Whose pain is it anyway? Comparability of pain reports from children and their parents

Published evidence points to the fact that musculoskeletal pain, and especially chronic pain in children and adolescents, is not only responsible for considerable personal suffering but also for a substantial economic burden [1]. In addition to its contemporaneous burden, pain in adolescence has also been shown to be an important predictor for pain in adulthood [2]. Musculoskeletal pain in children thereby represents a substantial societal problem, worthy of further investigation.

In comparison with the body of research pertaining to adults with musculoskeletal pain, there is a dearth of information relevant to children [3]. This is despite epidemiological evidence suggesting that prevalence rates for common conditions, such as back pain, in adolescents approach those in adults [4]. These factors point to the need for robust research into understanding the scope and consequences of pain in children. An important part of understanding more about pain in children and interpreting published research involves clarifying issues surrounding its measurement.

From a measurement perspective, the need to consider assessment of pain in children in a different manner to adults is well-accepted [5]. In recognition of this fact, the IMMPACT initiative considered their recommendations for the measurement of pain in children and adolescents separately from those for adults [6]. This is due to the different, and often rapidly evolving, physical, psychological and social factors that influence the report of pain.

Clinical assessment of children’s health conditions necessarily involves interaction with both children and their parents. The decision to seek care for a painful condition in pre-adolescent children typically rests with the parent(s) and further, once a decision to seek care has been made, parents are likely to be involved in the clinical encounter and have input into decisions about care. Thus while the child’s assessment holds a degree of inherent validity, the parents’ view is also relevant from a practical standpoint. In the context of epidemiological and clinical research into pediatric pain, a parent’s report is commonly used as a proxy for the child’s rating [7]. This is particularly relevant in longitudinal studies where children are too young to reliably self-report at the initial assessments although they might be able to do so at later follow-up points. It has been shown that the agreement between parent and child ratings is generally poor [8–10], although higher in the clinical samples than in the studies recruiting healthy children [11]. Nevertheless, reports of spinal pain in community or school based cohorts often fail to describe the degree of parents’ involvement in the data collection, lending uncertainty to the reported estimates [12]. Therefore, it is necessary to improve the understanding of the relationship between the two types of pain report, and draw attention to the importance of clear reporting.

The analyses in the present study are based on two different samples, both nested within the CHAMPS study [7]. This study was designed to assess the effects of increased physical activity in school children in a municipality in Denmark. The CHAMPS study includes twice-yearly assessment of the children with numerous clinical measures and physical assessment.

The Regional Committees on Health Research Ethics for Southern Denmark approved the CHAMPS study (ProjectID: S-20110042).

The concordance between parents’ and their child’s report of pain was generally quite poor. Children very often reported pain that was not reported by their parents, whereas the reverse was very rarely the case. When the parents did report pain, the agreement with the child regarding the site of pain was reasonably good, at close to 80 %. While it may be that concordance is better for some pain sites than others, the sample size was not large enough to investigate these differences.

The relationship between pain intensity and concordance, which reached statistical significance in the interviewed sample (Sample 1), reflects the unsurprising finding that parents were more likely to report their child’s pain when it was of greater severity. There were also differences in terms of pain duration between the concordant parent–child pairs and the discordant pairs in Sample 1. While this difference did not reach statistical significance, it was quite large (67 days) and in the expected direction i.e. better concordance for longer term pain. There was no influence of child gender or age.

Relevant research exploring the relationship between children’s and parents’ pain report comes from population based studies and from studies conducted on clinical populations. The latter generally include children experiencing acute surgical or procedural pain, or receiving medical care for a chronic condition. Population based studies are more directly comparable with the results of our study while the clinical studies often offers more in-depth insight into the issue in specific populations. In general, agreement between parent and child ratings is higher in the clinical samples than in the studies recruiting healthy children [11].

Chiwaridzo and colleagues [15] recently conducted a study addressing a similar question to ours in adolescent school-children and their parents in Zimbabwe. The level of agreement they report (kappa; 0.2) was very similar to that observed in our study, they also noted that the discordance was due to parents reporting no pain when their child reported pain. These similar findings provide support for the generalizability of these results, across countries with different socio-economic profiles. Kroner-Herwig et al. [9], Haraldstad et al. [8], Sundblad et al. [10] and Waters et al. [16] all report on community-based surveys of children and investigated the concordance between parents’ and children’s pain report. While the children filling in the questionnaires in these studies report a range of different types of pain e.g. musculoskeletal, headache, stomach-ache etc., overall the results from these studies are broadly similar to those of ours. Three of the four studies [8–10] reported generally poor agreement between parent and child ratings, while the last reported somewhat higher concordance [16]. Three studies [9, 10, 16] also found that children reported pain more often (and of greater intensity) than their parents, in the fourth study this relationship held for girls aged 12 and over but not for boys or younger girls. The influence of age and gender on concordance of the ratings was variable across the studies, but the finding that agreement between the ratings was better where the condition was more severe was reported in all studies, as in the present study.

Zhou and colleagues [17] performed a systematic review and meta-analysis of nine studies relating to the relationship between parents’ and children’s pain ratings and calculated a pooled correlation of 0.64. Studies included in the review recruited children experiencing pain due to surgery or medical procedures. Vetter [18], studying patients attending an outpatient chronic pain clinic and Cohen [11] on patients from rheumatology or pain treatment programs found moderate to good concordance between ratings of Health-related Quality of Life and physical functioning respectively. Counter to the findings of the population-based studies, in both cases parents tended to over-report severity of the impacts on their children. It is possible that the better concordance and switching from under- to over-reporting of pain reflects the greater sensitivity and concern paid by parents of sick children. Alternatively, the nature of concordance between measures of quality of life and physical function may be fundamentally different to that of pain.

In recent decades the ideal of ‘patient-centered practice’ has achieved increasing prominence. Central to this is the primacy of the patient’s own experience of their condition, treatment and outcome [19]. If we are to apply this principle to the results of our study, we assume that the child’s viewpoint is inherently valid, and may conclude that the parent’s rating is an unreliable substitute. However, the nature of the relationship, in particular the fact that parents almost never report pain when the child does not, invites further speculation. It is possible that parents are unaware of the minor aches and pains experienced by their children, or that they may apply a kind of ‘filter’, whereby they attend only to pain that they consider significant or meaningful. This conclusion is at least partly supported by the finding that higher pain intensity was associated with better concordance. If this is the case it would be of interest to know whether parent-reported, or child-reported pain served as a more useful indicator of more serious, ongoing, or future health conditions. Exploration thus of the predictive or prognostic validity of the two types of pain report could be a worthwhile target for further research.

From a clinical point of view, it can be said that parents’ report of their child’s pain is not likely to provide a reliable assessment of the child’s experience. Care should be taken though when generalizing this finding beyond the healthy population, since concordance may be better in groups of children receiving medical care for a painful condition.

Similarly the distinction between the findings from population based and clinical studies should be kept in mind when comparing the results from different studies. We can conclude based on findings from this, and previous studies, that in healthy populations parents’ report of their children’s pain is likely to underestimate the pain experience of the children. Obviously, the ‘best’ measure depends on the particular research question, but it is clear that the two measures should not be considered interchangeable. As regards to future research, exploration of the validity (e.g. concurrent, predictive) of the two types of report would inform decisions regarding whether one or the other is best suited to a particular research question.

This study has several strengths; it involves data collected from a representative sample of children attending schools in a western European country (Denmark). The data completeness is acceptable (<7 % missing data) and was collected within a large observational study, data were collected in two different ways (interview and questionnaire) and inclusion was based on availability of interviewers for Sample 1 and classroom time for Sample 2, and thus not subject to selection bias.

There are however some limitations, chief among which is the fact that mode of pain report was different for the parents and the children. Parents reported pain via SMS and children by answering questions in a face-to-face interview with a researcher or by answering a written questionnaire. The impact of this limitation is that part of the discordance observed in the study could be due to different data collection methods which would mean the real level of agreement between children and their parents is somewhat better than our findings indicate. The fact that this limitation is responsible for a systematic, rather than random bias means we can factor it in to our interpretation, that is to say that the concordance we observe in this study represents a minimum estimate of agreement between children and their parents regarding the existence of pain. The study also lacked power to explore detailed questions regarding reporting of pain at different sites and between age groups. Finally, since parents and children were not asked on the same day, there might be complaints reported by the parents which occurred after the interviews with the children. However, even if this occurred it would not alter the main conclusions of the study.

A parent’s report of the presence of pain in their child does not correspond well with the child’s own report in healthy children from 10 to 14 years of age. Children commonly report pain that is not reported by their parent, whereas the reverse is rarely the case. Not unexpectedly, parents and children are more likely to agree on the presence of pain if the pain is more intense and possibly if the pain is of a longer duration. Agreement is not influenced by age or gender of the child. Whether the one measure is more suitable than the other will depend on the particular research question, but it is clear that the two measures should not be considered interchangeable.