Journal of Pediatric Rehabilitation Medicine: An Interdisciplinary Approach 7 (2014) 197–206 DOI 10.3233/PRM-140289 IOS Press
197
Chronic pain in children and adolescents: 24–42 month outcomes of an inpatient/day hospital interdisciplinary pain rehabilitation program Gerard A. Baneza,b,c,∗, Thomas W. Fraziera , Andrea A. Wojtowicza , Kristen Buchannana, Douglas E. Henrya,c and Ethan Benorea,b a
Pediatric Pain Rehabilitation Program, Cleveland Clinic, Cleveland, OH, USA Center for Pediatric Behavioral Health, Cleveland Clinic, Cleveland, OH, USA c Department of Developmental and Rehabilitation Pediatrics, Cleveland Clinic, Cleveland, OH, USA b
Accepted 10 February 2014
Abstract. PURPOSE: The purpose of this research was to describe 24–42 month outcomes of a combined inpatient/day hospital interdisciplinary pain rehabilitation program for children and adolescents with chronic pain and functional disability. METHODS: One-hundred-seventy-three children and adolescents (mean age = 15.1 years, SD = 2.5) were treated in the threeweek program. Mixed effects regression models (MERM) examined changes over time in pain ratings (0–10), school days missed by children and work days missed by parents in the preceding week, and the number of days hospitalized in the preceding month. RESULTS: Participants reported a significant decline in pain from admission to 24–42 month follow-up. The largest declines in pain occurred from discharge to 1-month follow-up (6.20 vs. 4.81 on a 0–10 numerical rating scale) and from 12 months to 24–42 month follow-up (4.90 vs. 3.56). Two distinct trajectories of treatment response were identified using growth mixture modeling: children with initially high pain ratings exhibited large reductions in pain ratings, while those with lower pain ratings at admission showed minimal reductions. Treatment resulted in significant reductions in school and work days missed and the number of days hospitalized (all p’s < 0.001), with reductions evident at 1-month follow-up and maintained through 24–42 month follow-up. CONCLUSIONS: These results suggest that interdisciplinary pain rehabilitation is a promising approach to chronic pain and associated disability in children, with enduring improvements found 24–42 months following program completion. Distinct trajectories of treatment response were identified. Keywords: Adolescents, pain, interdisciplinary health team, rehabilitation, outcomes assessment
1. Background Chronic pain in children and adolescents is common, with a prevalence of at least 15% and as high as 40% [1,2]. A significant subset of these patients ∗ Corresponding author: Gerard A. Banez, Pediatric Pain Rehabilitation Program, Cleveland Clinic Children’s Hospital for Rehabilitation, 2801 MLK Jr. Drive/CR 11, Cleveland, OH 44104, USA. Tel.: +1 216 448 6253; Fax: +1 216 448 6207; E-mail: [email protected].
experiences a downward spiral of worsening disability. They do not attend school, interact with peers, or participate in sports, extracurricular and other personal/family activities. The existing literature on medical, physical, and psychosocial treatments suggests that these treatments may be helpful to some children with chronic pain, but they are not effective for all patients and do not typically eliminate pain [3–9]. For severely affected children, an interdisciplinary rehabilitation approach provides a useful model of care. In
c 2014 – IOS Press and the authors. All rights reserved 1874-5393/14/$27.50
198
G.A. Banez et al. / Chronic pain in children and adolescents
a rehabilitation model, pain is accepted as a symptom that may/may not be eliminated. The focus of care is on independent functioning, improved coping, and increased self-efficacy [10]. For children who have shown little response to unidisciplinary and symptomfocused strategies, this approach may be particularly helpful [11,12]. A number of studies, identified by computer and manual methods, have examined the utility of a rehabilitation approach to children and adolescents with chronic pain and functional disability [3,13–18]. These investigations have varied in terms of delivery setting (e.g., inpatient vs. day hospital), treatment modalities, patients served, and follow-up length. Three of the seven studies [13,15,17] examined an inpatient treatment program, while the remainder assessed a day hospital or outpatient treatment. In five studies [3,13, 15,17,18], multimodal treatment approaches, including physical and psychological therapies, were examined. The other investigations evaluated intense exercise therapy [16] and a combination of physical therapy and cognitive-behavioral therapy [14]. Patient types have varied. Four investigator groups [14,16– 18] treated only patients with complex regional pain syndrome (CRPS), while others treated a variety of pain diagnoses, including CRPS, headaches, abdominal pain, and fibromyalgia. Follow-up has ranged from three months to over five years post-treatment but has typically been two years or less [16]. Overall, the findings support the effectiveness of a rehabilitation approach. Numerous positive outcomes have been reported, including improvements in pain intensity, physical functioning, school attendance, anxiety, and depression; less parental anxiety, depression, and illness; and decreased medication usage. The limitations of the existing literature include: (1) no evaluations of interdisciplinary programs that combine inpatient and day hospital treatment, (2) limited attention to functional outcomes for parents (e.g., work days missed by parent due to children’s pain), and (3) limited longterm follow-up. There has been no investigation of individual response patterns or the influence of demographic/clinical characteristics (e.g., age, sex, pain chronicity) on treatment outcomes. The purpose of this report is to describe 24– 42 month follow-up outcomes of children and adolescents treated in a three-week, combined inpatient/day hospital interdisciplinary pain rehabilitation program. Children were treated as inpatients in their first two weeks and as day hospital patients in the third week. The combined inpatient/day hospital approach pro-
vided for the benefits of inpatient admission (e.g., increased control of activity, diet, sleep; consistent treatment philosophy; separation of children and parents to interrupt maladaptive interaction patterns and promote healthy behaviors), while facilitating transition into home, school, and community during the day hospital week. The program integrated rehabilitation therapies, behavioral health services, and medical subspecialty care in an individualized, coordinated manner. Children presented with a range of pain diagnoses, including CRPS, headaches, abdominal pain, and fibromyalgia. Follow-up focused on functional outcomes for children and their parents at discharge and at 1, 12, and 24–42 months following discharge. This research built on prior investigations and examined the following questions: (1) Do children and adolescents with varied chronic pain diagnoses show improvement in subjective pain ratings following participation in a combined inpatient/day hospital interdisciplinary pain rehabilitation program? As noted, past studies have focused more on day hospital treatment, and no studies have examined a combined inpatient/day hospital approach designed to facilitate a successful transition from inpatient treatment to home. (2) Can improvements be measured in functional outcomes for children and their parents (e.g., parent work days missed)? Previous research has focused primarily on child outcomes, whereas the present investigation assessed indices of everyday functioning for both children and their parents. (3) Are improvements in pain and functioning sustained beyond two years post-discharge? The 24– 42 month outcomes assessed in this research extend beyond the average two years or less followup in past studies. (4) Are there profiles of treatment response that suggest that demographic (age, sex) or clinical (pain chronicity, diagnosis) characteristics moderate outcomes? Investigation of patient profiles and subgroups of treatment response is important for understanding which individuals respond best to this particular treatment approach. We hypothesized that improvement in pain and functioning would occur during treatment and continue beyond the three-week treatment period. Improvements were expected to be seen in pain ratings and functional outcomes and to extend beyond two years. We also hypothesized that at least two distinct trajectories
G.A. Banez et al. / Chronic pain in children and adolescents
199
Table 1 Sample demographic and clinical characteristics for pain rehabilitation patients admitted from 2007 to 2009 Sample characteristics N Age (M, SD) [range] Sex (N, % male) Pain chronicity (median, inter-quartile range) Diagnosis (N, %) CRPS Headache Abdominal pain Fibromyalgia Other Pain rating (M, SD) [range] School days missed (last week) Caregiver work days missed (last week)
Admission 173 15.0 (2.5) [8–22] 46 (26.6) 16.5 (7–37)
Discharge 173 15.1 (2.5) [8–22] − −
60 (34.7) 38 (22.0) 22 (12.7) 21 (12.1) 32 (18.5) 5.8 (3.2) [0–10] 3.4 (2.1) [0–5] 2.7 (2.8) [0–7]
−
6.2 (2.9) [0–10] − −
Note: School days missed and caregiver work days missed are not relevant for discharge due to program participation.
of treatment response would be identified: some children and adolescents would show a strong sustained response and others would show modest treatment outcomes. Our investigation of demographic/clinical characteristics was exploratory. We hypothesized that children with lower baseline pain chronicity and those with CRPS would show a stronger treatment response.
2. Methods 2.1. Participants Our sample consisted of a consecutive patient series of one hundred seventy-three children and adolescents (mean age = 15.1 years, SD = 2.5; 46 male) treated in a three-week, combined inpatient/day hospital interdisciplinary pain rehabilitation program. CRPS (n = 60) was the most common referral diagnosis, followed by headache (n = 38), abdominal pain (n = 22), fibromyalgia (n = 21), and other pain conditions (e.g., back pain, limb pain; n = 32). The mean chronicity of pain, as assessed at admission, was 16.5 months. Nearly all patients had previously undergone extensive medical evaluation and attempted a broad range of surgeries, pain management procedures, and complementary and alternative strategies. The data presented in this report were collected as part of an Institutional Review Board-approved registry, and informed consent for all patients was obtained on the date of admission. Table 1 presents demographic and clinical characteristics of our sample at admission. 2.2. Program description The Cleveland Clinic Pediatric Pain Rehabilitation Program was designed to assist children and adoles-
cents with chronic pain that interferes with normal activities. The program goals are: (1) to help children manage their pain and (2) to restore daily activity. An emphasis is placed on return to age-appropriate activity despite pain. The program blends rehabilitation therapies, behavioral health, and pediatric subspecialty care, and differs from previously evaluated programs in that it consists of inpatient and day hospital components. In the first two weeks of the program, patients are admitted to an inpatient rehabilitation unit where they sleep and are often treated apart from their parents. The inpatient component of the program facilitates control of the therapeutic environment (e.g., amount of activity, diet, sleep schedule) and immerses patients into a shared treatment philosophy emphasizing functioning despite pain. Separation of children and parents also serves to interrupt maladaptive interactions that are maintaining or worsening pain behaviors and facilitates acquisition of more healthy behavioral habits. During the day hospital week, children participate as outpatients but return to their parents at the end of each day. They continue the activities of the first two weeks but have more opportunities to apply their new skills in situations outside the hospital. The final week emphasizes preparation for potential obstacles to normal functioning and successful transition to home, school, and other activities. Patients spend 7–8 hours in treatment daily, with services scheduled hourly from 8 a.m. to 4 p.m. Patients are seen individually or in groups for physical, occupational, and aquatic therapies three hours/day on average. Therapy services promote stretching, strengthening, and endurance. As indicated, sensory destimulation, postural alignment/body awareness, durable medical equipment assessment, and kinesiotaping are used. Each patient is given evening exercises to per-
200
G.A. Banez et al. / Chronic pain in children and adolescents
form, and an individualized home exercise program is developed prior to discharge. Patients are also scheduled for an average of three individual/family psychological treatment sessions/week and participate in a mind-body skills training group (e.g., progressive muscle relaxation, diaphragmatic breathing, cognitive self-statements, aromatherapy, acupressure) three times/week. These services are designed to support patients as they participate in their rehabilitation therapies, enhance pain management skills and resources, and facilitate improved emotional adjustment and familial functioning. To promote increased health and wellness, each patient has an individualized functional plan that identifies specific behavioral treatment goals. Patients also participate in a school program one to two hours/day and in leisure education and recreation therapy groups at least one hour daily. Parent/family education plays a critical role in the program and is provided individually and in group sessions. Emphasis is placed on the physical and psychological aspects of pain, associated disability, and its treatment. Guidelines for managing pain behaviors are presented, with attention to the importance of encouraging normal activity and discouraging pain behaviors. Medical and nursing visits occur daily. Medical/nursing staff assist in planning and supervising treatment and consult on daily care as needed. 2.3. Measure A 12-item clinical outcomes measure was designed for the purposes of this evaluation. Four items/outcomes were selected for this particular research investigation: (1) pain intensity rating, (2) number of school days missed by the patient over the past week, (3) number of work days missed by the parent over the past week due to adolescent’s pain, and (4) number of days in which the patient was hospitalized for pain in the past month. These outcomes were chosen because of their clinical importance and are consistent with PedIMMPACT’s recommendation that measures of pain intensity, physical functioning, and role functioning be considered when evaluating pediatric chronic pain treatments [19]. As recommended by IMMPACT [20], an 11-point (0–10) numerical rating scale (NRS) was used to measure pain intensity. Patients were asked to rate their pain during the last 24 hours on a scale from 0–10, with 0 meaning ‘No pain’ and ‘10’ meaning ‘Maximum or worst possible pain’ [21]. 2.4. Procedure The study design consisted of a restrospective chart review (admission and discharge data) and follow-up
telephone survey (1, 12, 24–42 months following discharge). Patients and parents completed the 12-item outcomes measure at admission, discharge, and by telephone post-discharge. 2.5. Statistical analyses Univariate analysis of variance and Pearson chisquare analyses examined whether baseline clinical characteristics differed across pain categories (CRPS, headache, abdominal pain, fibromyalgia, and other diagnosis). Due to the longitudinal and observational nature of this study, significant missing data were anticipated. To handle missing data, we examined the quantity and pattern of missingness. Because missing data were expected to be largest for later follow-up time points (24– 42 months), data from this time period were collapsed into a single set of observations for each patient, while data for admission, discharge, 1-month, and 12-months follow-up remained separate. Univariate analysis of variance and Pearson chi-square examined whether individuals with/without missing data differed on clinical/demographic characteristics. To accommodate missing data across time points and avoid the problems of last- observation-carriedforward, mixed effects regression models [22–24] were used to estimate changes over time in outcome variables. These models are advantageous to standard repeated measures analyses of variance because participants are not required to have data from all time points, unevenly spaced time points are treated appropriately, and covariance structures of outcome measures collected over time are directly estimated. Mixed effects regression models typically produce more accurate, powerful estimates of changes over time than repeated measures analyses using listwise deletion as they are robust when data are missing at random [25]. The missing at random assumption posits that the data are missing dependent on the covariates and outcomes [26]. Thus, we explicitly evaluated missing data for each outcome and included a priori relevant covariates in mixed effects models. To examine change over time in pain ratings, a mixed effects regression model (MERM) was computed with time (admission, discharge, 1-month post discharge, 12-months post discharge, and 24–42 months post-discharge) as a time-varying fixed-effects covariate and age at admission, gender, and baseline pain chronicity as time-invariant fixed-effects covariates. We also evaluated whether patients showed significant
G.A. Banez et al. / Chronic pain in children and adolescents
variability in pain ratings at admission (random intercept) and variability in change-over-time (random slope). Model fit was considered by iteratively examining the above random effects and alternative covariance structures [22,23,27]. Based on findings of significant variability in change -over-time (significant variability in slopes), a series of growth mixture models were computed to examine whether distinct patterns of change in pain ratings could be identified for sub-groups. Single class, 2-class, and 3-class growth mixture models were estimated. The Bayesian Information Criterion (BIC) and Akaike Information Criterion (AIC) were used to determine model fit. A growth model was computed to further delineate the pattern of change in pain ratings and to examine the effects of baseline characteristics on initial pain status and linear changes over time in pain ratings. To examine change over time in school and work days missed and hospitalization days in the past month, exploratory mixed effects regression models were computed with the same covariates as described above for pain ratings.
3. Results Missing pain ratings across study time points were as follows: admission n = 7 (4.0%), discharge n = 22 (12.7%), 1 month follow-up n = 124 (71.7%), 1 year follow-up n = 114 (65.9%), 24–42 month follow-up n = 94 (54.3%). Although missing data were substantial when viewed at individual time points, the majority of adolescents had data available from at least one follow-up time point (n=148 out of 173, 85.5%). Children and adolescents with any missing data at admission or discharge did not significantly differ from those with complete data in terms of sex (X2 [1] = 2.14, p = 0.144), diagnosis (X2 [4] = 6.42, p = 0.170), pain chronicity at admission (t[168] = 0.39, p = 0.701), school days missed (t[64] = 0.96, p = 0.363), or parent work days missed in the week prior to admission (t[50] = 1.08, p = 0.288). However, children with missing data at admission or discharge tended to be slightly older (t[171] = 2.15, p = 0.042) than those with complete data at those time points. Covarying for age, sex, and pain chronicity did not substantively alter the overall pattern of changes in pain ratings over time in mixed effects regression models. Results are presented without covariate adjustment. Children without data at any follow-up time point did not differ from those with at least one follow-up data point on age or sex (X2 [1]
201
Fig. 1. Pain ratings (M ± 95% CI) from admission through follow-up, separately by low and high pain chronicity. (Colours are visible in the online version of the article; http://dx.doi.org/10.3233/ PRM-140289)
= 0.10, p = 0.751). However, children with missing follow-up data tended to have shorter pain chronicity at admission, suggesting that follow-up data were biased toward those with longer pain chronicity. As such, outcomes analyses may underestimate the actual effects of treatment. Children with missing follow-up data were also more likely to be diagnosed with CRPS, abdominal pain, or fall in the other diagnosis category (X2 [4] = 13.53, p = 0.009). Overall, this pattern reinforces the choice of mixed effects regression models, which account for all available data rather than lastobservation-carried-forward or pairwise/listwise deletion approaches. 3.1. Pain ratings Figure 1 presents pain ratings from admission through 42 months, separately by low ( 12 months) and high (> 12 months) pain chronicity. There was significant variability in pain ratings at baseline (random intercept = 3.46, SE = 0.67, z = 5.19, p < 0.001). Overall, pain ratings showed a significant decline over time in both MERM and growth models (Time F[4,358] = 12.19, p < 0.001; Growth model slope = −0.08, SE = 0.012, t[10] = −6.22, p < 0.001; see Supplementary Table 1). Individual patient changes-over-time in pain ratings showed significant variability (random slope = 6.11, SE = 0.47, z =
202
G.A. Banez et al. / Chronic pain in children and adolescents
Pain Rang
Fig. 2. Pain ratings across the study follow-up period by growth mixture model classifications.
10 9 8 7 6 5 4 3 2 1 0
Pain Chronicity 12 Months
Fig. 3. Proportions of patients receiving specific diagnoses by empirical classifications. (Colours are visible in the online version of the article; http://dx.doi.org/10.3233/PRM-140289)
12.88, p < 0.001). The average pain rating at admission was estimated as moderate to high (MERM estimate = 5.78, Growth model estimate = 5.89). Substantial declines in pain ratings (0–10) were observed from discharge to 1 month follow-up (6.20 vs. 4.81) and from 12 months to 24–42-month follow-up (4.90 vs. 3.56). In the growth model, longer baseline pain chronicity was significantly associated with smaller reductions over time in pain ratings (t[10] = 3.96, p < 0.001; see Fig. 1). 3.2. Sub-groups of treatment response Growth mixture model results indicated that both AIC and BIC showed the lowest values, supporting a 2-class solution (2-class AIC = 2507, BIC = 2548; See Supplementary Table 2). Figure 2 presents estimated pain ratings from admission to 42-month follow-up across the empirically-identified classes. A narrow majority of the sample (Class 1−58.8%) showed high initial pain ratings with large reductions in pain ratings (3.4 point decrease – 7.60 to 4.20; a substantial 44% improvement) through follow-up. The remaining patients (Class 2–41.2%) had lower pain ratings at ad-
mission and showed minimal reductions in pain ratings (∼0.5 point decrease – 3.25 to 2.72). Class 1 (large improvement) had lower pain chronicity at baseline (p = 0.017) than Class 2 (small improvement), but these groups did not differ in age or gender (p’s > 0.30). The large improvement class had fewer days of inpatient hospitalization at 1 month follow-up (t[43] = 2.08, p = 0.044) relative to the small improvement class. However, there were no class differences for school days and work days missed (all p > 0.20). Figure 3 presents the proportion of individuals receiving each diagnosis by empirical class assignments. Individuals with CRPS and abdominal pain diagnoses were more likely to fall in the large improvement class, whereas patients with headache and fibromyalgia tended to fall in the small improvement class. Individuals with other diagnoses fell in between these groups (overall X2 [4] = 21.15, p < 0.001). 3.3. Child and parent functioning Figure 4 presents school days missed (panel A), work days missed (panel B), and number of days of hospitalization from admission through followup (panel C). Significant reductions in school and work days missed and hospitalization days were observed beginning by 1 month follow-up and maintained through 24–42 month follow-up (all p’s < 0.001). Pain chronicity did not modify reductions in any of these indicators (all p’s > 0.40).
4. Discussion These results provide evidence that children and adolescents with chronic, disabling pain demonstrated long-term improvement following participation in a combined inpatient/day hospital interdisciplinary pain rehabilitation program. Enduring improvements in subjective pain ratings and daily functioning were observed for children with a variety of pain diagnoses and their parents. Despite little initial change, pain ratings showed a decline over time, with substantial declines observed from discharge to 1-month and from 12 months to 24– 42 month follow-up. These results provide additional support for the benefits of interdisciplinary pain rehabilitation [3,13–18]. Importantly, the results provide evidence of enduring improvement in pain associated with a treatment program that combines inpatient and day hospital components. Critics of inpatient pain re-
G.A. Banez et al. / Chronic pain in children and adolescents
203
(a)
(b)
(c)
Fig. 4. School days missed (panel a), work days missed (panel b), and hospitalization days (panel c; M ± 95% CI) from admission through follow-up, separately by low and high pain chronicity. (Colours are visible in the online version of the article; http://dx.doi.org/10.3233/PRM140289)
habilitation have argued that inpatient treatment effects have limited generalizability. Our findings support not only generalization of effects but suggest that many patients continued to make improvement in pain long after discharge. As hypothesized, two subgroups of treatment response were identified. One group of children and adolescents showed initially high pain ratings with large reductions in pain ratings. The second group had lower pain ratings at admission and showed minimal reductions in pain ratings. This empirically-derived distinction is not likely due solely to psychometric artifacts because children with lower baseline pain chronicity as well as those with CRPS and abdominal pain diagnoses were more likely to fall in the large improvement group. Those with greater pain chronicity and di-
agnoses of headache and fibromyalgia tended to fall in the small improvement group, and individuals with other diagnoses fell between these two groups. The large and small improvement-in-pain groups did not differ in age or gender. Our results suggest that not all children and adolescents with chronic pain responded similarly to our program. The greater improvement in children with high baseline pain ratings may have been due to a mediating factor such as a change in pain perception (e.g., less pain-related anxiety), or a statistical phenomenon like a floor effect. In terms of baseline pain chronicity, past research has suggested that greater chronicity may predict a lesser treatment response in pain [28]. Our results also suggest that certain diagnostic groups may be more responsive to this particular treatment.
204
G.A. Banez et al. / Chronic pain in children and adolescents
Children with CRPS were expected to be large responders because self-management and functional rehabilitation, the program’s emphases, have been identified as keys to treatment of CRPS [29]. It is not clear whether the response trajectories in other groups were due to the clinical characteristics of those conditions, specific treatment components, or a combination of these. Importantly, significant reductions in school and work days missed and hospitalization days were observed by 1-month follow up and maintained through 24–42 month follow-up. In contrast to our findings for pain ratings, pain chronicity did not modify reductions in any of these functional indicators. These results are consistent with past studies that support the impact of interdisciplinary pain rehabilitation on daily functioning [3,13–18]. When the large and small improvementin-pain groups were compared, the large improvement group had fewer days of inpatient hospitalization at 1-month. No group differences were found on school and work days missed. These findings suggest that improvements in pain and functioning may be related but are separate. The functional progress in the small improvement group was similar to the progress of the large improvement group, suggesting that function can improve greatly even when there is little improvement in pain. This finding is consistent with the program’s philosophy.
ber of missing data. As noted, missing data were anticipated due to the nature of this study. Though mixed effects regression models accommodate missing data across time points, more complete data would facilitate more statistically powerful and accurate evaluations of program outcomes.
5. Conclusion The results of this research suggest that an interdisciplinary pain rehabilitation program that combines inpatient and day hospital treatment and focuses on functioning is a promising approach to the management of chronic pain and functional disability in children and adolescents. Improvements on multiple indices of pain and functioning were found 24–42 months following program completion, and distinct response trajectories were identified. Ongoing research will examine program effectiveness and treatment response patterns with standardized, psychometrically-sound measures of pain and functioning and objective physical data.
Conflict of interest The authors have no conflict of interest to declare.
4.1. Study limitations As a naturalistic study, certain factors that may influence outcomes could not be controlled. Some limitations need to be acknowledged. First, our findings were based on patient and parent-completed surveys designed for the purposes of this evaluation. Standardized, psychometrically-sound measures of physical functioning, role functioning, emotional functioning and other core outcome domains would provide additional support for the effectiveness of combined inpatient/day hospital pain rehabilitation. Second, the degree of pain reduction was less than reported by prior authors. This finding may be explained by the fact that the three-week program is fixed in length and not openended, the amount of physical therapy is less than reported in other studies, and the program philosophy prioritizes improvement in functioning. A third limitation was the absence of objective data on physical functioning. Data such as improvement in the number of abdominal crunches or step-ups completed would provide evidence of how these programs directly impact physical function. A final limitation was the num-
Funding source This work was partially supported by a grant from the CVS Caremark Charitable Trust.
Clinical Trials Registry Site and Number ClinicalTrials.gov# NCT01340261.
References [1]
[2]
[3]
King A, Chambers CT, Huguet A, MacNevin RC, McGrath PJ, Parker L, MacDonald AJ. The epidemiology of chronic pain in children and adolescents revisited: A systematic review. Pain. 2011; 152: 2729-38. Stanford EA, Chambers CT, Biesanz JC, Chen E. The frequency, trajectories and predictors of adolescent recurrent pain: A population-based approach. Pain. 2008; 138: 11-21. Eccleston C, Malleson PN, Clinch J, Connell H, Sourbut C. Chronic pain in adolescents: evaluation of inter-disciplinary cognitive-behaviour therapy (CBT). Arch Dis Child. 2003; 88: 881-5.
G.A. Banez et al. / Chronic pain in children and adolescents [4] [5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
Rothner AD. Headaches in children and adolescents. Semin Pediatr Neurol. 1991; 8: 2-6. Weydert, JA, Ball TM, Davis MF. Systematic review of treatments for recurrent abdominal pain. Pediatrics. 2003; 111: e1e11. Eccleston C, Morley S, Williams A, Yorke L, Mastroyannopoulou K. Systematic review of randomized controlled trials of psychological therapy for chronic pain in children and adolescents, with a subset meta-analysis of pain relief. Pain. 2002; 99: 157-65. Holden EW, Deichmann MM, Levy JD. Empiricallysupported treatments in pediatric psychology: Recurrent pediatric headache. J Pediatr Psychol. 1999; 24: 91-109. Huertas-Ceballos A, Logan S, Bennett C, Macarthur. Psychosocial interventions for recurrent abdominal pain (RAP) and irritable bowel syndrome (IBS) in childhood. Cochrane Database of Systematic Reviews. 2008; 1: CD003014. Janicke DM, Finney JW. Empirically supported treatments in pediatric psychology: recurrent abdominal pain. J Pediatr Psychol.1999; 24: 115-27. Townsend CO, Rome JD, Bruce BK, Hooten WM. Interdisciplinary pain rehabilitation programs. In Ebert MH, Kerns RD, eds. Behavioral and psychopharmacologic pain management. Cambridge, UK: Cambridge University Press; 2011: 114-28. Gatchel RJ, Okifuji A. Evidence-based scientific data documenting the treatment and cost-effectiveness of comprehensive pain programs for chronic malignant pain. J Pain. 2006; 7: 79-93. Turk DC. Clinical effectiveness and cost-effectiveness of treatments for patients with chronic pain. Clin J Pain. 2002; 18: 355-65. Hechler T, Dobe M, Kosfelder J, Damschen U, Hubner B, Blankenburg M, Sauer C, Zernikow B. Effectiveness of a 3week multimodal inpatient pain treatment for adolescents suffering from chronic pain. Clin J Pain. 2009; 25: 156-66. Lee BH., Scharff L, Sethna NF, McCarthy CF, ScottSutherland J, Shea AM, Sullivan P, Meier P, Zurakowski D, Masek BJ, Berde CB. Physical therapy and cognitivebehavioral treatment for complex regional pain syndromes. J Pediatr. 2002; 141: 135-40. Maynard CS, Amari A, Wieczorek B, Christensen JR, Slifer KJ. Interdisciplinary behavioral rehabilitation of pediatric
205
pain-associated disability: retrospective review of an inpatient treatment protocol. J Pediatr Psychol. 2009; 35: 128-37. [16] Sherry DD, Wallace CA, Kelley C, Kidder M, Sapp L. Shortand long-term outcomes of children with complex regional pain syndrome type 1 treated with exercise therapy. Clin J Pain. 1999; 15: 218-23. [17] Brooke V, Janselewitz S. Outcomes of children with complex regional pain syndrome after intensive inpatient rehabilitation. PM R. 2012; 4: 349-54. [18] Logan DE, Carpino EA, Chiang G, et al. A day-hospital approach to treatment of complex regional pain syndrome: initial functional outcomes. Clin J Pain. 2012; 28: 766-74. [19] McGrath, PJ, and the PedIMMPACT group. Core outcome domains and measures for pediatric acute and chronic/recurrent pain clinical trials: PedIMMPACT recommendations. J Pain. 2008; 9: 771-83. [20] Dworkin RH, and the IMMPACT group. Core outcome measures for chronic pain clinical trials: IMMPACT recommendations. Pain. 2005; 113: 9-19. [21] Cleeland CS, Ryan KM. Pain assessment: global use of the Brief Pain Inventory. Ann Acad Med. 1994; 23: 129-38. [22] Bryk AS, Raudenbush SW. Hierarchical linear models: applications and data analysis methods. 2nd ed. Newbury Park, CA: Sage, 1992. [23] Peugh JL, Enders CK. Using the SPSS mixed procedure to fit cross-sectional and longitudinal multilevel models. Edu Psychol Meas. 2005; 65: 717-41. [24] Verbeke G, Molenberghs G. Linear mixed models for longitudinal data. New York, N.Y: Springer, 2009. [25] Gibbons RD, Hedeker D, DuToit S. Advances in analysis of longitudinal data. Annu Rev Clin Psychol. 2010; 6: 79-107. [26] Little RJ, Rubin DB. Statistical analysis with missing data. 2nd ed. New York: John Wiley & Sons, 2002. [27] Kreft IGG. [Introduction]: Hierarchical linear models: problems and prospects. J Ed Behav Stat. 1995; 20: 109-13. [28] Buchner M, Neubauer E, Zahlten-Hinguranage A, Schiltenwolf M. The influence of the grade of chronicity on the outcome of multidisciplinary therapy for chronic low back pain. J Spine. 2007; 32: 3060-6. [29] Stanton-Hicks M. Plasticity of complex regional pain syndrome (CRPS) in children. Pain Med. 2010; 11: 1216-23.
206
G.A. Banez et al. / Chronic pain in children and adolescents
Supplementary material Supplementary Table 1 Mixed-effect regression model estimates for fixed-effects covariates predicting pain ratings Intercept Time Baseline Discharge 1-month follow-up 12-month follow-up 24–42 month follow-up Gender Age Baseline Pain Chronicity
Estimate (SE) 3.95 (1.16) 2.22 (0.41) 2.64 (0.41) 1.26 (0.48) 1.35 (0.50) 0 (0) −0.31 (0.42) −0.02 (0.07) < −0.01 (0.01)
F 22.84
DF 1, 190
p < 0.001
12.08
4, 356
< 0.001
0.55 0.76 0.02
1, 194 1, 191 1, 260
0.459 0.784 0.892
Note. SE = Standard error. DF = Degrees of Freedom. Supplementary Table 2 Growth mixture modeling fit statistics across classes 1-Class 2-Class 3-Class
LL −1251 −1241 −1239
Parameters 10 13 16
AIC 2523 2507 2510
BIC 2555 2548 2561
Note. LL = Log-Likelihood, AIC = Akaike Information Criterion, BIC = Bayesian Information Criterion.
Class proportions − 58.8%, 41.2% 55.0%, 37.2%, 7.8%
Copyright of Journal of Pediatric Rehabilitation Medicine is the property of IOS Press and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
197
Chronic pain in children and adolescents: 24–42 month outcomes of an inpatient/day hospital interdisciplinary pain rehabilitation program Gerard A. Baneza,b,c,∗, Thomas W. Fraziera , Andrea A. Wojtowicza , Kristen Buchannana, Douglas E. Henrya,c and Ethan Benorea,b a
Pediatric Pain Rehabilitation Program, Cleveland Clinic, Cleveland, OH, USA Center for Pediatric Behavioral Health, Cleveland Clinic, Cleveland, OH, USA c Department of Developmental and Rehabilitation Pediatrics, Cleveland Clinic, Cleveland, OH, USA b
Accepted 10 February 2014
Abstract. PURPOSE: The purpose of this research was to describe 24–42 month outcomes of a combined inpatient/day hospital interdisciplinary pain rehabilitation program for children and adolescents with chronic pain and functional disability. METHODS: One-hundred-seventy-three children and adolescents (mean age = 15.1 years, SD = 2.5) were treated in the threeweek program. Mixed effects regression models (MERM) examined changes over time in pain ratings (0–10), school days missed by children and work days missed by parents in the preceding week, and the number of days hospitalized in the preceding month. RESULTS: Participants reported a significant decline in pain from admission to 24–42 month follow-up. The largest declines in pain occurred from discharge to 1-month follow-up (6.20 vs. 4.81 on a 0–10 numerical rating scale) and from 12 months to 24–42 month follow-up (4.90 vs. 3.56). Two distinct trajectories of treatment response were identified using growth mixture modeling: children with initially high pain ratings exhibited large reductions in pain ratings, while those with lower pain ratings at admission showed minimal reductions. Treatment resulted in significant reductions in school and work days missed and the number of days hospitalized (all p’s < 0.001), with reductions evident at 1-month follow-up and maintained through 24–42 month follow-up. CONCLUSIONS: These results suggest that interdisciplinary pain rehabilitation is a promising approach to chronic pain and associated disability in children, with enduring improvements found 24–42 months following program completion. Distinct trajectories of treatment response were identified. Keywords: Adolescents, pain, interdisciplinary health team, rehabilitation, outcomes assessment
1. Background Chronic pain in children and adolescents is common, with a prevalence of at least 15% and as high as 40% [1,2]. A significant subset of these patients ∗ Corresponding author: Gerard A. Banez, Pediatric Pain Rehabilitation Program, Cleveland Clinic Children’s Hospital for Rehabilitation, 2801 MLK Jr. Drive/CR 11, Cleveland, OH 44104, USA. Tel.: +1 216 448 6253; Fax: +1 216 448 6207; E-mail: [email protected].
experiences a downward spiral of worsening disability. They do not attend school, interact with peers, or participate in sports, extracurricular and other personal/family activities. The existing literature on medical, physical, and psychosocial treatments suggests that these treatments may be helpful to some children with chronic pain, but they are not effective for all patients and do not typically eliminate pain [3–9]. For severely affected children, an interdisciplinary rehabilitation approach provides a useful model of care. In
c 2014 – IOS Press and the authors. All rights reserved 1874-5393/14/$27.50
198
G.A. Banez et al. / Chronic pain in children and adolescents
a rehabilitation model, pain is accepted as a symptom that may/may not be eliminated. The focus of care is on independent functioning, improved coping, and increased self-efficacy [10]. For children who have shown little response to unidisciplinary and symptomfocused strategies, this approach may be particularly helpful [11,12]. A number of studies, identified by computer and manual methods, have examined the utility of a rehabilitation approach to children and adolescents with chronic pain and functional disability [3,13–18]. These investigations have varied in terms of delivery setting (e.g., inpatient vs. day hospital), treatment modalities, patients served, and follow-up length. Three of the seven studies [13,15,17] examined an inpatient treatment program, while the remainder assessed a day hospital or outpatient treatment. In five studies [3,13, 15,17,18], multimodal treatment approaches, including physical and psychological therapies, were examined. The other investigations evaluated intense exercise therapy [16] and a combination of physical therapy and cognitive-behavioral therapy [14]. Patient types have varied. Four investigator groups [14,16– 18] treated only patients with complex regional pain syndrome (CRPS), while others treated a variety of pain diagnoses, including CRPS, headaches, abdominal pain, and fibromyalgia. Follow-up has ranged from three months to over five years post-treatment but has typically been two years or less [16]. Overall, the findings support the effectiveness of a rehabilitation approach. Numerous positive outcomes have been reported, including improvements in pain intensity, physical functioning, school attendance, anxiety, and depression; less parental anxiety, depression, and illness; and decreased medication usage. The limitations of the existing literature include: (1) no evaluations of interdisciplinary programs that combine inpatient and day hospital treatment, (2) limited attention to functional outcomes for parents (e.g., work days missed by parent due to children’s pain), and (3) limited longterm follow-up. There has been no investigation of individual response patterns or the influence of demographic/clinical characteristics (e.g., age, sex, pain chronicity) on treatment outcomes. The purpose of this report is to describe 24– 42 month follow-up outcomes of children and adolescents treated in a three-week, combined inpatient/day hospital interdisciplinary pain rehabilitation program. Children were treated as inpatients in their first two weeks and as day hospital patients in the third week. The combined inpatient/day hospital approach pro-
vided for the benefits of inpatient admission (e.g., increased control of activity, diet, sleep; consistent treatment philosophy; separation of children and parents to interrupt maladaptive interaction patterns and promote healthy behaviors), while facilitating transition into home, school, and community during the day hospital week. The program integrated rehabilitation therapies, behavioral health services, and medical subspecialty care in an individualized, coordinated manner. Children presented with a range of pain diagnoses, including CRPS, headaches, abdominal pain, and fibromyalgia. Follow-up focused on functional outcomes for children and their parents at discharge and at 1, 12, and 24–42 months following discharge. This research built on prior investigations and examined the following questions: (1) Do children and adolescents with varied chronic pain diagnoses show improvement in subjective pain ratings following participation in a combined inpatient/day hospital interdisciplinary pain rehabilitation program? As noted, past studies have focused more on day hospital treatment, and no studies have examined a combined inpatient/day hospital approach designed to facilitate a successful transition from inpatient treatment to home. (2) Can improvements be measured in functional outcomes for children and their parents (e.g., parent work days missed)? Previous research has focused primarily on child outcomes, whereas the present investigation assessed indices of everyday functioning for both children and their parents. (3) Are improvements in pain and functioning sustained beyond two years post-discharge? The 24– 42 month outcomes assessed in this research extend beyond the average two years or less followup in past studies. (4) Are there profiles of treatment response that suggest that demographic (age, sex) or clinical (pain chronicity, diagnosis) characteristics moderate outcomes? Investigation of patient profiles and subgroups of treatment response is important for understanding which individuals respond best to this particular treatment approach. We hypothesized that improvement in pain and functioning would occur during treatment and continue beyond the three-week treatment period. Improvements were expected to be seen in pain ratings and functional outcomes and to extend beyond two years. We also hypothesized that at least two distinct trajectories
G.A. Banez et al. / Chronic pain in children and adolescents
199
Table 1 Sample demographic and clinical characteristics for pain rehabilitation patients admitted from 2007 to 2009 Sample characteristics N Age (M, SD) [range] Sex (N, % male) Pain chronicity (median, inter-quartile range) Diagnosis (N, %) CRPS Headache Abdominal pain Fibromyalgia Other Pain rating (M, SD) [range] School days missed (last week) Caregiver work days missed (last week)
Admission 173 15.0 (2.5) [8–22] 46 (26.6) 16.5 (7–37)
Discharge 173 15.1 (2.5) [8–22] − −
60 (34.7) 38 (22.0) 22 (12.7) 21 (12.1) 32 (18.5) 5.8 (3.2) [0–10] 3.4 (2.1) [0–5] 2.7 (2.8) [0–7]
−
6.2 (2.9) [0–10] − −
Note: School days missed and caregiver work days missed are not relevant for discharge due to program participation.
of treatment response would be identified: some children and adolescents would show a strong sustained response and others would show modest treatment outcomes. Our investigation of demographic/clinical characteristics was exploratory. We hypothesized that children with lower baseline pain chronicity and those with CRPS would show a stronger treatment response.
2. Methods 2.1. Participants Our sample consisted of a consecutive patient series of one hundred seventy-three children and adolescents (mean age = 15.1 years, SD = 2.5; 46 male) treated in a three-week, combined inpatient/day hospital interdisciplinary pain rehabilitation program. CRPS (n = 60) was the most common referral diagnosis, followed by headache (n = 38), abdominal pain (n = 22), fibromyalgia (n = 21), and other pain conditions (e.g., back pain, limb pain; n = 32). The mean chronicity of pain, as assessed at admission, was 16.5 months. Nearly all patients had previously undergone extensive medical evaluation and attempted a broad range of surgeries, pain management procedures, and complementary and alternative strategies. The data presented in this report were collected as part of an Institutional Review Board-approved registry, and informed consent for all patients was obtained on the date of admission. Table 1 presents demographic and clinical characteristics of our sample at admission. 2.2. Program description The Cleveland Clinic Pediatric Pain Rehabilitation Program was designed to assist children and adoles-
cents with chronic pain that interferes with normal activities. The program goals are: (1) to help children manage their pain and (2) to restore daily activity. An emphasis is placed on return to age-appropriate activity despite pain. The program blends rehabilitation therapies, behavioral health, and pediatric subspecialty care, and differs from previously evaluated programs in that it consists of inpatient and day hospital components. In the first two weeks of the program, patients are admitted to an inpatient rehabilitation unit where they sleep and are often treated apart from their parents. The inpatient component of the program facilitates control of the therapeutic environment (e.g., amount of activity, diet, sleep schedule) and immerses patients into a shared treatment philosophy emphasizing functioning despite pain. Separation of children and parents also serves to interrupt maladaptive interactions that are maintaining or worsening pain behaviors and facilitates acquisition of more healthy behavioral habits. During the day hospital week, children participate as outpatients but return to their parents at the end of each day. They continue the activities of the first two weeks but have more opportunities to apply their new skills in situations outside the hospital. The final week emphasizes preparation for potential obstacles to normal functioning and successful transition to home, school, and other activities. Patients spend 7–8 hours in treatment daily, with services scheduled hourly from 8 a.m. to 4 p.m. Patients are seen individually or in groups for physical, occupational, and aquatic therapies three hours/day on average. Therapy services promote stretching, strengthening, and endurance. As indicated, sensory destimulation, postural alignment/body awareness, durable medical equipment assessment, and kinesiotaping are used. Each patient is given evening exercises to per-
200
G.A. Banez et al. / Chronic pain in children and adolescents
form, and an individualized home exercise program is developed prior to discharge. Patients are also scheduled for an average of three individual/family psychological treatment sessions/week and participate in a mind-body skills training group (e.g., progressive muscle relaxation, diaphragmatic breathing, cognitive self-statements, aromatherapy, acupressure) three times/week. These services are designed to support patients as they participate in their rehabilitation therapies, enhance pain management skills and resources, and facilitate improved emotional adjustment and familial functioning. To promote increased health and wellness, each patient has an individualized functional plan that identifies specific behavioral treatment goals. Patients also participate in a school program one to two hours/day and in leisure education and recreation therapy groups at least one hour daily. Parent/family education plays a critical role in the program and is provided individually and in group sessions. Emphasis is placed on the physical and psychological aspects of pain, associated disability, and its treatment. Guidelines for managing pain behaviors are presented, with attention to the importance of encouraging normal activity and discouraging pain behaviors. Medical and nursing visits occur daily. Medical/nursing staff assist in planning and supervising treatment and consult on daily care as needed. 2.3. Measure A 12-item clinical outcomes measure was designed for the purposes of this evaluation. Four items/outcomes were selected for this particular research investigation: (1) pain intensity rating, (2) number of school days missed by the patient over the past week, (3) number of work days missed by the parent over the past week due to adolescent’s pain, and (4) number of days in which the patient was hospitalized for pain in the past month. These outcomes were chosen because of their clinical importance and are consistent with PedIMMPACT’s recommendation that measures of pain intensity, physical functioning, and role functioning be considered when evaluating pediatric chronic pain treatments [19]. As recommended by IMMPACT [20], an 11-point (0–10) numerical rating scale (NRS) was used to measure pain intensity. Patients were asked to rate their pain during the last 24 hours on a scale from 0–10, with 0 meaning ‘No pain’ and ‘10’ meaning ‘Maximum or worst possible pain’ [21]. 2.4. Procedure The study design consisted of a restrospective chart review (admission and discharge data) and follow-up
telephone survey (1, 12, 24–42 months following discharge). Patients and parents completed the 12-item outcomes measure at admission, discharge, and by telephone post-discharge. 2.5. Statistical analyses Univariate analysis of variance and Pearson chisquare analyses examined whether baseline clinical characteristics differed across pain categories (CRPS, headache, abdominal pain, fibromyalgia, and other diagnosis). Due to the longitudinal and observational nature of this study, significant missing data were anticipated. To handle missing data, we examined the quantity and pattern of missingness. Because missing data were expected to be largest for later follow-up time points (24– 42 months), data from this time period were collapsed into a single set of observations for each patient, while data for admission, discharge, 1-month, and 12-months follow-up remained separate. Univariate analysis of variance and Pearson chi-square examined whether individuals with/without missing data differed on clinical/demographic characteristics. To accommodate missing data across time points and avoid the problems of last- observation-carriedforward, mixed effects regression models [22–24] were used to estimate changes over time in outcome variables. These models are advantageous to standard repeated measures analyses of variance because participants are not required to have data from all time points, unevenly spaced time points are treated appropriately, and covariance structures of outcome measures collected over time are directly estimated. Mixed effects regression models typically produce more accurate, powerful estimates of changes over time than repeated measures analyses using listwise deletion as they are robust when data are missing at random [25]. The missing at random assumption posits that the data are missing dependent on the covariates and outcomes [26]. Thus, we explicitly evaluated missing data for each outcome and included a priori relevant covariates in mixed effects models. To examine change over time in pain ratings, a mixed effects regression model (MERM) was computed with time (admission, discharge, 1-month post discharge, 12-months post discharge, and 24–42 months post-discharge) as a time-varying fixed-effects covariate and age at admission, gender, and baseline pain chronicity as time-invariant fixed-effects covariates. We also evaluated whether patients showed significant
G.A. Banez et al. / Chronic pain in children and adolescents
variability in pain ratings at admission (random intercept) and variability in change-over-time (random slope). Model fit was considered by iteratively examining the above random effects and alternative covariance structures [22,23,27]. Based on findings of significant variability in change -over-time (significant variability in slopes), a series of growth mixture models were computed to examine whether distinct patterns of change in pain ratings could be identified for sub-groups. Single class, 2-class, and 3-class growth mixture models were estimated. The Bayesian Information Criterion (BIC) and Akaike Information Criterion (AIC) were used to determine model fit. A growth model was computed to further delineate the pattern of change in pain ratings and to examine the effects of baseline characteristics on initial pain status and linear changes over time in pain ratings. To examine change over time in school and work days missed and hospitalization days in the past month, exploratory mixed effects regression models were computed with the same covariates as described above for pain ratings.
3. Results Missing pain ratings across study time points were as follows: admission n = 7 (4.0%), discharge n = 22 (12.7%), 1 month follow-up n = 124 (71.7%), 1 year follow-up n = 114 (65.9%), 24–42 month follow-up n = 94 (54.3%). Although missing data were substantial when viewed at individual time points, the majority of adolescents had data available from at least one follow-up time point (n=148 out of 173, 85.5%). Children and adolescents with any missing data at admission or discharge did not significantly differ from those with complete data in terms of sex (X2 [1] = 2.14, p = 0.144), diagnosis (X2 [4] = 6.42, p = 0.170), pain chronicity at admission (t[168] = 0.39, p = 0.701), school days missed (t[64] = 0.96, p = 0.363), or parent work days missed in the week prior to admission (t[50] = 1.08, p = 0.288). However, children with missing data at admission or discharge tended to be slightly older (t[171] = 2.15, p = 0.042) than those with complete data at those time points. Covarying for age, sex, and pain chronicity did not substantively alter the overall pattern of changes in pain ratings over time in mixed effects regression models. Results are presented without covariate adjustment. Children without data at any follow-up time point did not differ from those with at least one follow-up data point on age or sex (X2 [1]
201
Fig. 1. Pain ratings (M ± 95% CI) from admission through follow-up, separately by low and high pain chronicity. (Colours are visible in the online version of the article; http://dx.doi.org/10.3233/ PRM-140289)
= 0.10, p = 0.751). However, children with missing follow-up data tended to have shorter pain chronicity at admission, suggesting that follow-up data were biased toward those with longer pain chronicity. As such, outcomes analyses may underestimate the actual effects of treatment. Children with missing follow-up data were also more likely to be diagnosed with CRPS, abdominal pain, or fall in the other diagnosis category (X2 [4] = 13.53, p = 0.009). Overall, this pattern reinforces the choice of mixed effects regression models, which account for all available data rather than lastobservation-carried-forward or pairwise/listwise deletion approaches. 3.1. Pain ratings Figure 1 presents pain ratings from admission through 42 months, separately by low ( 12 months) and high (> 12 months) pain chronicity. There was significant variability in pain ratings at baseline (random intercept = 3.46, SE = 0.67, z = 5.19, p < 0.001). Overall, pain ratings showed a significant decline over time in both MERM and growth models (Time F[4,358] = 12.19, p < 0.001; Growth model slope = −0.08, SE = 0.012, t[10] = −6.22, p < 0.001; see Supplementary Table 1). Individual patient changes-over-time in pain ratings showed significant variability (random slope = 6.11, SE = 0.47, z =
202
G.A. Banez et al. / Chronic pain in children and adolescents
Pain Rang
Fig. 2. Pain ratings across the study follow-up period by growth mixture model classifications.
10 9 8 7 6 5 4 3 2 1 0
Pain Chronicity 12 Months
Fig. 3. Proportions of patients receiving specific diagnoses by empirical classifications. (Colours are visible in the online version of the article; http://dx.doi.org/10.3233/PRM-140289)
12.88, p < 0.001). The average pain rating at admission was estimated as moderate to high (MERM estimate = 5.78, Growth model estimate = 5.89). Substantial declines in pain ratings (0–10) were observed from discharge to 1 month follow-up (6.20 vs. 4.81) and from 12 months to 24–42-month follow-up (4.90 vs. 3.56). In the growth model, longer baseline pain chronicity was significantly associated with smaller reductions over time in pain ratings (t[10] = 3.96, p < 0.001; see Fig. 1). 3.2. Sub-groups of treatment response Growth mixture model results indicated that both AIC and BIC showed the lowest values, supporting a 2-class solution (2-class AIC = 2507, BIC = 2548; See Supplementary Table 2). Figure 2 presents estimated pain ratings from admission to 42-month follow-up across the empirically-identified classes. A narrow majority of the sample (Class 1−58.8%) showed high initial pain ratings with large reductions in pain ratings (3.4 point decrease – 7.60 to 4.20; a substantial 44% improvement) through follow-up. The remaining patients (Class 2–41.2%) had lower pain ratings at ad-
mission and showed minimal reductions in pain ratings (∼0.5 point decrease – 3.25 to 2.72). Class 1 (large improvement) had lower pain chronicity at baseline (p = 0.017) than Class 2 (small improvement), but these groups did not differ in age or gender (p’s > 0.30). The large improvement class had fewer days of inpatient hospitalization at 1 month follow-up (t[43] = 2.08, p = 0.044) relative to the small improvement class. However, there were no class differences for school days and work days missed (all p > 0.20). Figure 3 presents the proportion of individuals receiving each diagnosis by empirical class assignments. Individuals with CRPS and abdominal pain diagnoses were more likely to fall in the large improvement class, whereas patients with headache and fibromyalgia tended to fall in the small improvement class. Individuals with other diagnoses fell in between these groups (overall X2 [4] = 21.15, p < 0.001). 3.3. Child and parent functioning Figure 4 presents school days missed (panel A), work days missed (panel B), and number of days of hospitalization from admission through followup (panel C). Significant reductions in school and work days missed and hospitalization days were observed beginning by 1 month follow-up and maintained through 24–42 month follow-up (all p’s < 0.001). Pain chronicity did not modify reductions in any of these indicators (all p’s > 0.40).
4. Discussion These results provide evidence that children and adolescents with chronic, disabling pain demonstrated long-term improvement following participation in a combined inpatient/day hospital interdisciplinary pain rehabilitation program. Enduring improvements in subjective pain ratings and daily functioning were observed for children with a variety of pain diagnoses and their parents. Despite little initial change, pain ratings showed a decline over time, with substantial declines observed from discharge to 1-month and from 12 months to 24– 42 month follow-up. These results provide additional support for the benefits of interdisciplinary pain rehabilitation [3,13–18]. Importantly, the results provide evidence of enduring improvement in pain associated with a treatment program that combines inpatient and day hospital components. Critics of inpatient pain re-
G.A. Banez et al. / Chronic pain in children and adolescents
203
(a)
(b)
(c)
Fig. 4. School days missed (panel a), work days missed (panel b), and hospitalization days (panel c; M ± 95% CI) from admission through follow-up, separately by low and high pain chronicity. (Colours are visible in the online version of the article; http://dx.doi.org/10.3233/PRM140289)
habilitation have argued that inpatient treatment effects have limited generalizability. Our findings support not only generalization of effects but suggest that many patients continued to make improvement in pain long after discharge. As hypothesized, two subgroups of treatment response were identified. One group of children and adolescents showed initially high pain ratings with large reductions in pain ratings. The second group had lower pain ratings at admission and showed minimal reductions in pain ratings. This empirically-derived distinction is not likely due solely to psychometric artifacts because children with lower baseline pain chronicity as well as those with CRPS and abdominal pain diagnoses were more likely to fall in the large improvement group. Those with greater pain chronicity and di-
agnoses of headache and fibromyalgia tended to fall in the small improvement group, and individuals with other diagnoses fell between these two groups. The large and small improvement-in-pain groups did not differ in age or gender. Our results suggest that not all children and adolescents with chronic pain responded similarly to our program. The greater improvement in children with high baseline pain ratings may have been due to a mediating factor such as a change in pain perception (e.g., less pain-related anxiety), or a statistical phenomenon like a floor effect. In terms of baseline pain chronicity, past research has suggested that greater chronicity may predict a lesser treatment response in pain [28]. Our results also suggest that certain diagnostic groups may be more responsive to this particular treatment.
204
G.A. Banez et al. / Chronic pain in children and adolescents
Children with CRPS were expected to be large responders because self-management and functional rehabilitation, the program’s emphases, have been identified as keys to treatment of CRPS [29]. It is not clear whether the response trajectories in other groups were due to the clinical characteristics of those conditions, specific treatment components, or a combination of these. Importantly, significant reductions in school and work days missed and hospitalization days were observed by 1-month follow up and maintained through 24–42 month follow-up. In contrast to our findings for pain ratings, pain chronicity did not modify reductions in any of these functional indicators. These results are consistent with past studies that support the impact of interdisciplinary pain rehabilitation on daily functioning [3,13–18]. When the large and small improvementin-pain groups were compared, the large improvement group had fewer days of inpatient hospitalization at 1-month. No group differences were found on school and work days missed. These findings suggest that improvements in pain and functioning may be related but are separate. The functional progress in the small improvement group was similar to the progress of the large improvement group, suggesting that function can improve greatly even when there is little improvement in pain. This finding is consistent with the program’s philosophy.
ber of missing data. As noted, missing data were anticipated due to the nature of this study. Though mixed effects regression models accommodate missing data across time points, more complete data would facilitate more statistically powerful and accurate evaluations of program outcomes.
5. Conclusion The results of this research suggest that an interdisciplinary pain rehabilitation program that combines inpatient and day hospital treatment and focuses on functioning is a promising approach to the management of chronic pain and functional disability in children and adolescents. Improvements on multiple indices of pain and functioning were found 24–42 months following program completion, and distinct response trajectories were identified. Ongoing research will examine program effectiveness and treatment response patterns with standardized, psychometrically-sound measures of pain and functioning and objective physical data.
Conflict of interest The authors have no conflict of interest to declare.
4.1. Study limitations As a naturalistic study, certain factors that may influence outcomes could not be controlled. Some limitations need to be acknowledged. First, our findings were based on patient and parent-completed surveys designed for the purposes of this evaluation. Standardized, psychometrically-sound measures of physical functioning, role functioning, emotional functioning and other core outcome domains would provide additional support for the effectiveness of combined inpatient/day hospital pain rehabilitation. Second, the degree of pain reduction was less than reported by prior authors. This finding may be explained by the fact that the three-week program is fixed in length and not openended, the amount of physical therapy is less than reported in other studies, and the program philosophy prioritizes improvement in functioning. A third limitation was the absence of objective data on physical functioning. Data such as improvement in the number of abdominal crunches or step-ups completed would provide evidence of how these programs directly impact physical function. A final limitation was the num-
Funding source This work was partially supported by a grant from the CVS Caremark Charitable Trust.
Clinical Trials Registry Site and Number ClinicalTrials.gov# NCT01340261.
References [1]
[2]
[3]
King A, Chambers CT, Huguet A, MacNevin RC, McGrath PJ, Parker L, MacDonald AJ. The epidemiology of chronic pain in children and adolescents revisited: A systematic review. Pain. 2011; 152: 2729-38. Stanford EA, Chambers CT, Biesanz JC, Chen E. The frequency, trajectories and predictors of adolescent recurrent pain: A population-based approach. Pain. 2008; 138: 11-21. Eccleston C, Malleson PN, Clinch J, Connell H, Sourbut C. Chronic pain in adolescents: evaluation of inter-disciplinary cognitive-behaviour therapy (CBT). Arch Dis Child. 2003; 88: 881-5.
G.A. Banez et al. / Chronic pain in children and adolescents [4] [5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
Rothner AD. Headaches in children and adolescents. Semin Pediatr Neurol. 1991; 8: 2-6. Weydert, JA, Ball TM, Davis MF. Systematic review of treatments for recurrent abdominal pain. Pediatrics. 2003; 111: e1e11. Eccleston C, Morley S, Williams A, Yorke L, Mastroyannopoulou K. Systematic review of randomized controlled trials of psychological therapy for chronic pain in children and adolescents, with a subset meta-analysis of pain relief. Pain. 2002; 99: 157-65. Holden EW, Deichmann MM, Levy JD. Empiricallysupported treatments in pediatric psychology: Recurrent pediatric headache. J Pediatr Psychol. 1999; 24: 91-109. Huertas-Ceballos A, Logan S, Bennett C, Macarthur. Psychosocial interventions for recurrent abdominal pain (RAP) and irritable bowel syndrome (IBS) in childhood. Cochrane Database of Systematic Reviews. 2008; 1: CD003014. Janicke DM, Finney JW. Empirically supported treatments in pediatric psychology: recurrent abdominal pain. J Pediatr Psychol.1999; 24: 115-27. Townsend CO, Rome JD, Bruce BK, Hooten WM. Interdisciplinary pain rehabilitation programs. In Ebert MH, Kerns RD, eds. Behavioral and psychopharmacologic pain management. Cambridge, UK: Cambridge University Press; 2011: 114-28. Gatchel RJ, Okifuji A. Evidence-based scientific data documenting the treatment and cost-effectiveness of comprehensive pain programs for chronic malignant pain. J Pain. 2006; 7: 79-93. Turk DC. Clinical effectiveness and cost-effectiveness of treatments for patients with chronic pain. Clin J Pain. 2002; 18: 355-65. Hechler T, Dobe M, Kosfelder J, Damschen U, Hubner B, Blankenburg M, Sauer C, Zernikow B. Effectiveness of a 3week multimodal inpatient pain treatment for adolescents suffering from chronic pain. Clin J Pain. 2009; 25: 156-66. Lee BH., Scharff L, Sethna NF, McCarthy CF, ScottSutherland J, Shea AM, Sullivan P, Meier P, Zurakowski D, Masek BJ, Berde CB. Physical therapy and cognitivebehavioral treatment for complex regional pain syndromes. J Pediatr. 2002; 141: 135-40. Maynard CS, Amari A, Wieczorek B, Christensen JR, Slifer KJ. Interdisciplinary behavioral rehabilitation of pediatric
205
pain-associated disability: retrospective review of an inpatient treatment protocol. J Pediatr Psychol. 2009; 35: 128-37. [16] Sherry DD, Wallace CA, Kelley C, Kidder M, Sapp L. Shortand long-term outcomes of children with complex regional pain syndrome type 1 treated with exercise therapy. Clin J Pain. 1999; 15: 218-23. [17] Brooke V, Janselewitz S. Outcomes of children with complex regional pain syndrome after intensive inpatient rehabilitation. PM R. 2012; 4: 349-54. [18] Logan DE, Carpino EA, Chiang G, et al. A day-hospital approach to treatment of complex regional pain syndrome: initial functional outcomes. Clin J Pain. 2012; 28: 766-74. [19] McGrath, PJ, and the PedIMMPACT group. Core outcome domains and measures for pediatric acute and chronic/recurrent pain clinical trials: PedIMMPACT recommendations. J Pain. 2008; 9: 771-83. [20] Dworkin RH, and the IMMPACT group. Core outcome measures for chronic pain clinical trials: IMMPACT recommendations. Pain. 2005; 113: 9-19. [21] Cleeland CS, Ryan KM. Pain assessment: global use of the Brief Pain Inventory. Ann Acad Med. 1994; 23: 129-38. [22] Bryk AS, Raudenbush SW. Hierarchical linear models: applications and data analysis methods. 2nd ed. Newbury Park, CA: Sage, 1992. [23] Peugh JL, Enders CK. Using the SPSS mixed procedure to fit cross-sectional and longitudinal multilevel models. Edu Psychol Meas. 2005; 65: 717-41. [24] Verbeke G, Molenberghs G. Linear mixed models for longitudinal data. New York, N.Y: Springer, 2009. [25] Gibbons RD, Hedeker D, DuToit S. Advances in analysis of longitudinal data. Annu Rev Clin Psychol. 2010; 6: 79-107. [26] Little RJ, Rubin DB. Statistical analysis with missing data. 2nd ed. New York: John Wiley & Sons, 2002. [27] Kreft IGG. [Introduction]: Hierarchical linear models: problems and prospects. J Ed Behav Stat. 1995; 20: 109-13. [28] Buchner M, Neubauer E, Zahlten-Hinguranage A, Schiltenwolf M. The influence of the grade of chronicity on the outcome of multidisciplinary therapy for chronic low back pain. J Spine. 2007; 32: 3060-6. [29] Stanton-Hicks M. Plasticity of complex regional pain syndrome (CRPS) in children. Pain Med. 2010; 11: 1216-23.
206
G.A. Banez et al. / Chronic pain in children and adolescents
Supplementary material Supplementary Table 1 Mixed-effect regression model estimates for fixed-effects covariates predicting pain ratings Intercept Time Baseline Discharge 1-month follow-up 12-month follow-up 24–42 month follow-up Gender Age Baseline Pain Chronicity
Estimate (SE) 3.95 (1.16) 2.22 (0.41) 2.64 (0.41) 1.26 (0.48) 1.35 (0.50) 0 (0) −0.31 (0.42) −0.02 (0.07) < −0.01 (0.01)
F 22.84
DF 1, 190
p < 0.001
12.08
4, 356
< 0.001
0.55 0.76 0.02
1, 194 1, 191 1, 260
0.459 0.784 0.892
Note. SE = Standard error. DF = Degrees of Freedom. Supplementary Table 2 Growth mixture modeling fit statistics across classes 1-Class 2-Class 3-Class
LL −1251 −1241 −1239
Parameters 10 13 16
AIC 2523 2507 2510
BIC 2555 2548 2561
Note. LL = Log-Likelihood, AIC = Akaike Information Criterion, BIC = Bayesian Information Criterion.
Class proportions − 58.8%, 41.2% 55.0%, 37.2%, 7.8%
Copyright of Journal of Pediatric Rehabilitation Medicine is the property of IOS Press and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
