Table 3. Pooled odds ratios (ORs)a for diabetes mellitus and two associated risk factors.
Variable | Diabetes mellitus |
Impaired fasting glycaemia |
Impaired glucose tolerance |
||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
nb | OR (95% CI) | Pc | nb | OR (95% CI) | Pc | nb | OR (95% CI) | Pc | |||
All data sets | 35 | 1.01 (0.91–1.11) | 11 | 1.56 (1.20–2.03) | 21 | 0.84 (0.72–0.98) | |||||
Area of residence | 0.009 | 0.56 | < 0.001 | ||||||||
Combined | 9 | 1.17 (1.04–1.31) | 6 | 1.61 (1.14–2.26) | 7 | 0.69 (0.59–0.81) | |||||
Periurban | 2 | 0.70 (0.42–1.18) | 2 | 0.79 (0.46–1.37) | |||||||
Rural | 11 | 0.98 (0.80–1.20) | 2 | 2.21 (0.87–5.64) | 6 | 0.82 (0.61–1.09) | |||||
Urban | 13 | 0.86 (0.73–1.01) | 3 | 1.24 (0.71–2.19) | 6 | 1.33 (1.03–1.72) | |||||
Subregion of residence | < 0.001 | 0.019 | 0.001 | ||||||||
Middle Africa | 2 | 1.44 (1.31–1.59) | 1 | 1.04 (0.65–1.65) | 2 | 0.73 (0.49–1.09) | |||||
Eastern Africa | 21 | 1.08 (1.01–1.15) | 9 | 1.65 (1.35–2.02) | 11 | 0.71 (0.59–0.84) | |||||
Southern Africa | 12 | 0.80 (0.69–0.92) | 1 | 5.19 (1.75–15.38) | 8 | 1.30 (0.99–1.70) | |||||
Ethnicity of subjects | 0.012 | 0.066 | 0.002 | ||||||||
African | 24 | 1.12 (1.00–1.25) | 7 | 1.30 (0.96–1.74) | 11 | 0.81 (0.66–0.99) | |||||
Indian | 2 | 0.69 (0.52–0.94) | 2 | 1.66 (1.10–2.50) | |||||||
Multi-ethnic | 9 | 1.00 (0.87–1.14) | 4 | 1.93 (1.42–2.62) | 8 | 0.73 (0.60–0.89) | |||||
Study year | 0.125 | 0.81 | 0.61 | ||||||||
Before 1991 | 9 | 0.90 (0.73–1.11) | 1 | 1.94 (0.84–4.48) | 4 | 0.90 (0.62–1.31) | |||||
1991–1999 | 13 | 0.96 (0.83–1.12) | 4 | 1.41 (0.88–2.27) | 10 | 0.90 (0.68–1.19) | |||||
After 1999 | 13 | 1.13 (0.99–1.30) | 6 | 1.58 (1.09–2.30) | 7 | 0.74 (0.55–1.01) | |||||
Country income level | 0.008 | 0.006 | 0.028 | ||||||||
Low | 14 | 1.21 (1.06–1.37) | 6 | 1.18 (0.89–1.57) | 5 | 0.70 (0.52–0.95) | |||||
Lower middle | 4 | 1.16 (0.75–1.80) | 4 | 0.50 (0.29–0.87) | |||||||
Upper middle | 17 | 0.93 (0.83–1.03) | 5 | 2.05 (1.56–2.69) | 12 | 0.99 (0.80–1.23) |
CI, confidence interval.
a ORs represent the odds in men versus the odds in women.
b Number of data sets included in the analysis.
c P-value for the category, estimated in a Q-test.