Investigating Home-Based Opioid Use Among Youth with Sickle Cell Disease Using Ecological Momentary Assessment

Abstract

Background:

Opioids are a common and essential treatment for acute sickle cell disease (SCD) pain. However, opioids carry well-known adverse side effects, including potential development of hyperalgesia and nociplastic pain. We characterized opioid use in youth with SCD using ecological momentary assessment (EMA) data, and investigated the relationships between home-based opioid use, pain, and a range of biopsychosocial factors.

Method:

Eighty-eight youth with SCD (aged 8 to 17 years) completed EMAs assessing home-based opioid use, pain, and related factors. Analyses consisted of descriptive and multilevel logistic regression to predict daily home opioid use.

Results:

Youth averaged 3.64 weeks of EMAs. Approximately 35% of the sample (n=31) took an opioid during the EMA period, and used them on only 24% of reported pain days. Youth who took opioids reported a higher percentage of pain days (t =−2.67, p < .05) and mean pain severity scores (t=−2.30, p < .05) than youth who did not take opioids. Multilevel logistic regression analyses indicated that high daily pain severity (OR = 1.02, p < .01), older age (OR = 1.324, p < .01), and low positive affect (OR = 0.91, p < .01) were each related to an increased likelihood of opioid use.

Conclusion:

Youth with SCD take opioids appropriately in response to their pain, based on daily self-report. Beyond daily pain severity, age and daily variation in positive affect were related to home-based opioid use. This suggests that behavioral interventions that enhance positive affect may promote reduced opioid use among youth with SCD.

Introduction

Sickle cell disease (SCD) affects approximately 100,000 individuals in the U.S. and millions globally, with 515,000 infants born with the illness annually.^{1, 2} It is an inherited genetic disorder that is characterized by hemolysis and inflammation, which results in chronic anemia, vaso-occlusion, tissue ischemia, and chronic multiorgan dysfunction. Vaso occlusive crises (VOC) pain is the most prominent symptom, often experienced during childhood as acute episodes of severe pain lasting hours to days to weeks at a time.^{3, 4} Studies suggest that the frequency of acute pain episodes increase with age, possibly contributing to the development of chronic pain occurring in 40% of youth and 50% of adults, emerging primarily during adolescence.^{5, 6} A majority of pain episodes among youth living with SCD are treated at home, commonly using a combination of non-opioid and opioid analgesics.^7–9

Opioids are a common and essential treatment for acute pain among youth living with SCD, with an estimated 46.3% of youth with SCD aged 10–19 using opioids.¹⁰ In a 2020 daily diary study of 32 youth with SCD aged 10–18, 34% reported using opioid analgesics on at least one diary day over 14 days.¹¹ In a 2023 weekly survey study of 48 adolescents and young adults aged 14–35 with SCD, approximately 63% of participants reported using opioids¹² over 30 days. In a qualitative study of young adults with SCD and caregivers of youth with SCD (including some caregiver-youth dyads), participants reported that their (or their child’s) pain would have more significant adverse impact on daily functioning and mental health without access to opioids.⁹ However, the use of opioids may result in adverse side effects including vomiting, dizziness, and poorer sleep quality,^{13, 14} while long-term use has been associated with opioid non-responsiveness, physical dependence, and has been proposed to contribute to the development of nociplastic pain, defined by the International Association for the Study of Pain¹⁵ as “pain that arises from altered nociception despite no clear evidence of actual or threatened tissue damage”.

Proper opioid use in the home setting is essential to both optimize the effectiveness and reduce adverse side effects for youth living with SCD; thus, it is necessary that we better understand factors related to daily home-based opioid use. Daily diary studies have found more severe pain to significantly predict more home-based opioid use among youth with SCD.^8,12,16 However, findings from a qualitative study suggest that the decision-making process behind home-based opioid use among both young adults with SCD and caregivers of youth with SCD may be complex and involve a range of factors not related to pain severity, including age, stress, and negative emotions.⁹ The current study is focused on investigating biopsychosocial factors beyond pain associated with home-based opioid use patterns among youth with SCD.

Only four studies have investigated biopsychosocial factors beyond pain as predictors of opioid use among youth with SCD. One study found age to be a significant predictor of opioid use, with adolescents (ages 14–19) reporting significantly more opioid use than youth under the age of 14.⁸ In a daily diary study of 27 adolescents (aged 13–17) with SCD, baseline optimism did not significantly predict percentage of days youth used opioids, but moderated the association of average pain intensity and opioid use such that high pain was only associated with a higher percentage of days of opioid use at high and medium levels of optimism.¹⁶ A 2-week electronic daily diary study of 32 youth with SCD found that when controlling for daily pain intensity, higher negative affect was associated with an increased likelihood of opioid use.¹¹ Additionally, while youth reported pain catastrophizing (i.e., the tendency to magnify the threat of pain¹⁷) was not a significant predictor of home opioid use, parent-reported pain catastrophizing was associated with an increased likelihood of their child taking an opioid. In contrast, a prospective longitudinal study of 48 adolescents and young adults found youth reported pain catastrophizing significantly predicted opioid use, such that greater pain catastrophizing was associated with a decrease in opioid use.¹²

Overall, these findings suggest that beyond pain, a range of biopsychosocial factors may contribute to home-based opioid use patterns among youth with SCD. However, the sample sizes are relatively small, they focus on just a few biopsychosocial factors, and only one considers a positive, strength-based factor, optimism. Given potential side effects associated with prolonged opioid use, a greater understanding of biopsychosocial factors related to opioid use is needed to aid in the development and adaptation of multimodal interventions to treat pain among youth with SCD. The aims of the current study are to characterize opioid use in youth with SCD using ecological momentary assessment (EMA) data, and to investigate the relationships between home-based opioid use, pain, and a range of biopsychosocial factors, including age, sex, SCD genotype, sleep quality, positive and negative affect, and stress. We hypothesized that high daily pain severity would be related to more home-based opioid use. We also hypothesized older age, low positive affect, high negative affect, and high stress will be associated with more home-based opioid use when controlling for pain severity.

Method

Participants

We conducted a secondary analysis of data from a larger study of sleep, pain, and related factors in youth with SCD.¹⁸ The sample was recruited during their regularly scheduled clinic visits from three regional pediatric SCD clinics in the southeastern US. Inclusion criteria for the original study included having a diagnosis of SCD, aged 8 to 17 years, proficient in English, and having experienced a SCD pain episode within the past year – defined as pain due to SCD lasting at least 20 minutes.¹⁹ The current study also required that youth had at least 2 weeks (24 entries) of daily EMA data. Of the 123 youth in the original study, 88 (72%) meet the additional criteria. Previous investigations using the database have focused on assessing the daily relationships between sleep, pain, and mood. This will be the first to focus on opioid use. There were no differences in age, sex, SCD genotype, and hydroxyurea use (i.e., the only SCD modifying Unites States Federal Drug Administration approved drug at the time of data collection) between the current study sample and the larger study sample. Demographic characteristics are detailed in Table 1.

TABLE 1:

Demographic and Disease Characteristics

	n	%	N
Female	52	59%	88
SCD Genotype
HBSS	44	50%	88
HbSβ0	3	3%	88
HbSβ+	12	14%	88
HBSC	27	31%	88
Other	2	2%	88
% on Hydroxyurea	47	54%	88
% Used Opioids during EMA	31	35%	88
	M	SD	Range
Age	11.66	2.99	8 to 17
Person-Centered Mean EMA Variables
% of Opioid Use	8%	19%	0% to 100%
% of Pain Days	22%	27%	0% to 100%
Pain Severity	13.96	19.65	0 to 88.39
Sleep Quality	74.66	16.59	22.07 to 99.94
Positive Affect	17.12	4.14	8.33 to 25.00
Negative Affect	7.69	2.62	5.04 to 18.10
Stress	2.03	1.94	0 to 7.27

EMA = ecological momentary assessment.

Procedure

At baseline, youth’s guardians completed measures assessing youth’s demographic information, and youth’s medical charts were reviewed for disease information. Youth then completed twice daily EMAs, brief surveys administered via a mobile application, every morning and evening for up to 4 weeks. In the evening assessment, they reported on their pain, opioid use, positive affect, negative affect, and stress levels. In the morning assessment, they reported on their sleep quality the night before. The research was approved and monitored by the institutional review boards at all three data collection sites, and all data was collected between 2011 and 2015.

Measures

Demographic and Disease Information

Guardians provided information about the youth’s age, biological sex at birth, and race. Medical records were examined to determine SCD genotype and hydroxyurea use. For the analyses, SCD genotype was categorized into severe genotypes (HbSS and HbSβ⁰ = 1) and moderate (HbSC, HbSβ⁺, and other = 0).

EMA of Opioid Use

In the evening EMA, youth were asked “what pain medication did you take today?” and provided a list of pain medications used by youth with SCD during a previous diary study.¹⁴ They could also indicate “other” and provide the name of any pain medication not on the list. Responses were coded as either nonopioid or opioid analgesics. Previous studies have reported valid results when utilizing similar strategies to assess daily medication use in youth with SCD.^{8, 11, 20} A full list of opioids reported during the EMA period are in Supplement Table 1.

EMA of Daily Pain

In the evening EMA, youth indicated if they experienced any SCD pain that day. If they answered no, a pain severity score of 0 was automatically assigned. If they answered yes, youth were prompted to rate their average pain severity for the day using a 100 mm visual analog scale (VAS) ranging from “not hurting at all” (0 mm) to “hurting a whole lot” (100 mm). Previous research has established the reliability and validity of the VAS scale for assessing daily SCD pain in youth.^{14, 21}

EMA of Negative and Positive Affect

In the evening EMA, youth responded to 10 items from the Positive and Negative Affect Schedule for Children (PANAS-C) to assess positive and negative affect. Following the approach of previous EMA studies of youth^{22, 23, 24}, a subset of 10 items from the PANAS-C were administered, comprising five items measuring negative affect and 5 items measuring positive affect. Youth rated how much of each emotion they were feeling that day using a scale ranging from “very slightly or not at all” (1) to “extremely” (5). The scores for the five items in each category were then added, resulting in a total score between 5 to 25. The within-person Cronbach alpha mean for the negative affect was 0.83 (range = 0.69 to 0.88) and positive affect was 0.79 (range = 0.61 to 0.86) for the study sample.

EMA of Stress

In the evening EMA, youth completed the Daily Events Inventory²⁵ to assess stress. The inventory consists of 17 negative daily events and youth indicate whether they have experienced the event that day. Then, the daily stress total is calculated by adding the “yes” items. Previous research has supported the reliability and validity of the measure to assess daily stress.^{14, 25}

EMA of Sleep Quality

In the morning EMA, youth rated “how well they slept last night” using a 100 mm VAS ranging from “did not sleep well” (0 mm) to “slept very well” (100 mm). Daily self-reports are the gold standard for assessing sleep quality and previous research has supported the reliability and validity of daily reports for assessing sleep quality, including in youth.^{26, 27, 28}

Analytic Plan

Descriptive and multilevel logistic regression analyses were conducted using SPSS.^{29, 30} To assess factors influencing home-based opioid use, multilevel logistic regression models with two levels (Level 1: daily observations, Level 2: youth participant characteristics) were calculated. A null model predicting daily opioid use (level 1 variables) was fitted with no variables to estimate the intraclass correlation coefficient (ICC; proportion of variance of opioid use at the youth level) and the proportion of variance at the daily level. The multilevel logistic regressions used restricted maximum likelihood estimation, unstructured between-person error structures, and continuous-time autoregressive within-person error structures. The final multilevel logistic regression model predicting daily opioid use used the following level 1 variables: daily pain severity, self-reported sleep quality, positive affect, negative affect, and stress. It also included a random intercept; and the following level 2 covariates: age, sex, SCD genotype, and whether the youth were currently taking hydroxyurea.

Youth completed a mean of 81.87% of the possible EMA entries, resulting in a mean of 51 EMA entries per youth (approximately 25 morning entries and 25 evening entries each), which is equivalent to 3.64 weeks per youth participant.¹⁸ Missing data was handled using full information maximum likelihood in the multilevel logistic analyses.³¹

Results

Table 1 provides descriptive statistics for the demographic and disease characteristics, and person-mean centered EMA information for the sample. Youth reported taking opioids on 2% of non-pain days and 24% of pain days. Approximately 35% of the sample (n=31) took an opioid during the EMA period; with 1 participant taking an opioid every day. Distributions for percentages of opioid and pain days for this subsample are presented in Figure 1. Youth who took opioids reported taking one on an average of 22% of their EMA days; and reported a higher percentage of pain days (t =−2.67, p < .05) and mean pain severity scores (t=−2.30, p < .05) than youth who did not take opioids (see Table 2). For the opioid use subsample, the percentage of days youth took opioids was positively related to their percentage of pain days (r = .64, p < .01), person-centered mean pain severity scores (r=.54, p<.01), age (r=.59, p<.01), and negatively related to their person-centered mean positive affect scores (r=−.35, p<.05). There were no differences in the percentage of days youth took opioids based on sex, SCD genotype severity, or hydroxyurea use (p > .05).

FIGURE 1:

Percentage of Youth by % of Opioid Use and Pain Days (n = 31)

TABLE 2:

Comparing Youth Who Used Opioids Versus Not During EMA Period

	Opioid (n = 31)	Non-Opioid (n = 57)	Chi-square	p
Female	58%	60%	0.02	0.89
SCD Genotype - Severe	52%	54%	0.06	0.80
% on Hydroxyurea	58%	52%	0.32	0.57
			t	p
Age (years)	11.87	11.54	−0.49	0.63
Avg EMA Variables
% of Pain Days	32%	16%	−2.67	0.01*
Pain Severity (0–100)	20.92	10.17	−2.30	0.03*
Sleep Quality (0–100)	70.96	76.67	1.55	0.13
Positive Affect	16.63	17.40	0.83	0.41
Negative Affect	8.03	7.51	−0.89	0.37
Stress	2.30	1.89	−0.93	0.35

^*p < .05.

EMA = ecological momentary assessment.

Multilevel Logistic Regression Findings (Subsample who reported Opioid Use, N = 31)

The ICC was 0.40, indicating only 40% of variance in opioid use is due to between person variability. Conversely, this indicates that 60% of the variance is due to within person variability. Table 3 presents results from the full multilevel logistic regression analyses predicting daily opioid use using the subsample of youth who reported opioid use. High daily pain severity (Odds Ratio [OR] = 1.02, p < .01), older age (OR = 1.32, p < .01), and low positive affect (OR = 0.91, p < .01) were related to an increased likelihood of taking an opioid analgesic. Specifically, the ORs indicate that a 1-year increase in age was associated with 32% increased likelihood of taking an opioid, and a 1 unit increase in pain severity on the 100-point scale was associated with an approximately 2% increased likelihood of taking an opioid on that day.

TABLE 3:

Multilevel Logistic Regression Predicting Daily Opioid Use (n = 31)

Variable	Beta	SE	t	p	OR	95% CI
Intercept	−3.32	1.40	−2.37	0.02*	0.04	<0.01 to 0.57
Age	0.28	0.09	3.21	<0.01**	1.32	1.12 to 1.57
Female	−0.19	0.47	−0.41	0.68	0.82	0.33 to 2.08
SCD Genotype - Severe	−0.17	0.64	−0.26	0.79	0.84	0.24 to 2.98
Hydroxyurea Use	0.17	0.73	0.23	0.82	1.18	0.28 to 4.96
EMA Variables
Pain Severity	0.02	0.00	5.22	<0.01**	1.02	1.02 to 1.03
Sleep Quality	−0.01	0.01	−1.14	0.26	0.99	0.98 to 1.01
Positive Affect	−0.10	0.03	−2.75	<0.01**	0.91	0.85 to 0.97
Negative Affect	−0.04	0.04	−1.00	0.32	0.96	0.88 to 1.04
Stress	0.04	0.07	0.55	0.58	1.04	0.91 to 1.19

^*p < .05;

^**p < .01;

EMA = ecological momentary assessment. This analyses only includes the sample of youth who reported using opioids during the EMA period.

Of note, we conducted a supplemental multilevel logistic regression using the larger sample that included those who did and did not report opioid use during the e-diary period (See Supplementary Materials, Table 2). Those findings were consistent with the pattern of findings for this subsample. Also, one participant reported taking Oxycontin (an extended-release opioid that is intended to be used daily) three times during the EMA period. To ensure that their findings did not overly influence the models, we re-ran all of the primary analyses with that participant removed, and none of the findings significantly changed. Thus, the original findings with that participant included are presented above.

Discussion

The current study aimed to characterize opioid use among youth with SCD and examine the relationship between pain, biopsychosocial factors, and home-based opioid use. This study builds on previous examinations of opioid use in the pediatric SCD population and is the first to explore positive affect, a factor that has been connected to pediatric SCD pain and other symptoms, as a predictor of home-based opioid use among youth with SCD. Consistent with the previous daily diary study of opioid use in pediatric SCD¹¹, approximately 35% of youth in the current study reported using opioids during the EMA period. Findings suggest that youth with SCD are taking opioids appropriately in response to their pain, given that patterns of SCD pain and opioid use are strongly related in the sample. This is consistent with a recent retrospective study of Medicaid claims data that found youth prescribed opioids for the first time show a strong correlation between VOCs and days’ supply of opioids, and that patterns of opioid use are relatively low and not increasing over a 3-year period.³²

Beyond pain severity, daily opioid use was related to being older and low daily positive affect. These findings are consistent with a previous daily diary study finding age to significantly predict opioid use, with older youth reporting more opioid use than younger youth.⁸ It is likely that as youth with SCD get older, they experience changes in pain symptoms, resulting in greater daily opioid use. Among our subsample who took opioids, approximately 22% reported pain on 50% or more of their EMA days and approximately 13% reported taking opioids on 50% or more of their EMA days. This pattern of pain and opioid consumption during the 4-week EMA period is consistent with chronic SCD pain criteria.³³ More work is needed to determine the general timing of the transition from recurrent to chronic SCD pain patterns, though a 2-year longitudinal study of pediatric SCD patients suggests that it can happen as early as 12 years of age.⁵

As for positive affect, while no previous study has examined positive affect as a predictor of opioid use among youth with SCD, previous studies have found greater positive affect to be associated with lower pain intensity and better self-reported sleep quality among pediatric SCD patients.¹⁸ In other pain populations, positive affect has also been found to buffer the impact of pain on functioning and other health outcomes.³⁴ According to the fear-avoidance pain model³⁵, pain fuels pain catastrophizing, which induces pain–related fear and avoidance. The model further purports that avoidance leads to both disability and depression which then maintains the cycle of pain³⁵ perhaps leading to increased opioid use. Positive affect is proposed to widen attention and promote cognitive and behavioral flexibility, which reduces inflammation and cortisol production as well as countering fear-avoidance.³⁴These findings indicate that behavioral interventions that promote positive affect and aim to transform avoidance behaviors into healthy, preventative behaviors may be effective in promoting reduced opioid use among youth with SCD. Further research is needed to test this proposed effect.

Notably, within our sample, opioids were only used on 23.74% of reported pain days. It is unclear if this discrepancy between pain days and opioid use is due to pain severity experienced, psychological factors not assessed in the current study (i.e., stigma avoidance, personal beliefs), or availability. It may be that youth and their families are able to use other pain management strategies, including the use of non-opioid analgesics and behavioral strategies, to successfully deal with some SCD pain. Alternatively, this may suggest an undertreatment of SCD pain, which may contribute to increased pain frequency and severity. Overall, more data is needed on opioid use patterns in the context of other strategies to ensure that youth are optimizing their pain management.

The current findings should be interpreted in the context of a few limitations. First, opioid use was exclusively measured via self-report. Thus, rates of opioid use may have been influenced by recall and response bias. The EMA method allows for real-time, in context examination of pain and opioid use patterns. However, this method can be strengthened by pairing it with more objective measures of pain medication use. These include medication event monitoring system (MEMS) Caps, prescription refill histories, and biological indicators of opioid use. Second, participants were only asked to report if they took pain medication that day without reporting on medication dosage, timing or frequency throughout the day. If such information was reported, a more comprehensive description of opioid use and factors associated with timing could be reported. An examination of the timing at which individuals introduce opioid and non-opioid analgesics in pain control efforts may help optimize opioid use for individuals with SCD. This information would be informative for developing programs to promote proper opioid use among youth with SCD and their families. Third, the study only focused on youth-related factors. Results from a qualitive study of young adults with SCD and caregivers of children with SCD indicated that opioid use decision making often exists along a continuum from caregiver decision to caregiver-child collaborations to youth decisions based on developmental stage of the child and context.⁹ Thus, future research is needed to examine the influence of caregiver and child-caregiver dyadic influences on opioid use among youth with SCD.

Future research should also examine opioid use in the context of other pain management strategies, including behavioral strategies. As pain frequency and opioid use increase with age, drug tolerance is worth noting in the management of SCD pain. Routine use of opioids may lead to tolerance where higher doses are required to achieve the same level of relief.³⁶ If this tolerance is not addressed, it is possible that SCD pain episodes are undertreated and patients are not receiving adequate pain relief from well-established regimens. However, progressive dose escalation for long-term opioid therapy can lead to opioid-induced hyperalgesia and is not an optimal solution for a chronic illness. Our findings suggest that complementary interventions that target raising positive affect may be opioid sparing. Two interventions focused on enhancing positive affect are acceptance and commitment therapy (ACT) and mindfulness-based stress reduction (MBSR³⁷). ACT has been associated with reduced pain in other populations³⁸, and a case study of 16 year old with SCD saw improvement in quality of life, but did not see substantial changes in pain and did not assess for changes in opioid use.³⁹ There are currently still in progress clinical trials assessing ACT and MBSR in people living with SCD.^40–42 As mentioned earlier, clinical trials testing whether positive affect focused interventions may be useful for reducing opioid use among youth with SCD are needed.

In conclusion, current study findings indicate that youth with SCD take opioids appropriately in response to pain, based on daily self-report. Beyond daily pain severity, older age and low daily positive affect were related to an increased likelihood of home-based opioid use. This suggests that behavioral interventions that enhance positive affect may promote reduced opioid use among youth with SCD. Future research is needed to investigate this hypothesis, as well as to further investigate additional factors, such as caregiver factors, related to home-based opioid use among youth with SCD to best inform efforts to optimize opioid use among youth with SCD.

Abbreviations

SCD

Sickle cell disease

VOC

Vaso-occlusive crises

EMA

Ecological momentary assessments

VAS

Visual Analog Scale

PANAS-C

Positive and Negative Affect Schedule for Children

ICC

Intraclass correlation coefficient

OR

Odds Ratio

MEMS

Medication event monitoring system