Historically, all newborns, particularly preterm newborns, have been viewed as incapable of pain perception because of an immature central nervous system.1 There is now sufficient evidence to suggest that even fetuses are capable of nociception by 20–24 weeks of gestation.2 Pain is a complex, multidimensional phenomenon that is most often expressed subjectively; in preverbal patients, such as newborns, who cannot self-report pain, pain must be inferred through the observation of physiological and behavioural indicators. Detection of pain is particularly important in the setting of neonatal intensive care units (NICUs), where newborns are often hospitalised for several weeks and are subjected to multiple tissue-damaging noxious stimuli as part of diagnostic and therapeutic procedures.3 4 Pain scales have been developed with the aim of objectively detecting and quantifying pain expression, with scales for acute, procedural pain and for chronic, prolonged pain. However, it has been hypothesised that indicators of pain can be modified by factors internal and external to the individual. Gestational age and postnatal age,5^–8 illness severity,9 behavioural state9 and previous pain experiences10 11 have all been shown to influence the expression of acute pain. For this reason, gestational age and behavioural state have been used as indicators of pain in a validated scale for the detection of acute pain in newborns.12 The effects of potential influencing factors on the expression of chronic pain have not yet been investigated. The EDIN (Echelle Douleur Inconfort Nouveau-né) scale13 is presently the only validated scale to assess chronic pain in term and preterm neonates, and also in those on mechanical ventilation.

The aim of the present study was to evaluate whether neonatal maturity expressed as gestational age and/or postnatal age influences the EDIN score.

PATIENTS AND METHODS

This study was conducted in a tertiary care NICU where pain is routinely assessed on a daily basis by trained nurses using validated algometric scales. All nurses and neonatologists attend a theoretical and practical course given by two experienced neonatologists, one physiotherapist and one nurse, for basic knowledge of neonatal care and pain detection. With regard to pain detection, the EDIN scale was chosen for the measurement of chronic pain. The EDIN is a unidimensional scale made up of five behavioural indicators: facial expressions, body movements, quality of sleep, quality of contact with nurses or sociability and consolability.13 A total score >6 is considered indicative of pain. The nurses, at the end of an 8-h observation period, calculate the EDIN score while performing their usual work. In addition, our NICU has a set of guidelines for the management of pain in newborns.14

Between January 2006 and January 2007, all the 84 newborns referred to our NICU were evaluated by the EDIN scale, giving a total of 1571 scores. The number of EDIN scores obtained for each patient was also recorded. The scores were divided into two groups according to whether they were ⩽6 or >6. The reasons for referral were: prematurity (58 infants), sepsis (4 infants), malformations (4 infants), asphyxia (4 infants) and respiratory distress (14 infants). Newborns on mechanical ventilation for more than 24 h were routinely treated with fentanyl at a dosage ranging from 0.5 μg/kg/h to 2 μg/kg/h.

We reviewed the medical records of all the newborns and recorded the EDIN pain scores. For each EDIN score, the following data were also recorded at the same time: number of total previous painful procedures (including heel pricks, venous or arterial punctures, endotracheal intubations, catheter insertions), presence of sepsis and previous surgical operation, type of ventilator support (mechanical ventilation, nasal continuous positive airway pressure (nCPAP), nasal cannula, no respiratory support), fentanyl treatment, concomitant treatment with ibuprofen or indometacin for patent ductus arteriosus closure.

This study and data release were approved by the local ethics committee, and all parents/legal guardians of the babies studied gave informed written consent for data analysis and data publication.

STATISTICAL ANALYSIS

We entered the data into a specific Excel database and conducted the analysis using SPSS version 13.0 for Windows. The two groups of newborns with EDIN scores >6 or ⩽6 were compared for continuous variables (number of previous painful procedures) using one-way analysis of variance and for categorical variables (fentanyl treatment, surgery, sepsis, type of respiratory support, ibuprofen/indometacin administration) using χ² tests. Variables significantly influencing the presence of pain in the univariate analyses were included as independent variables in a logistic regression analysis model (backward elimination) together with gestational age and postnatal age in order to identify the individual impact of gestational age and postnatal age on the newborn’s ability to express pain. The model also included, as independent variable, the number of EDIN scores per newborn in order to account for the within-child variability. To estimate EDIN values at different gestational ages, we devised three categories of gestational age: 25–31, 32–37 and >37 weeks. For the same purpose, the following categories for postnatal age were created: 0–30, 31–60, 61–90, 91–120 and >120 days. Comparisons among groups were performed using the Kruskal–Wallis test. All hypotheses were tested using two-sided tests at the 0.05 level of significance.

RESULTS

Neonatal data are shown in table 1. A total of 194 EDIN scores were >6, indicating prolonged pain. EDIN scores >6 were more frequently detected after surgery (odds ratio (OR) 2.0, 95% CI 1.3 to 3.1; p = 0.002), in presence of sepsis (OR 2.8, 95% CI 1.8 to 4.3, p<0.001), when a nasal cannula was being used (OR 3.5, 95% CI 2.2 to 5.6, p<0.001) and when the infant was not receiving fentanyl (OR 1.8, 95% CI 1.3 to 2.5, p = 0.001). Neither ibuprofen/indometacin administration nor the number of previous painful procedures were associated with EDIN scores >6. When surgery, presence of sepsis, presence of respiratory support, fentanyl treatment and number of EDIN score/newborn were introduced in a logistic regression analysis model together with gestational age and postnatal age, only gestational age, postnatal age, sepsis and presence of respiratory support were considered significant predictors of the EDIN score. Fentanyl treatment did not show a significant independent effect. ORs and 95% CIs are reported in table 2.

When we categorised the infants by gestational age into three groups (25–31, 32–37 and >37 weeks), we found a significant increase in the EDIN scores (p<0.001) from the lower to the higher gestational age groups. The upper quartile for the EDIN score was 3 at 25–31 weeks’ gestational age, 4 at 32–37 weeks’ gestational age and 5 at >37 weeks’ gestational age. Therefore, an increase of about 1 point in the EDIN score may be expected with increase in gestational age, based on the three groups in this study. Moreover, the EDIN scores increased significantly from 0 to 60 days postnatal age (p<0.001) and from 61 to >120 days postnatal age (p = 0.002). Newborns who had a nasal cannula and were on nCPAP, but not those who were on mechanical ventilation, had significant higher EDIN scores than newborns without any respiratory support (Kruskal–Wallis test, p<0.001, p = 0.002, and p>0.05, respectively).

DISCUSSION

Algometric scales have been developed with the aim of helping caregivers to identify and objectivise pain signals in preverbal patients. However, especially in preterm, compromised neonates the expression of pain may not necessarily equate the experience of pain.15 Preterm, young critically ill newborns referred to the NICU may be too weak or too tired to express pain vigorously. Moreover, these preterm newborns are often unable to cry due to an endotracheal tube or are restrained/positioned in such a way as to preclude body movements. Therefore, preterm neonates may be at higher risk for the consequences of unrelieved pain than term and older neonates if they are not accurately assessed. For the above mentioned reasons, Stevens et al,12 when developing and validating a scale for the detection of acute pain, attributed higher pain scores to infants of lesser gestational age and to those who were not awake and alert. However, the potential effects of gestational age and/or postnatal age in a situation of prolonged pain have not been established. Debillon et al13 developed and validated a useful scale for the measurement of prolonged pain in newborns, also preterm or mechanically ventilated, based only on behavioural indicators. However, the authors suggested that further studies would be useful to investigate the potential influencing role of internal or external factors on the EDIN score. In the present study we investigated, in a homogeneous setting of neonatal pain and distress management, whether gestational age and postnatal age affect pain expression as detected by the EDIN score.

First, we performed univariate analysis to evaluate the correlation between EDIN scores (categorized as >6 or ⩽6, respectively) and several painful procedures or analgesic treatment. Presence of respiratory support, surgery, sepsis, opioid treatment but not ibuprofen/indometacin treatment or number of previous painful procedures significantly affected the risk of having an EDIN score >6. Previous pain experience has been explored as a factor influencing pain expression in newborns. The number of previous painful procedures has been shown to dampen the response to subsequent acute pain. Our data do not confirm this observation concerning prolonged pain. In our NICU, all critically ill newborns who undergo mechanical ventilation are treated by continuous fentanyl infusion; moreover, before performing minor invasive procedures (heel prick, venous and arterial sampling) oral sucrose is routinely given unless contraindicated. Pain control in our population may have avoided the dampened pain response due to uncontrolled pain reported in previous studies. Moreover, each newborn in our study underwent a mean of 10 painful procedures: the impact on the EDIN score of a higher number of painful procedures needs to be investigated.

Variables increasing significantly the risk of having an EDIN score >6 at the univariate analysis were included in a logistic regression analysis, together with number of EDIN scores per newborn, gestational age and postnatal age, in order to detect whether these last two factors could independently influence EDIN scores. Fentanyl treatment, which was associated with a lower risk of having an EDIN score >6 in the univariate analysis, also had no significant effect when introduced in the logistic regression analysis. The possible explanation for this result is that fentanyl treatment was not an independent variable as fentanyl was given only to mechanically ventilated newborns and to those who underwent surgical treatment. Therefore, the effect of fentanyl treatment could have been obscured by the effects of the surgery or by the type of respiratory support.

The most important determinant of the EDIN score in our population of preterm and term newborns was gestational age (OR per 1 week increase in gestational age: 1.166, 95% CI 1.123 to 1.211, p<0.001). The increase in the EDIN score with increase in gestational age can be interpreted in two ways. First, it may indicate a reduced ability to feel pain at lower gestational age. This hypothesis has been largely refuted in the past two decades and it has been extensively demonstrated that newborns, even if preterm, have anatomical and functional maturity for pain perception. The second and more reliable explanation is that less mature newborns have a reduced ability to express pain. Craig et al reported marked changes in pain response across the age range of prematurity, with the most premature infants showing the lowest activity.6 Johnston et al, who compared the response to heel stick in the same infants across 8 weeks of development, found that infants demonstrated increasingly more behavioural responses as they matured.5 Also, Xia et al found that preterm infant pain behaviours (particularly facial activity and limb movements) increased with increasing gestational age.7 All these studies referred to acute pain; to our knowledge this is the first study demonstrating an influence of gestational age on the expression of prolonged pain. This finding suggests that the EDIN score could be improved by adding categories for gestational age and attributing higher basal scores to less mature newborns in order to better detect prolonged pain in newborns less able to respond vigorously to chronic/prolonged pain. On the basis of our data a basal score of 2, 1 and 0 could be attributed to newborns of gestational age 25–32 weeks, 33–37 weeks and >37 weeks, respectively. Comparison of construct validity, inter-rater reliability and internal consistency of the EDIN scale applied with or without categories for gestational age could be useful in confirming our hypothesis. Our results were obtained on a mixed population of newborns, with only a few newborns in the postoperative period. Further studies in a more selected population undergoing chronic pain could be useful.

Another factor influencing the EDIN score in our population of newborns was postnatal age. EDIN scores increased significantly about every 2 months of age so that the effect of postnatal age over EDIN in the neonatal period may be considered negligible. Type of respiratory support, sepsis and surgical treatment also affected the EDIN score in our population. EDIN scores in newborns with a nasal cannula and on nCPAP, but not EDIN scores in newborns on mechanical ventilation, were significantly higher than EDIN scores in newborns without any respiratory support. A possible explanation of this result is that all mechanical ventilated newborns, but not those with a nasal cannula or on nCPAP, receive routine opioid treatment in our unit. In addition, sepsis, which is not routinely treated with analgesics in our unit, and that is clearly a source of distress, correlated with a higher risk of having an EDIN score >6 (OR 3.74, 95% CI 2.30 to 6.10, p<0.001).

CONCLUSIONS

It was already well known that gestational age influenced the expression of acute pain in newborns; now it seems that gestational age also influences the expression of prolonged pain. In the present study the EDIN scores were influenced by gestational age; the EDIN scores were lowest in the least mature gestational age neonates and highest in the most mature.

Therefore, creating categories for gestational age may improve the content validity of the EDIN scale.