Deterioration of exercise capacity after neonatal extracorporeal membrane oxygenation

RESULTS

Between January 1992 and July 2004, 240 neonates received ECMO support within 28 days after birth at the Erasmus MC–Sophia Children’s Hospital. 68 (28%) of them died before age 5 yrs. 59% of those children were born with CDH. In addition, five children who received ECMO support elsewhere were included in our follow-up programme. 32 children were lost to follow-up due to refusal to participate (n=16), not being traceable (n=10) or follow-up elsewhere (n=10). Their baseline characteristics did not differ from those who were included in the final analysis.

145 children participated in our follow-up programme (82% of all survivors). 14 children did not perform the exercise test because of neurological problems, such as hemiplegia and seizure disorder (n=10); chromosomal disorder with mental retardation (n=2); or behavioural problems (n=2). These 14 children had been ventilated longer than those who were included in the analysis (Mann–Whitney U-test p=0.027).

Thus 131 children performed the Bruce treadmill test (214 measurements). We excluded 23 measurements because maximal performance could not be achieved due to balance problems (n=5) and other reasons such as fear, pain in the legs and concentration problems (n=18). The 11 children who did not perform any maximal exercise test had longer oxygen support and more severe CLD than the 120 participants (Mann–Whitney U-test p=0.007 and p=0.012, respectively). The final analysis included 191 exercise tests performed by 120 children (68% of all survivors) (fig. 1). The primary diagnoses of these 120 children were MAS (n=66), CDH (n=18) and other diagnoses (n=36). The diagnoses of the children in the “other diagnoses” group were: persistent pulmonary hypertension of the newborn (n=20); sepsis (n=7); pneumonia (n=6); cardiorespiratory problems (n=2); and congenital cystic adenomatoid malformation of the lung (n=1). The perinatal characteristics and ECMO treatment characteristics of all survivors are presented in table 1. Inhaler medication was used by 3.4% of the 5-yr-olds, 10.4% of the 8-yr-olds and 5.9% of the 12-yr-olds.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1–

Flowchart of infants included in the follow-up programme. ECMO: extracorporeal membrane oxygenation; CDH: congenital diaphragmatic hernia; MAS: meconium aspiration syndrome.

At time of follow-up, the mean sd-score for FEV₁ was significantly below zero at ages 5, 8 and 12 yrs. This also applies to mean sd-score for FEV₁/FVC at ages 8 and 12 yrs. The mean change in FEV₁ from baseline was 8.6%, 6.3%, and 7.2% at 5, 8, and 12 yrs, respectively. We observed significant reversibility of airflow obstruction (i.e. a change in FEV₁ >11% [22]) in 13, 12 and six children aged 5, 8 and 12 yrs, respectively. The characteristics at time of follow-up of the children of the study group are presented in table 2.

Exercise capacity

At ages 5, 8 and 12 yrs, ANOVA resulted in mean±se sd-score endurance time on the Bruce treadmill protocol of -0.5±0.1, -1.1±0.1 and -1.5±0.2, respectively, all significantly less than 0 (all p<0.001). Standard deviations of the mean sd-score endurance time at the three measurement points were 1.06, 1.18 and 1.18, respectively. The mean difference between sd-score endurance time at ages 8 and 5 yrs was -0.5 (95% CI -0.9– -0.2); that between ages 12 and 8 yrs was -0.4 (95% CI -0.8–0.0). The sd-score was abnormally low (< -1.96) in six (7%) out of 90 measurements at age 5 yrs, in 10 (15%) out of 67 measurements at age 8 yrs and in 12 (35%) out of 34 measurements at age 12 yrs. At the end of the test, heart rate was reliably recorded in 156 (82%) measurements and the 10th and 90th percentiles were 168 and 200 beats·min⁻¹, respectively. Technical problems precluded reliable recording of heart rate in the final stage of the test for the other 35 measurements, but, based on heart rate in the pre-final stage or loss of coordination, we considered performance in those measurements to be maximal. In five out of 156 measurements, there was a decrease in transcutaneous oxygen saturation of ≥5% from baseline (two out of 29 measurements in children with CDH and three out of 107 measurements in children with MAS). None of the children had transcutaneous oxygen saturation <90%.

Figure 2 shows the longitudinal evaluation of the sd-score for endurance time at ages of 5, 8 and 12 yrs for the total group and for the different initial diagnoses. 62 children underwent at least two measurements at different ages. All together, there were 80 paired measurements. 34 of these paired measurements showed a deterioration of >1 in sd-score. In 41 paired measurements, the change in sd-score ranged between -1 and +1 (suggesting no significant change in exercise capacity). Improvement in sd-score of ≥1 was observed five times. The mean outcome at age 5 yrs was higher than at ages 8 and 12 yrs (both p≤0.001), with a marginal difference between ages 8 and 12 yrs (p=0.050). The underlying diagnosis had no significant influence. Further analysis using ANOVA did not show significant relationships with: time on ECMO; duration of ventilatory support before start of ECMO; total duration of ventilatory support; oxygen support after decannulation; prevalence of CLD; sd-score for weight at follow-up; sd-score for height at follow-up; sd-score for BMI at follow-up; sd-score for FEV₁; sd-score for FEV₁/FVC; and sports participation (data not shown). The levels of exercise capacity as estimated by the parents were positively correlated with the measured endurance sd-scores (p=0.002).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2–

Exercise capacity at ages 5, 8 and 12 yrs. Data are presented as ANOVA estimates of mean values, with 95% confidence intervals. sd-score: standard deviation score; CDH: congenital diaphragmatic hernia; MAS: meconium aspiration syndrome.

DISCUSSION

We evaluated exercise capacity at ages 5, 8 and 12 yrs in children who had undergone neonatal ECMO treatment. Exercise capacity declined significantly over time, irrespective of the underlying primary diagnosis. Parents’ estimations of the children’s exercise capacity correlated positively with the endurance sd-scores. None of the clinical characteristics correlated with exercise tolerance.

14 children did not perform the exercise test because of neurological or behavioural problems. These 14 children seemed to be the sickest individuals, as reflected by the longer duration of ventilation, and should therefore be considered as poor outcome. Boykin et al. [5] tested the exercise capacity of 10–15-yr-old children who as neonates had received ECMO treatment for MAS [5]. The cross-sectional study revealed that the 17 ECMO-treated children had similar aerobic capacity as age-matched healthy controls. Duration of oxygen use following decannulation proved the most significant factor in predicting long-term pulmonary outcome. In our study, the mixed linear model analysis revealed deterioration of exercise capacity irrespective of the underlying diagnosis. Our study population presumably included more children with initial pulmonary hypoplasia and residual pulmonary sequelae, because Boykin et al. [5] studied only MAS patients, who have intrinsically normal lungs.

Hamutcu et al. [9] showed that 48 children (mean age 11.1 yrs), treated with ECMO for various conditions, including MAS and CDH, had lower peak oxygen consumption than healthy age-matched controls. Differences in the protocols used hamper comparison of the results. Nevertheless, we would like to point out that the oxygen desaturation (arterial oxygen saturation measured by pulse oximetry <90%) observed during exercise in almost 25% of the ECMO-treated children in the study of Hamutcu et al. [9] was not observed in any of our participants. Unlike Hamutcu et al. [9], we did not measure gas exchange parameters for various reasons. First, wearing a mask may lead to loss of cooperation and to submaximal results, especially in younger children. Secondly, Cumming et al. [23] reported a strong correlation between maximal endurance time and maximal oxygen uptake. They concluded that maximal endurance time might be used as a sole criterion of exercise capacity.

Recently, Gischler et al. [24], from our group, evaluated maximal exercise performance in 16 5-yr-old children born with CDH, of whom nine had undergone neonatal ECMO treatment. The mean sd-score for endurance time was -0.84 (also significantly different from zero, p=0.012) [24]. In a study by Peetsold et al. [25], 36 out of 53 survivors of CDH who had not received ECMO treatment (mean±sd age 11.9±3.5 yrs) reliably performed the Bruce treadmill test. In most of them, the exercise capacity agreed with the reference values of Binkhorst et al. [26] established in 1987. There was a positive but small correlation between the sd-score for peak oxygen uptake and the sd-score for FEV₁ (r² = 0.27; p=0.001). In our study, the correlation between maximal endurance time and sd-score for FEV₁ was not significant. ECMO offers survival to those children with CDH with more severe lung hypoplasia. The children in the study of Peetsold et al. [25] who did not need ECMO at all represent survivors with a milder form of CDH.

Vrijlandt et al. [27] reported on exercise capacity in young adults who had been born prematurely and in age-matched controls. As bronchopulmonary dysplasia was over-represented in the born-prematurely group, the exercise capacity of participants with and without bronchopulmonary dysplasia was compared. No significant difference was found.

There is a question over whether bronchodilators would have improved our results. We do not think so, because we observed a significant reversibility of airway obstruction in only a small number of children (n=31). In 20 (64.5%) of them, lung function assessment with bronchodilation was performed prior to the exercise test. So, better performance during the exercise test could have been expected in 11 children only.

Previously, we have found balance-skill problems in 5-yr-old children treated with neonatal ECMO [7]. These could perhaps (partly) explain the decline in exercise capacity between ages 5 and 8 yrs. In the present study, 5-yr-old children were allowed to hold the guardrail to maintain body position near the centre of the moving belt, unlike older children [14, 15]. Hence, part of the deterioration between ages 5 and 8 yrs could be explained by possible balance problems. However, this cannot explain deterioration between ages 8 and 12 yrs.

Another explanatory factor could be lack of exercise; however, reported sports participation had increased at 8 yrs; this seems to contradict lack of exercise as explanatory factor. At 12 yrs, on the contrary, 26% of the 12-yr-old children did not report sports participation.

Regrettably, we have no data from computed tomography scans, so progressive fibrosis as a reason for deterioration over time cannot be ruled out.

In our study group, 18 patients had CDH with pulmonary hypertension and 20 were treated with ECMO for persistent pulmonary hypertension. In both conditions, pulmonary vasoreactivity may persist and result in ventilation–perfusion mismatch. We did not perform cardiac evaluations to confirm this speculation. For CDH patients, ventilation–perfusion mismatch has been described previously [28].

In contrast to the UK Collaborative ECMO trial [1], our study was not set up as a randomised study with controls suffering from severe neonatal cardiorespiratory failure who were treated conventionally. With only two ECMO centres in the Netherlands, covering a relatively small geographical area, the large majority of neonates with similar severity of illness who are not born prematurely (i.e. born after 34 weeks’ gestation with birth-weight >2,000 g) are treated with ECMO. Therefore, it was not possible to include a sufficient number of children who survived without ECMO to serve as a control group, since use of historical data is not desirable. Recruiting controls from other countries with less easy access to ECMO treatment could have been an option. However, differences in the definitions of optimal care for specific diseases and differences in quality of care will almost certainly create bias.

At follow-up at age 7 yrs, the conventionally treated children in the UK Collaborative ECMO trial had more respiratory morbidity than those treated with ECMO [3]. Thus, we speculate that imaginary disease-matched controls in our study would have performed worse than the ECMO-treated children.

Another possible limitation of our study is the lack of healthy controls. We do not think that this invalidates our findings, as our study aimed to prospectively evaluate exercise capacity with advancing age. We used reference data for the Bruce treadmill protocol recently obtained in healthy Dutch children [14, 15]. Furthermore, testing at the different ages was done with exactly the same protocol and equipment as in that study and for the most part even by the same investigator.

In conclusion, neonates treated with ECMO are at risk of decreased exercise capacity at school age. We therefore propose prolonged follow-up. Proactive advice on sports participation or referral to a physical therapist is recommended, especially when either the parents or the children themselves report impaired exercise capacity.