Using a national digital EQA platform to identify laboratories at risk of recurrent underperformance in Malawi

Recurrent laboratory underperformance in External Quality Assessment (EQA) programmes remains a practical challenge in Malawi, where delayed recognition of analytical instability can slow corrective action and reduce the value of participation. Within the national Full Blood Count (FBC) EQA scheme, this project assessed whether routinely collected digital EQA data could be used to identify laboratories at risk of repeated poor performance and support earlier, targeted intervention.

The Kamuzu University of Health Sciences Quality Assurance Program (KUQAP) implemented EQALite as part of routine EQA operations. The platform supports participant registration, electronic result submission, automated statistical analysis using Algorithm A, and generation of laboratory-specific performance reports. It is embedded in the existing KUQAP workflow and operates within governance arrangements that include controlled user access, anonymised participant identifiers, confidentiality safeguards, report authorisation, and documented follow-up of unsatisfactory performance.

The dataset comprised 10 standardised EQA rounds from April 2025 to 2026 and approximately 1,600 analyte-level submissions from 38 registered laboratories. Overall round acceptability ranged from 84.0% to 93.2%; the newest H4/2026 round achieved 92.3% overall acceptability, including 100.0% for Haemoglobin concentration (HGB) and 96.8% for Platelet count (PLT). Mean Cell Volume (MCV) remained an important signal for monitoring, with two unsatisfactory results in H4/2026 despite otherwise strong round performance. Participant-level longitudinal analysis showed a strong association between Z-score variability and unacceptable result frequency (Spearman ρ = 0.847, p < 0.001), and between mean absolute Z-score and unacceptable results (ρ = 0.840, p < 0.001). These findings support the use of variability metrics as practical early warning indicators for recurrent underperformance.

Implementation barriers included occasional incomplete submissions, localized reagent stock-outs, intermittent internet connectivity, and the absence of automated real-time alerting. Despite these constraints, digital EQA data enabled risk-based performance review without additional data collection burden. The next improvement step is to embed automated alert rules in EQALite so that rising variability, repeated warning scores, or analyte-specific instability can trigger earlier technical follow-up before recurrent failure becomes established.