India Health Action Trust (IHAT) needed a foolproof data collection method that could gather large sets of data – their challenge being that the scenarios were very complex, and each survey could take up to 28 hours. Collect wasn’t capable of some of the requirements for the survey, but our technology team at SocialCops adapted accordingly and came up with innovative solutions for the various requirements at hand. Collect now helps IHAT cover 1113 health care facilities across 150 blocks in Uttar Pradesh.
Introduction to Project
India Health Action Trust (IHAT) is a registered secular trust working on public health issues focusing on HIV and AIDS in different states of India. In Uttar Pradesh, IHAT were contacted by the Bill & Melinda Gates Foundation to set up a Technical Support Unit (TSU). The TSU’s goal is to support the government to increase the efficiency, effectiveness and equity of the delivery of key RMNCH+A (Reproductive, Maternal, Newborn, Child, Adolescent Health) services to improve its outcomes in UP.
To meet this goal, one of the primary objectives of the TSU is to improve the quality of primary care services at health facilities across the state. This involves continuous assessment of the current state of facilities across the state and the competency of the corresponding staff. For this, IHAT decided to monitor these facilities over 4 cycles with an approximate time period of about 6 months per cycle.
To be able to implement the right kind of training modules and infrastructure improvements, it was essential for IHAT to get an exhaustive assessment of the current situation at the facility level. The questionnaire they ended up designing to ensure for exhaustive assessment was extremely complex and time consuming – with surveys taking up to 28 hours each. Continuous assessment also meant that data needed to be layered on top of the earlier collected data. Paper based data collection for such a questionnaire would make errors inevitable, and thus cost them a lot of time and effort for data that wouldn’t be of the highest quality.
To execute such a questionnaire using an Android device was definitely preferable – however, there were a few major challenges even on the technological front for us to conquer.
Challenges We Faced on the Technology Front and How We Adapted
Considering how exhaustive IHAT needed the data to be, there were bound to be a lot of challenges in surveying.
Challenge 1: While some of these cases might be referred to another facility midway through the process, some of them would be paused midway to take up another – either way, the data needed to be captured.
Our solution: We introduced a separate referral system wherein another survey could be linked to the current survey and switched to – all while ensuring that the data for the primary questionnaire isn’t lost. We also introduced the concept of saving drafts of responses – so that a new response could be taken midway between another while ensuring that the data of the original isn’t lost.
Challenge 2: Each patient could be monitored by different health workers at different points of time, and it was imperative to capture the corresponding action to the corresponding skill worker. This meant that the platform had to register not just the answer for each question/parameter, it also needed to be capable of simultaneously registering the health worker in charge of the patient.
Our solution: We added another element to our monitoring feature, which was an ‘Observation survey’ that could monitor certain entities over each question instead of questionnaires.
Challenge 3: Certain actions during the assessment of a patient in the postpartum period requires for them to be checked on a regular timed basis. This required the platform to be able to register data at pre-set time intervals.
Our solution: We created interval questions that could be created based on flexible pre-set time intervals. We didn’t however incorporate an alarm feature keeping in mind the battery longevity of the device.
Challenge 4: Since it involved monitoring a large number of entities (facilities, workers etc.), we realised our monitoring feature would be put to the task, and could be a lot less time consuming.
Our solution: We changed our monitoring from a funnel based system (state→block->facility->worker) to a filtering based system (city&block&facility&worker) which would reduce the total number of queries put to the system, and hence decrease the time required drastically.
Quality data collection helped us to drive our survey with precision, this was possible with the help of SocialCops.
Thirumalai, India Health Action Trust