Main Findings
In our cohort, the diagnostic performance of POC-StatStrip® was generally good at both fasting and 2-hour time points. It provided a good diagnostic yield as evidenced by ROC-curve and Youden index analysis. However, on statistical analysis for method comparison, POC-StatStrip® showed a positive bias in both fasting (0.13 mmol/L) and 2-hours (0.55 mmol/L) compared to Lab-VPG. This resulted in a low PPV (68%) and moderate specificity (77.8%) for POC- StatStrip® in OGTT.
If POC StatStrip® were to replace all Lab-VPG measurements, it would increase the proportion of women in our cohort diagnosed with GDM by 15.3% (from 32 % to 47.3%), which would have implications for the pregnant women and for health resources.
An alternative approach that can safely utilise the advantages of POC testing is to identify samples within a specific glucose range with acceptable specificity and sensitivity that will not require laboratory confirmation. We therefore tested the utility of POC- StatStrip® in replacing conventional Lab-VPG testing when fasting POC-StatStrip® ≤ 5.0mmol/L and 2-hour glucose POC-StatStrip® is <7.5 or >9.5 mmol/L (7.5 to 9.5 mmol/L being around the diagnostic threshold and our analysis showed weaker correlation between the two methods within this range). The specificity and the NPV for fasting POC-StatStrip® ≤ 5.0 mmol/L and for 2-hour POC-StatStrip® < 7.5 mmol/l were 100% which means that we can confidently exclude GDM when POC-StatStrip® glucose concentrations are below these concentrations. The sensitivity and the PPV for 2-hour POC >9.5 mmol/L was 100% which means that we can confidently diagnose GDM when POC-StatStrip® glucose is > 9.5 mmol/L. We cannot comment on the threshold for fasting glucose because in our cohort very few women were above diagnostic threshold.
Based on the above we propose an OGTT protocol in which POC-StatStrip® is performed first, with immediate venous sampling only required at fasting time-point if POC-StatStrip® ≥5.1 mmol/L and at 2-hour time-point if POC-StatStrip® is in the range of 7.5-9.5 mmol/L . This would, in our cohort, reduce the requirement for venepunctures and laboratory glucose testing by 75% for fasting samples and by 48 % for 2-hour samples.
The protocol could improve patient experience and reduce the cost of laboratory tests. Furthermore, as a proportion of women can be given a definitive diagnosis of GDM/no GDM immediately, this would improve staff productivity by reducing time spent on phoning patients to provide results and will allow those with a diagnosis of GDM to receive equipment and counselling sooner.