Article Text

Download PDFPDF

Impact of functional somatic symptoms on 5–7-year-olds' healthcare use and costs
  1. Troels Græsholt-Knudsen1,
  2. Anne Mette Skovgaard2,3,
  3. Jens Søndergaard Jensen1,
  4. Charlotte Ulrikka Rask1,4
  1. 1 Research Clinic for Functional Disorders and Psychosomatics, Aarhus University Hospital, Aarhus N, Denmark
  2. 2 Department of Public Health, University of Copenhagen, Copenhagen, Denmark
  3. 3 National Institute of Public Health, University of Southern Denmark, Odense, Denmark
  4. 4 Regional Centre for Child and Adolescent Psychiatry, Risskov, Aarhus University Hospital, Aarhus N, Denmark
  1. Correspondence to Dr Troels Græsholt-Knudsen, Marienlystvangen 129, Aarhus N 8200, Denmark; troelsgk{at}gmail.com

Abstract

Objective Knowledge about childhood functional somatic symptoms (FSS) and healthcare costs is scarce. This study aims to assess whether FSS in children aged 5–7 years are associated with increased future primary healthcare.

Design At baseline of the observational cohort study, between years 2005 and 2007, 1327 children from the Copenhagen Child Cohort were assessed at ages 5–7 years for FSS and chronic physical diseases using the Soma Assessment Interview. Information on primary healthcare use was obtained from the National Health Insurance Service Register, and measured as the price of all medical services outside the hospital during a 4.5-year follow-up period from the day of assessment. Regression with bootstrap bias-corrected and accelerated CIs were performed.

Results 1018 (76.8%) children had no FSS with primary healthcare use adjusted for other child health problems, maternal education and family changes of €448.2, 388.2–523.8 and number of face-to-face contacts: 11.90, 10.71–13.25; 250 (18.9%) had FSS with healthcare use €441.0, 355.0–550.3 and face-to-face contacts: 11.22, 9.60–12.91, and 58 (4.4%) had impairing FSS with healthcare use: €625.9, 447.9–867.8 and face-to-face contacts: 14.65, 11.20–19.00. In unadjusted regression analysis, impairing FSS were associated with increased healthcare use (increased costs: €246.0, 67.6–494.3). The adjusted association was slightly attenuated (increased costs: €177.8, 1.3–417.0).

Conclusions Impairing FSS in children aged 5–7 years is a predictor for the child's future primary healthcare use. More research on complex predictive models is needed to further explore the clinical significance of these results, and to contribute to the underpinning of early interventions towards impairing FSS in children.

  • Epidemiology
  • General Paediatrics
  • Health services research
  • Somatoform disorders
  • Medically unexplained symptoms

Statistics from Altmetric.com

What is already known on this topic?

  • Functional somatic symptoms (FSS) are common health complaints in adults, which are related to chronic health complaints and frequent and costly use of health services.

  • Prospective studies on young children with FSS and their healthcare use are lacking.

What this study adds?

  • FSS in children aged 5–7 years involving functional impairment are associated with increased primary healthcare use later in childhood.

  • Further research is needed to explore contributing factors to healthcare use, and feasibility of early intervention.

Introduction

Functional somatic symptoms (FSS) are defined by subjective symptoms that cannot be explained by well-defined organic pathology.

In adults, FSS, often diagnosed as somatoform disorders or functional somatic syndromes, are associated with frequent and costly use of health services,1 ,2 which may lead to increased morbidity due to excessive investigations and overtreatment.3

In childhood, FSS are common health complaints;4 a population-based study found a 1-year prevalence among children aged 5–7 years of 23%, with symptoms impairing daily life functioning in 4% of the sample.5

FSS in children accounts for 8%–10% of visits in primary care facilities,6 and may be associated with costly and potentially harmful medical investigations and treatments.7 Thus, it is important to elucidate the prospective relationship between early-onset FSS and healthcare use. However, studies in this area are few, and prospective findings are scarce but needed.8 Previous studies have found increased use of overall healthcare9 ,10 or primary healthcare11 ,12 in children and adolescents with various FSS. Other health factors associated are multisymptomatic presentation of FSS,5 past healthcare use,5 health during the past year5 ,13 and well-defined physical illness and mental problems.9 ,11 General factors are the child’s gender,14 age,8 ,15 absence from school,9 ,11 parental health and psychopathology, parents' own healthcare use and socioeconomic status8–10 ,13–19 as well as ethnicity.10

In sum, these findings suggest a complex explanatory model of healthcare use that includes both FSS and child and family covariates. However, most available studies are limited by their cross-sectional design and some focus only on specific symptoms rather than the full spectrum of FSS,11 ,12 not taking into account that various FSS co-occur in young children20 ,21 and different symptom presentations may represent a common underlying phenomenon.22

The present study aimed to investigate the prospective healthcare use in children assessed with various FSS at ages 5–7 years.

We hypothesised that (1) FSS would predict a higher future healthcare use and that (2) the size of this effect would be influenced by the degree of FSS impairment and other mental and physical health problems in child and family related factors.

Method

Sample

The basic study population was the Copenhagen Child Cohort 2000. The cohort comprises all 6090 children born in the former county of Copenhagen in 2000. Details of the cohort are reported elsewhere.23 For a 5-year to 7-year follow-up, a random sample of 3000 cohort children was selected for comprehensive assessment, as this sample size was considered sufficient to estimate the population prevalence of psychiatric disease (data used elsewhere).24 Complete data on FSS were available for 1327 children; one child died during the following 4.5 years, leaving 1326 in the final sample, which has been described previously.5 ,21 ,25 ,26

Measures and constructs

An overview of all the variables used in the present study is given in table 1.

Table 1

Variables, definitions, sources and time of measurement; N=1326

Baseline variables assessed in children aged 5–7 years

Child variables.

Functional somatic symptoms

FSS were assessed by the Soma Assessment Interview for parents. The layman-conducted Soma Assessment Interview combine highly structured and open-ended questions concerning the child’s physical health and physical complaints, including the duration and impact of possible medically unexplained symptoms. Information is subsequently reviewed by an experienced clinician to decide on FSS status. The Soma Assessment Interview has been validated elsewhere; the full instrument is available online.28 A previous categorisation of FSS as no FSS, non-impairing FSS and impairing FSS was used.21 Impairment was defined as parent’s report of a substantial, negative influence on at least one of eight different aspects of the child’s life, including visits to the doctor and hospitalisation due to the symptoms, distress, absence from school or day care and/or interference with social, leisure, family life and/or learning activities.

Child's health during the past year

The child’s health during the past year was based on the corresponding item in the Soma Assessment Interview, and dichotomised into ‘good’ and ‘bad’, with ‘bad’ including ‘poor’ and ‘bad’.

Chronic physical disease

In the Soma Assessment Interview, parents were asked if the child had received any diagnosis of a serious and chronic physical disease by a medical doctor. A predefined checklist, including ‘other serious physical diseases’ was used in the interview. If any item was answered with yes, chronic physical disease was defined to be present.

Mental health problems

The parent’s form of the Strengths and Difficulties Questionnaire (SDQ) was used to assess the child’s mental health status. The SDQ is based on 25 items, which generates a total score that can be used for screening for mental problems. The psychometric properties of the Danish version have been evaluated29 and our data were interpreted using Danish norms.30 In accordance with previous practice,14 we dichotomised the score into ‘no’ and ‘yes’, where ‘yes’ consisted of children with ‘borderline’ and ‘abnormal’ total scores.

Family variables

Information on parental country of birth (used as a proxy to estimate ethnicity), maternal education, family income and family changes since the child’s birth was obtained from national population registers (table 1). All Danish citizens are given a personal identity number, which enables the linkage of individual data across registers.31

Outcome variable: healthcare use

Primary healthcare use, including general practitioner (GP) and other medical specialties outside hospital settings, was used as measure of healthcare use. All Danish citizens are entitled to free medical care from public insurance. The GP serves as gatekeeper of referrals to all other health services. Data on reimbursements, maintained by Statistics Denmark, are considered to have good coverage.32

In order to obtain a variable equivalent to the direct costs of primary healthcare, we used the reimbursed price of all consultations and services, including laboratory analyses, delivered by GPs and other medical and surgical specialties outside hospital settings. Currency was converted into Euro, and prices were adjusted to 2014 price levels using data from historic pricing agreements (unpublished data, obtained from the Danish Medical Association, 2014). For laboratory tests, historic prices were not available, and the general consumer index at the 2013-level from Statistics Denmark was used instead.33 Data were acquired from the time of the Soma Assessment Interview and 4.5 years prospectively.

Face-to-face contacts during the same follow-up and including all services covered in the primary outcome were also used as an alternative outcome. As part of reimbursement data, a unique code describing the service given is included in the register. All such codes in the sample given during follow-up were categorised to determine whether a face-to-face consultation had taken place. Any such contact, during which more than one service given to the patient, was counted as only one face-to-face visit.

The variable past healthcare use was constructed in the same way as the primary outcome, but using data on healthcare use from birth and 4 years prospectively.

Statistical analyses

To assess selection bias, attrition analyses were carried out, based on the whole random sample and conducted after running the main model. Comparisons were carried out using either generalised linear models estimating relative risk or Wilcoxon rank-sum test, where assumptions of linearity were violated on continuous variables.

To study covariates separately, crude analyses of healthcare use were made using the non-parametric Kruskall-Wallis test.

Linear regression was used to estimate whether FSS predicted the use of healthcare, measured as costs and number of face-to-face contacts. To study FSS both as an individual clinical predictor of increased healthcare use, and the relationship to other predictors, two models were applied: an unadjusted regression model with FSS as the only predictor; and an adjusted model simultaneously adding FSS and child covariates (parent reported child's health during past year, chronic physical disease, mental health problems, age and sex) and family covariates (family changes and maternal education). In the adjusted model, only records with data on all variables were used. Because of the relatively low number of cases with impairing FSS (n=58), most covariates were dichotomised. Cut-off points were chosen aiming to use clinically relevant categories and at the same time avoiding categories with very few cases. The number of covariates was restricted to those judged to be potentially important explanatory factors based on prior knowledge.8–15 ,34–38 As we have previously found that higher past healthcare use is associated with being a consulter for FSS in children aged 5–7 years, previous healthcare use was omitted in the main model as a covariate in order to avoid overadjustment bias,39 but included in a sensitivity analysis. Model validity was assessed using graphical inspection of residuals. Due to the skewed distribution of the residuals, as could be expected from economic consumption data, bootstrapping was used as recommended1 to produce bias corrected and accelerated 95% bootstrap CIs (using 5000 repeats). Accordingly, the resulting CIs instead of p values were used to determine strength of evidence.

To assess robustness of findings, sensitivity analyses were performed focusing on variables suspected of biasing results or representing subpopulations: excluding children with both parents born outside Denmark; excluding children with chronic disease; excluding the SDQ item on pain complaints in the total SDQ score (pain is also assessed to determine the outcome); using an altered FSS variable excluding GP and hospital use for FSS as criteria for impairment and including family income as a covariate.

Two-tailed p values <0.05 were considered strong evidence. In attrition analyses, owing to multiple comparisons, two-tailed p values <0.01 were considered strong evidence. All statistical analyses were carried out in Stata V.13 except parts of attrition analyses, which were carried out using R project software.

Ethics

The study was approved by the Scientific Ethics Committee of Copenhagen County (KA-05103) and the Danish Data Protection Agency (2012-58-0004, RHP-2013-001). All parents gave written, informed consent and received no stipend.

Results

FSS and attrition

Data on FSS and healthcare use during 4.5 years were available on 1326 (44.2%) out of the 3000 children. Of those, 250 (18.9%) children were categorised with non-impairing FSS and a further 58 (4.4%) children categorised with impairing FSS, in total 308 (23.2%) children with FSS. Among these, 16 cases with non-impairing FSS (6%) and 21 with impairing FSS (36%) reported frequent pain, with the extremities being the most common location. Participants more often had mothers with longer education, higher family income, non-immigrant parents and increased healthcare costs during the follow-up (table 2).

Table 2

Characteristics of participants and non-participants

Crude analyses

There was strong evidence of an association between FSS categories and primary healthcare use over the next 4.5 years, p=0.04 (table 3).

Table 3

Crude analysis of prospective healthcare costs during a 4.5-year follow-up in relation to different baseline variables

Regression analyses

Among children with no FSS, absolute, adjusted numbers showed a mean use of healthcare of €448.2 (388.2–523.8). Among children with FSS, use of healthcare was €441.0 (355.0–550.3) and among children with impairing FSS, €625.9 (447.9–867.8) (further results available on request).

Unadjusted regression analyses showed accumulated, increased costs amounting to €246.0 (67.6–494.3) during the 4.5 years in children with impairing FSS (table 4).

Table 4

Regression on FSS and other potential explanatory covariates for prospective healthcare use (unadjusted: N=1326, adjusted: N=1216)

In the adjusted model with relevant covariates included, the association between impairing FSS and future healthcare costs was attenuated, but remained strong with increased costs of €177.8 (1.3–417.0) (table 4).

Among children with no FSS, absolute, adjusted numbers showed a mean number of face-to-face contacts of 11.90 (10.71–13.25). Among children with FSS, face-to-face contacts was 11.22 (9.60–12.91) and among children with impairing FSS, 14.65 (11.20–19.00) (further results available on request).

Face-to-face contacts in unadjusted regression analyses showed accumulated increased number of visits by 4.42 (1.05–8.79); when adjusted, the association between impairing FSS and accumulated number of visits was attenuated: 2.76 (−0.43–7.09) (table 5).

Table 5

Secondary outcome: regression for frequency of prospective healthcare face-to-face contacts (N=1216)

Only small differences were seen in sensitivity analyses excluding children with immigrated parents, using an alternative classification of impairing FSS, using SDQ score without the specific pain item and using family income as a covariate (analyses available on request). Exclusion of children with chronic physical disease weakened the adjusted association: €113.0 (−34.6–316.2) (analyses available on request).

Discussion

The main finding was that impairing FSS but not non-impairing FSS in children aged 5–7 years were associated with increased primary healthcare costs during the subsequent 4.5 years, after adjustment for covariates.

To our knowledge, this is the first prospective general population study of healthcare use in young children based on a thorough assessment of the full spectrum of FSS, contemporary mental health problems and a highly reliable record of healthcare use. The study adds to current knowledge by supporting that impairing FSS in young children are associated with increased healthcare use later in childhood.

Other research are cross-sectional studies,6 ,8–12 ,15 ,19 ,34 ,36 ,37 or studies on particular symptoms rather than the full spectrum of FSS.10 ,12 ,13 ,15 ,16 ,18 ,19 ,36 ,37 Furthermore, all other studies used the number of contacts to various specialties or any contact versus no contact as outcome5 ,6 ,8–14 ,16 ,18 ,19 ,34 ,35 ,38 as opposed to financial costs like the present study (table 5 can be used to ease comparability). We chose accumulated costs because they combine qualitatively different services into a single variable.1 In our data, number of contacts represents a number of highly different health-related events, with all events counting one, regardless of related stress and use of resources for patient and healthcare system.

Our results on additional covariates are largely in concordance with findings from other prospective studies—showing that other health problems, measured as physical disease13 and mental health problems,14 predict increased healthcare use.

Our findings regarding childhood FSS correspond to findings from studies of adults with FSS in whom moderate and severe chronic FSS are associated with high healthcare use.40 It has been indicated that FSS may be stable across time, but can be ameliorated with early action.41 Intervention on health behaviour in affected families might potentially prevent progression of childhood FSS into chronic and impairing symptoms.

Strengths and limitations

The strengths of the study are the prospective investigation of a large population-based sample of young children with detailed FSS assessment, including evaluation of several areas of associated impact, and the independent measurement of healthcare costs, based on highly reliable register data, within a long follow-up period.

Potential limitations are the attrition and the participation bias. The results might most accurately describe children from families with both parents born in Denmark, medium-income to high-income family and a mother with a high educational level, and thus might not be generalised to high-risk families. However, a relatively low participation of families of higher social risks, and an overall rate of participation below 50% (in this study 44.2%) is seen in many population-based epidemiological studies.42

The adjusted regression analysis included only 1216 observations of 1326 available, due to missing data. The distribution of FSS classification among observations with missing data was similar to the distribution among included observations (data available on request). No evidence of differences in the median use of healthcare between observations with missing data and those with all data were seen in a Mann-Whitney U test, p=0.53. Furthermore, as the authors had no reason to believe that missings in any of the covariates would be related to healthcare use, bias from missing cases is unlikely, but they could influence the final analysis with regard to a loss of strength.

The weakened association between impairing FSS and healthcare use when using frequency in the analyses is a cause of concern, as it complicates the interpretation of our results. Therefore, our findings need replication in larger studies.

The weakened association in analyses excluding children with chronic physical disease might indicate that children with high medical costs were influencing our results. However, the size of the impairing FSS group was reduced from 58 to 47, which could weaken the strength of evidence for the association by itself. Supporting this explanation, the association between impairing FSS and healthcare use remained strong if chronic physical disease was included as a covariate.

Private healthcare services not covered by the public insurance were excluded as these services constitute only a minority of all provided healthcare services to children in Denmark.

Due to the lack of information on parents' own healthcare use and physical and mental health status, we could not explore potential effects of these issues on the child’s healthcare pattern.

Finally, potential limitations due to small sample size have to be considered. The width of the CIs for impairing FSS may have been caused by (1) lack of power because of the small number of children (n=58) and/or (2) overadjustment bias39 caused by including the potential mediator mental health problems as an explanatory variable. A post hoc model was run, which supported this explanation (analysis available on request).

Conclusion

The present study adds to the still small body of research on healthcare use in children with FSS. The main findings indicate that like in adults, impairing FSS may independently entail increased future healthcare costs in young children. The findings suggest that interventions focusing on change of illness behaviour in families with children suffering from impairing FSS could potentially reduce future morbidity and costs of FSS. However, more large-scale longitudinal studies are needed to replicate our findings and further examine the contribution of personal, family and environmental risk factors.

Acknowledgments

We want to thank Janne Tidselbak Larsen, research assistant, MSc, affiliated with National Centre for Register-Based Research, Aarhus University, Aarhus V, Denmark and Lundbeck Foundation Initiative for Integrative Psychiatric Research, iPSYCH, Aarhus University, Aarhus, Denmark, for her valuable contributions of methods to calculate the variables on family changes, family income and mother's years of education.

References

Footnotes

  • Contributors TGK took part in designing the study, carried out data preparation, took part in the analysis and interpretation of data, drafted the initial manuscript and prepared the final manuscript as submitted. AMS was initiator of the CCC2000, took part in gathering the data, took part in designing the study and critically revised the manuscript. JSJ took part in the statistical design of the study, supervised data preparation, took part in the statistical analysis and interpretation of data and critically revised the manuscript. CUR took part in gathering the data, took part in designing the study, supervised data preparation, analysis and interpretation and critically revised the manuscript.

  • Funding Steenbeck's Grant, Ulla and Mogens Folmer Andersen's Fund (798815 JGL/bkh), Mrs C Hermansen's Memorial Fund, Ville Heise's Grant (M01-13), The Beatrice Surovell Haskell Fund for Child Mental Health Research of Copenhagen (11531), Rosalie Petersen's Fund (020432-0001 hvh/hvh/sra), Fund of 1870 (200652).

  • Competing interests TG-K reports grants from Fund of 1870, grants from Mrs C. Hermansen's Memorial Fund, grants from Steenbeck's Grant, grants from The Beatrice Surovell Haskell Fund for Child Mental Health Research of Copenhagen, grants from Ulla and Mogens Folmer Andersen's Fund, grants from Ville Heise's Grant, during the conduct of the study.

  • Ethics approval Scientific Ethics Committee of Copenhagen County (KA-05103) and the Danish Data Protection Agency (2012-58-0004, RHP-2013-001).

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Additional data on frequency of use within subspecialties can be obtained from the corresponding author on request.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • Atoms
    Martin Ward Platt