Sampling Procedure
This will be a prospective cohort study with a pre-post design. It will be divided into two phases:
(i) Pilot phase
This will involve recruitment of patients on Afyapap and each patient followed for four months to
assess the acceptability of and feasibility of using the service module and home based management
of hypertension or diabetes. Within the cohort, we aim to test patient retention in care as primary
outcome. However, secondary outcomes include the optimal frequency of home blood-pressure
measurements and uptake of the behavioral intervention and patient satisfaction with the use of the
applications. This pilot group will receive BP devices plus behavior intervention and measure their BP
daily (to be recruited in October 2018) and followed for four months. Depending on the lessons learnt
on the feasibility of scaling up, this group could be followed for one year. The lessons learnt from the
pilotwill provide insights into the scale up phase.
The patients will be screened for eligibility (the inclusion and exclusion criteria is outlined above). The
reasons that patients are not eligible will be recorded. If patients are eligible, they will be asked to
participate in the study. If patients are eligible and unwilling to participate in the study, they will be
asked to fill out a short quantitative questionnaire describing their reasons for not participating, they
will however not be forced to fill in the questionnaire. In the pilot phase, 500 eligible patients with
hypertension or diabetes will be enrolled consecutively on to AFya pap and will be asked to monitor
their BP daily or fasting blood sugar weekly if they have diabetes. This arm also contains a behavioral
intervention in the form of reminders to measure their BP prior to each prescribed measurement and
is treated as a pilot arm to examine the feasibility of the study procedures in the first 4 months.
11
We additionally propose to pilot a behavioural intervention that entails using a mini-tracker by
patients enrolled on Afyapap. The mini-tracker is a card including a small calendar, instructions on
how to enter measurements into Afya Pap in the USSD platform, and a goal chart where patients can
indicate their BP or blood glucose goal for the end of the month. The calendar will display one month’s
worth of days and times for which patients can indicate their measurement results after each reading.
Every month new cards will be issued through patient group leaders while cards of their previous
months are returned. Patients will also receive various reminders to measure, based on their chosen
day and time of measuring during the onboarding phase. To determine the most effective source of
messages on measurement adherence, we will provide reminders coming from their preferred clinic,
their patient group leader, their chosen physician, or a family member of their choice. These more
personalized messages serve to increase the patient’s motivation to continue monitoring their blood
pressure and/or blood sugar levels and to engage with the Afya pap platform.
Patients will also be assigned a patient group based on their geographic location. These groups exist
to provide support to each patient through group discussions of chronic disease management,
practice and information on measuring their BP or blood sugar, and social support through other
patient members dealing with either HTN or Diabetes. We anticipate including approximately 1,000
Afya Pap users who are not receiving the SMS intervention and each will be followed for 6 months to
test these iterations of the tracker. Based on previous research with providing calendars to help new
or expectant mothers save for health insurance, we postulate that the tracker will provide a similar
function for Afya Pap users and can be accomplished at a reduced cost compared to providing full
annual calendars.
(ii) Scale up phase
After the learning from the pilot phase of 4 months, the subsequent enrolments will be randomized
to two armsin the ratio of 1:1. Thepatients will be randomly assigned to 2 groups defined by frequency
of measuring BP or blood sugar and the presence or absence of behavioral interventions. The study
design is outlined in figures 2 and 3. In arm A, patients with hypertension or diabetes will be asked to
monitor their BP or blood sugar weekly using a BP device or glucometer provided to them by the study
team and will enter the measurements on afyapap. Patients with diabetes using smartphones will be
provided with glucometers that are connected by a device that automatically relays blood sugar
readings to the smartphone. Those with non-smartphones (feature phones) will use a USSD code to
activate Afyapap and enter blood sugar mearurements manually. Arm A has behavioral intervention
in the form of weekly SMS remiders to increase adherence to BP or blood glucose measurements. In
arm B devices will also be provided for measurement of BP and blood sugar at home, however the
particpants will choose when to take the measurements. There will be no SMS reminders to the
participants.
12
Study population
The study population will be lower income and middle income adult patients with HTN or diabetes or
both, recruited from selected health facilities: divided into the two arms of the intervention. The
patients will be selected based on the inclusion and exclusion criteria outlined before.
Description of the steps in enrolment and follow up.
Following randomization to one of the two study arms, day 0 of the study starts with a baseline study
assessment and onboarding of the patient on M-TIBA and in the self-management application.
Healthcare providers or a study nurse will assist patients with installing the application on their smart
phone, and instruct them in its use. During the patient set-up in the application, patients will complete
a personal profile that allows for tailoring of the behavioral intervention.
Patients will then receive training in the use of a home based BP or blood sugar monitoring device.
Patients will be instructed to perform weekly BP measurements (or more, if national guidelines
prescribe more frequent monitoring). Once at home and using the BP monitor, patients will enter their
measurement results in the application. We will set up protocols for extreme BP and blood sugar
values based on best practice and national guidelines.
Study setting
The study will take place in health care clinics servicing the low and middle- income populations in
Nairobi, Kiambu and Vihiga Counties in Kenya. The choice of this population is based the challenges
of affording care and inability to enrol in private health insurance. But the patient population included
will be those patients who have some means to afford HTN or diabetes treatment or who are not
insured whose access can be enhanced through the financing model and home based management.
There will be selected health facilities from Nairobi, Kiambu and Vihiga Counties in Kenya where
patients will be enrolled from. The health facilities are; Langata hospital, Access Afya kwa Njenga clinic,
Access Afya Taasia Clinic, Limuru Nursing Home, Kalimoni Mission Hospital, Coptic Mission Hospital,
Kima Mission Hospital, Jumuia Friends Kamusinga Hospital, Mungoma Hospital and St. Marys hospital.
In the event that we will not be able to get the required sample in the nine health facilities, we will
recruit more facilities. These facilities are purposely selected because they have patient populations
of interest to the study, as informed by the experiences of PHARMACCESS. Currently, PharmAcces is
rolling out the service model for management and financing hypeternsion in three counties (Nairobi,
Kiambu and Vihiga) targeting approximately 4000 patients in total. Permission will also be sought from
relevant authorities including local authorities, county and sub-county to allow entry into the selected
health facilities in Nairobi, Kiambu and Vihiga counties. The evaluation will be conducted in a
representative sample as shown below.
Sample size calculation and recruitment
Sample size for patients with hypertension
From Van De Vijver et al. 10 ’s study, we estimate that among people who are in treatment, 50% were
able to achieve glycemic control. We assume that the current intervention will increase the BP control
by 20% in one year of follow-up. Thus, we posit a 20% meaningful difference, and 80% power with 5%
level of significance (one-sided test) to detect that difference. Since our study design is a pre-post
design, we estimate the sample size following Connor’s study 64. The conservative necessary sample
13
size is 75 hypertensive patients per study arm. Furthermore, we assume a dropout rate of 20%,
implying that the final sample size is 94 hypertensive patients to be recruited at the pre-intervention
phase (baseline), and 94 hypertensive patients to be surveyed at the post-intervention phase (midline
and end line) in each of the 3 selected counties. Hence for baseline, midline and end line, the total
sample size required for patients with HTN is 282 (94×3) respondents at each time point. When it is
not feasible to recruit the minimum number from a particular county, we will oversample in the other
counties to achieve the required total sample.
Figure 2 represents the survey design and distribution of patients by study arm in each county
Figure 2 represents the survey design and distribution of hypertensive patients by study arm. Group
A will receive the BP device plus behavior intervention and measure their BP daily (to be recruited
from January 2019- scale up phase) and each patient followed for one year.
Group B will receive the BP device plus behavior intervention and measure their BP weekly (to be
recruited from January 2019- scale up phase) and each patient followed for one year. Group C will
receive the BP device no behavior intervention and measure their BP at their discretion (to be
recruited from January 2019- scale up phase) and each patient followed for one year.
Sample size and recruitment of patients with diabetes
In reference to a study by Otieno C et al.,22 we estimate that among patients with diabetes who are in
treatment, 30% were able to achieve glycemic control. We assume that the current intervention will
increase the BP control by 30% in one year of follow-up based on findings from a randomized control
trial in 3 Low and Middle Income countries including Democratic Republic of Congo65. Thus, we posit
a 30% meaningful difference, and 80% power with 5% level of significance (one-sided test) to detect
that difference. Since our study design is a pre-post design, we estimate the sample size of 49 patients
with diabetes per study arm. Furthermore, we assume a dropout rate of 20%, implying at baseline,
midline and end line survey, 62 patients with diabetes wil be enrolled in each arm, a total 124 (62X2)
in each of the 3 counties at baseline, midline and endline surveys respectively. In the event that we
can not get the minimum number of patients in a particular county, we will oversample in the other
counties to achieve the total number required per survey round.
Figure 3 represents the survey design and distribution of diabetic patients by study arm. Group A will
receive the glucometer device plus behavior intervention and measure their blood sugar weekly (to
be recruited from January 2019- scale up phase) and each patient followed for one year. Group B will
receive the glucometer device no behavior intervention and measure their blood sugar at their
discretion (to be recruited from January 2019-scale up phase) and each patient followed for one year.
