In the last years, the survival rates following the birth of extremely low birth weight (ELBW; < 1,000 g) infants have increased (Arpino et al., 2010), although the risk of neonatal morbidities has also increased with decreasing gestational age and very low birth weight (Manuck et al., 2016; Mukherjee et al., 2023; Platt, 2014). Different health difficulties are more frequent in these children than in term-born children (Larroque et al., 2008; Marsh et al., 2023; Taylor and O’Shea, 2022; Chung et al., 2020). Extremely preterm infants are at risk of deficits with potential consequences in functionality regarding general health. Nevertheless, the impact of the unhealthy condition on the well-being of these children and their families has been poorly described (Vederhus et al., 2010; Shaw et al., 2023).

This effect can be studied on health-related quality of life (HRQoL), which is a dynamic, multidimensional, and subjective concept, associated with health status in physical, emotional, and social functioning (Hays and Reeve, 2008). HRQoL assessments provide valuable additional information to traditionally reported medical evaluations (Saigal and Doyle, 2008). Some HRQoL assessments following individuals born preterm or with low birth weight and their families have applied multi-dimensional health profile measures, including physical, psychological, and social well-being (Pal et al., 2020; Petrou et al., 2020).

Research on quality of life during the school years of very preterm or very low birth weight (VLBW) children is scarce. Two separate systematic reviews about HRQoL in preterm children include only one study each at school age (Vieira and Linhares, 2016; Zwicker and Harris, 2008). The first study (Saigal et al., 1994a; Saigal et al., 1994b) compared the HRQoL of extremely low birth weight children (ELBW) with that of a reference group, showing that the global long-term burden experienced by children with ELBW was higher, with the latter presenting more functional limitations. In the second study, the health score of ELBW children was similar to that of normal birth weight children, while parents reported significantly poorer health for their ELBW children (Hack et al., 2011).

In addition to the studies included in the systematic reviews, Peart et al. (2021) studied HRQoL in a sample of 8-year-old extremely preterm children compared with three cohorts of term-born children with different birth years. In this study, parent-reported HRQoL was lower for children born extremely preterm with respect to all control groups (Huhtala et al., 2016) concluded that HRQoL was not significantly different for healthy VLBW children at 8 years of age when compared to a control group.

Few studies have examined factors related to a better or worse HRQoL for these children during their school years. Studies included in Mottram and Holt (2010) found that preschool preterm children with lower GA had lower scores in HRQoL. Other studies have found no differences in HRQoL related to GA in preschool children (Schiariti et al., 2007), or in birth weight in healthy preterm VLBW children at school age (Huhtala et al., 2016) compared with full-term children.

There is little evidence of the mechanism through which the effects of neonatal morbidities are transmitted to HRQoL in preterm children. One study found that, among children born extremely preterm, perinatal, and neonatal morbidity was mostly unrelated to HRQoL (Vederhus et al., 2010). Recent studies have clarified the mediating effects of bronchopulmonary dysplasia (BPD) and severe non-respiratory morbidity between GA and HRQoL (Kim et al., 2023). More specifically, the mediation analysis revealed that BPD and severe non-respiratory morbidity contributed to significant reductions in HRQoL, with indirect effects translating into HRQoL score decrements for both at the gestational ages studied (<26, 26–27, and 28–29 weeks). The effect was more detrimental when the GA was lower. In this case, as the authors indicate, the worse HRQoL score is mainly due to the complications of preterm birth rather than the preterm birth itself.

Furthermore, studies about families of children with neurocognitive developmental disorders and disabilities (NDD) have highlighted a clear relationship between children’s quality of life, parents’ quality of life, and family environment, concluding that the importance of family well-being in the well-being of the child should be recognized (Puka et al., 2020). The similarity in parental quality of life among parents of children with different NDD indicates that quality of life is influenced by parental and family characteristics rather than by the disease or specific factors (Hatzmann et al., 2008; Puka et al., 2018). To the best of our knowledge, little is known about how family variables influence the quality of life of children born preterm at school age. Much more is known about maternal stress before or immediately after delivery (Hoffman et al., 2016). Another study highlighted the importance of social support as a protective factor for perinatal depression in the 2 years following postpartum and adverse child development (Milgrom et al., 2019). Socioeconomic level, ethnicity, family conflicts, maternal distress and maternal education are some risk factors identified for development in studies about children born prematurely (Moreira et al., 2014). There is greater knowledge of how family variables behave when children are newborns or toddlers, but not how they behave in the long-term during school age. It remains unclear how these variables act on each other as these children grow up, as well as the real impact on their quality of life, knowing their true mediating role.

Therefore, we aimed to (1) study HRQoL in a sample of children born preterm with VLBW at 8 years of age, (2) examine the relationship between medical and socio-family factors and HRQoL, and (3) explore the mediation effects of maternal factors between perinatal variables, socio-family characteristics and HRQoL.

We proposed a theoretical model, testing a direct effect of neonatal medical and socio-family variables, as well as a mediating effect of maternal variables, on the HRQoL of preterm children (see Figure 1). This model draws on other studies (Hatzmann et al., 2009; Isa et al., 2016; Wang et al., 2022), although they did not refer specifically to preterm samples at school age. According to the model, risk factors related to children’s perinatal status and socio-family situation have direct negative effects on their HRQoL. Additionally, parental stress and perceived support would play a mediating role in the association between risk factors and quality of life. This model can be applied not only to the general quality of life, but also to its subdimensions. Within this model, partial models will be considered if the theoretically proposed model is not confirmed.

We tested the following hypotheses: (1) children in the study will show lower HRQoL than the normed mean, (2) there will be a negative association between neonatal medical and socio-family risk and the level of HRQoL, and (3) this relationship would be mediated by the level of stress and the support perceived by the mothers.

A total of 147 VLBW children were followed up until they reached 8 years of age. Table 3 presents descriptive results of KIDSCREEN-27 and KIDSCREEN-10 scores. The means of our sample were higher than the KIDSCREEN normed mean (M = 50, p < 0.001). Cohen’s d ranged from 0.506 to 0.898, indicating medium-to-high effect sizes (see Table 3).

Correlations between the variables in the study are shown in Table 4. First of all, the scores of self-reported KIDSCREEN-10 showed positive correlations with the scores of parent-reported KIDSCREEN-27, being significant only with the psychological well-being and social support and peers subdimensions. Moreover, non-significant correlations were found between KIDSCREEN-10 and the other variables, except the negative correlation with maternal stress. NMRI did not correlate with KIDSCREEN-27 total scores, although it did correlate with its physical well-being subdimension. Total KIDSCREEN-27 score and the score of some of its subdimensions were also negatively correlated with SFRI and maternal stress. Maternal perceived support correlated negatively with KIDSCREEN-27. SFRI correlated significantly and negatively with functional social support.

Analyses of the initial theoretical model (Figure 1) did not show the expected results. Some relationships between variables were not significant, as was the total effect of the model (F = 1.69, p > 0.05, R² = 0.12). Based on these results, two partial models were tested with the two risk indexes. The first partial model (with the neonatal medical risk index) did not show significant direct neither indirect effects between most variables (NMRI with social support, p = 0.70; NMRI with maternal stress, p = 0.61, NMRI with HRQoL, p = 0.08 and social support with HRQoL, p = 0.75). The second partial model (with the socio-family risk index) was validated, and it is graphically presented in Figure 3.

Table 5 and Figure 3 show the results of the mediation analysis with the direct and indirect effects. It is shown in Figure 3 that there was a significant influence of SFRI and functional social support (a₁ = −0.71, p = 0.018). Maternal stress scores were explained by functional social support (d = −0.31, p < 0.001), but not by SFRI (a₂ = −0.08, p = 0.762). Furthermore, SFRI (c’ = −0.80, p = 0.036) and maternal stress (b₂ = −0.39, p = 0.001) exerted a statistically significant influence on the dependent variable of this model (parent-reported HRQoL). However, this result was not significant for functional social support (b₁ = 0.04, p = 0.733). Finally, regarding the total model effect, there was a significant influence of SFRI on HRQoL (c = −0.88, p = 0.023).

Three possible indirect effects were established in this model (Table 5). The first indirect effect did not explain the relationship between SFRI and HRQoL through social support. The results of the second indirect effect were also non-significant, thus SFRI had no effect on HRQoL through maternal stress. Only the third indirect effect was significant, in which SFRI was related to HRQoL through the indirect effects of the two mediating variables: functional social support and maternal stress.

This study aimed to examine the HRQoL of children born preterm with VLBW at 8 years of age and to explore the mediation effects of maternal variables on the relationship between NMRI, SFRI, and HRQoL. This study adds evidence to the existing results in the preterm child population and illustrates relationships between variables that affect children born preterm and their families.

Contrary to our first hypothesis, we found that self-and parent-reported HRQoL of school-aged VLBW children was better than that of the population norm of KIDSCREEN (Ravens-Sieberer et al., 2007). These findings are in disagreement with those of previous studies; however, few works have examined HRQoL in school-aged VLBW children (Peart et al., 2021; Stahlmann et al., 2009; Vederhus et al., 2010). Huhtala et al. (2016) found that healthy VLBW children at 8 years of age did not differ significantly in HRQoL from a control group. Moreover, Hack et al. (2011) found that self-reported health results ELBW children were comparable to those of normal birth weight children. To explain this result, possible hypotheses to be tested in the future would be the existence of greater family involvement from the beginning due to the condition of their children or the fact that they were attending the early care service (greater professional intervention from the beginning). In fact, a large proportion of the participants in this study received care in early intervention services (physiotherapy, psychology and speech therapy), with two thirds of the sample being discharged between the ages of 0 and 3 years old (Lacalle et al., 2023). It has been found that the higher maternal sensitivity is explained by being in early childhood care (Sánchez-Sandoval and Jiménez-Luque, 2023).

The second hypothesis, related to the negative association of NMRI and SFRI with HRQoL, was only partially confirmed. The correlations between variables indicated that they were related to each other before proposing the predictive models. Regarding children’s self-perception, this was barely associated with SFRI and NMRI, but association was significant in the case of parents’ perception. Thus, the predictive models were carried out on parent-reported HRQoL. Although the use of the SFRI and NMRI is limited in previous literature, some studies yield results that align with those obtained in this work. For example, the systematic review by Jardine et al. (2014) shows that the self-perceptions of children with various medical conditions (including some considered in the construction of the NMRI) are similar to those of their reference group. In contrast, from the families’ perspective, the health-related quality of life of these children tends to be rated lower (Cremeens et al., 2006; Jardine et al., 2014). According to these authors, those differences may be related to many variables, including family-related factors.

With regard to perinatal variables, in contrast to other studies (Vederhus et al., 2010), our preliminary correlation analysis showed that GA did not correlate with HRQoL outcomes. Birth weight did not correlate either, but inconclusive results were found in Vieira and Linhares’ (2016) systematic review, where some studies found correlations and others did not. Finally, the cumulative NMRI did not correlate with the total HRQoL score, only weakly with the subdimension of physical well-being at school age. This is consistent with results found in another study, where VLBW children’s neonatal morbidity was mostly unrelated to HRQoL (Vederhus et al., 2010).

On the other hand, this study confirmed that higher socio-family risk correlates with lower HRQoL, both in their total score and in the physical, psychological, and school well-being subdimensions. Other studies have also shown that different variables included in our SFRI (parental education status, higher socioeconomic deprivation) were negatively associated with children’s HRQoL (Berbis et al., 2012; Chien et al., 2006; Liu, Harding, and for the PIANO Study Team, 2021). The final model demonstrated a direct and negative effect of SFRI on HRQoL. The impact of negative social circumstances on children’s quality of life has also been highlighted in studies of children born at term (Kim et al., 2018; Von Rueden et al., 2006).

The third hypothesis stated that two maternal variables, i.e., maternal stress and perceived social support, could influence the relationship between these risk factors and HRQoL. Maternal stress and social support, which are considered risk and protective factors, respectively, did correlate with different dimensions of children’s HRQoL. Treyvaud (2014) emphasized the importance of social and parental factors in influencing child development. The focus on parental role stress was based on research suggesting a negative impact of high parenting stress on VLBW child outcomes (Huhtala et al., 2012). Although the initially proposed model could not be fully validated, a partial model could be confirmed.

The results showed evidence of the mediating effects of functional social support and maternal stress. A higher SFRI was related to lower scores in functional social support, low perceived social support was associated with higher levels of maternal stress, and finally, high levels of maternal stress involve lower perception of HRQoL in their children. Parents’ social support did not find a significant direct association with HRQoL, as was the case of SFRI with parental stress. The effect of social support on HRQoL is mediated by maternal stress. In a study by Howe et al. (2014), mothers reported high levels of stress, as well as high levels of social isolation and little support from their partners.

It should be noted that the variables included in this study did not correlate with the variables most closely related to preterm births (gestational age and birth weight), and that no statistically significant partial model was found for NMRI. This suggests that, the older the children born preterm, the lower the impact of NMRI on quality of life, and the more important SFRI becomes. These mediating effects need to be studied further, in larger populations and at different ages, to confirm this new hypothesis and establish it with certainty. Vieira and Linhares (2016) suggested in their systematic review that there was a need to identify predictive models that assess mediating and moderating effects on HRQoL outcomes with more challenging statistical analyses in people born preterm. Kim et al. (2023) confirmed evidence of neonatal morbidity effects, although they stated that mediation analyses are limited and only considered two types of morbidities. To the best of our knowledge, there are studies investigating risk and protective factors in children born preterm and their relationship with children’s developmental outcomes, but not the mediating effects of family variables on HRQoL outcomes.

This study provides evidence of how some of these family variables mediate the HRQoL of VLBW children. Even though the children in our sample reported, on average, good HRQoL, the results may be important for targeting interventions to improve the quality of life of preterm children scoring below average. It is difficult to intervene in variables intrinsic to the individual (gestational age, birth weight, medical variables, or socio-family factors) or in the quality of life directly. On the other hand, achieving good practice in intervention on mediating variables that positively or negatively affect children’s HRQoL, such as parental stress (Martínez-Shaw and Sánchez-Sandoval, 2023) or social support (Guralnick, 2012), would significantly improve the lives of these children and their family context.

One of the strengths of this study include both self-and parent-reports of HRQoL. Systematic reviews by Eiser and Morse (2001) and Zwicker and Harris (2008) highlight the importance of obtaining information on HRQoL from various sources whenever possible. This work follows these recommendations, having assessed the HRQoL from both perspectives (children and parents). Another strength is the opportunity to shed light on poorly studied variables in the VLBW population and their relationship with family variables. Risk and protective factors for the adequate development of these children have been studied, but not as much as the influence or weight of certain variables on HRQoL. Greater knowledge about medical, psychological, and family variables and their relationship and influence on quality of life will allow developing better intervention programs by combining efforts on what is important.

The limitations include the small sample size, which may have reduced our statistical power. However, it is close to the recommendations made by (Sim et al., 2022) (n = 160). The neonatal medical risk index is based on a composite score and does not take into account the impact of each chronic condition on HRQoL, and these results should be interpreted with caution. The effect of bronchopulmonary dysplasia (Kim et al., 2023) is being studied, but would need to be studied with each of the medical conditions of prematurity. The measures used in this study were self-reported. The use of these type of measures could introduce different sources of bias in the results, such as social desirability, although these instruments have been widely used in research. One reason for the loss of some items could be the size and duration of the global questionnaire for parents. Although the children in this sample were, on average, 8 years old and sufficiently able to express their thoughts and emotions (Kim et al., 2023), some may have found it difficult to understand some items due to their conditions. This is the reason why different versions of the KIDSCREEN were chosen to minimize participant fatigue. This made comparisons between them statistically unfeasible. This study used a prospective two-wave design. Longitudinal designs with only two data collection points may overlook certain temporal changes, whereas multi-wave longitudinal studies with at least three data points provide a more robust approach to capturing changes over time. With only two data points, this mediation analysis cannot confirm temporal order or account for potential confounding variables that may influence outcomes between these two time points, reducing its robustness for a full temporal mediation. Data collection was conducted during the COVID-19 pandemic, which made it difficult to access participants and some families refused to participate. Demographic characteristics of non-participating families are not available, so comparisons with participating families could not be studied. Furthermore, loss to follow-up at 8 years may limit the generalizability of the results, although the distribution of sex, gestational age and birth weight did not differ between participants and non-participants. Moreover, it would be interesting to include participants from other hospitals in Spain and even make comparisons with preterm children and their families from other countries.

It would have been valuable to have a control group of full-term children. Our results may be due to the bias introduced by not having said control group. It should be considered that test norms may be outdated (Wolke and Sohne, 1997), thus the quality of life of these preterm children may be overestimated if there is no contemporary comparison group. This would increase the risk of misidentifying fewer children and families in need of support. Therefore, the argument that a control group in the same geographical area would be needed is reinforced for a future line of research.

This study provides evidence on the HRQoL of VLBW preterm school-aged children, which has been understudied in the literature. Although prematurity is often presented as a major risk factor for child development, this study shows a favorable picture for these children during middle childhood. On average, these children and their parents reported better HRQoL than the mean reference value of the KIDSCREEN scales. Probably, medical advances, as well as the involvement of parents and professionals during the first years of life of these children, have been fundamental in overcoming some initial difficulties. In fact, the possible variability in HRQoL is not explained by variables related to extreme prematurity (gestational age, birth weight) or increased neonatal medical risk. Other circumstances common to other families come into play, such as socio-familial factors and the role of the mothers. Our analyses corroborate that when socio-familial circumstances are more adverse, quality of life in childhood suffers. In addition to this direct effect, the socio-familial circumstances of risk weaken the social support that mothers perceive, thereby escalating the stress with which they face their motherhood, also negatively affecting the quality of life of their children in this way. These findings underline the need to implement social policies that support parenthood, especially in adverse social circumstances; and in families with children with higher medical needs where maternal stress has been found to be higher.