In order to ensure their survival, premature born infants hospitalized in a neonatal intensive care unit (NICU) are subjected to many painful diagnostic and therapeutic procedures [
1‐
3]. Although there have been efforts in recent years to quantify, and most importantly, reduce the number of procedural exposures to pain in preterm infants, procedural acute pain remains a challenge in the NICU setting [
3‐
5]. Often, these treatment interventions take place during a crucial period in the development of the nociceptive and central nervous systems [
6‐
8]. There is more and more alarming evidence that repeated painful stimuli at this early age may induce both structural and functional reorganization of the nervous system [
7,
9‐
13] and result in an altered pain response [
14‐
16]. As a consequence, the motor and cognitive development of premature infants may be impaired [
9,
13,
17‐
22]. In premature infants requiring intensive care, the frequency of exposure to pain and systematic implementation of preventive pain measures are therefore of key importance for their later development [
4,
5]. Accurate pain measurement is the first step toward effective pain management.
Pain assessment in neonates
Clinical pain assessment in neonates, particularly those delivered preterm, is highly challenging [
4,
23]. In the clinical setting, their pain responses have to be observed and assessed using behavioral and physiological indicators, which can vary across premature infants depending on their physiological and neurological development stages [
23]. Behavioral indicators used as pain assessment tools include body movements, facial expressions and crying [
24]. Some pain assessments also include behavior status indicators, e.g., sleep-wake state [
25,
26]. Physiological responses to pain include, for instance, changes in heart rate, respiratory rate, blood pressure, oxygen saturation, vagal tone, and peripheral blood flow [
25,
27]. Recently, researchers have begun to investigate more objective approaches to pain assessment, such as measurement of heart rate variability, skin conductance and cortisol as a biomarker of stress [
23,
25]. To better understand and assess neonatal pain responses at cortical level, newer brain-oriented techniques, such as electroencephalography (EEG) [
28,
29] and functional magnetic resonance imaging (fMRI) [
30,
31], are used [
11,
32‐
34]. However, for systematic clinical pain assessment, exclusively observable indicators need to be considered.
Because of the complex nature of pain, multidimensional pain measures that include behavioral and physiological indicators are generally assumed to be most appropriate for the clinical setting [
23]. Although most infants show both types of pain response indicators, the correlation between these two indicators is often low [
25,
35]. Moreover, no consistent associations between behavioral, physiological and cortical measures of pain have been detected so far [
36]. In the face of inconclusive associations between different indicators of pain, the validity of existing multidimensional tools and their choices of indicators are currently being questioned, and, to date, no universally accepted gold standard exists for neonatal pain assessment [
23].
More than 40 pain assessment scales for premature and full-term infants exist to date [
25,
37]. The majority were designed for research purposes and are inappropriate for routine clinical procedures (e.g., because they require extended observation periods) [
25,
38]. Furthermore, only a few have undergone extensive psychometric testing and are both reliable and valid [
25,
39]. Of the pain assessment scales compiled for clinical application, few have been validated in premature infants and even fewer consider individual contextual factors, e.g. gestational age (GA) and health status [
23,
40].
The Bernese pain scale for neonates
The Bernese Pain Scale for Neonates (BPSN; [
41]) was developed by nurses of the University Hospital of Berne primarily for clinical use. Since its development in 1996, it has been widely used for bedside pain assessment in NICUs in the German speaking areas of Europe. Several hospitals in Switzerland have fully integrated the BPSN into their daily routine.
The BPSN is a 9-item multidimensional pain assessment tool that includes behavioral and physiological indicators. The instrument consists of seven subjective (alertness, crying, consolation, skin color, facial expression, posture, and changes in respiratory rate) and two physiological (i.e. objective) (changes in heart rate and oxygen saturation) indicators. Each item is rated on a four point Likert scale (0, 1, 2, and 3). Higher scores indicate greater pain-related distress, and a total score of 11 or higher is considered to indicate pain.
In the year 2004, the BPSN was validated to differentiate between pain and non-pain status in neonates between 27 and 41 weeks of gestation [
41]. The results suggested that the BPSN is a valid and reliable pain assessment instrument for assessing acute pain in term and preterm neonates. A shortcoming of this first validation study of the BPSN is the small study population of 12 infants. Furthermore, increasing evidence indicates that pain reactions of neonates are probably influenced by more than noxious stimulation alone; individual contextual factors might also impact pain reactivity [
40,
42‐
44]. Currently, the BPSN focuses entirely on physiological and behavioral indicators.
Individual contextual factors
Individual contextual factors encompass individual infant characteristics (e.g., GA, gender, health status, and weight), previous pain experience, or the duration of hospitalization [
23,
44]. The variability in pain responses between and within premature infants as well as the low association between behavioral and physiological pain responses may be explained by the influence of individual contextual factors [
35,
42,
45,
46].
Neonatal age is the most commonly examined individual contextual factor associated with neonatal pain response [
44]. Premature neonates generally seem more sensitive to painful stimulation than full-term newborns. In addition to having low reflex thresholds [
47,
48], newborns lack the inhibitory control that mature brain structures would exert [
49]. As a result, premature neonates display diffuse responses to noxious stimuli rather than more complex affective reactions [
50]. Moreover, the association between behavioral and physiological stress responses may differ depending on GA [
35]. Although older GA infants displayed a positive association between the extent of behavioral pain reaction and heart rate levels, Lucas-Thompson et al. (2008) found no association between physiological and behavioral responses in the youngest GA infants. Despite the high variability in behavioral and physiologic pain responses in premature neonates, their responses are less intense [
42,
45,
51,
52].
The results of several studies suggest that facial expression in response to pain increases with GA [
45,
52‐
55]. This difference is manly influenced by the older infants’ increased facial expressiveness, which results from their more developed nervous system and facial muscles [
53,
54]. In contrast, several studies have reported no significant relationship between GA and facial expression in response to pain [
44,
56]. However, the consideration of reduced facial movement in response to pain in premature neonates is important. Using pain assessment scales which rely only on facial expressions may lead clinicians to the incorrect conclusion that younger premature infants do not feel or feel less pain [
57]. In addition, the presence of endotracheal tubes in premature neonates impedes using facial reaction and crying as indicators of pain because endotracheal tubes are typically secured by taping them to the skin of the face [
52,
54,
57]. Therefore, the consideration of other behavioral pain indicators encoded in specific body movements (e.g., hand on face), may provide further information about pain in premature infants with extremely low GA [
52,
56,
58].
Several studies have examined the influence of previous pain exposure on reaction to pain, but the findings do not provide a clear answer [
44]. Some studies report that infants subjected to frequent painful procedures during their hospitalization display less intense behavioral responses to heel sticks than those who have undergone fewer procedures [
46,
52,
59]. The dampened pain responses in very premature neonates may be a sign of exhaustion or a state of passivity resulting from the numerous procedures they experience during their stay in a NICU [
43,
60,
61]. Contrary to those findings, other studies suggest that repeated exposure to pain may lead either to increased pain response (hyperalgesia) or to pain responses without painful stimulus (allodynia) [
15,
62].
Few studies have investigated the influence of other contextual factors (e.g., gender, health status) on pain reactions in neonates, and of those that have, the results are inconsistent [
44]. This might be explained by methodological limitations (e.g. the comparison of different GA groups and the use of a variety of pain assessment tools) [
44]. One challenge in examining the influence of contextual factors on pain response is the associations between the individual factors [
44]; for example, extremely low GA infants have a longer stay in a NICU and are exposed to a higher number of painful procedures than more mature infants. Due to the fact that contextual factors can lead to underestimation or misjudgment of pain severity [
54,
63‐
65], further research is needed to better understand the factors that influence pain responses in neonates. Relevant contextual factors should also be considered in future pain assessment.
Study aims
The aim of this observation study is the validation of the BPSN, using a large sample of neonates spanning a full range of GAs. The validation will involve the detection of the underlying structure of the data and the examination of the concurrent validity of the BPSN with the Premature Infant Pain Profile-Revised (PIPP-R; [
26]), construct validity, interrater reliability, specificity and sensitivity. Furthermore, the variability of pain reactions over time related to behavioral and physiological patterns will be analyzed and the relationship between behavioral and physiological indicators examined. In addition, the influence of contextual factors on the variability of pain reactions across GA groups will be explored. Finally, the results of this analysis will be used for modification of the BPSN, to account for individual contextual factors in future clinical pain assessment in neonates.
Based on a previous validation study of the BPSN [
41], we hypothesize that the BPSN will be a valid and reliable pain assessment tool for premature and term infants. In addition, we expect that the impact of single contextual factors on infants’ pain reaction will be described and considered for future pain assessment. In particular, we anticipate finding a difference in pain reaction depending on GA. Moreover, we hypothesize that behavioral and physiological indicators will show low association across time and that this low association may be explained by the influence of individual contextual factors.