Role of surveillance data in detecting malaria outbreaks in an epidemic-prone region in Kenya: findings from an investigation of a suspected outbreak in Nandi County.

Abstract

Background: Approximately 70% of the Kenyan population is at risk for malaria, including 19 million people in highland epidemic-prone and seasonal transmission areas. Surveillance data showed a 288% increase in malaria cases and an incidence rate of 10.5 per 1000 population between January and May 2021 in Nandi County. We investigated the increased incidence of malaria in Nandi County.

Methods: We abstracted demographic and clinical data from the laboratory register in health facilities with high malaria burden. Key informant interviews using a structured questionnaire collected healthcare worker perceptions on malaria interventions and personnel capacity. We calculated means and medians for continuous variables and frequency and proportions for categorical variables. Data quality assessment (DQA) was conducted to evaluate timeliness and completeness, data accuracy, and overall system assessment.

Results: We reviewed 19,526 records from 12 health facilities. Females contributed 61% cases (11,862). A majority of cases, 21% (4111), were between the age group 15-24 years. Of the 19,498 tested, 2662 tested positive (test positivity rate, TPR = 13.7%). Microscopy accounted for 39% (1041) and RDT for 61% (1620) of tests conducted, with some patients being double tested using both tests. Kapsabet County Referral contributed 26% (5051) suspected cases, TPR 3.2%, and Chemase Health Centre TPR was 33.2%. Facilities experienced major RDTs stock-outs in the preceding 3 months while three (30%) of the 10 facilities assessed conducted laboratory Internal Quality Control (IQC) programmes. Of the 12 facilities assessed, four (33%) facilities had an over-reporting of suspected cases in the monthly summary, while three (25%) facilities were over-reported in the online tool. On reporting confirmed malaria cases, over-reporting was noted in three (25%) facilities in both the monthly summaries and the online tool. Data completeness was 77% and timeliness 93%.

Conclusion: The increase in malaria cases in Nandi County displayed a seasonal pattern that coincided with either the long or short rainy seasons, the investigation did not reveal an active outbreak at the time of the inquiry. Sub-county hospitals in Tinderet and Aldai sub-counties had malaria cases exceeding both the alert and action thresholds at specific times during the year under review, suggesting a potential occurrence of unidentified outbreaks, while several other facilities had an increase of cases reaching alert thresholds, indicating upsurges. In healthcare settings, we noted there were problems with data quality. We advised routine data review, analysis, and feedback; mentorships for data analysis and on the job and support supervision; mentorships for malaria diagnosis; and installation of laboratory quality assurance.