1. A follow up study meets some of the criticisms of earlier studies, but this design has some methodological limitations as it is mainly possible to test what has been called the accumulation hypothesis in contrast to the immediate causation hypothesis.25 In our study, the final information about distress was given in the 30th week of pregnancy, and a state of distress after this point was not registered. Acute distress after the 30th week of pregnancy, though not recognised in our study, might cause preterm delivery
2. study, women who answered the second questionnaire but not the third had higher general health questionnaire scores in the 16th week than women who answered all questionnaires. Although it is thus unlikely, selection bias could be introduced by the non-response: if non-responders in the 30th week (with a high incidence of preterm delivery) had low scores on the general health questionnaire the association of the study might be explained by bias (differentiated non-response). Assuming the extreme situation, in which all non-responders had low general health questionnaire scores, we conducted additional analyses. In this hypothetical example, the risk associated with high scores on the general health questionnaire remained (adjusted relative risk;VOLUME, B.M.J.,1
3. We found a distinct association between distress in the 30th week of pregnancy and risk of preterm delivery, but distress in the 16th week was not significantly associated with risk of preterm delivery. The general health questionnaire is a "here and now" assessment and does not seek to integrate distress over time. The absence of association between the measurement in the 16th week and preterm delivery is therefore consistent. The findings seem to depict a change in vulnerability during pregnancy
4. A large prospective cohort study using validated scales, confounder control, and a well defined outcome measure was conducted by Brooke et al, who studied the relation between several psychosocial factors, including the general health questionnaire, and birth weight in 1800 pregnancies.27 As gestational age was controlled for in the analysis, the results were presented in terms of fetal growth. When smoking was controlled for, no relation with general health questionnaire scores was found, but the relation with preterm delivery was not investigated, and to our knowledge no previous studies conceming preterm delivery have met all components ofvalidity
5. The contribution of low birth weight to infant mortality and childhood morbidity;McCormick, M.C.;NEnglJMed,1985