Table 4.
Effects of early life conditions on diabetes overall and by groups
| Model 1 | IMR1 | Times 1 | |||
|---|---|---|---|---|---|
| Model 2a | Model 2b | Model 3a | Model 3b | ||
| Lower | Higher | Good | Bad | ||
| OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) | |
| Female | 0.94 (.72–1.23) | 1.11 (.73–1.68) | 0.83 (.58–1.17) | 1.02 (.74–1.41) | .80 (.49–1.30) |
| Age 60–64 (ref) | |||||
| Age 65–69 | 1.18 (.89–1.56) | 1.39 (.91–2.13) | 1.05 (.72–1.53) | 1.24 (.90–1.70) | 0.98 (.52–1.84) |
| Age 70–74 | 1.27 (.94–1.71) | 1.82 (1.11–2.99) | 1.02 (.68–1.52) | 1.18 (.80–1.76) | 1.07 (.59–1.94) |
| Education (yrs) | 0.99 (.96–1.02) | 1.01 (.97–1.06) | 0.98 (.94–1.02) | 0.99 (.96–1.02) | 0.99 (.94–1.05) |
| Obesity | 1.42 (1.06–1.90) | 1.11 (.74–1.67) | 1.67 (1.15–2.42) | 1.48 (1.05–2.07) | 1.29 (.76–2.17) |
| Poor child health | 1.22 (.93–1.61) | 1.33 (.85–2.08) | 1.22 (.85–1.74) | 1.28 (.92–1.79) | 1.11 (.66–1.86) |
| Low knee height | 1.28 (.97–1.69) | 1.25 (.77–2.02) | 1.33 (.94–1.87) | 1.40 (1.00–1.96) | 1.09 (.66–1.81) |
| Poor child SES | 1.03 (.79–1.35) | 1.47 (.99–2.19) | 0.79 (.54–1.17) | 1.19 (.87–1.61) | 0.75 (.47–1.21) |
| Family member w/diabetes | 4.08 (2.74–6.07) | 4.17 (2.23–7.78) | 3.91 (2.35–6.50) | 4.12 (2.50–6.77) | 3.88 (1.96–7.69) |
| Exposure period | 1.86 (1.19–2.90) | 2.09 (1.08–4.07) | 1.62 (.86–3.05) | 1.91 (1.08–3.37) | 1.78 (.85–3.73) |
| Family member X exposure period interaction | 0.54 (.32–.93) | 0.44 (.20–.99) | 0.71 (.33–1.51) | 0.51 (.26–.99) | 0.63 (.25–1.60) |
| Never smoked (ref) | |||||
| Previous smoker | 0.78 (.57–1.06) | 1.01 (.63–1.62) | 0.63 (.42–.94) | 0.85 (.59–1.22) | 0.64 (.37–1.09) |
| Current smoker | 0.79 (.52–1.22) | 0.44 (.19–1.02) | 0.94 (.55–1.60) | 0.72 (.42–1.23) | 0.86 (.39–1.90) |
| Exercises | 0.67 (.52–.86) | 0.63 (.42–.93) | 0.66 (.47–.92) | 0.70 (.52–.94) | 0.58 (.37–.92) |
| Range of n2 | 1429–1444 | 570–577 | 851–860 | 998–1007 | 421–434 |
| Range of LL2 | −832, −859 | −341, −332 | −503, −484 | −586, −571 | −264, −251 |
Source: Puerto Rican Elderly: Health Conditions project (PREHCO 2007), imputed; all 60- to 74-year olds who lived in the countryside as children. Results shown are based on combining multiple imputation results.
Lower IMR is defined as the lowest tercile of annual IMR at respondent’s year of birth and Higher IMR is defined as the second and third tercile of annual IMR at respondent’s year of birth. Good/Bad Times defined according to if the observed IMR was greater than 1 standard deviation from the mean trend in a particular municipio.
The multiple imputation procedure required us to work with five alternative completed data sets. In this case it was not clear how to calculate conventional statistics, such as chi-square, all of which are functions of data-specific log-likelihood functions. As a partial resolution to the conundrum, we present in this table the range of values for the chosen statistics obtained after estimating models for each of the imputed data sets. Sample sizes for groups vary slightly because differences between Low/High IMR and Good/Bad Times groups resulted in slightly different division into groups.