Reducing stock-outs of life saving malaria commodities using mobile phone text-messaging: SMS for life study in Kenya

Abstract

Background: Health facility stock-outs of life saving malaria medicines are common across Africa. Innovative ways of addressing this problem are urgently required. We evaluated whether SMS based reporting of stocks of artemether-lumefantrine (AL) and rapid diagnostic tests (RDT) can result in reduction of stock-outs at peripheral facilities in Kenya.

Methods/findings: All 87 public health facilities in five Kenyan districts were included in a 26 week project. Weekly facility stock counts of four AL packs and RDTs were sent via structured incentivized SMS communication process from health workers' personal mobile phones to a web-based system accessed by district managers. The mean health facility response rate was 97% with a mean formatting error rate of 3%. Accuracy of stock count reports was 79% while accuracy of stock-out reports was 93%. District managers accessed the system 1,037 times at an average of eight times per week. The system was accessed in 82% of the study weeks. Comparing weeks 1 and 26, stock-out of one or more AL packs declined by 38 percentage-points. Total AL stock-out declined by 5 percentage-points and was eliminated by the end of the project. Stock-out declines of individual AL packs ranged from 14 to 32 percentage-points while decline in RDT stock-outs was 24 percentage-points. District managers responded to 44% of AL and 73% of RDT stock-out signals by redistributing commodities between facilities. In comparison with national trends, stock-out declines in study areas were greater, sharper and more sustained.

Conclusions: Use of simple SMS technology ensured high reporting rates of reasonably accurate, real-time facility stock data that were used by district managers to undertake corrective actions to reduce stock-outs. Future work on stock monitoring via SMS should focus on assessing response rates without use of incentives and demonstrating effectiveness of such interventions on a larger scale.

Figure 1. A mobile phone displaying the stock request message sent to registered health workers.

Figure 2. Screenshots of SMS for Life web-based reporting tool.

(A) SMS for Life System dashboard. (B) Display of health facilities showing stock levels and stock-out alerts.

Figure 3. Weekly proportion of health facilities that responded to stock request messages and SMS formatting errors.

Legend. Black bars show responses within 0–24 hrs; dark grey bars responses within 24–27 hrs (after reminder but within incentive period); light grey bars responses within 27 hrs-7 days (after the incentive period); white bars shows no responses, and black line shows SMS formatting errors.

Figure 4. Accuracy of SMS reported stock parameters.

Legend. Black bars show correctly reported stock counts (+/−0–10%); dark grey bars show counts with discrepancy +/−10–20%; light grey bars show counts with discrepancy +/−20–30%, and white bars show counts with >30% discrepancy.

Figure 5. AL stock-out trends in study districts and stock-out comparisons with national trends.

(A) Proportion of health facilities stocked out of all four AL packs and at least one AL pack over 26 weeks. Legend. Blue bars show stock-outs of at least one AL pack; red bars show stock-outs of all four AL packs. (B) Proportion of health facilities stocked out of AL 6, AL 12, AL 18 and AL 24 packs over 26 weeks. Legend. Yellow bars show stock-out of AL 6; blue bars stock-out of AL 12; red bars stock-out of AL 18 and green bars stock-out of AL 24 (C) Stock-out trends of all four AL packs and at least one AL pack in study districts compared to a nationally representative sample. Legend. The 7 consecutive bars show stock-outs in August, September, October, November, December, January and February. (D) Stock-out trends of AL 6, AL 12, AL 18 and AL 24 packs in study districts compared to a nationally representative sample. Legend. The 7 consecutive bars show stock-outs in August, September, October, November, December, January and February.

Figure 6. RDT stock-out trend over 26 weeks monitoring period.