We present the results of three years of BH functionality monitoring. We hypothesized that groundwater quality, and therefore the level of satisfaction with BH water quality, would have a bearing on functionality. However, we found that in any given year, approximately 20% of the BHs were broken in the study towns, with no differences across clusters. This is consistent with other studies conducted in rural Ghana [6–7]. In a complementary study conducted in the same 15 towns, we found that groundwater quality also did not reduce self-reported BH use, but rather increased the use of other water sources such as hand-dug wells and surface water . However, this finding was based only on categorical self-reported use and not on the relatively volume of water used from each source.
Here, we found that 67% of the BHs observed at all three time points were consistently functional and only three BHs were consistently broken (one in each water quality cluster), showing a concerted effort to finance and repair BHs over time. We also found that some study towns switched from not paying for water at all or paying reactively when a BH breaks to a proactive payment mechanism over the course of the monitoring period. This shift was more common in the control and high salinity clusters. In the high iron cluster, four of five towns still were not charging for BH water in 2016. This is consistent with the contextual knowledge of our study team: residents of towns with elevated iron content are generally unwilling to pay for BH water and prefer to use free alternative water sources widely available in their neighborhoods.
We also observed that the number of water users per BH was higher in the control cluster, as compared to towns with water quality problems. This observation is also consistent with the overall higher number of BHs in towns with water quality problems. In the case of the high iron cluster, iron concentrations in the water samples are more than ten times higher (up to 4.5 mg/L) than the water quality standard in Ghana (0.3 mg/L) . We suspect that while BHs are being constructed, and water testing results reflect unsatisfactory quality, more BHs may be drilled to find a depth or location with lower iron levels. This may result in a higher number of BHs in these locations as compared to places where water quality standards are met.
We note that we intentionally excluded some important predictors that were significant in other studies: namely, presence of a water committee and the availability of and distance to BH mechanics and spare parts [12, 17]. Presence of a water committee was excluded because it was highly correlated with proactive payment mechanism. Distance to BH mechanics and spare parts was excluded because all towns were located relatively close (< 25 km) to district capitals that had these resources, exhibiting no variability in the dataset.
Historically, government resources in sub-Saharan African countries have been insufficient to repeat large-scale data collection efforts and track infrastructure over time . More recently, there is a shift toward mapping and monitoring of existing water supplies. For example, Liberia conducted extensive water point mapping in 2011 and 2017 (https://wash-liberia.org/raw-water-point-data/). The expansion of digital technologies may provide further opportunity for cost-effective near real-time monitoring of water sources in remote and rural areas. Further availability of longitudinal data, new standardized data-sharing platforms, such as the Water Point Data Exchange (https://www.waterpointdata.org/), and advances in analytical methods  will allow governments, funding agencies, and implementing organizations to study infrastructure functionality and its correlates and direct resources toward sustainability.