Suitability of recommended limits for fasting glucose tests in women with polycystic ovary syndrome

Abstract

Background: The Canadian and American Diabetes Associations recommend the use of an oral glucose tolerance test to screen for abnormal glucose tolerance among women with polycystic ovary syndrome when their fasting plasma glucose level is 5.7 mmol/L or more (Canadian guideline) and 5.6 mmol/L or more (American). Our objective was to determine the predictive value of 5.6 mmol/L as a fasting plasma glucose cutoff for detecting abnormal glucose tolerance in women with polycystic ovary syndrome, and then to define the optimal cutoff for this population.

Methods: An oral glucose tolerance test was administered to 105 consecutive women with polycystic ovary syndrome referred to an academic reproductive endocrine clinic. We calculated sensitivity, specificity and likelihood ratios.

Results: The sensitivity of a 5.6 mmol/L cutoff was 48% (95% confidence interval [CI] 30%-67%); its specificity, 98.7% (95% CI 96.1%-100%). With this cutoff, 52% of women with polycystic ovary syndrome whose glucose tolerance is abnormal would be missed. The prevalence of abnormal glucose tolerance was 28%, with a positive predictive value of 93% (95% CI 81%-100%) and a negative predictive value of 83% (95% CI 76%-91%). The likelihood ratio for a positive test was 36.7 (95% CI 5.0-267), and for a negative test, 0.5 (95% CI 0.4-0.7). The optimal fasting plasma glucose cutoff value was 5.0 mmol/L, with a 79% sensitivity (95% CI 65%-94%) and 66% specificity (95% CI 55%-77%). If this cutoff were used, 24% of women with abnormal glucose tolerance would still be missed.

Interpretation: The Canadian and American recommendations--of screening for abnormal glucose tolerance with an oral glucose tolerance test only when the results of a fasting plasma glucose test are 5.7 mmol/L (or 5.6 mmol/L) or more--are inappropriate for women with polycystic ovary syndrome. We therefore recommend that all women with polycystic ovary syndrome have an oral glucose tolerance test.

Fig. 1: A 2 × 2 table showing the performance characteristics of the 2-h glucose tolerance test (cutoff ≥ 7.8 mmol/L) compared with the fasting plasma glucose test (cutoff ≥ 5.6 mmol/L). PPV = positive predictive value, NPV = negative predictive value.

Fig. 2: Receiver operating characteristic curve of the association between fasting plasma glucose and abnormal glucose tolerance. The area under the curve was 0.828 (95% confidence interval 0.735–0.921). The optimal cutoff for the test is the point closest to the upper-left corner of the graph, which corresponds to a fasting plasma glucose measurement of 5.0 mmol/L.

Fig. 3: A logarithmic representation of the likelihood ratio of having abnormal glucose tolerance, as a function of fasting plasma glucose test results. At a glucose concentration of 4.4 mmol/L, the likelihood of finding that test result among patients with abnormal glucose tolerance is equal to its likelihood among those with normal glucose tolerance.