Emerging Project Lessons
If CARE had to implement the scheme again, they woulduse a mobile phone based solution that has a biometric card reader and thermal printer. The connectivity of this solution would likely be better and the cost would be about half ($300 versus $600 for HHD). CARE may roll out mobile solutions once the HHDs are at capacity.
Development and implementation of CDSS software is complex, both from a technology and a human perspective. It thus requires adequate time and money to allow for variablessuch as:
- Bi or multi-lingual requirements between the field and the back office. In the case of CARE, the field staff work in Marathi, a Hindi keypad is used, and the back end is in English.
- Doctors (especially pediatricians), who require additional time to evaluate the treatment protocols and to reach consensus. It turned out that pediatricians were reluctant to support the piloting of a CDSS for pediatric patients, perceiving higher risks to young patients who may be more difficult to diagnose and treat with a telemedicine model. It was agreed that pediatric CARE, CDSS software would not be implemented in the first phase, until a successful pilot for adults was observed.
- Training of doctors to use a CDSS for treatment, record health data and generate prescriptions must include both the rational for any behavior change (e.g. changing prescribing patterns) and the use of the technology itself. The rule of thumb being adopted by CARE will be to prescribe no more than a 3 day drug supply per prescription; to avoid antibiotics unless warranted and then when used, for no more than 5 days per prescription. CARE plans to do regular audits of physicians to monitor their prescribing behavior to guide corrective actions.
- Adapting traditional wisdom about systems for insurance, which does not fit in the context of microinsurance packaged with basic health services delivered in a rural setting by a VHC using a HHD. CARE’s understanding of the attributes to be tracked went through a series of expansion and contraction, leading to the recasting of a number of database tables and their relationships.
- Building in sufficient flexibility so that users can adapt the software ongoing, for example to accommodate new drugs or improved treatment pathways, or to modify the insurance offering.
Identification of insured patients using biometrics is much more challenging when it is being done with a HHD with functional limitations driven by cost, user, location and technology constraints (e.g. memory and connectivity). Given limits of the HHD, the system validates identity based on one to one matching of a stored fingerprint on a biometric card to a registered identification number. It cannot identify a client based on a fingerprint alone in the way that a governments or other organizations with much more computing power can). Thus the technology requires a system to map identification number to a finger template. All 4 fingerprints (2 each on left and right hand) for each client will be stored to optimize the ability to identify clients via the HHD. This process is implemented to reduce enrollment time as well as improve ease of use.
Back-up options for vendors are an essential part of good project management and essential for technology driven solutions. CARE had several vendor issues arise, notably a dispute with the hardware supplier of the HHD for payments for unauthorized services, and another vendor for insurance literacy that did not deliver as promised. Finding a backup to support for the HHD technology was difficult because of the skill requirements (a C-language programmer was required; this programming skill is difficult to find, even in India). CARE minimized the impact of the delay to get a replacement programmer by proceeding with field testing using the existing version, with the intention to bring in the updated, final version when available.
Back up options for a critical functionality of the HHD, i.e. connectivity to the server via GPRS, are required. In rural India, the reliability of GPRS (global positioning via satellite) in the HHD can vary by location and by carrier. This also requires additional training, and operational procedures with the hub clinic and server to ensure seamless service regardless of what technology interface is deployed. Solutions identified by CARE to improve connectivity included:
- A software patch to be developed by HDD vendor and implemented in April to boost connectivity
- Antennas were added to HDDs
- An SMS based application was implemented for back up. Thus, if GPRS does not function, connectivity via a mobile phone will enable VHCs to complete a transaction. All enrolled member data will be backed up, so even if VHC has to call in request, the doctor can access patient info required for health transactions.
- Creating a system (called CHIPS) to back up the technology solution during transition from manual environment and ongoing. This could have been anticipated earlier in the project implementation to smooth the workload during testing and going live, and to enhance process improvements as a result of more extensive documentation.
Hardware solutions developed in an urban context must be tested in a rural/local context to uncover potential issues. CARE could have discovered many of the technology challenges earlier, had preliminary testing in the actual context been done instead of relying on testing done at the location of the hardware vendor. Additionally, it would have been helpful to study lessons learned by other users of biometric data, and to better integrate work of the developers of the front end modules with the software development by the vendor.
Date of last Learning Journey update: April 2012