Morphine use in the neonatal period and later neuropsychological development: a systematic review

With the latest advances in neonatal intensive care, many preterm and seriously ill term-born infants now survive; however, they undergo multiple painful procedures while being treated in the neonatal intensive care unit (NICU).^1-4 Furthermore, there is a growing body of evidence highlighting the detrimental impact of uncontrolled and repetitive neonatal pain on later neuropsychological outcomes in terms of social behaviour, cognitive, motor, and executive function development.^{5, 6, 8-10} Therefore, neonatal pain management has intrigued a variety of clinicians, scientists, and care providers.

Morphine is the most commonly used opioid in the NICU for procedural pain, postoperative pain, and sedation with prolonged mechanical ventilation.^11-17 Of note, neonates born preterm have an immature brain and are thought to be more vulnerable to morphine treatment.¹⁸ Moreover, neonatal rodent models have reported that morphine can play a modifying role in neurodevelopmental processes such as maturation and proliferation of important regions in the brain.^19-22 This could lead to long-term sequelae.²³ Therefore, the association between morphine use in the neonatal period and later neuropsychological development is a critical issue in current neonatal pain management. Recently, Zwicker et al. indicated that intravenous morphine had a negative correlation with cognition and motor development at 18 months.²⁴ However, Grunau et al. found an association between higher morphine exposure and poorer motor outcomes at 8 months instead of 18 months in a different cohort of 137 children born very preterm.⁸ A 5- to 7-year pilot follow-up of a subset of participants previously recruited in the Neurological Outcomes and Pre-emptive Analgesia in Neonates trial demonstrated that there was no significant difference in overall IQ between the morphine and placebo groups,²⁵ while one cohort study at the 7-year follow-up reported neonatal morphine infusion exposure had an adverse effect on internalizing behaviours.⁹

So far, there have been no systematic reviews identifying and evaluating the evidence demonstrating the association between neonatal morphine and neuropsychological development during childhood. Thus, the objective of our systematic review was to identify and assess studies examining the relation between morphine and neuropsychological development, to provide health-care professionals with the necessary information for neonatal pain management.

Method

Results

Discussion

To the best of our knowledge, this is the first systematic review containing evidence about the impact of neonatal morphine on subsequent neuropsychological development. Using PRISMA guidelines, eight individual studies (12 separate reports) were included in this review (Fig. S1).

Evidence of this systematic review indicated that neonatal morphine was not associated with total behaviour problems at 5 to 7 years and 8 to 9 years, or externalizing behaviour problems at 2 years, 5 to 7 years, and 8 to 9 years. Three studies showed that neonatal morphine was not associated with internalizing behavioural problems at 18 months, 2 years, and 8 to 9 years.^{10, 33, 36} However, two RCTs consistently reported that the use of morphine in the neonatal period had no adverse effects on internalizing behavioural problems at 5 to 7 years,^{25, 32} while one cohort study showed that neonatal morphine was significantly related to more internalizing behavioural problems at 7 years.⁹ This discrepancy might be explained by the following reasons. First, different study designs: the reason a child needs morphine may have a far stronger association with later neuropsychological development than morphine itself, and without controlling for this immense confounder, it is difficult to draw any conclusions about the potential impact of morphine. RCTs are the only studies that are not clearly subject to the problem of potential confounders related to the need for the morphine, while the cohort study could not explain the causality and its result was susceptible to confounding factors. Second, different ventilation periods and morphine medication protocols. This was 15.9 days in the cohort study,⁹ 71 hours in the study by Graff et al.,³² and at most 14 days in the study by Ferguson et al.²⁵ The daily morphine exposure (in milligrams adjusted daily weight) was 2.3 in the cohort study,⁹ but a loading dose 100μg/kg for all gestational ages, followed by 10μg/kg/h continuous infusion (maximum 7d), or 10 to 30μg/kg/h continuous infusion depending on gestational age (maximum 14d) in the two RCTs.^{25, 32} Therefore, a possible timing and dosing effect may be related to poor brain development and then lead to more internalizing behavioural problems.^37-40 Third, there were different time points for assessing behavioural development. Participants were measured at 7 years of age in the cohort study,⁹ and at 5 years or 5 to 7 years in the two RCTs.^{25, 32} Taken together, we could not draw definitive conclusions about whether neonatal morphine use had an adverse effect on internalizing behavioural problems at 5 to 7 years.

Regarding cognitive and motor development, the evidence of this systematic review indicated that neonatal morphine use did not affect children’s cognitive development at 8 months, 12 months, 2 years, 3 years, 5 to 7 years, and 8 to 9 years,^{8, 25, 30-34, 36} and children’s motor development at 12 months, 20 months, 2 years, 3 years, 5 to 6 years, and 7 years.^{30, 31, 34-36} However, cumulative morphine exposure may be associated with poorer motor development at 8 months⁸ and cognitive development at 20 months.³⁵ Regarding whether morphine use in the neonatal period affects cognitive and motor development at 18 months, the two included cohort studies had inconsistent results,^{8, 24} which was probably for the following reasons. First, the duration of morphine exposure was much shorter in the study by Grunau et al.⁸ than in the study by Zwicker et al.²⁴ Experimental study with animal models demonstrated that prolonged morphine exposure may trigger apoptosis and necrosis in Purkinje cells and result in decreased cerebellar growth.^{37, 38} Moreover, one clinical study indicated that morphine had not been associated with altered development of the cerebrum.³⁶ Second, in the study by Zwicker et al., researchers did not systematically say whether morphine was given in response to pain.²⁴ Animal studies found that morphine may be neurotoxic when given in the absence of pain.⁴¹ Third, participants in the higher-dose group were completely different from those in the lower-dose group for medication (except morphine), gestational age, birthweight, score for Neonatal Acute Physiology (version II), frequency of painful procedures, hypotension, and time on the ventilator. This could have a confounding effect on evaluating the relation between morphine and later neurodevelopmental outcome.⁴²

With regard to executive functions, neonatal morphine did not affect global executive function assessed by BRIEF at 2, 7, and 8 to 9 years.^{33, 36} Morphine-treated children (continuous morphine infusion of 10μg/kg/h in the neonatal period) showed fewer problems on the inhibition of parental subscales and organization of materials of BRIEF.³³ This suggests that morphine might have a potential neuroprotective effect on the painful stimuli to which neonates in NICU are routinely exposed. However, given the high number of outcome measures with no control for overall statistical error, and the lack of differences in teacher reports or standardized assessment of executive functions, the positive association between morphine exposure and executive functions appears tentative.

It is worth nothing that the results of this systematic review should be considered and generalized with caution owing to the following limitations of the review and the included studies. First, the total number of included trials was low. Moreover, many included studies used a smaller subset from the original population and had a lower follow-up rate which could have produced selection bias, and in turn affected the evaluation of the study results. Second, there was considerable heterogeneity among the included studies in terms of the study design (RCT, prospective or retrospective cohort study), participants’ characteristics (gestational age, birthweight, severity of illness, brain injury, sepsis), morphine protocol (dose, duration, indications), and outcome measurements (tools, follow-up time). Furthermore, two RCTs had high or unclear risks for allocation concealment, blinding of carers and personnel, or follow-up (Table 1). Four cohort reports only had moderate quality (Table 2), and some cohort studies did not report detailed information about the indication, frequency, dose, or duration. Also, some RCT reports did not present the information on morphine use after the original study period of the maximum days of treatment or the cumulative dose of morphine use. All of these factors may have had an impact on determining the association between neonatal morphine and later neuropsychological development outcomes. This makes it difficult for us to synthesize the findings obtained to make a more definite conclusion about the morphine quantity, indication, and gestational age that are safe for later childhood neuropsychological development. Third, some included studies were not designed to address the effects of specific morphine protocols on later childhood neuropsychological development, which greatly lowers the power of the studies. Finally, this review included RCTs and cohort studies published in English or Chinese and excluded trials published in other languages, which may have caused publication bias.

More prospective, multi-centred, large studies with a standardized study design are required to examine the correlation between neonatal morphine and future neuropsychological development outcomes. Given that morphine is the standard of care in many NICUs, using RCTs to examine whether neonatal morphine contributes to poor later childhood neuropsychological development outcomes would probably be just as difficult. A potential solution could be to allow open-label boluses of morphine in the placebo groups just as in the two major morphine trials (Neurological Outcomes and Pre-emptive Analgesia in Neonates trial and Simons et al. trial).^{43, 44} However, this can lead to crossover which complicates these trials.⁴⁵ In addition, future prospective cohort studies with more detailed information about morphine use and participants’ characteristics, adequate follow-up, and comprehensive assessments of neuropsychological development might be able to provide much valuable information for addressing this question. Beyond that, researchers may place more emphasis on the investigation of novel pharmacological and non-pharmacological treatments. Lastly, Chau et al. found that individual genetic differences in morphine biotransformation biomarkers play a vital role in the association of exposure to neonatal morphine with long-term adverse effects.⁴⁶ Therefore, studying how to identify neonates born preterm or at term who are genetically very sensitive to morphine will be another alternative method, to prevent the adverse effects of morphine on neuropsychological development in infants in the NICU.

Conclusions

This review demonstrated that neonatal morphine use had no adverse effects on internalizing, externalizing, and total behaviour problems. This was also true for cognitive and motor development, and executive function at 8 to 9 years and earlier. This would not be true if there were inconsistent conclusions about whether the use would affect children’s internalizing behavioural development at 5 to 7 years of age as well as cognitive and motor development at 18 months of age. Nevertheless, considering the small number, heterogeneity, and limitations of the included studies, the results of this systematic review should be interpreted with caution, and more research is needed to further examine the association of neonatal morphine with later childhood neuropsychological development.