data

The Botswana Combination Prevention Project (BCPP) Data Linkage Tool

What was the problem?

Data management systems that support individual-level patient tracking are often neither affordable nor easy to implement. However, being able to track patients at the individual level can facilitate timely and accurate HIV service delivery to those in need. For example, real-time follow of antiretroviral treatment (ART) patients who have not picked up their medications reduces the amount of time an ART patient is without medication.

Researchers in a large-scale, PEPFAR-funded project in Botswana (known as the Botswana Combination Prevention Project, or BCPP), needed to follow project participants over time, using an electronic platform, in order to measure the impact of offered HIV services. That is, researchers needed data demonstrating that persons diagnosed with HIV were linked to treatment, initiated and retained on ART, and are virally suppressed in a timely and efficient manner. As such, the requirements of the electronic tracking system included:

  • Trace persons who tested positive for HIV, but did not initiate ART (i.e., persons who were lost to follow up)

  • Monitor individuals through the clinical cascade, from HIV testing through viral suppression

  • Uniquely identify persons (either through unique identifiers (UIDs) or another approach)

  • Monitor movement of persons across communities and geographic spaces

  • Record age and sex at the individual level

The investigators undertook an extensive search to identify and purchase an affordable data management system that supported individual patient tracking over time. However, they were not successful in finding a system that met their needs. As a result, the Botswana Harvard AIDS Institute Partnership developed a system rather than purchasing one.

What is the tool?

The flexible, affordable open-source data management system supported the longitudinal tracking of patients on an individual basis for the duration of the project.

Key steps in building the data linkage system included:

  1. Data Security

    Key elements to ensure data security include a firewall, hash keys to ensure encrypted data cannot be unencrypted, user logins, physical security and access, and routine data back-up measures.

  2. Data Quality

    Tablets have built-in validation tools to limit errors (e.g., invalid, implausible numbers). Staff are also trained to review data for completeness and errors.

  3. Confidentiality of Data

    Data were ‘hashed’ or encrypted to protect the privacy of patients in this project. As a result, data became “unusable, unreadable, or indecipherable” when unauthorized persons attempted to view the data. An Omang, the Botswana national identification card for citizenship, UID was encrypted with a one-way hash prior to being transferred to the project server. When there was consent, or a waiver of consent, from the client to extract identifiable data from the clinic medical record systems (electronic and web-based), clinic data were linked with HIV testing data using the ‘hashed’ Omang.

  4. Costing

    The average costs for establishing the project’s infrastructure for the 30 communities was $14,394 per community (total = $431,815). The annual bandwidth and maintenance cost for each of the three communities where this was implemented was $4,974 (total = $14,922). The project bought 18 laptops in 2016 to replace agency laptops at the project sites. The cost was $20,625, or $1,146 per laptop for the 18 Dell Latitude 3470 laptops purchased. Estimates for the total cost of the equipment used for the database hardware were as follows:  $5,500 for Ubuntu server, $1,200 for tape backup and $3,500 for UPS for a total of $10,200.  Please note, with typical information technology systems, the standard replacement cycle for hardware is approximately every 3 years.  The annual cost to maintain the hardware and network is $28,000/year.

Figure 1. BCPP Data Linkage System

Figure 1. BCPP Data Linkage System

How to Use the System:

After BCPP research assistants completed their daily data entry of patient level data collected at the community level, they reported to BCPP ‘base operation village’ trailers (i.e., their main community-based headquarters). The research assistants’ laptops automatically connected to the secure Wi-Fi when in 15 meters range. Data were synchronized, uploaded, and transmitted to the server at the Botswana Harvard AIDS Institute Partnership offices.

Complete details on this data linkage approach can be found at: https://aids.harvard.edu/after-the-capture/#more-6903. Countries can access the BCPP data management system (Python modules) at GitHub (https://github.com/botswana-harvard).

Key Outcomes:

This BCPP data linkage management system facilitated the rapid examination of the effects of HIV clinical interventions on timely patient-level linkage to treatment, retention, and viral suppression rates; and follow-up of patients within 24 hours as opposed to months later.

Figure 2. Flow of information from the first test of HIV to cleaning and analyzing the data.

Figure 2. Flow of information from the first test of HIV to cleaning and analyzing the data.

Expedited Biobehavioral Surveys and Data Availability

What is the problem?

Data is necessary for targeted service delivery. Key population (KP) programs need relevant and timely data about KPs—sex workers, men who have sex with men, transgender individuals, people who inject drugs, and people in prison and other closed settings—engaged and not engaged in services in order to target their services. Biobehavioral surveys (BBS) provide population-based information about KP risk behaviors, HIV prevalence and, importantly, the 90-90-90 cascade to inform service provision and policy. The time BBSs take to plan, implement, and make available the results for remediated actions, is a major limitation to their utility.

What is the tool? 

These tools aim to make BBS planning, implementation, and results provision efficient and timely.

1.       Key factors and interventions to consider for timely and successful implementation of the BBS are summarized in the Technical Considerations Fast Tracking Quality Key Population Biobehavioral Surveys (doc).

2.       Data management and cleaning for a BBS survey can take upwards of three months. It is therefore essential to begin these processes during data collection. This also allows problems to be addressed in near real-time and means that data analysis can begin soon after the end of data collection. The Priority Results Table (doc) is a minimum set of variables whose results should be shared with key stakeholders within two months of the end of data collection. The survey team should collaborate with stakeholders to identify context-specific modifications to the Priority Results Table, including of response categories and additional priority variables.

3.       The sample size and thus duration of BBS implementation can heavily influence the budget. The WHO Blue Book (external link) offers a budget template for costing BBS. As the bulk of the budget is associated with implementation, expediting BBS does not necessarily result in substantial cost savings as survey duration may remain unchanged.

Examples of outcomes related to use of BBS data: BBS data can be used to advocate for, and inform, the development of a Ministry of Health key populations policy and provide new insight into areas where key populations congregate. For example, a BBS in South Sudan revealed additional areas where service providers can find sex workers, through the geospatial mapping component of BBS. In Papua New Guinea, early release of BBS data prompted the use of social-network methods for reaching and testing KPs.

Additional Resources:

WHO BBS Survey Guidelines for Populations at Risk for HIV: Supplementary Materials (pdf)