Stepwise multivariate linear regression using log hydroxybutyrate:glucose ratio as dependent variable
| Independent variable | Coefficient and 95% CI | p Value |
| Age (hours) | 1.02 (1.00 to 1.03) | 0.002 |
| Gestation | ||
| Post-term compared with term | 1.94 (1.37 to 2.75) | <0.001 |
| Birthweight (kg) | −1.57 (−1.99 to −1.23) | <0.001 |
| Rectal temperature (°C) | 1.31 (1.13 to 1.51) | <0.001 |
| Maternal free T4 | −1.04 (−1.07 to −1.01) | 0.006 |
| Infant free T4 | 1.02 (1.01 to 1.03) | <0.001 |
| Infant TSH (log values) | −1.95 (−3.03 to −1.25) | 0.003 |
| Sex | 1.33 (1.11 to 1.60) | 0.002 |
| Female compared with male | ||
| Antenatal care | −1.24 (−1.51 to −1.01) | 0.04 |
| Compared with none | ||
The estimated effect on blood hydroxybutyrate/glucose ratio is the change in level for a unit increase in a continuous variable, or the stated categorical variable. The maternal and infant variables entered into the regression were maternal age, height, weight, haemoglobin, parity, social score, antenatal attendance and thyroid function; infant sex, birthweight, length, head circumference, haemoglobin, Apgar scores at one and five minutes, free T4 and TSH. Infant age at sampling was entered as an important potential confounder.
Number of observations = 515
Geometric mean of hydroxybutyrate ratio = 41 (min 0.2 to max 135)
Adjusted R-squared = 0.25