRB Approval Number
Post Trial Information
Is the intervention completed?
Is data collection complete?