Table 2.
Results and bias of 18 selected publications
| Publication: First author (ref) | Main conclusion (based on tables or according to authors in abstract/discussion) | Outcomes by subgroups (sex/disability) and/or subscales | Selective non‐response bias or other biases/remarks |
|---|---|---|---|
| Batsvik, et al 4 |
Inconclusive: Overall, scores for EPT adults were similar to those of controls on the SF36 at age 24, but the authors only show comparisons for EPT with and without disabilities. EPT without disabilities scored significantly lower on 3 of the 8 SF36 scales |
Disability: EPT without disabilities (n = 35) scored lower on social functioning, emotional role, mental health (3 of 8 scales SF36) and psychological complaints. SF36‐scales of EPT with disabilities (n = only 8) were comparable to the scale scores of controls | Nothing was reported on non‐response bias; response rate was high. However, low sample size, especially in EPT born adults with disabilities (n = 8) |
| Baumann, et al 14 |
Lower HRQoL than controls: At age 26, a higher percentage of VPT/VLBW had more severe levels of disability on the HUI3 (self‐report and parent report) |
No subgroups reported. Self‐reported HRQoL was higher than parent‐reported HRQoL. Subscales were reported only with regard to change from 13 to 26 y of age and; parent‐reported HRQoL worsened into adulthood |
Dropout was not random; socially disadvantaged participants dropped out more. SES and sex were included in the multivariate analyses as covariates. Lower HRQoL was related to economic and social functioning problems in adulthood |
| Bjerager, et al 15 |
No difference in HRQoL: Total objective and subjective HRQoL did not differ between the non‐handicapped VLBW adults aged 18‐20 and NBW controls. |
Disability: Adults born with VLBW reported lower objective and subjective HRQoL than controls on the Elementary Biological Needs subscale | Nothing was reported on non‐response bias; authors stated that both groups had comparable dropout rates |
| Björqvist, et al 17 |
No difference in HRQoL: On total 15D score or any of the profile dimensions, HRQoL did not differ between the whole VLBW group and the control group at age 18‐27 |
Sex: Women reported lower a 15D score than men (P 0.001) in all areas except mobility, hearing, eating, speech, usual activities and sexual activities. NSI a : VLBW participants with NSI did not differ on the total 15D score, but had lower scale scores on mobility, vision, sexual activities and eating. SGA vs AGA: the SGA VLBW group reported a significantly lower total 15D score. Compared to controls, AGA VLBWs scored lower on mobility and higher on depression. Compared to controls, SGA VLBWs scored lower on eating; compared to AGA VLBWs, they scored lower on mental functions, depression and vitality |
The authors hypothesised that dropout in participants with NSI might have been high, but emphasised that including the NSI participants in the analyses and reporting separately on this group were strengths of this study>? |
| Cooke 11 |
No difference in HRQoL between VPT/VLBW and controls: HRQoL did not differ between groups except for the Physical Functioning scale of the SF36 at ages 19‐22 |
Sex: Male VPT’s had lower physical functioning and general health perception than male controls. In females, no differences were found on subscales | 50% of the cohort could not be traced, and more females and preterms returned the questionnaires |
| Darlow, et al 3 |
No difference in HRQoL: At age 22‐23, HRQoL did not differ between the VLBW and control groups on all 8 SF‐35 subscales and on the physical and mental component score |
No subgroups were reported. No differences were found on subscales (not reported in article; sent after contact with the authors) | The authors reported that although 71% of the VPT/VLBW cohort had participated, there were no differences in the basic demographic characteristics of those who did and did not participate |
| Dinesen & Greisen 16 |
Lower objective HRQoL: Non‐handicapped VLBWs scored lower on objective HRQoL, but at age 18 age their subjective HRQoL did not differ from that of controls |
Disability: Handicapped VLBW and LBW had lower subjective and objective HRQoL than NBW controls | Five people could not be assessed due to a handicap. There was also a high participation rate in the VLBW group. There was no further report of selective non‐response or other bias |
| Gaddlin, et al 12 |
No difference in HRQoL: At age 20, HRQoL did not differ between VLBW and controls on all 8 SF‐35 subscales and on the physical and mental health scores. |
Disability b : The n = 15 handicapped VLWBs had significantly lower scores on the physical functioning subscale only (P<.001) |
The authors reported that non‐responders did not differ from responders in BW and GA. Univariate and multivariate regression analyses showed an association between physical functioning and some perinatal and neonatal factors |
| Hack, et al 18 |
No difference in poor health profiles: At age 20, similar proportions of VLBW and NBW participants reported excellent, average or poor health profiles on the Chip‐AE |
20 subdomains of Chip‐AE: Although the VLBW group had lower scores on resilience (physical activity and family involvement) and more disorders (acute minor, long‐term medical, long‐term surgical and psychosocial) than controls, they also had better work performance and less risk behaviour. |
A higher IQ in the control group might have led to a higher probability of participation |
| Hallin & Stjernqvist19 |
No difference in HRQoL: At age 18, HRQoL did not differ between groups: 71.7 in EPTs compared to 74.8 in FT controls |
No subgroups reported | High response rate. It was not evident which specific item or items were used to measure HRQoL on a visual scale from 0 to 100 |
| Husby, et al 6 |
Lower HRQoL than controls: VPT/VLBW had lower HRQoL at age 23 (on 6 of 8 SF36 scales and also the physical and mental component scales; no difference after correction for CP and low IQ) |
Subscales: VLBW participants (n = 35) reported significantly lower values on the physical component (physical functioning, role‐physical and bodily pain) and on the mental component scales (social functioning and role‐emotional). Disability: On SF‐36 subscales, VLBWs without cerebral palsy and/or low IQ (n = 25) did not show any significant differences from controls |
Noting the limited sample size, the authors state that because they did not know the reasons for non‐participation, selection bias may have resulted. On perinatal data, however, participants did not differ from non‐participants |
| Lund, et al 5 |
No difference in HRQoL: At age 20, HRQoL did not differ between VLBWs and controls, except on Mental Health on 8 SF36 scales) |
Subscales: VLBWs scored lower than controls on mental health |
In the VLBW group, non‐participants were more often male, no difference on GA, BW and HC. A term SGA group (n = 55) had lower scores than controls on SF36 scales for mental health, social functioning and emotional role |
| Natalucci, et al 10 |
Inconclusive: Lower HRQoL on mental HRQoL but higher HRQoL on physical HRQoL: ELBW’s had lower HRQoL total scores on Mental Component Summary, but higher HRQoL on Physical Component Summary on the SF36 at 22‐26 y, compared to community norms |
Subscales: Compared to community norms, ELBWs had lower HRQoL scores on social functioning, but higher HRQoL scores on bodily pain and general health. Groups at risk: multiple regression analysis showed that females were more at risk for lower mental HRQoL, but ELBW with BPD were less at risk |
The authors reported that the high dropout rate and consequent small size of the group studied may have let to non‐response bias, even if there was no difference between participants and dropouts with regard to perinatal and socio‐demographic variables. It is also unclear where the community norms originated. After contact with the authors, we concluded that there was an overlap with participants in Baumgard et al, where, on all SF36 scales, there were no differences between adults at age 23 with a BW < 1250 g and term controls at the same age. But males did score lower than controls on physical functioning |
| Poole, et al 8 |
No difference in HRQoL: HRQoL did not differ between ELBW and NBW adults at 22‐26 years on all 8 SF36 scales |
Subscales: there were no significant differences on SF35 subscales. Subgroups: no subgroups were reported. In ELBWs, motor coordination was significantly associated with Physical and Social Functioning. This was also the case in the NBW group |
The authors did not report non‐response bias, but there was a low response rate and a higher percentage of females. The authors warn about the broad age range, which made it difficult to base conclusions on a specific ages. |
| Roberts, et al 13 |
No difference in HRQoL: HRQoL did not differ between EPT/ELBW and controls at 18 years (not on overall HUI3 score and only on Physical Functioning of 8 SF36 scales). |
Subscales: there was only one significant difference on physical functioning. Disability: Additional analyses without individuals with a major disability did not alter the results (data not shown). Low GA/BW: Regression analyses revealed no evidence of an association between GA and BW on HUI3 and SF36 physical functioning scale scores. |
Non‐responders had had more problems than responders, such as PVL after birth, and disabilities and lower IQ at age 8 |
| Saigal, et al 7 |
Inconclusive: At age 23, HUI2 mean utility scores did not differ between the ELBW and control groups, but did differ after substitution of missing values. No differences were found on the preferences for hypothetical health states based on the HUI2. |
NSI a : Within ELBW, there were no differences between those with NSI and those without NSI (n = 38). Sex: No effect of sex. Subscales: HUI2 attributes Sensory and Cognition (out of 7 HUI2 attributes) were significantly more affected in the ELWB group than in controls. |
Except for lower maternal education, ELBW non‐responders and ELBW responders had a similar prevalence of NSIs and parental socio‐demographic features. HRQoL decreased over time (from age 12‐16 to age 23), but this happened in both the ELBW group and the control group, and there was no group effect |
| Saigal, et al 9 |
Lower HRQoL than controls: ELBWs had lower HRQoL on the HUI3 both in their 20s and 30s than controls and the HUI3 score decreased more in the ELBW groups (with and without NSI) from their 20s to 30s |
NSI a : The HUI3 score was even worse for ELBW with NSI (n = 37) than for ELBW without NSI and NBW controls, and more HUI3 attributes were affected. A participant's sex or socio‐economic status had no significant impact on HUI3 trajectories. HUI attributes: at age 22‐26, there were differences between VPT/LBW adults and controls on 5 subscales of HUI3 and at 29‐36 y on 3 attributes |
More males and more people whose mother had lower SES or lower maternal education had missing data on one of the measurements |
| Vederhus, et al 20 |
No difference in HRQoL: At age 18, HRQoL did not differ on the Child Health Questionnaire between the EPT/ELBW group and term controls. |
Sex and subscales: EPT/ELBW boys significantly improved on 3 subscales from age 10‐18 y and at 18 y HRQoL on subscales did not differ from controls anymore. While differences between the EPT/ELBW and control girls remained significantly lower in the EPT/ELBW study group on 3 subscales (mental health, physical functioning and general health), compared to controls | The authors reported nothing on non‐response bias, but, due to the small sample sizes, warned about any interpretation of the results |
NSI (neuro‐sensory impairments) definitions. Björqvist, et al 17 : cerebral palsy, developmental disorders and severe visual impairment. Saigal, et al 9 : cerebral palsy, blindness, deafness and microcephaly. (Saigal, et al 7 report no definition of NSI, but it is assumed to be the same as Saigal, et al 9 ).
Definition of handicap by Gaddlin, et al 12 : moderate or severe cerebral palsy, moderate or severe attention deficit hyperactivity disorder or mental retardation (IQ < 70).