Article Text


Two-year neurodevelopmental outcome of preterm born children ≤750 g at birth
  1. M J Claas1,
  2. H W Bruinse1,
  3. C Koopman2,
  4. I C van Haastert2,
  5. L M Peelen3,
  6. L S de Vries2
  1. 1Department of Obstetrics and Gynaecology, University Medical Centre, Wilhelmina Children's Hospital, Utrecht, the Netherlands
  2. 2Department of Neonatology, University Medical Centre, Wilhelmina Children's Hospital, Utrecht, the Netherlands
  3. 3Julius Centre for Health Sciences and Primary Care, University Medical Centre Utrecht, Utrecht, the Netherlands
  1. Correspondence to M J Claas, Department of Obstetrics and Gynaecology, University Medical Centre, Wilhelmina Children's Hospital, KE 04.123.1, PO Box 85090, Utrecht 3508 AB, the Netherlands; m.j.claas{at}


Objectives To describe 2-year neurodevelopmental outcome (NDO) in a cohort of extremely low birthweight infants, and compare NDO between two consecutive 5-year periods and between appropriate (AGA, ≥p10) and small for gestational age (SGA, <p10) infants.

Design Retrospective cohort study.

Setting Wilhelmina Children's Hospital, Utrecht, the Netherlands.

Patients 146 children, born between 1996 and 2005, with a birth weight ≤750 g and a gestational age ≥24 weeks, admitted to the neonatal intensive care unit. 111 children (76%) survived the neonatal period.

Interventions At 2 years corrected age, 101 children (cohort I: born in 1996–2000, n=45 and cohort II: born in 2001–2005, n=56) were assessed with either the Griffiths Mental Developmental Scales or the Mental Scale of the Bayley Scales of Infant Development, second edition.

Main outcome measures NDO, classified as normal (≤−1 Z score ≥0), mildly delayed (>−1 Z score ≤−2) or severely delayed (Z score >−2).

Results 74.3% of the children had a normal NDO at 2 years corrected age, 20.8% a mildly and 5% a severely delayed outcome. Although survival significantly increased with time (65.8% to 88.1%, p=0.002), significantly fewer children in cohort II (66.1% vs 84.4% in cohort I, p=0.042) as well as fewer SGA children (64.3% vs 86.7% of AGA children, p=0.012) had a normal NDO.

Conclusions Increased survival of infants with a birth weight ≤750 g coincided with more children with an impaired NDO at 2 years corrected age. SGA infants are especially at risk of impaired NDO.

Statistics from


Changes in perinatal and neonatal care, such as increased use of prenatal steroids, early assisted ventilation in the delivery room and advanced techniques for mechanical ventilation in combination with surfactant therapy, have resulted in improved survival rates for extremely preterm and extremely low birthweight (ELBW) infants.1,,10 Although it is promising that the survival of ELBW infants has improved, it is well known that survivors are at increased risk of impaired neurodevelopmental outcome (NDO).2 3 11,,15 Studies on the NDO of ELBW children show contradictory results, with the prevalence of cognitive impairment varying between 10.6% and 50%, which either increased, decreased or remained unchanged over time.1,,3 11 13,,18

The objectives of this retrospective cohort study of children with a birth weight ≤750 g were to assess NDO at 2 years corrected age, and to compare NDO between two consecutive 5-year periods of birth and between children who were either appropriate (AGA) or small for gestational age (SGA).

What is already known on this topic

  • Extremely low birthweight (ELBW) infants are at risk of perinatal death and severe neonatal morbidities.

  • ELBW infants are at risk of neurodevelopmental impairment.

  • Studies on the survival and neurodevelopmental impairment of ELBW infants show contradictory results with varying prevalences which either increased, decreased or remained unchanged over time.

What this study adds

  • Data on survival, neonatal morbidity and neurodevelopmental outcome at 2 years corrected age in a cohort of ELBW infants ≤750 g born during a 10-year period.

  • Comparison of survival, neonatal morbidity and neurodevelopmental outcome at 2 years corrected age of two consecutive 5-year periods and of appropriate and small for gestational age children.



The study population consisted of a cohort of 272 infants with a birth weight ≤750 g and a gestational age ≥24 completed weeks, born in 1996 through to 2000 (cohort I) and 2001 through to 2005 (cohort II). Ninety-three (34.2%) infants were intrauterine deaths and 179 (65.8%) were live born infants. A total of 146 infants were admitted to the neonatal intensive care unit (NICU) of Wilhelmina Children's Hospital in Utrecht in the Netherlands (130 infants (89%) were born in our university hospital, while the remaining 16 infants were transported to our NICU after delivery in a regional hospital). Of the 111 survivors, 91% (cohort I, n=45 and cohort II, n=56) were available for follow-up at 2 years corrected age (figure 1).

Figure 1

Cohort of 272 infants with a birth weight ≤750 g born in 1996–2005.

Data collection and definitions

Data were collected by reviewing the medical charts. Gestational age was based on the last menstrual period and early ultrasound examination. Birth weight percentiles were determined according to the Netherlands Perinatal Registry.19 SGA was defined as a birth weight less than the 10th percentile (p10). NICU admission was divided into short-term (≤28 days) or long-term (>28 days) stay. Mechanical ventilation was recorded as short-term (<2 weeks), intermediate (2–4 weeks) or long-term (>4 weeks). Oxygen requirement was recorded. Infant respiratory distress syndrome (IRDS) grades I–IV were defined according to Giedion.20 Bronchopulmonary dysplasia (BPD) was defined as the need for oxygen at 36 weeks postmenstrual age according to Northway.21 Antenatal bethametasone and postnatal hydrocortisone use was registered. Hypotension was defined according to postnatal age specific blood pressure standards and treatment with inotropes was registered. Persistent ductus arteriosus (PDA) was diagnosed clinically and confirmed by cardiac ultrasound. Treatment of PDA with indomethacin or surgery was recorded. Periventricular leukomalacia (PVL) and intraventricular haemorrhage (IVH) were graded according to de Vries et al.22 Septicaemia was defined as clinical signs in combination with a positive blood culture. Necrotising enterocolitis (NEC) was classified according to Bell.23 Surgical treatment of NEC was recorded. Hyperbilirubinaemia was registered as needing phototherapy according to postnatal age specific bilirubin levels.24 Hypothyroidism was diagnosed according to postnatal age specific standards for free T4 and TSH.25 Hypoglycaemia and hyperglycaemia were defined as a plasma glucose of <2.6 and >8.0 mmol/l, respectively.26 Retinopathy of prematurity (ROP) was classified according to the international classification.27 28 Parental educational levels were recorded according to the occupational classification standard of Statistics Netherlands.29 Socio-economic status (SES) was recorded according to the zip code estimated income of The Netherlands Institute for Social Research.30

Neurodevelopmental assessments

For our primary outcome, data on NDO at or near 2 years corrected age were collected. NDO was evaluated using either the Griffiths Mental Developmental Scales (GMDS) or the Bayley Scales of Infant Development, second edition (BSID-II), administered by certified investigators. Both tests are most reliable when performed at or around 24 months corrected age. The GMDS consists of five subscales: locomotion, personal-social, hearing-speech, eye-hand and performance. This test is designed to yield both global (sum of five subscales) and subscale developmental quotients (DQ) with a mean (±SD) DQ score for the general population of 100 (±12).31 Assessment of neurodevelopment with the GMDS in our study is based on four subscales, excluding locomotion.32

The BSID-II consists of a Mental Scale and a Psychomotor Scale; for neurodevelopmental assessment the mental developmental index (MDI) was used, with a mean of 100 (±15).33 In case of a MDI <55, 54 was entered in the dataset.

For calculation of developmental scores, the accurate gestational age in fractions (such as 25 weeks and 1/7 week) was used. These fractions were transformed into decimals by SPSS (resulting in 1 day=0.14 week, etc). Developmental scores were calculated both for chronological age and corrected age. A correction for prematurity was made by subtracting the amount of prematurity (40 minus gestational age at birth) from the actual age at testing.

From December 2000 onwards, all but six children were assessed with the BSID-II due to recent implementation of guidelines of the Dutch national follow-up working group. Z scores were calculated for both GMDS (DQ without locomotion subscale) and BSID-II (MDI) outcomes in order to compare these neurodevelopmental scores.

NDO was classified as normal (≤−1 Z score ≥0), mildly delayed (>−1 Z score ≤−2) and severely delayed (Z score >−2).

Statistical analysis

To check for accuracy, data entered were double checked. All analyses were performed using SPSS v 15.0. Statistical comparisons for continuous variables were made with Mann–Whitney tests. Dichotomous and categorical variables were tested using the χ2 test or Fisher's exact test. Univariate and multivariate analyses for continuous variables were performed by linear regression, and for dichotomous variables logistic regression was used. A p value <0.05 was considered to be statistically significant.


Figure 1 shows a flow chart of the initial study population (n=272). The survival rate of all NICU admissions (n=146) was 76% and increased with time from 65.8% (52/79) in cohort I to 88.1% (59/67) in cohort II (p=0.002). No significant difference in survival was noted between AGA and SGA infants (73.4% (47/64) and 78% (64/82), p=0.561). However, the survival of SGA infants significantly increased with time (61.4% (27/44) to 97.4% (37/38), p<0.001), whereas the survival of AGA infants remained unchanged (71.4% (25/35) to 75.9% (22/29), p=0.780).6 In 101/111 (91%) of the surviving children, NDO was assessed at 2 years corrected age.

Table 1 shows the most relevant characteristics of these 101 infants. A significantly higher birth weight was noted in cohort II (685 vs 649 g, p=0.025). The median gestational age of both cohorts I and II was 28 weeks. As expected, SGA infants had a significantly lower birth weight than AGA infants (635 vs 720 g, p<0.001), whereas a significantly shorter gestational age was found in AGA infants (26.70 vs 28.84 weeks, p<0.001). Maternal education was significantly lower in cohort I (p=0.019). All SGA infants were delivered by caesarean section compared to 60% of AGA infants (p<0.001).

Table 1

Characteristics of 101 surviving children assessed at 2 years corrected age: total cohort, cohort I and II, and AGA and SGA infants

Neonatal morbidity during NICU admission

The majority of the infants required NICU admission for at least 4 weeks (table 2). One in five infants did not need mechanical ventilation. IRDS grade I/II and III/IV were both diagnosed in almost 30% and the majority of the infants with IRDS received surfactant. BPD developed in 56.4%, and over 70% of the BPD cases received hydrocortisone treatment. Hypotension was present in 62.4% and the majority needed treatment with inotropes. PDA was diagnosed in one third, almost 56% received indomethacin and about 15% required surgical closure. Cranial ultrasound showed PVL grade I in 43.6%, IVH grade I/II in 20% and severe intracranial lesions (IVH grade III/IV or cystic-PVL grade II) in 5%. None developed cystic-PVL grade III. Septicaemia occurred in 61.4%. NEC was diagnosed in 8.9%, and the majority required a laparotomy. Furthermore, a high prevalence of hyperbilirubinaemia was noted. ROP (any stage) developed in 46.5%, but more severe ROP (stage III, IV and V) was present in only 5% of the children, none of whom required laser surgery (although one child received a cerclage because of retinal detachment).

Table 2

Neonatal morbidity and interventions during NICU admission

In cohort I a significantly higher prevalence of IRDS grade III/IV (p=0.042), BPD (p=0.002) and mechanical ventilation for >4 weeks (p=0.008) was present, whereas significantly more hyperbilirubinaemia was found in cohort II (p=0.022).

A significantly higher prevalence of IRDS (p=0.033), mechanical ventilation for >4 weeks (p=0.002) and PDA (p=0.019) was noted in AGA infants, as well as treatment with indomethacin (p=0.038).

Neurodevelopmental outcome at 2 years corrected age

Overall, 101 children were assessed by either the GMDS (n=49) or the BSID-II (n=52). The mean age at testing was 23.4 months (SD 1.8 months) corrected age.

In table 3 the mean neurodevelopmental scores and Z scores are presented. No significant differences were noted between the corrected Z scores of the GMDS and BSID-II (p=0.661). (For completeness uncorrected scores are shown in the tables as well.) The total cohort performed within 1 SD below the population mean (Z score −0.37). The scores were not significantly different between cohort I and II (p=0.164), but SGA children scored almost significantly lower compared to AGA children (−0.54 vs −0.15, p=0.050).

Table 3

Neurodevelopmental outcome at 2 years corrected age of 101 children birth weight ≤750 g

Table 4 shows a classification of Z scores into three categories: 74.3% of the children had a normal NDO, 20.8% a mildly delayed and 5% a severely delayed outcome. Significantly more children in cohort I performed within normal limits (84.4% vs 66.1%, p=0.042), fewer had a mildly delayed (15.6% vs 25%) outcome and none a severely delayed performance compared to 8.9% in cohort II.

Table 4

Neurodevelopmental outcome of total cohort, cohorts I and II, and AGA and SGA infants

The difference in NDO between cohort I and II slightly attenuated after adjustment for potential confounding by the variables ventilation for >4 weeks, IRDS grade III/IV, BPD, hydrocortisone, oxygen and hyperbilirubinaemia (p=0.040 for linear analysis, p=0.123 for logistic analysis, p=0.055 using propensity score).

A significantly poorer outcome was found in SGA children (p=0.020). Significantly fewer SGA children performed within normal limits (64.3% vs 86.7%, p=0.012).

No significant difference in NDO was found between AGA children in cohort I and cohort II, nor between SGA children in cohort I and II (p=0.906 and p=0.129, respectively; table 5).

Table 5

Outcome of AGA and SGA infants between the two cohorts


Survival and neonatal morbidity

Of the initial cohort of infants in our study with a birth weight ≤750 g, 40.8% (111/272) survived, whereas the survival rate of the infants who were admitted to the NICU was 76% (111/146). Survival rate improved and birth weight increased significantly over the two time periods studied. Although the increasing survival rate is promising, considerable neonatal morbidity persisted. The majority of infants required long-term NICU admission and mechanical ventilation, developed BPD and received hydrocortisone, mostly to wean them off the ventilator. Mild intracranial lesions (IVH grade I/II and PVL grade I) were diagnosed in over 65%, but severe intracranial lesions (IVH grade III/IV or PVL grade II) were found in only 5%. The low prevalence of severe brain injury in our NICU has also been confirmed in a recent study by Groenendaal et al who showed that IVH grade III and IV were both diagnosed in 5.7% and cystic PVL was detected in 1.6% of infants (gestational age 25–30 weeks) born between 2002 and 2006. The prevalence of IVH remained unchanged, but cystic PVL significantly decreased over time. Increased antenatal use of antibiotics and nasal continuous positive airway pressure (CPAP) and less mechanical ventilation are suggested as two possible explanations for the decreasing prevalence of cystic PVL.34 However, these explanations could not be confirmed in our data from a subgroup (≤750 g) of this population.

Intensive care was withdrawn from only five of the 146 infants who had been admitted to our NICU during the study period. These five infants had severe cerebral lesions (such as IVH grade III with severe acute ventricular dilatation of the lateral ventricles or a large unilateral grade IV with a midline shift or bilateral grade IV haemorrhage).

Improvement was noted with respect to mechanical ventilation, mainly because ventilation lasting for >4 weeks was significantly less often required in cohort II. The significantly lower prevalence of IRDS grade III/IV and BPD is a plausible explanation for the decreased need for prolonged mechanical ventilation and oxygen supply, and increased use of continuous positive airway pressure (introduced to the NICU in August 2000) in this cohort. However, the decreased prevalence of IRDS cannot be explained: gestational age was similar in the two cohorts and no differences in administration of prenatal steroids and surfactant were found, whereas the decreased prevalence of BPD probably resulted from the reduced need for mechanical ventilation in cohort II.

It is also difficult to explain the increased prevalence of hyperbilirubinaemia in cohort II as gestational age was not significantly different between the two cohorts and there was no difference in the prevalence of IVH.

The significantly shorter gestational age of AGA infants probably resulted in significantly more IRDS, mechanical ventilation lasting for >4 weeks, PDA and indomethacin administration.

According to our results and the results of others, it is clear that ELBW infants are prone to severe neonatal morbidities.6 35 36 However, the number of infants with severe respiratory problems who required mechanical ventilation for >4 weeks decreased with time.

Neurodevelopmental outcome

Neurodevelopmental outcome in total cohort

Of the 101/111 (91%) children who were assessed at 2 years corrected age, the majority had a normal NDO, 20.8% a mildly and 5% a severely delayed outcome.

Our NDO data are partly in agreement with data presented by others. A review of Lorenz et al reported a MDI of <70 in approximately 14.3% of the infants with a birth weight ≤800 g.2 However, the percentage of ELBW children with a MDI of <70 assessed with the BSID (at 18–24 months of age) ranged from 10.6% to 50%. Hack et al and Casiro et al found a MDI of <70 in 20% and 23%, respectively, in infants weighing 500–750 g at birth.1 16 A review of Hack et al demonstrated a MDI of <70 in 13–47% of children with a birth weight of <750–800 g.18 Hack et al found a MDI of <70 at 20 months corrected age in 42% of children with a birth weight of <1000 g.11 In our study population, none of the 49 children assessed with the GMDS had a DQ of <76 (>−2 SD), whereas 5/52 (9.6%) children assessed with the BSID had a MDI of <70.

The lower prevalence of severe developmental delay in our study population compared to other studies most likely results from the guidelines in the Netherlands stating intensive care should be provided from 25 weeks gestational age onwards. The prevalence of severe brain injury was also low in our population. Intensive care was withdrawn in five infants because of severe cerebral lesions as stated above and this may also have affected the eventual number of infants with severe developmental delay.

Neurodevelopmental outcome over time

NDO was significantly worse in cohort II in comparison with cohort I. In view of the reduced need for ventilation in cohort II, this was an unexpected finding. According to the higher birth weight and the analysis of neonatal morbidity, showing respiratory problems to be less common, we would have expected to find an improved outcome for cohort II. However, the poorer NDO of cohort II may be explained by the higher prevalence of hyperbilirubinaemia. Oh et al showed that serum bilirubin level directly correlated with impaired NDO.37

We furthermore speculate that an explanation for the lower NDO in cohort II might be due to differences in ethnic background, parental educational level, SES, or obstetric and neonatal care. However, there was no difference in ethnic background between cohort I and II. According to Gross et al and Weisglas-Kuperus et al, parental educational level, especially of the mother, is an important indicator for NDO.38 39 In the present study, maternal and paternal educational level was only recorded in 87.1% and 79.2%, respectively, of the children. Overall, 23.9% of the mothers indicated they were housewives and were not included in the analysis. Significantly lower maternal education was found in cohort I (p=0.019), whereas paternal educational level was not significantly different between the two cohorts (p=0.381). The lower NDO in cohort II therefore could not be explained by parental educational level; however, a considerable amount of data were missing.

Hille et al found a sevenfold increase in the need for special education at 9 years of age among children with a low SES, but in our cohort no significant difference in SES between cohort I and II was found (p=0.515).40

Schmidt et al reported that three neonatal morbidities (BPD, brain injury and severe ROP stage IV and V) strongly predict the risk of neurodisability at 18 months. Since our prevalence of severe ROP is very low at 1%, we feel that it is not possible to draw conclusions on the predictive value of the presence of severe ROP for neurodisability at 2 years of age in our cohort.41

Furthermore, no important differences in perinatal and neonatal care (such as antenatal steroids, number of caesarean sections, and use of surfactant, postnatal steroids and high frequency oscillation) were shown in our data. Nevertheless, due to more active measures in obstetric and neonatal care, more severely compromised infants can be kept alive, which may have resulted in a protracted neonatal course with a higher prevalence of neurodevelopmental impairment. Thus, in our population of ELBW infants, an increased survival rate was not accompanied by an improvement in NDO.

Advances in developmental assessments resulted in the use of two different tests. From the literature and experience, we know that some children perform better on the GMDS, because prolonged attention is required for the BSID-II.33 Use of the BSID-II in the majority of cohort II infants could partly explain the poorer NDO. However, a comparison of the BSID-II and GMDS Z scores showed no significant difference (table 3). Furthermore, a pilot study of 29 children with a birth weight of <1000 g or a gestational age of <30 weeks, assessed with both the GMDS and BSID at 2 years corrected age confirmed no significant difference between the two tests (p=0.287, unpublished data).

Adverse NDO in ELBW infants is also reported by others. Hack et al demonstrated similar survival rates and a significant increase in the percentage of ELBW infants (birth weight of <750 g) with a MDI of <70 at 20 months corrected age (20% in 1990–1992 to 48% in 1993–1995, p<0.02).3 Wilson-Costello et al also showed an increase in the number of children (birth weight 500–749 g) with a MDI of <70 at 20 months corrected age (28% in 1982–1989 to 34% in 1990–1998, p=0.529) and significantly increased survival (27% to 48%, p<0.001).13

In contrast to these data, Vohr et al showed significantly improved survival rates and a significantly decreased rate of MDI of <70 at 18 months corrected age (birth weight 401–1000 g, 29.9% in 1993–1994 to 25.5% in 1995–1996 and 22.8% in 1997–1998, p<0.01).14 Also a recent study by Wilson-Costello et al demonstrated a significant decrease in the number of children (birth weight 500–999 g) with a MDI of <70 at 20 months corrected age (35% in 1990–1998 to 23% in 2000–2002, p=0.01).15

In accordance with other reports on NDO in preterm infants, we have also presented corrected scores. A correction for gestational age is used, as it is implied that correction for prematurity assists in differentiating between developmental delay associated with prematurity from that caused by brain damage, and results in comparability of the NDO of preterm and full-term infants.1,,3 11,,18 42,,44 However, full correction for prematurity results in a more favourable outcome, as has been noted in our ELBW cohort as well. In our opinion and that of others (Miller et al, den Ouden et al and Barrera et a)42,,44) corrected scores overestimate NDO in preterm infants and uncorrected scores better estimate NDO at 2 years of age and more reliably predict the future outcome of these children. Moreover, the degree of prematurity of infants admitted to the NICU has increased due to improvements in perinatal and neonatal care. This increased degree of prematurity requiring correction over a wide gestational range may have resulted in reduced reliability of corrected scores. A poorer NDO for cohort II is shown when using corrected scores. However, uncorrected NDO scores were lower, but not significantly different between the two cohorts.

Neurodevelopmental outcome in AGA and SGA children

Comparison of AGA and SGA children in our cohort showed a similar survival rate, but significantly more SGA children appeared to have an impaired NDO. An expected finding is the significantly longer gestational age of SGA infants. Previous reports state that gestational age has a major influence on survival and it is universally agreed that long-term morbidity increases with decreasing gestational age.11,,14 However, the NDO of our SGA children does not seem to be positively affected by their greater gestational age. The significantly higher percentage of males among SGA children could possibly explain their poorer NDO, as other studies support a better outcome for female children.45 46

As male gender was significantly over-represented among SGA children, gender was tested in a multivariate analysis, but it was not predictive for adverse NDO, nor was delivery by caesarean section.

Furthermore, SES and parental educational level did not differ between SGA and AGA infants. The most plausible explanation is that brain development was adversely affected in these severely growth restricted ELBW infants by chronic intrauterine malnutrition.

There are only a few recent reports comparing NDO at 2 years corrected age between AGA and SGA ELBW children. Hack et al and Procianoy et al found similar BSID-II outcomes for AGA and SGA children (birth weight <1500 g and <p10).18 47 Gortner et al also found similar GMDS outcomes (SGA <p10),48 whereas Anderson et al found SGA infants (birth weight <1000 g and <p3) had more cognitive, educational and behavioural impairments.49 Also Sung et al demonstrated a poorer performance on the BSID by SGA children (birth weight <p10).50 Stoelhorst et al found higher anxious/depressed and/or withdrawn behaviour at 2 years of age in SGA children (birth weight <p10).51 Thus, reports on neurodevelopmental performance comparing AGA and SGA children show contradictory results. Moreover, varying definitions of SGA and differences in neurodevelopmental assessment policy make it difficult to compare our data with other studies.

Limitations of this study include the fact that this was a retrospective analysis. Nevertheless, we were able to see 91% of the children, born during a 10-year study period, for a neurodevelopmental assessment at 2 years corrected age. All 10 children who were lost to follow-up for the neurodevelopmental assessment in our hospital, survived until 2 years of age. These children were all Caucasian singletons, 40% male and 80% SGA, and their neonatal morbidity was comparable to that of the children who were available for follow-up. Interviews with the parents revealed that the main reason for being lost to follow-up was their preference for visiting a local paediatrician. We have no reason to suspect that these 10 children would significantly alter our NDO results.

In conclusion, children with a birth weight ≤750 g remain at risk for serious neonatal morbidity and adverse NDO. Although survival of these infants increased with time, and the degree of respiratory problems decreased, this was not associated with a better NDO, especially not in SGA children. Long-term follow-up of ELBW infants is essential as they are known to grow into their deficits.


The authors thank H Brouwers and JUM Termote, neonatologists in Wilhelmina Children's Hospital of Utrecht, for their participation in collecting research data.


View Abstract


  • Competing interest None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.