Prolonged Sedation and/or Analgesia and 5-Year Neurodevelopment Outcome in Very Preterm Infants: Title and subTitle BreakResults From the EPIPAGE Cohort FREE

Very preterm neonates are exposed to many painful procedures.¹ Repeated and prolonged pain exposure may affect the subsequent development of pain systems and potentially contribute to altered long-term development and behavior in very preterm infants.^{2 - 8} This knowledge has provided evidence-based arguments for the development of professional guidelines on preventing and managing infant pain.^{9 - 10}

Recently, the short- and long-term effects of sedation and/or analgesia have been discussed. The Neopain study¹¹ and a Dutch study¹² revealed that continuous morphine therapy given for a maximum of 14 or 7 days, respectively, does not increase neonates' poor short-term neurological outcome or neonatal death. The Neopain study showed, however, that continuous morphine infusion was associated with a higher rate of severe intraventricular hemorrhage or a composite outcome, including neonatal death, intraventricular hemorrhage, and periventricular leukomalacia in a subgroup of very preterm infants born at 27-29 weeks' gestation and in those who were very hypotensive before morphine therapy.^{11 ,13} The only published study to date regarding long-term effects suggests that exposure to morphine in the neonatal period for fewer than 7 days does not seem to have adverse effects on intelligence or motor function.¹⁴ Regarding sedation, midazolam is widely used in neonatal intensive care units. However, a recent meta-analysis has concluded that further research on the effectiveness and safety of midazolam in neonates who received mechanical ventilation is needed before its routine use can be recommended in clinical practice.¹⁵

Although data on adverse effects are still controversial, analgesia and/or sedation therapies are routinely used in neonatal intensive care units worldwide, and some very preterm infants may receive prolonged treatment with sedation and/or anesthesia for more than 7 days. Given the lack of data regarding the neurological development outcome of these very preterm infants, we have performed an analysis to evaluate the relationship between prolonged sedation and/or analgesia exposure and the 5-year cognitive and neuromotor outcome of a large French cohort of very preterm infants.

We included all surviving very preterm infants enrolled in the Etude EPIdémiologique sur les Petits Ages GEstationnels (EPIPAGE) cohort study who received mechanical ventilation and/or surgery. The EPIPAGE study is a prospective population-based cohort study including all infants born between 22 and 32 weeks' gestation in 1997 in the maternity wards of 9 French regions, accounting for about one-third of all births in France.¹⁶ Initial data about pregnancy, delivery, and medical care of infants was collected from medical records in the maternity and neonatal units with the intention of testing the association between obstetric and neonatal management and outcome. One aspect of these data concerned “exposure to prolonged sedative and/or analgesic drugs,” defined as exposure of more than 7 days to sedative and/or opioid drugs, with infants receiving sedation each day. The status “no prolonged sedation/analgesia” included children who received no sedation and/or analgesia at all or for whom sedation and/or analgesia was administered for fewer than 7 days.

Once consent was obtained from the parents, the infants were enrolled in a longitudinal follow-up until 5 years of age. At 2 years children were examined by a physician who completed a standardized questionnaire. At 5 years children were evaluated with a clinical examination performed by a trained pediatrician and a neuropsychological assessment performed by a trained psychologist. The neuropsychological assessment was performed using the Kaufman Assessment Battery for Children.¹⁷ This assessment was carried out when appropriate for the patient's condition and when accepted by the patient. The Kaufman Assessment Battery for Children yields 4 global test scores, including the Mental Processing Composite (MPC), which is a global measure of cognitive ability. The MPC score was standardized to a mean (SD) of 100 (15), based on results in children born in the late 1970s. Mental Processing Composite scores of less than 70 define severe cognitive impairment and scores of between 70 and 85 indicate moderate cognitive impairment.

The main outcome criteria were the presence of disability at 5 years of age or death after discharge. Children were classified into 3 groups: no disability, moderate disability, and severe disability. Disabilities were assessed in 4 areas: neuromotor, cognition, hearing, and vision. A disability was defined as severe if it was considered likely to make the child highly dependent on caregivers and included nonambulant cerebral palsy, an MPC score of less than 70, profound bilateral sensorineural hearing loss (> 70 dB), blindness, or attending a special-needs school. Cerebral palsy was defined as a nonprogressive central nervous system disorder, as proposed by the European Cerebral Palsy Network.¹⁸ The severe disability group also included children who were classified into the cerebral palsy group at 2 years of age and were lost to follow-up at 5 years. A disability was defined as moderate if reasonable independence was likely to be reached, including ambulant cerebral palsy, independent walking with abnormal gait, age of walking greater than 24 months, an MPC score between 70 and 84 (< 1 SD), hearing loss less than 70 dB, or unilateral impaired vision without blindness.

Median and quartiles are reported for continuous variables and frequencies for categorical variables. The Mann-Whitney U test and χ² test with a Yates correction, if necessary, were used to compare infant characteristics and 5-year outcome between the group of infants who received prolonged sedation and/or exposure to analgesia during the neonatal period and the group who did not.

Because EPIPAGE is an observational study, we used the propensity score method to reduce bias.^{19 - 23} The propensity score is defined as a conditional probability between 0 and 1 that a patient will be treated, based on an observed group of covariates. This score is then used as if it were the only confounding covariate. Thus, the collection of predictors is collapsed into a single predictor, which may better adjust covariates between the groups and reduce bias. A full nonparsimonious logistic regression model was developed to derive a propensity score for prolonged sedation and/or analgesia exposure. This model included true confounders, variables that are potentially associated with both treatment assignment and outcome.²³ The variables included characteristics of the newborns (gestational age, birth weight, malformation, Clinical Risk Index for Babies score), characteristics of pregnancy (antenatal corticosteroids, multiple pregnancies, complications of pregnancy), place of birth (inborn/outborn, large maternity, region of birth), and characteristics of neonatal hospitalization (duration of ventilation, patent ductus arteriosus, severe respiratory failure including air leak and refractory hypoxemia, hemodynamic failure, necrotizing enterocolitis, neonatal surgery, bronchopulmonary dysplasia defined by supplemental oxygen at 36 weeks, and postnatal corticotherapy). Patients with missing data were excluded from multivariable analysis. The Hosmer-Lemeshow goodness-of-fit test and the area under the curve were used to assess model fit.

We have used a log-binomial model to estimate adjusted relative risk (RR) of disability.²⁴ As described above, we integrated the propensity score into the analysis. The propensity score itself was included with exposure as a covariate in a log-binomial model predicting outcome, with gestational age as an additional covariate, and we estimated the adjusted RR of disability with 95% confidence intervals (CI). This was performed for all infants for whom propensity scores were calculated. Finally, we carried out 2 other analyses to independently estimate the adjusted RR of the risk of neuromotor disability and the risk of an MPC score of less than 85. All P values were based on 2-tailed tests. All analyses were performed using SPSS 15.0 (SPSS Inc, Chicago, Illinois).

Of the 1956 very preterm infants hospitalized in neonatal intensive care units who received mechanical ventilation and/or surgery, the status of prolonged sedation and/or exposure to analgesia was known for 1941 (99%) of the infants; of these, 369 died and 1572 survived. Prolonged sedation and/or exposure to analgesia was more frequent among infants who died during hospitalization than among survivors (72 of 369 [19.5%] vs 115 of 1572 [7.3%]; P < .001). Of the 1572 survivors, infants who were exposed to prolonged sedation and/or analgesia (n = 115) had significantly lower gestational ages, lower birth weights, and higher Clinical Risk Index for Babies scores than infants who did not (n = 1457) (Table 1). In the exposed group, antenatal corticosteroids, multiple pregnancies, birth in a large maternity (a maternity hospital in which > 98 very preterm infants are delivered per year), and presence of congenital malformations were significantly more frequent (P < .05). Prolonged sedation and/or analgesia was also significantly associated with neonatal complications such as severe respiratory failure, bronchopulmonary dysplasia, patent ductus arteriosus requiring treatment, hemodynamic failure, necrotizing enterocolitis, longer duration of mechanical ventilation, postnatal corticotherapy, and surgery (Table 1). Prolonged sedation and/or analgesia exposure rates varied according to region of birth from 0% to 9.3% (P < .001). Data on long-term outcome were available for 86% of the 1572 infants (84% in the exposed group vs 86% in the nonexposed group; P = .46). Disability rates in the exposed and nonexposed groups were 42.3% and 26.0%, respectively (P < .001).

It was possible to calculate the propensity score for the very preterm infants receiving prolonged sedation and/or analgesia for 1497 of the 1572 very preterm infants receiving mechanical ventilation and/or surgery who were alive at the time of discharge from the hospital and whose status regarding exposure to prolonged sedation and/or analgesia exposure and outcome were known. Of the 1497 infants with a propensity score, 113 were exposed to prolonged sedation and/or analgesia and 1384 were not. The propensity scores ranged from 0.000 to 0.975. The Hosmer-Lemeshow goodness-of-fit test was 6.2 (P = .62). The receiver operating characteristic curve area was 0.92. Among the 20 variables included in the score, 6 were significantly associated with prolonged sedation and/or exposure to analgesia: surgery (P < .001), duration of mechanical ventilation (P = .002), postnatal corticotherapy (P < .006), severe respiratory failure (P < .02), multiple pregnancies (P < .003), and region of birth (P = .001) (Table 2).

Of the 1497 very preterm infants for whom propensity scores were calculated, data on long-term outcome was available for 87% of the 1384 infants in the exposed group vs 86% of the 113 in the nonexposed group (P = .73). Disability rates in the exposed and nonexposed groups were 42.3% and 25.7%, respectively (P < .001). Relative risks of disability for prolonged sedation and/or analgesia are indicated on Table 3.

To limit bias, 2 log-binomial models were constructed. The first model included 3 covariates: gestational age, propensity score, and prolonged sedation and/or exposure to analgesia. Prolonged sedation and/or analgesia exposure was not associated with an increased risk of survival with any disability or risk of death after discharge (adjusted RR, 1.0; 95% CI, 0.8-1.2; P = .57), but propensity score was (adjusted RR, 2.2; 95% CI, 1.3-3.8; P = .005). The second model included the 3 previous covariates and 8 other covariates significantly associated with outcome: intrauterine growth, maternal socioeconomic status, maternal educational status, sex of preterm infant, surgery, duration of ventilation, postnatal corticotherapy, and presence of malformation. Prolonged sedation and/or analgesia exposure was not significantly associated with disability (adjusted RR, 1.0; 95% CI, 0.8-1.2).

To better assess the relationship between disability and prolonged sedation and/or analgesia exposure, we conducted 2 other analyses where the dependant variables were an MPC score lower than 85 and neuromotor disability. The covariates were the 11 previous covariates. Prolonged sedation and/or analgesia exposure was not associated with an increased risk of survival with neuromotor impairment (adjusted RR, 1.1; 95% CI, 0.9-1.2), an MPC score of less than 85 (adjusted RR, 1.0; 95% CI, 0.7-1.2), or an MPC score lower than 85 or not available (adjusted RR, 1.1; 95% CI, 0.9-1.2).

In this observational study, univariate analysis showed that very preterm infants requiring prolonged sedation and/or analgesia during neonatal hospitalization had a higher risk of developing a disability. Prolonged sedation and/or exposure to analgesia was also significantly associated with need for surgery, congenital malformation, necrotizing enterocolitis, severe respiratory failure, hemodynamic failure, and duration of ventilatory support. It is likely that these factors were the reason for the use of prolonged sedation and/or analgesia. The association of some of these factors with a poor outcome is well known.^{25 - 26} Because the prolonged sedation and/or analgesia treatment assignment in this observational study may have been subject to bias, we used the propensity score method to reduce it. The propensity score method is a way of ensuring that the effect of a treatment is compared only between individuals who are equally likely to receive it.^{21 - 22} After adjustment for gestational age and propensity score, we did not observe any association between prolonged sedation and/or analgesia exposure and poor outcome, whereas most of the nonexposed infants did not receive any analgesic or sedative drugs in 1997.

We cannot rule out the possibility that prolonged sedation and/or analgesia exposure has an effect on outcome for 4 reasons. First, our findings should be interpreted with caution, because this study has several limitations. Despite the large size of this cohort, the power of this study was not high enough to detect small differences between the groups, because exposure to prolonged sedation and/or analgesia is a relatively rare event. Thus, this study did not have the power to detect a difference of 10% in the rate of disability. Like regression modeling, propensity score methods cannot control for unknown confounders. Some covariates, like the duration of mechanical ventilation, could be the cause of prolonged sedation and/or analgesia as well as the consequence. Nevertheless, we performed a model without including duration of ventilation in propensity score, bronchopulmonary dysplasia, and hemodynamic failure, and we observed similar results. Second, the odds ratio of the risk of developing a disability increased as the propensity score increased. This association could be the result of an insufficient reduction of bias by the propensity score or a cumulative effect of prolonged sedation and/or analgesia exposure that appeared when the severity of illness increased. Third, MPC scores were similar between exposed and nonexposed infants, while white matter damage was observed less frequently. One hypothesis could be that prolonged sedation and/or exposure to analgesia reduced the rate of cerebral lesions but itself had an effect on cerebral development. No difference was observed between groups when we included all preterm infants. Fourth, we cannot differentiate and assess the potential effect of any individual sedative or anesthetic agent in this study.

Our results suggest that prolonged sedation and/or analgesia exposure has no major effect on neurological outcome. This is currently a critical and controversial issue, because the recent understanding that the neonate is capable of sensing pain and mounting a stress response after a surgical stimulus²⁷ has led to widespread use of anesthesia and analgesia for painful procedures; however, exposure to opiates, anesthetics, or sedatives may have adverse effects on neurological neonatal outcome.²⁸ Several investigators have demonstrated that prolonged administration of opiates^{29 - 30} and other anesthetic drugs produce increased neurodegeneration in rat pups.³¹ Similarly, repetitive pain in newborn rat pups leads to long-term changes in behavior, and some of these changes are prevented by analgesic therapy.^{32 - 35} This data suggests that the long-term effects of anesthetics and analgesic drugs depend on whether they are given in the presence or absence of painful stimulation.²⁸ Extrapolation of animal data are problematic, because a rat brain develops over a matter of weeks, while a human brain develops over years. Six hours of anesthesia in a neonatal rat pup is thought to be equivalent to a month in a human infant.³⁶ Moreover, this rodent data has not been replicated in monkeys.³⁷ Therefore, cautious interpretation of experimental data is prudent before extrapolating data across species. Although the animal data may not be applicable to human beings, it would be unwise to ignore it.³⁶ Evidence must be sought from human data. MacGregor et al¹⁴ have assessed the outcome at 5 to 6 years in a cohort of neonates born before 34 weeks' gestation to determine whether brief exposure to exogenous opiates had an effect on subsequent neurological or behavioral performance. Their results are reassuring; morphine did not show a detrimental effect on the intelligence of children who received it in the first days of neonatal life. Our data are consistent with these results, though it pertained to a population that was sicker and exposed to more prolonged sedation and/or analgesia.

In this population-based cohort study, prolonged sedation and/or analgesia exposure was not associated with a poor 5-year neurological outcome after adjustment for the propensity score.

Correspondence: Jean-Christophe Rozé, Department of Neonatal Medicine and Centre d’Investigation Clinique INSERM CIC004, Hôpital de la Mère et l’Enfant, Boulevard Jean Monet, 44000 Nantes, France (jcroze@chu-nantes.fr).

Accepted for Publication: January 24, 2008.

Author Contributions:Study concept and design: Rozé, Kaminski, Truffert, André, Larroque, and Bréart. Acquisition of data: Rozé, Arnaud, Truffert, Marret, Matis, Thiriez, Cambonie, André, Larroque, and Bréart. Analysis and interpretation of data: Rozé, Denizot, Carbajal, Ancel, Kaminski, Arnaud, Thiriez, Larroque, and Bréart. Drafting of the manuscript: Rozé, Denizot, Carbajal, Ancel, and Bréart. Critical revision of the manuscript for important intellectual content: Carbajal, Ancel, Kaminski, Arnaud, Truffert, Marret, Matis, Thiriez, Cambonie, André, Larroque, and Bréart. Statistical analysis: Rozé, Carbajal, Ancel, Kaminski, and Bréart. Obtained funding: Kaminski, Larroque, and Bréart. Administrative, technical, and material support: Carbajal, Marret, and Larroque. Study supervision: Kaminski, Cambonie, André, Larroque, and Bréart.

Additional Contributions: Etude EPIdémiologique sur les Petits Ages GEstationnels (EPIPAGE) Study Group: B. Larroque (national coordinator), P. Y. Ancel, B. Blondel, G. Bréart, M. Dehan, M. Garel, M. Kaminski, F. Maillard, C. du Mazaubrun, P. Missy, F. Sehili, K. Supernant, INSERM U149, Paris; M. Durand, J. Matis, J. Messer, A. Treisser, Hôpital de Hautepierre, Strasbourg; A. Burguet, L. Abraham-Lerat, A. Menget, P. Roth, J.-P. Schaal, G. Thiriez, CHU of Nantes, Besançon; C. Lévêque, S. Marret, L. Marpeau, Hôpital Charles Nicolle, Rouen; P. Boulot, J.-C. Picaud, Hôpital Arnaud de Villeneuve, Montpellier; A.-M. Donadio, B. Ledésert, Observatoire Regional de la Santé Montpellier; M. André, J. Fresson, J. M. Hascoët, Maternité Régionale, Nancy; C. Arnaud, S. Bourdet-Loubère, H. Grandjean, INSERM U558, Toulouse; M. Rolland, Hôpital des Enfants, Toulouse; C. Leignel, P. Lequien, V. Pierrat, F. Puech, D. Subtil, P. Truffert, Hôpital Jeanne de Flandre, Lille; G. Boog, V. Rouger-Bureau, J.-C. Rozé, Hôpital Mère-Enfant, Nantes; P. Y. Ancel, G. Bréart, M. Kaminski, C. du Mazaubrun, INSERM U149, Paris; M. Dehan, V. Zupan-Simunek, Hôpital Antoine Béclère, Clamart; M. Vodovar, M. Voyer, Institut de Puériculture, Paris, France.

Financial Disclosure: None reported.

Funding/Support: This study was supported by grants from Institut National de la Santé et de la Recherche Médicale (French National Institute of Health and Medical Research), Merck-Sharp, Dohme-Chibret, Medical Research Foundation, Directorate General for Health of the French Ministry for Social Affairs, and the French Hospital Program of Clinical Research, Paris, France.