Pain assessment tools for use in infants: a meta-review

Pain occurs across the continuum of life. However, for those who cannot self-report their pain, such as pre-verbal children, those with intellectual disability and those with neurological disorders who have lost the ability to communicate (e.g., people living with advanced dementia), there is evidence that their pain often goes under-recognized and under-treated [1‐3]. Research clearly indicates that accurate pain assessment in infants is a challenge for health care professionals [2, 4]. The problem of differentiating or discriminating pain from other expressions of unpleasant feelings or experiences, such as hunger, fear or distress adds to the complexity of assessing pain in this group [5].

In the absence of infants’ ability to self-report, observation of behaviours such as facial expression, crying, hand and leg movements is a valid approach for pain assessment [3, 4, 6]. However, pain behaviours are not specific reflections of pain intensity or distress, and the context of infant’s behaviour must be considered when using this approach [6].

Herr et al. [6], report that typically children as young as 3 years may be able to report pain using basic tools, however it is not until 8 years of age that children are generally considered reliable in using a numerical rating scale. In clinical practice, Eccleston et al. [7], in support of Beltramini et al. [8], suggest that self-reported measures of pain (e.g., numerical scales) can be used for children aged 6 years and older, while behavioral scales can be used for children aged below 6 years or those unable to verbally report. According to the American Society of Pain Management Nursing (APSMN) position statement on pain assessment in the patient unable to self-report, health care professionals should search for potential causes of pain and use both primary behavioral categories and behavioral pain tools for initial and ongoing assessment of pain [6].

An essential element in assessment and management of pain is the identification of the type and source of pain, both of which can be challenging in the infant population [8]. The following types of pain have been identified in infants: acute pain that occurs in response to recent illness, injury or following procedures or operations, and has a limited duration; and chronic pain which commonly persists beyond the time of curing of an injury or illness, and often lasts beyond 6 months [9]. Validated assessment tools in infants have been developed mostly within research into acute pain. These included comprehensive identification of primary behavioral categories, such as facial expression, body activity/motor movement, and crying/verbalization [2]. For assessment of chronic pain, infants may demonstrate different behaviours, such as withdrawal, lack of expression, lack of interest in surroundings, and a decrease in their ability or their willingness to play [10].

Eccleston et al, [7] proposed four transformative goals to improve the lives of children and adolescents with pain and their families. These were to 1) “make pain matter”, 2) “make pain understood”, 3) “make pain visible”, and 4) “make pain better”. They pointed out that whilst on face value these goals might seem obvious and simple, the ongoing occurrence of acute pain, pain post-surgery and procedures, and chronic pain in this population suggest otherwise. Considering the above goals highlighted by Eccleston et al,[7] of particular importance remains the issue of adequately identifying and therefore assessing pain using evidence-based tools. In this regard, previous reviews have identified that while there are a number of tools that can be used in infant population, the strength of psychometric evidence to support their use varies [8, 11],. Therefore, there is an ongoing need to review evidence and therefore progress made in this area. Meta-reviews, also known as systematic reviews of systematic reviews, provide a rigorous approach for synthesizing evidence from available systematic reviews and summarizing knowledge into one accessible document [12]. Given the wide variety of pain assessment tools and uncertainty as to which assessment tools are more suitable when used in clinical or multiple settings and within different populations, we felt that a meta-review was particularly useful for our purpose. Undertaking this meta-review would provide an updated systematic identification and appraisal of the evidence concerning the validity and reliability of pain assessment tools available for use in infants aged 1-12 months old. This would then allow an evaluation of the appropriateness of these pain tools for use in different clinical settings. This age range was selected in order to exclude tools which are used exclusively for pre-term and full-term neonates.

The 10 reviews that met our inclusion criteria were assessed for methodological quality by two reviewers independently (DA, MA) using the JBI Critical Appraisal Checklist for Systematic Reviews and Research Syntheses [23]. Disagreements between the two reviewers were resolved through discussion and if needed involvement by a third reviewer (EM).

Quality assessment according to the JBI checklist tool involved 11 assessment criteria: clarity and explicitly of the review questions, inclusion criteria, search strategy, adequacy of sources and resources used, criteria for study appraisal, number of reviewers (2 or more), methods to minimize errors in data extraction, methods used for combined studies, assessment of publication bias, recommendation for policy and practice, and direction for new research (Table 2). Every criterion in the checklist was given a rating of ‘yes’, ‘no’, ‘unclear’ or ‘not applicable,’ and one point was given to every criterion rated ‘yes.’ Using the JBI checklist tool, methodological quality can be judged in terms of “low” if they failed to reach a score of > 50% on critical appraisal, the predetermined cut off score agreed upon by the research team.

Eccleston et al. [7] have set making pain visible as one of their translational goals to improve pediatric pain management. To achieve this there is a need for valid and reliable observational pain assessment tools which inform clinical decision making and can be easily incorporated into clinical practice. Therefore, there is a need for evidence which supports the psychometric properties (validity and reliability) and feasibility as well as clinical utility of such tools, which is what this meta-review aimed to synthesize. What was demonstrated is that there are over 36 pain assessment tools which can been used to evaluate pain in infants aged 1 to 12 months. However, based on currently available data, no single tool can be recommended as the “gold standard”. This confirms the need for further research and development with the focus on arriving at a pain assessment tool which has sound psychometric and clinical utility properties. For our study we adopted the NICE definition as a gold standard (https://​www.​nice.​org.​uk/​glossary?​letter=​g): “A method, procedure or measurement that is widely accepted as being the best available to test for or treat a disease”. In our case to achieve the status of gold standard it had to be considered the best available tool to measure pain in terms of reliability, validity, feasibility and usability.

Only one of the 11 reviews included in the current meta-review focused specifically on infants [30], the remainder included across a range of aged groups, such as premature and full-term neonates, toddlers, young children, and adolescents. From these reviews we extracted and analyzed data for 36 tools. The majority of the tools were multidimensional (exceptions being the NFCS-R, FPS-R, MAX and the ObVAS) and varied in their derivation, content, and rating of the pain scores. Many of the tools have been applied to children who are intubated, sedated and ventilated, including infants. Further, they have been used for acute pain, pre-, peri- and post-operative pain, procedural pain, including that associated with immunization, and chronic pain. An important lesson learnt that pain assessment tools in infants were mostly focused on acute pain, whereas chronic pain was relatively ignored. Of the tools identified in this review, the COMFORT scale was the main tool that can differentiate stages of pain and assess pain temporality. However, traditional definition of chronic pain may not apply in infant population. Therefore, our discussion in this regard has been hampered by difficulties in rectifying differences between acute and persistent/prolonged pain in definitions, as well as in their clinical picture. The fact that infants have not lived enough to meet chronic pain criteria applied to older children, and that most studies were designed to measure acute pain in predictable situations, such as painful procedures, may have contributed to the researcher’s inability to predict or adequately measure chronic pain [38]. The use of explicit pain definitions and the integration of biological and behavioral/observational evidence could ultimately inform a more accurate assessment of infants who suffer from prolonged and persistent pain and assist in pain management.

The most cited tools in the included reviews were the FLACC, COMFORT, CHEOPS, followed by the MBPS, POPS, TPPPS and PPP. The findings of this meta-review in terms of the most commonly cited tools were in line with those identified by an earlier meta-review by Andersen et al. [11]. They reported that the previous systematic reviews had most frequently recommended the FLACC, COMFORT, and CHEOPS. Yet differences between the reviews in terms of their inclusion criteria, rating system and incomplete data synthesis made it impossible to recommend one particular tool to use. Given the variety of clinical settings in which the tools may be used and the different etiologies of pain this may not be unexpected. However, further evidence of the psychometric properties and clinical application and utility of these tools in practice would allow such recommendation to be made. Routine application of the COSMIN “Guideline for selecting outcome measurement instruments for outcomes included in a Core Outcome Set” [39] in the evaluation of current and future pain assessment tools would allow potential aggregating of data, thus strengthening the evidence pool.

This meta-review found that measures of inter-rater reliability, face/content validity and construct validity are the types of measurement properties most commonly and comprehensively reported, whereas information on clinical utility and cross-cultural validity is seldom and poorly reported. While it may be argued that certain aspects of validity and reliability are more important than others, we believe that evidence should be looked at as a whole. To date, there is no hierarchy of evidence described for the different psychometric properties described by earlier studies. Therefore, the collective opinion of earlier studies is regarded as the best available source of advice on offering an evaluation about what constitutes a validated and a reliable tool. Nevertheless, while many pain scales have focused on validity related to inter-rater and intra-rater reproducibility, it could be argued that looking at more scale properties, such as responsiveness to a known treatment may be more fundamental.

As suggested by Duhn & Medves [30] further evaluation of clinical utility and feasibility of existing pain assessment tools is needed to enhance our ability to accurately assess pain in the infant population. Furthermore, evaluation of clinical utility and feasibility would provide important information in regard to the appropriateness of existing tools across different clinical settings. As suggested by Eccleston et al. [7] in clinical practice pain in infants tends to be both underdiagnosed and undertreated by clinicians. This is compound by the lack of a gold standard tool used to measure pain in infants. Certainly, a gold standard cannot be achieved without the tool having a clearly evidenced clinical utility and feasibility, in addition to sound psychometric properties.

Studies focused on tools’ clinical utility and feasibility generally have not moved forward in a systematic nor rapid way because of the logistical problems connected with use of pain assessment tools in vulnerable populations. It could also be related to the lack of clarity about what constitutes clinical utility in the literature and how it can be evaluated [40]. The length of items used in pain assessment tools, ease of use, time needed to complete/ document the assessment, and clear user instructions are examples of aspects that have the potential to influence both feasibility of use and clinical utility in clinical practice. Therefore, these should be reported by studies, in addition to other evaluation of reliability and validity. Additionally, future studies should also consider current conflicting data on utility and feasibility of tools [24, 25, 27, 29, 32, 34] by addressing issues arising from heterogenicity of studies, dimensions evaluated, heterogeneity of populations and settings that makes it difficult to conclusively evaluate clinical utility and feasibility.

It is worth noting that Patient Related Outcomes Measures (PROMs) [39] are critical to evidencing the clinical utility of pain assessment tools. Establishing valid and reliable assessment of pain in clinical practice leads to improved patient outcomes (e.g., less pain, better pain management, better quality of life). Proctor’s conceptual model of implementation science outcomes is another approach that could be used in future studies to present evidence of the feasibility of use of current pain assessment tools in infants and related outcomes [41]. The described implementation science methodology and the PROMs could be used as a framework for future studies.

We noted also that nine of the 11 reviews were conducted in Western countries and cultural validity of the pain assessment tools were only addressed in four; one review by Bai & Jiang [25] with special focus Chinese children and included the FLACC, COMFORT-B and POCIS, and those of Crellin et al. [24, 26‐28] which covered the FLACC, TPPPS and MBPS. There are several variations in the way different individuals interpret pain assessment tools as a function of their age, literacy levels, and cultural background [42]. We can argue, therefore, that the appropriateness of using one measure created in one culture or setting for another setting is a matter of concern, especially in developing countries where the social, educational, and economic context differs dramatically from Western countries. Exploration of cross-cultural validity provides suggestive evidence of potential for use within diverse cultural and linguistic context.

Given that there is still no ground truth for the measurement of pain in infants, certain aspects of validity testing are hard to estimate. COSMIN guideline[35] for systematic reviews of patient-reported outcome measures is increasingly seen as measurements that may help overcome of the above-mentioned methodological problems. Future studies can achieve greater rigor and relevance by (i) ensuring that a tool’s ‘responsiveness’ is tested against known analgesics; (ii) testing ‘construct validity’ such as ‘hypothesis testing validity’ and (iii) ensuring that the Importance/Assessment of ‘Interpretability’ is considered when constructing and testing new tools.

Considering difficulties in assessing pain in populations unable to communicate, including infants, parental involvement remains an important component for assessment of pain. Nurses and clinicians often rely on parent’s input to document infant’s behavioral responses, especially when the child is too young to verbalize their pain or symptoms[43]. Inclusion of providers and parental input to identify the early stages of pain is similar to that made previously by an earlier review[38] and is based on pediatric committee consensus [43]. There is also a growing interest in addressing such challenges by leveraging from technological advances. In this regard, a review by Zamzmi et al. [3] showed that the use of automated methods for pain analysis and recognition may be a better fit for pain assessment in infants. Therefore, considering developments in machine learning based techniques for measuring pain in this population group, it can be expected that automation of pain assessment could have a number of implications in clinical practice that will result in a) development of pain assessment tools whose automation is driven by artificial intelligence; b) digitization of pain assessment documentation; c) enabling of rapid point-of-care assessment and d) continuous monitoring of pain. Continuous monitoring and predictive algorithms developed by artificial intelligence (AI) is another approach that may assist in measuring aspects of pain, including chronic pain[42]. These developments may well have a positive impact on addressing the four transformative goals proposed by Eccleston et al. [7] and therefore improve the lives of children experiencing pain as well as their families.

The methodology approach adopted for this meta-review was a strength as it provided a comprehensive and a practical way of obtaining information on currently used pain assessment tools compared to undertaking reviews of primary studies. Another strength is the rigorous approach adopted in the review process as we attempted to minimize bias by having each review screened independently by at least two reviewers. The main limitation of this study was that identification of all pain assessment tools used in infants could not be guaranteed, as not all of them may have been included in identified systematic reviews. Furthermore, pain assessment tools not published in English may also have been missed. Whilst we checked further available data when readily available, such as in supplementary files, we did not communicate with the authors to obtain missing data. Finally, our search did not specifically target reviews reporting on the feasibility or clinical utility of the pain assessment tools, therefore our evidence may be more limited on these aspects than it might have been otherwise.

The assessment of pain in infants remains complex and challenging for practitioners and researchers, mainly because infants are unable to self-report their pain. The 11 published reviews included in this meta-review evaluated 36 tools of which seven of these were included at least by three reviews. The level of evidence reported in regard to each tool’s psychometric properties varied widely across published reviews, and only a few of these reviews reported on feasibility and clinical utility and had conflicting results or measured different dimensions of utility. Therefore, we were unable to conclude that any one tool is better than the other without proper caution. To address this there is a need for standardization of the evaluation of psychometric properties of pain assessment tools, together with a greater focus feasibility and clinical utility. Underpinning the latter is the need for accompanying PROMs data, indicating the systematic use of the tool results the best possible pain management outcomes.