The HOMA-IR Performance to Identify New Diabetes Cases by Degree of Urbanization and Altitude in Peru: The CRONICAS Cohort Study
Table 3
Area under the receiver operating characteristic curve for new diabetes cases.
Population
AUC (95% CI)
Empirical cutoff point estimation
Empirical threshold (95% CI)
Youden’s index
Sensitivity
Specificity
AUC (95% CI)
+LR
−LR
PPV
NPV
Overall
0.6860 (0.6366–0.7355)
2.8 (2.1337–3.5417)
0.31
0.53
0.77
0.65 (0.6099–0.6962)
2.30
0.61
14.59
95.67
Study site
Lima
0.7114 (0.6294–0.7933)
2.9 (2.1050–3.6879)
0.38
0.62
0.75
0.69 (0.6225–0.7546)
2.48
0.51
12.68
97.12
Urban Puno
0.6233 (0.5171–0.7295)
0.9 (0.0000–1.9610)
0.23
1.00
0.23
0.62 (0.5578–0.6746)
1.30
0.00
8.78
100
Rural Puno
0.6218 (0.4885–0.7550)
1.2 (0.7160–1.6160)
0.36
0.82
0.54
0.68 (0.5687–0.7871)
1.78
0.33
5.46
98.93
Tumbes
0.7281 (0.6464–0.8098)
2.8 (1.8282–3.8472)
0.40
0.67
0.74
0.70 (0.6356–0.7674)
2.58
0.45
22.80
95.14
Altitude above the sea level
Low
0.7191 (0.6614–0.7768)
2.8 (2.1979–3.4774)
0.39
0.64
0.74
0.69 (0.6460–0.7402)
2.46
0.49
17.43
95.99
High
0.6335 (0.5531–0.7139)
1.2 (0.6240–1.7129)
0.27
0.85
0.42
0.63 (0.5842–0.6848)
1.47
0.36
7.17
98.15
Rurality
Urban
0.6752 (0.6081–0.7424)
2.5 (1.6665–3.2595)
0.30
0.60
0.70
0.65 (0.5919–0.7088)
2.00
0.57
11.34
96.47
Semiurban
0.7281 (0.6464–0.8098)
2.8 (1.7997–3.8757)
0.40
0.67
0.74
0.70 (0.6355–0.7674)
2.58
0.45
22.80
95.14
Rural
0.6218 (0.4885–0.7550)
1.2 (0.7569–1.5751)
0.36
0.82
0.54
0.68 (0.5757–0.7801)
1.78
0.33
5.46
98.93
values for equality between populations in terms of their areas under the receiver operating characteristic curve were 0.306, 0.090, and 0.366 for study site, altitude, and rurality, respectively; the null hypothesis is that all areas are equal. +LR refers to the positive likelihood ration, whereas −LR refers to the negative likelihood ratio. PPV and NPV refer to positive and negative predictive values, respectively (results presented in %); these were computed with specific prevalence estimates according to the level of the variable (e.g., diabetes prevalence at low altitude).