Adolescent predictors of young adult pain and health outcomes: results from a six-year prospective follow-up study

Introduction

Chronic pain is a prevalent and burdensome health issue affecting approximately 12% of young adults globally[43]. Young adults with chronic pain are a vulnerable population who have strikingly high rates of continued chronic pain from childhood and heightened rates of comorbid depression, anxiety, and sleep disturbance [2; 8; 46; 53; 55; 60]. The transition from adolescence to young adulthood is a vulnerable phase marked by notable health risks, including the onset and exacerbation of psychiatric and behavioral health challenges. Despite advances in understanding the long-term trajectories of individuals with chronic pain transitioning into young adulthood, important questions remain regarding identifying those most at risk for adverse outcomes.

To improve long-term outcomes, it is imperative to identify adolescent vulnerabilities linked to adverse outcomes in young adulthood. Palermo’s (2020) lifespan model of pediatric pain highlights biopsychosocial vulnerabilities that may confer risk for adverse adult outcomes [48]. Rooted in social-ecological theory, Palermo’s (2020) lifespan model delineates multiple childhood vulnerabilities that may be associated with an elevated risk for unresolved young adult chronic pain and negative health outcomes, including individual risk factors (e.g., emotional distress) as well as interpersonal social and family factors (e.g., parenting behaviors). For instance, emotional vulnerabilities in adolescents with chronic pain may amplify disability [20; 36; 68] and lead to persistent pain and poorer mental and physical health in adulthood.

While some studies have linked childhood vulnerabilities including pain severity, sleep quality, anxiety, and depression with an increased risk of continued chronic pain in adulthood [29; 30; 33; 60; 67], most focus narrowly on pain outcomes and overlook other key dimensions of young adult health. As a notable exception, longitudinal research on juvenile-onset fibromyalgia by Kashikar-Zuck and colleagues found worsening trajectories of depression in adolescence predicted reduced psychosocial functioning [60], and family conflict predicted subsequent depressive symptoms [57] in young adulthood. Nonetheless, to our knowledge, comprehensive studies examining a range of adolescent predictors and their impact on various dimensions of young adult health remain scarce. Such investigations hold promise for informing targeted intervention strategies to reduce pain and enhance health and well-being in this population.

This study aims to identify adolescent pain and psychosocial factors associated with young adult pain and health outcomes among individuals with chronic pain unrelated to chronic disease. Initially recruited from 15 tertiary pain clinics in the United States and Canada during adolescence (ages 11–17), participants were followed up during young adulthood (ages 18–24) approximately six years later. We hypothesize that more severe pain and individual (i.e., anxiety, depression, sleep quality) and interpersonal (i.e., social impairment, parent protective behaviors, family dysfunction) psychosocial vulnerabilities in adolescence will predict adverse young adult outcomes.

Method

Participants and Procedures

This is an observational, prospective cohort study of adolescents with chronic pain with long-term follow-up approximately 6 years later, when all youth were in the young adult age range (between 18–25 years old). We have published one paper from this dataset to document healthcare transition outcomes [44]. The current study aimed to determine adolescent vulnerabilities that predict long-term young adult pain and health outcomes. The Seattle Children’s Hospital Institutional Review Board approved this study before any research procedures. Participants received gift cards upon completion of surveys.

Baseline cohort (adolescents).

The original cohort included 273 adolescents ages 11–17 with chronic pain unrelated to chronic disease. Adolescents were recruited from 15 interdisciplinary pediatric pain clinics across the United States and Canada to take part in a clinical trial [49]. Inclusion criteria for the original study were: (a) ages 11–17 years, (b) pain present for at least 3 months duration, (c) pain occurred at least 1 time per week and interfered with daily functioning, (d) pain was not comorbid with chronic disease (e.g., cancer, sickle cell disease), (e) received evaluation as a new patient in a specialized tertiary pediatric pain clinic, and (f) had access to the Internet through a personal computer at home, work, school, or a public library. Exclusion criteria included (a) serious comorbid chronic condition in the patient (e.g., diabetes, arthritis, cancer) and (b) non-English speaking.

Follow-up cohort (young adults).

The only eligibility criteria for the follow-up study were that participants were 18–25 years old at the time of recruitment and had taken part in the original study. All 273 youth who took part in the original study were between the ages of 18–25 and thus were eligible to participate in the follow-up survey. There were no additional eligibility criteria for participation, e.g., young adults did not need to meet any of the original pain-related inclusion criteria. Participants were re-contacted via email, text, or phone. Those agreeing to participate in the study signed an online consent form through REDCap, a secure web-based data collection platform, and completed survey measures [24; 25]. Participants completed follow-up surveys between October 2018 and July 2021, which was on average 6.3 years (SD = 0.79; range = 4.1–8.1 years) after their baseline assessment. We could not reach 37 young adults, and 6 young adults declined participation due to a lack of interest or availability. 230 young adults consented, however, one young adult failed to complete their surveys after consenting. Thus, 84% (N = 229) of the original cohort completed the follow-up survey. There were no significant differences concerning baseline demographic information (adolescent age, sex, ethnicity, parent education, family income, country of residence [U.S. or Canada]), pain (intensity, disability), and psychosocial vulnerabilities between those who were eligible and participated at the follow-up time survey vs. those who did not take part in the follow-up survey.

Data Collection

Adolescent baseline survey measures:

Demographic Characteristics.

Parents completed a background form to collect information on adolescent age, sex, race, ethnicity, caregiver education, and household income.

Pain intensity and interference.

We assessed pain intensity through a 7-day diary report using an 11-point numerical rating scale (NRS, 0 = no pain, 10 = worst pain) [11; 65], and the average pain intensity score was computed. We assessed pain interference using the prospective version of the Child Activity Limitations Interview (CALI) [51], which measures a child’s perceived difficulties in performing typical daily activities because of pain. They rated difficulties using a 5-point scale (0 = no difficulty, 4 = extremely difficult) for 8 selected activities. Total scores ranged from 0 to 32, with higher scores representing more difficulties. We computed average daily activity limitation scores across 7 days. Reliability, validity, and responsiveness to changes have been demonstrated in school-aged children [51].

Pain location.

Pain locations were indicated in a daily pain diary where participants reported if they experienced pain in the head, face, shoulders, arms, hands, spine, legs, feet, chest, abdomen, and pelvis. We classified pain locations into 7 categories, including head or face pain, shoulder pain, arm or hand pain, spine pain, leg or foot pain, chest pain, and abdominal or pelvic pain. We created a variable indicating the total number of pain locations reported on the first day of the pain diary, ranging from 0 to 7.

Depression.

The Bath Adolescent Pain Questionnaire (BAPQ)–Depression subscale measured adolescents’ depressive symptoms within the past 2 weeks. The Depression subscale has 6 items measuring feelings and experiences of negative mood and sadness using a Likert scale (0 = never, 4 = always). Total scores ranged from 0 to 24, with higher scores indicating greater depression. Sufficient internal consistency, re-retest reliability, and construct validity for the Depression subscale have been demonstrated in adolescents with chronic pain [17].

Anxiety.

The BAPQ–General anxiety subscale measured adolescents’ anxiety within the past 2 weeks. The General anxiety scale has 7 items measuring excessive and uncontrollable worry and associated physical symptoms using a Likert scale (0 = never, 4 = always). Total scores ranged from 0 to 28, with higher scores representing greater anxiety. Sufficient internal consistency, re-retest reliability, and construct validity for the General anxiety subscale have been demonstrated in adolescents with chronic pain [17].

Social impairment.

We measured adolescent social impairment using the BAPQ–Social functioning subscale, which includes 9 items with a Likert response format (0 = never, 4 = always); sample items include “I go out and meet friends”, “I spend time talking to people” and “I enjoy social activities”. Total scores ranged from 0 to 36 with higher scores representing greater impairment in social functioning. Sufficient internal consistency, re-retest reliability, and construct validity for the Social functioning subscale have been demonstrated in adolescents with chronic pain [17].

Family functioning.

We measured adolescent family functioning using the BAPQ–Family functioning subscale, which measures conflicts, interactions, and relationships within the family, including items like “Family life is stressful”, and “There are fights between members of my family.” There are 12 items and total scores ranged from 0 to 48 with higher scores indicating greater family dysfunction. Sufficient internal consistency, re-retest reliability, and construct validity for the Family functioning subscale have been demonstrated in adolescents with chronic pain [17].

Sleep quality.

We measured adolescent perception of sleep quality using the Adolescent Sleep Wake Scale (ASWS) [35], a 28-item measure assessing 5 subscales (going to bed, falling asleep, maintaining sleep, re-initiating sleep, and returning to wakefulness) using a 6-point Likert scale (1 = always, 6 = never). We used the total average score in our analysis, ranging from 1 to 6, with a higher score representing better sleep quality. Acceptability, reliability, and validity have been shown in adolescents with comorbid sleep and medical conditions [19].

Parent protective behavior.

Adolescents completed the adolescent version of the Adult Responses to Children’s Symptoms (ARCS) [47], which was adapted from the parent-report ARCS [66], and assesses adolescents’ perception of parents’ responses to their chronic pain. Items were rated on a 5-point Likert-type scale (0 = never, 4 = always). We averaged responses, with higher scores representing more protective parent behaviors. Adolescent ARCS-Protect has demonstrated acceptability, reliability, validity, and responsiveness to change among youth with chronic pain [45; 47].

Young adult follow-up survey measures

Pain intensity and interference.

Young adults reported pain intensity and interference using the Brief Pain Inventory (BPI) [12]. The BPI is an 11-item instrument designed to assess the severity of pain and the extent to which it interferes with various activities. The pain severity subscale comprises 4 items assessing: (1) pain at its worst in the last week; (2) pain at its least in the last week; (3) pain on average; (4) pain right now. Each of these items is scored 0 (“no pain”) to 10 (“pain as bad as you can imagine”). The severity subscale score is the mean of these 4 items, where a higher score represents higher pain severity. The pain interference subscale comprises 7 items, each scored 0 (“does not interfere”) to 10 (“completely interferes”). Finally, the interference subscale score is the mean of these 7 items, where a higher score represents greater interference with daily activities. The BPI has shown good internal consistency in a variety of pain populations and concurrent validity with other pain instruments [52]. Participants also reported on pain location, duration, and frequency over the past 3 months.

Pain locations.

Pain locations were reported on the electronic Collaborative Health Outcomes Information Registry (CHOIR) body map which includes 74 body sites. We classified these body sites into 7 pain locations, aligning them with the pain locations reported in adolescence. We then created number of pain locations variable from these 7 locations, ranging from 0 to 7.

Anxiety.

We measured anxiety using the Generalized Anxiety Disorder 7-item Questionnaire (GAD-7) [37]. Each item is rated 0 (“not at all”) to 3 (“nearly every day”) with ratings summed to create a total score ranging from 0 to 21. Higher scores on the GAD-7 represent greater severity of generalized anxiety symptoms. For descriptive purposes, a score of 10 or higher was used to identify individuals with clinically significant (moderate to severe) symptoms of anxiety, who are likely to meet the diagnostic criteria for generalized anxiety disorder [58]. Psychometric studies have demonstrated that the GAD-7 has good psychometric properties [58].

Depression.

Young adults reported depressive symptoms using the Patient Health Questionnaire-9 (PHQ-9), which has demonstrated validity [32]. The PHQ-9 includes 9 items where participants rate how often depressive symptoms were bothersome over the past two weeks. Responses range from 0 (‘not at all’) to 3 (‘nearly every day’) with higher scores representing more severe depressive symptoms. Scores of 10 or higher indicate clinically significant depressive symptoms (range = 0–27) [32].

Social isolation.

We measured social isolation with the Patient-Reported Outcomes Measurement Information System (PROMIS) Social Isolation–Short Form 6a [23]. Young adults reported how often they felt excluded, detached, unknown, or avoided by others (1 = never, 5 = always). We calculated T-scores with a mean of 50 and an SD of 10 using a diverse sample of adults in the U.S. [23]. For descriptive purposes, T-scores > 60 (1 standard deviation [SD] above the national mean) were used to identify young adults experiencing higher than normative levels of social isolation.

Global physical health.

We measured physical health using the 2-item PROMIS Global Physical Health short form (v1.2) [26; 27]. The two items on this scale included: “In general, how would you rate your physical health?” (scored 5 = “Excellent” to 1= “Poor”) and “To what extent are you able to carry out your everyday physical activities such as walking, climbing stairs, carrying groceries, or moving a chair?” (scored 5 = “Completely” to 1= “Not at all”). Higher scores on this measure represent better global physical health. We calculated T-scores with a mean of 50 and an SD of 10 using the general population as the norm. For descriptive purposes, a cut-off score of 42 was used to identify individuals with fair to poor physical health [28].

Sleep disturbances.

We measured sleep quality using the Pittsburg Sleep Quality Index (PSQI) [9]. The PSQI is a 19-item questionnaire evaluating sleep disturbances over a 1-month interval. The global PSQI score ranges from 0 to 21 with higher scores representing higher levels of sleep disturbance. For descriptive purposes, a global PSQI score of > 5 was used to identify “poor sleepers” or individuals with poor sleep quality [9]. The PSQI has a well-established validity and reliability [7; 9; 10].

Data analysis plan

We described sample characteristics using means and SDs or numbers and frequencies. Given the small percentages of missing values across variables, we did not input missing values. Multiple linear regression models examined adolescent vulnerability factors (e.g., pain intensity, pain interference, number of pain locations, depressive symptoms, anxiety symptoms, social impairment, sleep quality, family dysfunction, parental protectiveness) as predictors for 8 young adult outcomes: pain intensity, pain interference, number of pain locations, global physical health, sleep disturbances, depressive symptoms, anxiety symptoms, and social isolation.

For preliminary analyses, we first ran multiple linear regression models including all adolescent demographic, pain, and psychosocial vulnerability variables as predictors (the full model) to understand the unique contribution of each predicting variable while accounting for the potential influence from other variables. To overcome limitations related to reduced power in detecting significant effects especially when predictors were highly correlated with each other, and further to identify a set of variables more likely to predict each adult outcome, our next step was to use a stochastic search variable selection (SSVS) approach [21; 22]. SSVS is a Bayesian variable selection method that provides information about the relative importance of predictors accounting for uncertainty in other predictors included in the model, which both increases power and decreases false-positive results [6; 61]. Through sampling thousands of regression models, the variable selection was based on the proportion of times each predictor was selected (i.e., marginal inclusion probability [MIP]) [6]. We specified a flat prior for the intercept, a Gamma (α= 0.01,β = −.01) prior for the parameters, and a common prior probability of 0.5 for each parameter so that each predictor had a 50/50 prior probability of being selected. We ran the Markov chain Monte Carlo (MCMC) sampler for 10,000 iterations, discarding the first 1,000 samples (a burn-in period) to ensure convergence [6]. For each outcome, we deemed variables with MIP ≥ 0.5 from the SSVS procedure as important predictors.

To avoid potentially missing any important predictors, we included any significant predictor (p < 0.05) in the full model (preliminary analyses, with all baseline variables), and further examined variables with 0.4 ≤MIP < 0.5 from the SSVS procedure to see whether these “borderline” variables significantly predicted the outcome of interest when including other selected variables. The final set of selected variables, for each outcome, included variables significant in the full model, variables with MIP ≥ 0.5 from the SSVS procedure, as well as variables with 0.4 ≤ MIP < 0.5 and significant in the final model. We computed adjusted R² for each outcome. Past research has used similar methods of variable selection to improve efficiency in detecting meaningful predictors [6]. We performed descriptive analyses in SPSS version 19 and regression analyses using R version 4.0.3 [1].

Results

Participant characteristics.

We present participant characteristics in Table 1. The mean age of the sample at baseline (during adolescence) was 14.7 years (SD = 1.6) and at follow-up (during young adulthood) was 21.0 years (SD = 1.6). Most participants were female sex (77.3%), with 83.0% living in the United States and 17.0% in Canada. The majority identified as White (87.1%) and non-Hispanic (93.3%). Most participants (63.4%) reported pain in more than 1 location.

Table 1.

Participant Characteristics

Characteristics	n	Mean (±SD), median (range), or n (%)
Adolescent age (years)	229	14.7 (±1.6)
Young adult age (years)	229	21.0 (±1.6)
Sex at birth: female	229	177 (77.3%)
Race	225
American Indian or Alaska Native		5 (2.2%)
Asian		2 (0.9%)
Black or African American		10 (4.4%)
Native Hawaiian or Other Pacific Islander		1 (0.4%)
White		196 (87.1%)
Othera		6 (2.7%)
More than one race		5 (2.2%)
Ethnicity	209
Hispanic or Latino		9 (4.3%)
Not Hispanic or Latino		195 (93.3%)
Unknown		5 (2.4%)
Household income (adolescence)	218
Less than $10,000		6 (2.8%)
$10,000 - $29,999		21 (9.6%)
$30,000 - $49,000		23 (10.6%)
$50,000 - $69,000		78 (35.8%)
$70,000 - $100,000		25 (11.5%)
More than $100,000		65 (29.8%)
Parent education	226
High school or less		28 (12.4%)
Vocational school/some college		56 (24.8%)
College		90 (39.8%)
Graduate/Professional school		52 (23.0%)
Adolescent pain location	227
Head or face pain		88 (38.8%)
Shoulder pain		77 (33.9%)
Arm or hand pain		52 (22.9%)
Spine pain		120 (52.9%)
Leg or foot pain		114 (50.2%)
Chest pain		37 (16.3%)
Abdominal or pelvic pain		87 (38.3%)
Adolescent number of pain locations	227	2 (1–7)
Adolescent multisite painb	227	144 (63.4%)
Young adult pain location
Head or face pain	229	109 (47.6%)
Shoulder pain		95 (41.5%)
Arm or hand pain		72 (31.4%)
Spine pain		152 (66.4%)
Leg or foot pain		135 (59.0%)
Chest pain		33 (14.4%)
Abdominal or pelvic pain		71 (31.0%)
Young adult number of pain locations		3 (0–7)
Young adult multisite painb		167 (72.9%)

Notes. Percentages calculated for non-missing values.

^aParticipants selected “Other” when answering the race question and did not specify the racial identity in an subsequent open-ended question.

^bMultisite pain defined as pain in 2 or more locations.

At follow-up (in young adulthood), participants reported moderate levels of pain intensity (M = 3.8, SD = 2.4) and pain interference (M = 3.4, SD = 2.7) on average, which was further categorized as 63.8% reporting moderate-to-severe pain interference. Most young adults (71.6%) reported chronic pain (pain for 3 months or longer, at least once per week) and 45.9% reported daily pain over the past 3 months. Compared to adolescence, a slightly greater percentage of participants (72.9%) reported pain in more than 1 location. Persistence rates of same pain location from adolescence to young adulthood are presented in Supplemental Table 1. Crude Spearman correlation coefficients between number of adolescent pain locations and young adult outcomes are presented in Supplemental Table 2. Participants rated their physical health lower than the national average (T-score M = 46.1, SD = 8.5), with 36.7% classified as with poor or fair physical health (i.e., T-scores of 42 or lower). The mean global score for PSQI was 9.6 (SD = 4.8), with over three-quarters of the sample (76.5%) reporting significant sleep disturbance (i.e., scores higher than 5). In terms of psychological health, 48.0% reported clinically elevated anxiety (i.e., scores of 8 or higher on GAD), and 40.2% reported clinically elevated depression (i.e., scores of 10 or higher on PHQ-9). Over half (55.9%) reported clinically elevated depression or anxiety, and 32.3% reported clinical elevations of both. Social isolation was similar to the national average (T-score mean = 51.1, SD = 10.2), with 19.7% classified as having higher than normative levels of social isolation (i.e., T-scores higher than 60).

Regression results

Regression analyses using the variable selection approach identified several adolescent predictors of young adult outcomes, as shown in Table 2 and Table 3. For each young adult outcome, the full prediction model where all predictors were included did not suggest high level of multicollinearity (variance inflation factor [VIF] < 5 for all predictors). The largest VIF was seen for adolescent anxiety symptoms (VIF = 3.1). The final model for each young adult outcome based on the SSVS approach did not suggest an issue of multicollinearity (VIF for all predictors across all outcomes < 1.5). Below we report the results from the final models.

Table 2.

Adolescent predictors of young adult pain and physical health outcomes, results from multiple linear regression models based on stochastic search variable selection (SSVS)^a

Adolescent Predictors	Pain intensityb		Pain interferencec		Number of pain locationsd		Physical functioninge
	Beta (95% CI)	P	Beta (95% CI)	P	Beta (95% CI)	P	Beta (95% CI)	P
Age
Female sex
Race
Household income			−0.32 (−0.56, −0.07)	0.011			1.06 (0.25, 1.86)	0.010
Caregiver education
Pain intensity	0.47 (0.17, 0.77)	0.003	0.45 (0.10, 0.79)	0.012
Pain interference
Number of pain locations			0.26 (0.05, 0.47)	0.018	0.47 (0.32, 0.62)	<0.001	−1.11 (−1.78, −0.45)	0.001
Depression
Anxiety			0.67 (0.30, 1.04)	<0.001
Sleep quality	−0.52 (−0.82, −0.21)	0.001			−0.44 (−0.69, −0.18)	0.001
Social impairment
Family dysfunction
Parental protectiveness
Adjusted R 2	0.10		0.16		0.25		0.07

Notes. CI = confidence interval. Race = Other racial and ethnic group vs non-Hispanic White.

^aSSVS: Predictors with marginal inclusion probability (MIP) ≥ 0.5 were selected, with a prior inclusion probability (for each predictor) of 0.5, 10,000 iterations of samples, and 1,000 burn-in (discarded) iterations; predictors with 0.4 ≤ MIP < 0.5 were examined and retained if significant in the final model; predictors significant in the full model were automatically retained (regardless of their MIP value).

^bFor young adult pain intensity, adolescent pain intensity and sleep quality were selected by the SSVS procedure; no other adolescent variables were significant predictors from the full model.

^cFor young adult pain interference, adolescent household income, pain intensity, number of pain locations, and anxiety symptoms were selected by the SSVS procedure; no other adolescent variables were significant predictors from the full model.

^dFor young adult number of pain locations, adolescent number of pain locations and sleep quality were selected by the SSVS procedure; no other adolescent variables were significant predictors from the full model.

^eFor young adult physical functioning, adolescent household income and number of pain locations were selected by the SSVS procedure; no other adolescent variables were significant predictors from the full model.

Table 3.

Adolescent predictors of young adult psychological health outcomes, results from multiple linear regression models based on stochastic search variable selection (SSVS)^a

Adolescent Predictors	Depressionb		Anxietyc		Social isolationd		Sleep disturbancese
	Beta (95% CI)	P	Beta (95% CI)	P	Beta (95% CI)	P	Beta (95% CI)	P
Age
Female sex							1.98 (0.39, 3.56)	0.015
Race
Household income
Parent education
Pain intensity	1.13 (0.22, 2.04)	0.015					0.76 (0.12, 1.41)	0.021
Pain interference
Number of pain locations
Depression
Anxiety			1.46 (0.47, 2.45)	0.004	2.06 (0.47, 3.65)	0.011	0.84 (0.07, 1.61)	0.032
Sleep quality	−1.46 (−2.40, −0.53)	0.002					−0.87 (−1.62, −0.11)	0.024
Social impairment
Family dysfunction	1.69 (0.77, 2.61)	<0.001	0.74 (−0.22, 1.71)	0.132	2.19 (0.64, 3.74)	0.006
Parental protectiveness
Adjusted R 2	0.15		0.09		0.12		0.17

Notes. CI = confidence interval. Race = Other racial and ethnic group vs non-Hispanic White.

^aSSVS: Predictors with marginal inclusion probability (MIP) ≥ 0.5 were selected, with a prior inclusion probability (for each predictor) of 0.5, 10,000 iterations of samples, and 1,000 burn-in (discarded) iterations; predictors with 0.4 ≤ MIP < 0.5 were examined and retained if significant in the final model; predictors significant in the full model were automatically retained (regardless of their MIP value).

^bFor young adult depressive symptoms, adolescent pain intensity, sleep quality, and family dysfunction were selected by the SSVS procedure; no other adolescent variables were significant predictors from the full model.

^cFor young adult anxiety symptoms, adolescent anxiety symptoms was selected by the SSVS procedure; adolescent family dysfunction was additionally included because it was a significant predictor in the full model.

^dFor young adult social isolation, adolescent anxiety symptoms and family dysfunction were selected by the SSVS procedure; no other adolescent variables were significant predictors from the full model.

^eFor young adult sleep disturbances, adolescent female sex, pain intensity, anxiety symptoms, and sleep quality were selected by the SSVS procedure; no other adolescent variables were significant predictors from the full model.

Adolescent sociodemographic characteristics.

Female sex and household income emerged as the only significant sociodemographic factors associated with young adult pain and health outcomes in the multivariable models. Higher household income in adolescence was associated with lower pain interference and better global physical health in young adulthood. Female sex was associated with greater sleep disturbances in young adulthood.

Adolescent pain.

Higher pain intensity in adolescence was associated with greater pain intensity, pain-related interference, depressive symptoms, and sleep disturbances at young adult follow-up. Adolescent pain-related interference was not associated with any outcomes. Number of pain locations in adolescence was associated with greater pain-related interference, higher number of pain locations, and poorer global physical health in young adulthood.

Adolescent emotional and behavioral health factors.

Multivariable analyses indicated that greater symptoms of anxiety during adolescence predicted greater pain-related interference, anxiety symptoms, social isolation, and sleep disturbances during young adulthood. Adolescent depressive symptoms were not associated with any young adult outcomes. However, poorer adolescent sleep quality was associated with greater sleep disturbances, higher pain intensity, higher number of pain locations, and greater symptoms of depression in young adulthood.

Adolescent social and family factors.

Family functioning during adolescence emerged as a significant predictor for several young adult outcomes. Specifically, lower family functioning (i.e., greater family conflict) was associated with greater depressive symptoms, anxiety symptoms, and social isolation in young adulthood. Adolescent social impairment and parent protective behaviors were not associated with any outcomes.

Discussion

Few studies have sought to understand the long-term impact of pediatric chronic pain and early risk factors for adverse adult outcomes. In line with previous research, findings reveal that around three-quarters (72%) of our sample continued to experience chronic pain in young adulthood. Among them, half experienced daily pain, and two-thirds reported moderate-to-severe pain interference. Moreover, our study uncovered several health challenges faced by youth once they reach young adulthood. Approximately one-third of the sample rated their physical health as below average, while half reported clinically elevated levels of anxiety or depression. Over three-quarters experienced sleep disturbances. Additionally, our analysis identified several significant adolescent predictors of young adult outcomes, including pain intensity, greater pain locations, anxiety, poor sleep quality, and family dysfunction. These findings underscore the challenging transition and poor prognosis of youth with chronic pain into adulthood and highlight several promising avenues for further research and interventions aimed at improving the health and well-being of this population.

Expanding upon prior research, our study revealed several adolescent vulnerabilities linked to impairments in young adulthood. First, adolescents with more severe pain and a greater number of pain locations subsequently experienced worse pain and physical functioning in young adulthood. Specifically, higher pain intensity was associated with more severe and disabling pain in young adulthood, while a greater number of pain locations predicted greater pain-related disability, a higher number of pain locations, and poorer global physical health. Drawing from quantitative sensory testing and neuroimaging research [4; 42; 59; 62] more severe and widespread pain may signify neurobiological alterations in pain modulation and sensory integration, contributing to a persistent trajectory of pain as youth grow older. Pain interference in adolescence did not predict any young adult outcomes, possibly because psychosocial vulnerabilities such as anxiety (see below) as well as maladaptive lifestyle factors (e.g., avoidance of physical activity) may be more robust predictors of long-term persistence of pain-related disability. Replicating these findings and expanding understanding of the interplay of psychological, biological, environmental, and lifestyle factors on persistent trajectories of pain and disability in adulthood is a research priority [48].

Additionally, our study uncovered a significant association between poor sleep quality during adolescence and increased pain intensity, number of pain locations, and depressive symptoms in young adulthood. These findings echo prior research examining daily relationships between sleep, pain, and mood in pediatric pain populations [3; 38; 63] but present novel prospective data on the enduring consequences of adolescent sleep on future pain and depression. Researchers have coined the interplay of sleep, pain, and mood as the “unhappy triad,” theorizing that poor sleep may be a unifying risk factor for both pain and depression [16]. Specifically, behaviors used to cope with pain, such as reduced physical activity and extended periods spent in bed, may disrupt sleep patterns, thus increasing pain sensitivity and weakening pain inhibitory systems. Moreover, sleep deficiency can directly affect emotional processes, potentially contributing to chronic depressive symptoms and mood disorders [41].

Our findings also highlight that adolescent anxiety symptoms confer risk for several long-term pain and health outcomes in young adulthood. Specifically, adolescent anxiety predicted greater pain-related disability, more pain locations, poorer physical health, elevated anxiety, worse sleep disturbances, and greater social isolation. Anxiety sensitivity has been identified as a salient emotional vulnerability fueling the fear-avoidance cycle of pain, thereby perpetuating a downward spiral of disability, social isolation, and maladaptive sleep habits and physical health [39]. Notably, anxiety disorders (e.g., generalized anxiety disorder, post-traumatic stress disorder or PTSD) co-occur at high rates in adolescents with chronic pain and are more prevalent in this population than any other psychiatric condition [31; 54]. Long-term persistence of pain and anxiety symptoms from adolescence to adulthood may be explained by the presence of shared neurobiological, cognitive, somatic, and behavioral factors theorized to drive development and long-term maintenance of both symptoms [5; 64].

Finally, our results underscore the potentially critical role of family functioning in shaping youths’ future health and well-being. Greater family dysfunction in adolescence predicted greater depression, anxiety, social isolation, and sleep disturbances in young adulthood. We know of only one other prospective study investigating the impact of family environment on long-term outcomes of youth with chronic pain in secondary data analyses: Sil et al. found that greater family conflict reported by children with fibromyalgia increased the risk for depression in adulthood [57]. Interestingly, social impairment and parent protective behaviors did not emerge as significant interpersonal predictors of young adult outcomes. To inform future research and intervention efforts, more detailed evaluations of social functioning, parenting behaviors, and family functioning are needed to understand more which social and family dimensions (e.g., parent-child conflict, family communication) are linked to long-term outcomes.

This research has important clinical implications. Healthcare providers can play an essential role by recognizing that youth with chronic pain and comorbid anxiety or sleep disturbance have increased risk of adverse outcomes. Psychological therapies, including cognitive-behavioral therapy (CBT) for pain management, have demonstrated success in youth with chronic pain [18]. Unfortunately, some individuals do not respond to CBT treatment (~25%; [18; 56]), and comorbid mood and sleep disturbances may attenuate treatment response [14; 40; 50]. Treatments for pediatric chronic pain must move away from a “one-size-fits-all” approach towards personalized or “hybrid” treatment approaches to effectively target pain and co-morbid mood and sleep disorders, and the underlying mechanisms that maintain these conditions. A recent pilot trial of a hybrid psychological intervention addressing pain and anxiety among youth with func15tional abdominal pain was found to be feasible and acceptable and lead to reductions in anxiety and pain-related disability [15]. Moreover, a recent study found that sleep impairments led to poorer executive function which, in turn, led to lower engagement in pain management [34], highlighting novel targets for intervention.

The study has several strengths, including a prospective cohort design with baseline and follow-up data collected during the crucial developmental periods. To our knowledge, this study is one of the few to evaluate a range of adolescent vulnerabilities linked to multiple dimensions of young adult health. Finally, we had a high response rate (80% participation) at the young adult follow-up. Despite these strengths, limitations should be acknowledged. First, we initially recruited our sample from tertiary pain clinics to participate in an internet-delivered CBT intervention, which limits generalizability. Second, although our predominantly White and female sample is representative of patients presenting to tertiary pain centers, the lack of racial and ethnic diversity may limit generalizability. Future research should aim to replicate findings in community and primary care samples. Third, these results should be replicated using longitudinal research that collects data at multiple time points from childhood to young adulthood to distinguish whether youth following specific trajectories are at greater (or less) risk for adverse outcomes. Fourth, adolescent and young adult measures used in the current study were different although they were intended to measure similar constructs. The use of developmentally appropriate measures maximizes content validity; however, this inconsistency may make it difficult to identify shared variability and complicate the interpretation of adolescent predictors for young adult outcomes. Finally, measures of trauma (i.e., PTSD) and protective factors (e.g., pain acceptance, self-efficacy, positive emotion) were lacking. Future research may integrate existing social-ecological and resilience-based frameworks of pediatric pain [13] could enhance understanding of risk and resilience factors that confer risk for, or protection against, adverse trajectories of pain during the transition to young adulthood.

In summary, we identified several adolescent risk or vulnerability factors associated with young adult pain and health outcomes among a cohort of adolescents who presented to tertiary pain clinics across the U.S. and Canada. These data represent an important first step toward identifying ways to optimize current psychological pain interventions. Psychological pain interventions can individualize treatment to target each adolescents’ vulnerabilities, including mood, sleep, and family risk factors, with the potential to disrupt a lifelong trajectory of pain and suffering. Indeed, adolescence and the transition to young adulthood represent critical years of rapid change and, thus, an opportune time to deliver interventions that can positively alter the course of health and well-being in adult life.

Data availability:

The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.