TABLE 9.

Summary of indicator characteristics relative to predicting MPA > 0.50, for NPNL women across all sites

Food group cutoffs	Sensitivity1	Specificity	False positives, %	False negatives, %	Total misclassified, %	Cutoff yielding closest estimate of % of women with MPA > 0.50 (percentage point difference)2
FGI-9R
≥3	86–100	4–32	33–69	0–4	33–69
≥4	57–98	30–65	16–51	1–18	34–54	Mozambique (−3), Philippines (+14)
≥5	14–67	66–100	0–25	9–52	23–52	Burkina Faso (+15), Mali (−4), Bangladesh (+1), Philippines (−12)
≥6	2–36	94–100	0–5	18–60	21–60	Burkina Faso (−13)
FGI-13R
≥3	86–100	2–32	33–69	0–4	34–69
≥4	59–98	18–62	18–53	1–15	33–56	Mozambique (+4)
≥5	30–79	59–93	3–30	6–43	23–46	Mali (+12), Bangladesh (+8), Philippines (−5)
≥6	10–50	84–100	0–9	14–55	22–55	Burkina Faso (−6), Mali (−16), Bangladesh (−12)
≥7	2–19	95–100	0–3	19–60	21–60
FGI-21R
≥4	83–98	13–46	21–58	1–9	30–61	Mozambique (+13)
≥5	59–87	36–83	7–34	5–24	29–40	Mozambique (−17), Philippines (+9)
≥6	22–75	71–100	0–16	7–48	23–48	Burkina Faso (+9), Mali (+3), Bangladesh (+2)
≥7	11–40	89–100	0–6	16–54	21–54

¹Sensitivity assesses the proportion of all those who have MPA > 0.50 who are identified by the indicator.

²The difference is the site-specific prevalence for MPA > 0.50, less the prevalence estimated by the FGI at a given cutoff. When 2 cutoffs yielded estimates that were similar (within 5 percentage points) in distance from “true” prevalence, both are listed.