High frequency of monitoring of fertility trends in Kenya using mobile data
collection system
Mary Thiongo (1), Peter Gichangi (1), Bartilol Kigen(2), Joseph Kiseli(1), Scott Radloff(3)
1) International Centre for Reproductive Health Kenya
2) Ministry of Health Kenya
3) John Hopkins Bloomberg School of Public Health
Abstract: Demographic and Health Survey (DHS)
is responsible for collecting and reporting
nationally representative data on health and population in the developing countries. The data
are reported in five-year intervals, a gap that restricts the ability of stakeholders’ to
plan, make
timely adjustments to policies and programs
. Performance Monitoring Accountability
(PMA2020) Kenya report use of mobile data collection system to conduct a nationally
representative survey that are intended
to fill the gaps by providing current and
reliable
information for family planning, water and sanitation at household and health facility levels.
Open data kit (ODK) free open-source software was used to author, field and manage data
collection. Smart phone equipped with camera and augment GPs deployed as a data collection
instrument. Inbuilt skip logic, validation checks ensured that data uploaded had minimal or no
errors. Survey data were uploaded real time to a secured cloud server using mobile data
connections, data quality and completeness was monitored close to real time. Reduced turnaround time from data collection to analysis allows rapid feedback to program and policy
makers.
Keywords: Demographic Health Survey, mobile phone, Rapid, feedback
1. Introduction
A deep technical literature shows that high fertility rates are inversely related to the
incidence of extreme poverty and per capita economic growth, gender inequality,
maternal mortality and poor child health, environmental degradation, and other
dimensions of sustainable development [1]. Programme of Action adopted by the
International Conference on Population and Development (ICPD) highlights also that
reducing population growth through voluntary transition to lower fertility levels is one
component of achieving sustainable development [2].
Timely monitoring of these fertility indicators is therefore critical in addressing these
factors. However current routine health systems in place are unable to monitor regularly
while large scale surveys are costly and infrequent; Demographic and health survey
(DHS) which is known to be the gold standard for collecting such data in the developing
county data are reported in five year intervals; a gap that restricts the ability of
stakeholders to plan, make timely adjustments to policies and programs based on these
data. We all know Data are the lifeblood of decision-making and the raw material for
accountability. Without high-quality data providing the right information on the right
things at the right time; designing, monitoring and evaluating effective policies becomes
almost impossible [3]. Existing gaps can only be overcome through innovative data
collection methods and use that can eventually contribute to saving money and creating
economic social and environmental value. Paper based data collection has been the
standard method for years with its many shortcomings. I.e. errors are frequent, storage
and transportation costs are unaffordable, and the costs involved in data entry are also
pretty high. The electronic methods of data collection have been developed in order to
merge data collection and data entry step and to also increase the quality of the data
collected. With advancement of the mobile technology in Africa there is an evolution of
data collection for health systems and other sectors. The technology is increasingly being
used instead of paper and pencil methods of data collection due to the fact that mobile
phone networks have penetrated to most areas in Kenya and Africa as a continent.
Performance Monitoring and Accountability (PMA2020) is being implemented in 9
countries so far by using mobile phone data collection system to contribute to global
monitoring and evaluation by conducting national representative survey that are intended
to fill gaps in the availability of current and reliable information on population dynamics;
family planning reproductive health services delivery; water sanitation and hygiene [4].
The focus of PMA2020/Kenya is to monitor Kenya’s contribution to the Family Planning
2020(FP2020) goal of adding 120 million new contraceptive users globally by 2020 [4].
This paper will report a threefold: (i) Need for frequent monitoring? (ii) Highlights on
advancement of mobile phone technology. (iii) Discussion on lessons lessons learnt.
2. Methodology
PMA2020/Kenya is a five-year project that supports rapid turnaround surveys in 9
(Bungoma, Kericho, Kiambu, Kilifi, Kitui, Nandi, Nairobi, Nyamira and Siaya) out of 47
counties in the country. The project in partnership with the Kenya government and
International Centre for Reproductive Health Kenya utilizes a network of resident
enumerators (RE) who lives in or near probability sample of selected enumeration areas
to collect nationally representative surveys from households and service delivery points.
The survey provides semi-annual updates to key FP2020 indicators of contraceptive
need, use quality, choice, and access as well as questions on WASH in household and
health facilities.
Sample design: PMA2020/Kenya survey used a two-stage cluster design with urbanrural and country as strata. A sample of 120 enumeration areas (EAs) was drawn by the
Kenya National Bureau of Statistics; from its master sampling frame. In each EA
households and service delivery points were listed and mapped, with 42 households
randomly selected. Households were surveyed and occupants enumerated. All eligible
females aged 15 to 49 were contacted and consented for interviews. The final sample
included 4810 households from a target of 5040, 4396 females and 270 health facilities
[].
Training
The project recruited a team of 9 supervisors that were expected to take charge of their
respective counties as fieldwork supervisors. They were trained for 2weeks as Training
for trainers (TOT) and they were to cascade similar two week training to 120 resident
enumerators. Several teaching methods were used which included power point
presentation, group work discussions, plenary sessions, and paired interviews and mock
field exercises. We also conducted fieldwork to a pre identified mock EA where the
enumerators practiced boundary identification, mapping, listing and also tested the tools
through mock interviews. The responses from these interviews were recorded in the
smart phone and submitted to test server using mobile data connection as would be the
case during real data collection. In general the training included comprehensive training
on using mobile technology, using ODK application in data collection, knowledge on
survey tool content and more so training on contraceptive methods.
Data Processing
Open data Kit (ODK) free open-source software was used to author, field and manage
data collection. Smart phone equipped with a camera and augmented GPS deployed as a
data collection instrument to the project staff. Survey data were uploaded real time to a
secured cloud server using mobile data connection. The instantaneous aggregation of
data also allowed for real-time monitoring of data collection progress and concurrent
data processing while data collection was still active in the field and course corrections
could be made [4]. Data manager routinely monitored the incoming data and notified
field staff of any potential errors, missing data or problems.
3. Need for frequent monitoring
If data was to be useful and support good decision-making, it has to be ready at the time
when decisions are being made or where the opportunity for influencing the outcomes is
there. Trade-offs between timelines and other quality dimensions depend on the purpose
to which data is being put. In 2014 Kenya registered a striking increase in its
contraceptive prevalence rate (CPR) for married women which rose from 39% in 2009 to
55% representing a 16 % increase; additionally the survey found marked increase in use
of long acting methods with the proportion of females who are using implants increasing
from 1.9% to 20.1%. [5, 4] The survey findings also realized the Total Fertility Rate
(TFR) the decline slowed in the 1990s, but the decrease in TFR from 4.6 in the 2008-09
KDHS to the current 3.5 may indicate that Kenya’s fertility is returning to the decline
observed from the mid-1970s through 1990s. The TFR of 3.5 for the whole country is the
lowest ever recorded. The unmet need for family planning dropped from 26% 2008 to
21% in 2014, with 12% of the women wanting to delay their next births and 9% wanting
no more children. The unmet need is highest for women in poorest wealth quintiles. The
survey also found a 17% point increase in demand satisfied by modern methods, from
55% in 2008 to 72% in 2014. Based on these results there seems to be a wide gap in
the measures and frequent monitoring should be considered to aid project planning
monitoring and management.
4. Benefits of mobile phone advancement
Data collection and Entry step merged: The use of mobile technology as a data
collection instrument represented itself as an improvement we diverted from the
traditional way where data entry and management is done after data collection. Many
surveys take time to enter and clean the data. The data collection was translated to one
step and data cleaning was running concurrently. The data manager was monitoring the
data submissions real time and raising any discrepancies realized with the supervisors in
the field. This resulted to reduction of time used for processing the data as well as
enhancing the quality of data by preventing future errors.
Ability to collect new data formats: The multimedia capabilities of the smart phone
presented a new way of collecting new data formats. GPRS for recording GPS coordinates
and Camera for taking the photo. Integration of GPS coordinates directly into the form
and database is an aspect that cannot be replicated using paper forms without requiring
an extra gadget and more time. With the mobile technology the non-text data was
integrated with the text data in real time. The contraceptive methods were displayed as
media file where the enumerator could show the woman the media while conducting an
interview.
Data entry forms: Interviewers using paper may enter inconsistent data. This may go
unnoticed until the data-cleaning phase. To determine also if the survey skip patterns are
met will involve manual visual checks by the supervisors, which are time consuming.
Data analyst will also be expected to write code that will review if the skip patterns were
obeyed. Many are the times when data is reviewed it come out that men were recorded
to be pregnant and even imaginable outliers realized. With mobile technology the forms
were programmed with inbuilt skip logic, validation checks and inbuilt response
constraints to prevent data entry errors. The inbuilt set logical question flow restricted
the enumerators to only questions which were applicable to the rules; thereby making
non-applicable questions hidden from the enumerator. Set validation checks for
responses entered also provoke or prompt enumerator if responses did not match prefilled previous data. The enumerators were also provided with hits to guide them on how
to answer questions when need presented.
Data submissions, Data backup and security
Traditionally data are recorded on paper questionnaires. These forms are then
transported to one or more central locations for further processing. Various procedures
will be employed to ensure that forms are not lost but this is not always the case. With
the mobile technology the survey data was uploaded on real time in most of the cases.
Survey data was uploaded real time to a secured cloud server using mobile data
connections, allowing generation of data monitoring progress reports in real time. This
data in the cloud could be accessed anywhere all over the world enabling progress
monitoring. To restrict data loss multiple protocols were put in place, the data was stored
in the phone internal storage and an automated cloud server Google drive. The mobile
phone and the cloud server had restricted access.
Facilitated communication, management & payment
Supervision of fieldwork is very critical to the success of the project. Time and resources
for training are invariably limited and even with the best possible training some lessons
learned would be forgotten as fieldwork proceeds. Fieldwork may also raise questions or
involve difficulties that were not anticipated in the training. Supervision of field workers
as fieldwork proceeds is necessary to ensure that they are doing the best possible job of
obtaining accurate information. During our survey the supervisor was able to call the Res
and resolve most of the issues over phone. This resulted in reducing travelling cost and
time. This was only possible since each RE had a mobile phone where they were to be
reached on. On completion of the exercise, the RE were paid via Mpesa, a mobile money
transfer technology saving them from either the rigorous bank account opening process
or having to make long distances to receive the payment.
Quality Control Measures
” No study is better than the quality of its data” this asserts the magnitude of
collecting, processing and storing data which is complete, accurate, consistent and
auditable(Friedman et al (1998). Paper and pencil surveys mostly depends on the
supervisory work done by the supervisors trusting and hoping that surveys take place
within the selected households and are collected within the required time allocated to
conduct an interview. While smart phone does not eliminate the supervisory step the
advancement functionalities contribute to better ways of monitoring data collection.
Time-stamping facilitating in monitoring the time take to complete a survey which is then
monitored to see the average time the enumerators are taking to conduct an interview
progressively and those with unexpected patterns are identified and informed. By design
the tracking starts the moment the enumerator registers an interview in the phone and
when the survey are submitted the data manager flags out the funny cases and its
communicated back to the supervisor who cascades the information to the team to
improve. This is identified in real time.
The GPS coordinates also enhanced the quality of survey data collected. This was
crosschecked to monitor the location of the enumerators at the time of interview and
ensured that the correct households were interviewed. Since the coordinates were
recorded at listing and mapping level then it was easy to compare and monitor the
consistency of data recorded at the 2 levels. Since we collected the coordinates in the
service delivery points also the recording were used for further analysis to calculate the
distance the women travels to access the services and this can also further be evaluated
to see the cost and time implications and associations in the circumstances that the
women travels and they find a method out of stock.
5. Discussions
We have highlighted several benefits that mobile data collection brings to collecting
large-scale survey data. However, we must agree that there are challenges that come
with it and it will also be important to discuss.
Survey trainings: The trainings take longer by the fact that it had to comprehensively
cover the survey content, smart phone functionality and survey software. In our case
more than 60% of the Resident enumerators did not have an experience with the
Smartphone prior to the training. These necessitated the trainers to take more time to
train on smartphone functionalities. Women aged <35 years were able to manipulate the
phone faster than their counterparts who were aged >35 years. Age of the enumerator
can be considered during recruitment. To ensure that the RE is comfortable with using
the technology we ensured that our training quizzes and mock surveys were using the
survey software and were done using the phone. This facilitated on increasing the mobile
and software use skills uptake. We also had to train the Res on how to handle the smart
phone and understanding the gravity of losing it and what it meant to the survey
implementation. The supervisors also required extra training to troubleshoot Smart
phone issues in the field.
Quality control: We mentioned that the advanced capabilities of smart phone, such as
GPS recording and time-stamping were used to enhance quality in our survey, this only
cannot be used to ensure high quality. It fully depends on several other factors, with
time-stamping the date and time setting on the phone has to be correct and monitored
as survey progresses. The GPs readings especially during adverse weather conditions
may take longer to attain the accuracy level, which was noted as one of our challenges.
We addressed this by installing ActiveGPS app in the smart phone.
Network coverage: Though there is increased mobile phone penetration and coverage
in Kenya, some areas are still encountering network issues. The survey data was to be
submitted after completion of the interview but in some of the areas it was not possible.
The RE had to move outside the enumeration area to look for a hotspot that brought
cost implications. In some circumstances forms were lost in transit as a result of poor
connectivity and they had to be retrieved from the internal storage of the phone or our
extra external storage. We also encountered duplicate submissions due to network
coverage in the circumstances when the RE clicked the button more than once if the
network was slow to submit. This meant taking more time to clean the data.
Paper not completely eliminated: Though the survey was mobile phone assisted we
still had to carry consents forms and progress sheets in paper form to the field. A
consent form is a mandatory requirement by the IRB in Kenya requiring consenting on
paper. We were required keep one copy and the respondent is left with the other copy.
Consenting on phone is possible but may require a lot of community sensitization
because even conducting the mobile phone assisted survey was associated in some
circumstances associated with land grabbing, devil worshipping especially in the rural
areas and the enumerators had to really explain to the particular respondents. The
enumerators to document and track their fieldwork activities used the progress sheet,
which was noted as a very important document for monitoring fieldwork activities.
Other challenges remains regardless of the mode of data collection; the survey areas
presented various geographical terrains, different weather conditions were experienced
and enumerators had to be accompanied by police for their security.
Conclusions
Innovation in mobile technology continues to advance at rapid speed and with it we can
improve data collection for survey and Health Research and Development. The model
can also be integrated into national Monitoring & Evaluation systems by offering a low
cost, rapid turnaround survey plat form to be used for various data needs. The rapid and
frequent information can be used to advocate for program improvements and inform
policy change in all sectors.
We also demonstrated that mobile phone technology smart features have revolutionized
survey data collection. GPS coordinates can be used for quality control (location and
verification) bringing a possibility of mapping the geographical spread of the interviews
and showing the location of the enumerators at the time of data collection. The GPS
coordinates can also be used for proximity on calculating the distance between
households and health facilities, which can be used for further analysis to estimate other
accessibility needs.
Acknowledgement
We acknowledge the Government of Kenya, the Funding agent and technical support we
got from Gates Institute John Hopkins School of Public Health, Baltimore. Not forgetting
the Fieldwork team and the survey respondents.
References
1. For a comprehensive review of the evidence linking population growth and fertility rates
to sustainable development see UN Population Division (2011). Seven Billion and
Growing: The Role of Population Policy in Achieving Sustainability. Technical Paper No.
2011/3. New York.
2. SDSN (2013).
3. Independent Expert Advisory Group on a Data Revolution for Sustainable Development
(IEAG), (2014),A World That Counts: Mobilizing The Data Revolution for Sustainable
Development
4. Performance Monitoring and Accountability 2020 (PMA2020) Project, International
Centre for
Reproductive Health Kenya (ICRHK). 2014. Detailed Indicator Report: Kenya 2014.
Baltimore,
MD: PMA2020. Bill & Melinda Gates Institute for Population and Reproductive Health,
Johns
Hopkins Bloomberg School of Public Health.
5. Kenya National Bureau of Statistics (KNBS and ICF Macro. Kenya Demographic and
Health Survey 2014: Key Indicators. Calverton,Maryland: KNBS and ICF Macro, 2015.
6. The use of mobile phones as a data collection tool: A report from a household survey in
South Africa
7. The use of personal digital assistants for data entry at the point of collection in a large
household survey in southern Tanzania
8. A comparison of smartphones to paper-based questionnaires for routine influenza
sentinel surveillance, Kenya, 2011–2012
9. The use of continuous surveys to generate and continuously report high quality timely
maternal and newborn health data at the district level in Tanzania and Uganda