HUI3 quantifies disability in eight domains of functioning and also quantifies the preference of the general public for each of the health states defined by the HUI3 system. As HUI3 thus incorporates preferences for health states, we feel this is an appropriate measure of quality of life. Furthermore, use of HUI3 had the great advantage of making our results directly comparable to those reported from other countries, for instance Canada and Germany . We strongly favour standardization of HRQoL measurement, even though other measures might have generated relevant disease-specific information . Respondent burden is also at issue here.
Horsman et al  found a 0.03 difference in MAU to be clinically important. Our comparison of HRQL at age 14 and 19 showed a 0.01 MAU difference. Clearly then, at the group level no important changes in HRQL were found in our VLBW subjects. HRQL was fairly high at both ages, and almost similar to results reported for the general US population [37, 38] and self-reported HRQL in ELBW young adults in Canada [39, 40]. It should be remembered though, that participation was related to SES and to level of handicap at 5 years of age . Our results represented less than half of the original cohort. Our data showed non-participants had lower SES and more handicaps and also that, in participants, these factors were negatively related to HRQoL. We hypothesize we only saw a positive tip of the iceberg in our data, due to loss to follow-up.
Saigal et al  found a 0.05 HRQL decrease in ELBW subjects between adolescence and young adulthood. Matched controls showed the same decrease. A decrease in HRQL between age 10 and 40 was also reported by Chen et al.  in a study of HRQL among 752 persons born between 1965 and 1975 in the US. Perhaps HRQL decreases between adolescence and young adulthood independently of health conditions, due to the increasingly difficult developmental tasks most young adults are confronted with (e.g. choosing their studies or profession, living on their own, and finding a partner). This is consistent with one Dutch study on the psychological well-being of Dutch adolescents, that tended to decrease gradually in the period from 12 to 23 years of age .
These findings from the literature are inconsistent with the results we found in the present study, showing no decrease in HRQL between age 14 and 19 at the group level. Since no matched control data of children born at term were available, we have no way of knowing whether VLBW children differed from children born at term in this respect. One explanation for the fact our findings differed from those reported by Saigal et al  may be that they used self-perceived utility, whereas we used a MAU function representing preferences of the general Canadian population. Maybe self-perceived HRQL is more sensitive to change. Futhermore, Saigal’s cohort included ELBW children exclusively, whereas the POPS VLBW cohort included only 15% ELBW children. Maybe ELBW children are more vulnerable in growing up, due to their relatively unfavourable start. Our findings may also be the result of social and cultural factors compensating for perinatal disadvantage. As children grow older, the impact of biological and perinatal risk factors diminishes and demographic and psychological factors have a greater influence on the cognitive performance of LBW and preterm children [3, 43]. Indeed, our regression analysis corroborated the importance of psychological factors in HRQL. Furthermore, the wider social policy and cultural context may have an impact on HRQL and well-being of children and young adults. A recent UNICEF report  on the well-being of children in 21 rich countries found that the Netherlands ranked first place in the overall educational, social, health wellbeing in children, whereas Canada, for example, ranked 12th.. Thus the general favourable conditions of care for children in the Netherlands may also be reflected in the stable HRQL of our VLBW children .
Although HRQL was stable at the group level, our analyses of separate HUI3 attributes showed considerable individual change over time. Was this the result of measurement error or was it true change? Part of the changes observed may be due to random error of measurement. Nevertheless, we do not want to exclude the possibility that clinically important changes in HRQL actually took place, especially in the psychological attributes of HRQL. A considerably proportion of subjects were better off in these attributes, but a comparable proportion were worse. Especially the increased proportion of subjects reporting pain is puzzling and needs further research.
Unlike Hack , we found that SES was only weakly related to HRQL at age 19. Sigmond-de Bruin suggested that the lack of influence of SES in our cohort might result not only from the high mean level of the SES in the Netherlands, but also from the country’s high accessibility of care, and its relatively low levels of social and economic inequality .
The relationship of AGA to HRQL at age 19 was weak. Since AGA is a strong predictor of several health and psychological outcomes at younger age, the impact of AGA on HRQL may diminish with age [2, 4]. However, level of handicap at age five was still a good predictor of HRQL at age 19. Assessment of level of handicap early in life may therefore help parents to understand what HRQoL later in life may be.
The importance of physical problems was underlined by the fact that handicap at age five and neuro-motor problems at age nineteen were both related to HRQL.
As mentioned, Saigal et al  found no difference between the mean HRQL of young adults born preterm and that of young adults born at term, and concluded that young adults born with a handicap have adapted to their disabilities and view their lives fairly positively. We found handicap measured at age five and neuro-motor score at age nineteen both to be significantly related to HRQL at age 19. Whereas 68% of the young adults without a handicap reported a high HRQL (MAU > 0.90), only 38% of the young adults with a mild to severe handicap reported a HRQL that high. The high mean score for HRQL might thus be explained not by handicapped young adults having a high HRQL, but by the non-handicapped young adults compensating for their handicapped peers in our cohort, thereby raising the mean HRQL to the same level as that in young adults born at term. Our results do not support the assumption that all young adults with a handicap have learned to cope with their handicaps .
Our finding that non-adaptive coping strategies were negatively associated with HRQL is consistent with other studies that found an association between a lower HRQL and non-adaptive coping strategies for various diseases [45–47]. Use of strategies such as self-blame, rumination, catastrophizing and blaming others may lead to a lack of confidence in the ability to cope with health problems. In its turn, this might cause a lower HRQL, consistent with previous reports on the reduced activity that results from non-adaptive coping .
Future research must create greater clarity on the relationship between psychological problems and HRQL. For instance, do psychological problems cause lower HRQL, or is it the other way around? If it turns out that such problems have an important effect on the HRQL of young VLBW adults, it might be possible to detect and address such problems early. Physicians may be trained in detecting children with non-adaptive coping styles. Interventions could then be designed to teach these children how to cope adaptively, and thereby to smooth the impact of their handicaps.