Abstract
Background
Pediatric cancer survivors often have pain, which may be managed with opioids. We examined the prevalence of opioid prescriptions, potential misuse, and substance use disorders (SUDs) among pediatric cancer survivors during the first year posttherapy.
Methods
Using MarketScan Commercial Database, we identified 8969 survivors (aged 21 years or younger at diagnosis) who completed cancer therapy in 2009-2018 and remained continuously enrolled for at least 1 year posttherapy and 44 845 age-, sex-, and region-matched enrollees without cancer as a comparison group. Outcomes included opioid prescriptions, any indicator of potential prescription opioid misuse, and SUDs within 1 year posttherapy. Outcomes were compared between survivors and noncancer peers in bivariate and adjusted analyses, stratified by off-therapy age (children: 0-11 years; adolescents: 12-17 years; young adults: 18-28 years). All statistical tests were 2-sided.
Results
A higher proportion of survivors than noncancer peers filled opioid prescriptions (children: 12.7% vs 2.0%; adolescents: 22.9% vs 7.7%; young adults: 26.0% vs 11.9%). In models adjusting for sociodemographic factors and health status, survivors remained 74.4%-404.8% more likely than noncancer peer to fill opioid prescriptions (P < .001). The prevalence of potential misuse or SUDs was low, with 1.4% of child, 4.7% of adolescent, and 9.4% of young adult survivors fulfilling at least 1 criterion; however, it was higher than noncancer peers (0.1%, 1.4%, and 4.3%, respectively). In adjusted models, the likelihood of potential misuse among survivors remained at least 2 times higher than that among noncancer peers (P < .001), and the difference in SUDs became nonstatistically significant.
Conclusion
Statistically significantly higher rates of opioid prescriptions and potential misuse were found among pediatric cancer survivors within 1 year posttherapy as compared with peers without cancer.
In the United States, there is a growing population of nearly 500 000 survivors of pediatric cancer (1,2). Survivors are at risk for physical and psychosocial late effects (3,4), as well as pain during and after cancer treatment (5,6). Approximately 1 in 3 pediatric cancer survivors experiences moderate to severe pain during adulthood (7). Pain during survivorship may result from cancer treatment and/or treatment-related late effects, such as subsequent malignancies and neuropathy (5,8).
For moderate to severe cancer-related pain, opioids are recommended to improve pain and function (9). However, opioid use among youth can have side effects and may increase the likelihood of misuse, abuse, and addiction (10), because opioid use at young ages is associated with misuse of prescription opioids later in life (11). Furthermore, adolescent and young adult (YA)–aged survivors of pediatric cancer may be at particular risk for opioid use and misuse, given their high prevalence of physical and mental health problems (12-14) and unique challenges faced during developmental periods (15).
Pediatric cancer survivors must receive adequate pain management with minimized risk for opioid misuse, abuse, and addiction. Current opioid-prescribing guidelines exclude people surviving cancer (10,16,17). To inform future guidelines and practice targeting these high-risk cancer survivors, the need to identify opioid use and misuse patterns in this population is critical.
We aimed to examine the prevalence of opioid prescription, potential misuse, and substance use disorders (SUDs) in pediatric cancer survivors during the first year posttherapy as compared with peers without cancer. Among pediatric cancer survivors, we also examined salient factors associated with the risk for potential misuse and SUDs.
Methods
Data and Sample
Data were from the 2009-2019 IBM MarketScan Commercial Claims and Encounters Database available at Emory University, which contains a nationwide convenience sample of de-identified inpatient, outpatient, and pharmacy insurance claims data provided by private health plans and medium-to-large employers in the United States for approximately 50 million enrollees per year (18,19). This study was deemed exempt by the Emory University institutional review board.
Pediatric cancer survivors were defined as enrollees who completed therapy for hematologic (including leukemia and lymphoma), central nervous system, bone or connective tissue, and/or gonadal cancers between January 1, 2009, and December 31, 2018, and were aged 21 years or younger at index cancer diagnosis visit (20). Survivors were required to have at least 2 outpatient and/or inpatient health-care claims with cancer diagnoses on distinct dates; the first claim observed was defined as the index visit. Consistent with prior research (20), cancer diagnoses were identified using International Classification of Diseases (ICD) diagnosis codes within the respective clinical classifications software categories (Supplementary Table 1, available online) developed by the Agency for Healthcare Research and Quality (21). The types of cancer included were the most prevalent cancers in the US pediatric population (22). We further required survivors to have health-care claims associated with cancer treatment (ie, surgery, chemotherapy, and/or radiation therapy) identified using procedure (Current Procedural Terminology/Healthcare Common Procedure Coding System) codes and National Drug Codes (20). We defined the end of therapy (EOT) date as 30 days following either the date of the last inpatient or outpatient claim for a cancer therapy or the date when the last oral chemotherapy prescription was concluded, whichever was later (20). We restricted survivors to those continuously enrolled in a health plan for at least 1 year following EOT, during which there was no health-care encounter related to cancer treatment.
Additionally, we identified a cohort of enrollees who had no claim indicating any cancer diagnosis or personal cancer history [using ICD diagnosis codes (21,23,24)] throughout our study period as a comparison group. Enrollees in this group were matched to survivors based on age, sex, and geographic region and randomly sampled at a 5-to-1 ratio (25,26). Matched noncancer peers were then assigned with the EOT date of the matched survivor. Noncancer peers were also required to have at least 1 year of continuous enrollment from their assigned EOT date. Our final analytic sample comprised 53 814 enrollees, including 8969 survivors and 44 845 matched noncancer peers.
Outcome Measures
Opioid Prescription
Pharmacy claims associated with prescription opioids were identified using National Drug Codes compiled by the Centers for Disease Control and Prevention (27). We created an indicator dichotomizing whether enrollees filled any opioid prescription during the first year posttherapy. Of opioid recipients, we calculated the number of prescriptions and days of supply from opioid claims filled. We calculated daily morphine milligram equivalents (MMEs) for each opioid prescription using the medication strength, days of supply, quantity dispensed, and relevant conversion factor (27).
Potential Misuse of Prescription Opioids
We constructed 4 dichotomous variables to denote whether, over the first year posttherapy, enrollees experienced 1) high daily opioid dose (ie, ≥1 opioid prescription with daily dose of ≥100 MMEs), 2) opioid overlap (ie, multiple opioid prescriptions that overlapped for ≥7 days), 3) opioid and benzodiazepine overlap (ie, prescriptions that overlapped for ≥7 days; Supplementary Table 2, available online), and 4) opioid dose escalation (ie, ≥50% increase in mean MMEs per month twice consecutively over the year). Although these conditions can occur in monitored and prescribed clinical care, they have been applied in previous research as indicators for potential misuse of prescription opioids among the general population (28). Additionally, we created a variable dichotomizing whether the enrollee met at least 1 of the 4 criteria for potential misuse during the first year posttherapy.
Substance Use Disorder
Three dichotomous variables were created to denote whether enrollees received a diagnosis of 1) opioid use disorder (OUD), 2) nonopioid SUD (eg, alcohol, tobacco, marijuana, other psychotherapeutic drug, or illicit drug use disorder), and 3) any SUD (ie, OUD and/or nonopioid SUD) during the first year posttherapy. OUD was identified following the Chronic Condition Warehouse Condition Algorithms (29). A nonopioid SUD was defined as having at least 2 outpatient and/or inpatient claims associated with the SUD ICD diagnosis codes (Supplementary Table 3, available online) over the first year posttherapy (30–32).
Covariates
Sociodemographic characteristics included sex, geographic region, and rural vs urban residence. Health plans were categorized using monthly enrollment information over the first year posttherapy (see Table 1). To assess health status, we created dichotomous indicators to denote the presence of mental health disorders and painful conditions, defined as at least 2 outpatient and/or inpatient claims with the respective diagnosis codes (Supplementary Tables 3 and 4, available online) within 1 year posttherapy. Finally, cancer-related factors included 2 mutually exclusive categorical variables capturing cancer type and treatment modalities (hematopoietic stem cell transplantation [HSCT], radiation therapy [no HSCT], chemotherapy [no radiation or HSCT], or surgery only). HSCT was identified using relevant procedure codes (33,34).
Table 1.
Sample characteristics
| Characteristics | Aged 0-11 yearsa |
Aged 12-17 yearsa |
Aged 18-28 yearsa,b |
||||||
|---|---|---|---|---|---|---|---|---|---|
| Noncancer peers, %c | Survivors, %c | P d | Noncancer peers, %c | Survivors, %c | P d | Noncancer peers, %c | Survivors, %c | P d | |
| (n = 16 240) | (n = 3248) | (n = 12 795) | (n = 2559) | (n = 15 810) | (n = 3162) | ||||
| Sociodemographic factors | |||||||||
| Female (vs male) | 46.9 | 46.9 | >.99 | 42.0 | 42.0 | >.99 | 38.4 | 38.4 | >.99 |
| Region | >.99 | >.99 | >.99 | ||||||
| Northeast | 19.6 | 19.6 | 21.3 | 21.3 | 21.8 | 21.8 | |||
| North Central | 24.4 | 24.4 | 22.4 | 22.4 | 24.1 | 24.1 | |||
| South | 36.1 | 36.1 | 35.9 | 35.9 | 34.4 | 34.4 | |||
| West | 19.9 | 19.9 | 20.4 | 20.4 | 19.7 | 19.7 | |||
| Rural (vs urban) residence | 12.5 | 11.7 | .22 | 13.3 | 13.4 | .86 | 13.9 | 12.8 | .10 |
| Health plan type | .58 | .08 | .09 | ||||||
| Health maintenance organization alone | 13.0 | 13.8 | 12.6 | 11.3 | 13.3 | 13.9 | |||
| High-deductible plan alone | 12.0 | 11.9 | 12.2 | 11.6 | 12.4 | 11.5 | |||
| Preferred provider organization alone | 56.3 | 56.3 | 56.5 | 59.2 | 55.4 | 57.0 | |||
| Other or multiple plan types | 18.7 | 18.0 | 18.7 | 17.9 | 19.0 | 17.5 | |||
| Health status | |||||||||
| Painful conditions (vs none) | 5.4 | 28.4 | <.001 | 15.4 | 40.8 | <.001 | 13.9 | 32.3 | <.001 |
| Mental health conditions | |||||||||
| Mood disorder (vs none) | 0.4 | 1.4 | <.001 | 3.3 | 6.2 | <.001 | 4.1 | 8.2 | <.001 |
| Anxiety disorder (vs none) | 1.1 | 3.4 | <.001 | 3.1 | 6.6 | <.001 | 3.6 | 7.3 | <.001 |
| Other mental health disorder (vs none) | 4.3 | 7.7 | <.001 | 6.6 | 10.7 | <.001 | 4.3 | 7.4 | <.001 |
| Cancer-related factors | |||||||||
| Site and type of cancer | |||||||||
| Hematologic | — | 49.4 | — | — | 45.6 | — | — | 47.7 | — |
| Bone or connective tissue | — | 10.5 | — | — | 13.7 | — | — | 10.7 | — |
| Central nervous system | — | 29.6 | — | — | 27.0 | — | — | 16.0 | — |
| Gonadal | — | 2.7 | — | — | 7.9 | — | — | 22.1 | — |
| Multiple types of cancer | — | 7.8 | — | — | 5.8 | — | — | 3.5 | — |
| Cancer treatment | |||||||||
| Surgery only | — | 22.9 | — | — | 30.2 | — | — | 25.9 | — |
| Any HSCT | — | 7.0 | — | — | 7.7 | — | — | 5.6 | — |
| Any radiation therapy (no HSCT) | — | 19.0 | — | — | 24.4 | — | — | 23.9 | — |
| Any chemotherapy (no radiation or HSCT) | — | 51.0 | — | — | 37.7 | — | — | 44.6 | — |
Age at the end of cancer therapy. HSCT = hematopoietic stem cell transplantation; — = data not applicable.
There were 7 survivors aged 26-28 years at the end of cancer therapy; these survivors (diagnosed younger than age 22 years) remained on cancer treatment for a range of 4.2-6.4 years.
Column percentage reported.
P values were calculated using a 2-sided χ2 test to compare percentages between survivors and noncancer peers.
Statistical Analyses
We performed bivariate analyses to characterize outcome measures and covariates. Fisher exact tests or χ2 tests were used to compare percentages, and student t tests were used to compare means between survivors and noncancer peers, with P values reported. Next, we estimated multiple logistic regression models to identify adjusted differences between survivors and noncancer peers in the likelihood of any opioid prescription, indicator of potential misuse, and SUD. For each outcome, we ran 2 regression models: the first model adjusted for sociodemographic factors, and the second adjusted for sociodemographic and health status factors. All models also adjusted for calendar year of EOT to control for national secular trends that may confound the association between cancer experience and opioid-related behaviors. All analyses were limited to the first year posttherapy and were stratified by age at EOT (ie, children aged 11 years or younger, adolescents aged 12-17 years, or YAs aged 18-28 years). Among survivors, Wald tests and multiple logistic regressions were performed to identify the sociodemographic, health status, and cancer-related (ie, cancer type, treatment modality) factors that were associated with the outcomes. P values from Wald tests and regression models were calculated. The threshold of a P value less than .05 was used to determine statistical significance, and all statistical tests were 2-sided.
For ease of interpretation, we reported marginal effects (MEs), with 95% confidence intervals (CIs), calculated for survivors vs noncancer peers at the observed values of all other model covariates. The MEs reflect the adjusted percentage point difference in the likelihood that the outcome of interest occurs in survivors as compared with noncancer peers (35).
Results
Sample Characteristics
There were 3248 (36.2%) child survivors, 2559 (28.5%) adolescent survivors, and 3162 (35.3%) YA survivors at EOT. The most prevalent cancers were hematologic and central nervous system cancer in children (49.4% and 29.6%, respectively) and adolescents (45.6% and 27.0%, respectively); the most prevalent cancers among YAs were hematologic (47.7%) and gonadal cancer (22.1%) (Table 1). Compared with noncancer peers, survivors were more likely to experience painful conditions and mental health disorders (P < .001; Table 1).
Opioid Prescriptions
In bivariate analyses, a higher proportion of survivors than noncancer peers filled at least 1 opioid prescription during the first year posttherapy across all age groups: children (12.7% vs 2.0%; P < .001), adolescents (22.9% vs 7.7%; P < .001), and YAs (26.0% vs 11.9%; P < .001; Table 2). This finding persisted in multivariable regression models (Table 3 and Figure 1). For example, in models controlling for sociodemographic factors and health status, MEs suggested that the model-adjusted percentage of child, adolescent, and YA survivors filling at least 1 opioid prescription was 8.5 (95% CI = 7.3 to 9.6), 10.4 (95% CI = 8.8 to 12.0), and 9.3 (95% CI = 7.8 to 10.8) percentage points (ppt) higher, respectively, than noncancer peers (P < .001). These differences between survivors and noncancer peers represent a 404.8%, 126.8%, and 74.4% relative increase in the likelihood of filling at least 1 opioid prescription among children, adolescents, and YAs, respectively.
Table 2.
Bivariate comparisons in opioid prescription, potential misuse, or SUD between survivors and noncancer peers
| Outcomes | Aged 0-11 yearsa |
Aged 12-17 yearsa |
Aged 18-28 yearsa,b |
||||||
|---|---|---|---|---|---|---|---|---|---|
| Noncancer peers, %c | Survivors, %c | P | Noncancer peers, %c | Survivors, %c | P | Noncancer peers, %c | Survivors, %c | P | |
| (n = 16 240) | (n = 3248) | (n = 12 795) | (n = 2559) | (n = 15 810) | (n = 3162) | ||||
| Any (≥1) opioid prescription | 2.0 | 12.7 | <.001d | 7.7 | 22.9 | <.001d | 11.9 | 26.0 | <.001d |
| Any indicator of potential misuse | 0.04 | 1.1 | <.001d | 0.5 | 3.6 | <.001d | 1.1 | 5.8 | <.001d |
| Any SUD | 0.02 | 0.2 | <.001e | 0.9 | 1.1 | .26d | 3.3 | 4.9 | <.001d |
| Any OUD | 0.01 | 0.1 | .003e | 0.1 | 0.2 | .25d | 0.5 | 1.1 | <.001d |
| Any other SUD | 0.02 | 0.2 | .001e | 0.9 | 1.0 | .68e | 3.2 | 4.6 | <.001d |
| Any indicator of potential misuse or SUD | 0.1 | 1.4 | <.001d | 1.4 | 4.7 | <.001d | 4.3 | 9.4 | <.001d |
Age at the end of cancer therapy. OUD = opioid use disorder; SUD = substance use disorder.
There were 7 survivors aged 26-28 years at the end of cancer therapy; these survivors (diagnosed younger than age 22 years) remained on cancer treatment for a range of 4.2-6.4 years.
Column percentage reported.
P values were calculated using a 2-sided χ2 test to compare percentages between survivors and noncancer peers.
P values were calculated using a 2-sided Fisher exact test to compare percentages between survivors and noncancer peers.
Table 3.
Adjusted comparisons in opioid prescription, potential misuse, or SUD between survivors and noncancer peers
| Outcomes | Adjusted percentage point difference (95% CI) |
|||||
|---|---|---|---|---|---|---|
| Aged 0-11 yearsa |
Aged 12-17 yearsa |
Aged 18-28 yearsa |
||||
| (n = 3248 survivors, 16 240 noncancer peers) |
(n = 2559 survivors, 12 795 noncancer peers) |
(n = 3162 survivors, 15 810 noncancer peers) |
||||
| Adjusted for sociodemographicsb | Adjusted for sociodemographics and health statusc,d | Adjusted for sociodemographicsb | Adjusted for sociodemographics and health statusc | Adjusted for sociodemographicsb | Adjusted for sociodemographics and health statusc | |
| Any opioid prescription | ||||||
| Noncancer peers | Referent | Referent | Referent | Referent | Referent | Referent |
| Survivors | 10.7 (9.5 to 11.9) | 8.5 (7.3 to 9.6) | 15.2 (13.5 to 16.9) | 10.4 (8.8 to 12.0) | 14.1 (12.5 to 15.7) | 9.3 (7.8 to 10.8) |
| P e | <.001 | <.001 | <.001 | <.001 | <.001 | <.001 |
| Any indicator of potential misuse | ||||||
| Noncancer peers | Referent | Referent | Referent | Referent | Referent | Referent |
| Survivors | 1.1 (0.7 to 1.5) | 0.7 (0.4 to 1.0) | 3.1 (2.4 to 3.8) | 1.6 (1.1 to 2.1) | 4.6 (3.8 to 5.5) | 2.6 (2.0 to 3.2) |
| P e | <.001 | <.001 | <.001 | <.001 | <.001 | <.001 |
| Any SUDd | ||||||
| Noncancer peers | Referent | Referent | Referent | Referent | Referent | Referent |
| Survivors | 0.2 (0.04 to 0.4) | 0.1 (-0.03 to 0.3) | 0.2 (-0.2 to 0.7) | −0.10 (-0.5 to 0.3) | 1.6 (0.8 to 2.4) | −0.02 (-0.7 to 0.6) |
| P e | .01 | .11 | .31 | .55 | <.001 | .94 |
| Any indicator of potential misuse or SUD | ||||||
| Noncancer peers | Referent | Referent | Referent | Referent | Referent | Referent |
| Survivors | 1.3 (0.9 to 1.7) | 0.8 (0.5 to 1.1) | 3.2 (2.4 to 4.1) | 1.6 (1.0 to 2.2) | 5.1 (4.1 to 6.2) | 2.2 (1.4 to 3.1) |
| Pe | <.001 | <.001 | <.001 | <.001 | <.001 | <.001 |
Age at the end of therapy. CI = confidence interval; SUD = substance use disorder.
Regressions adjusted for sociodemographic factors and calendar year of the end of therapy.
Regressions adjusted for sociodemographic factors, health status, and calendar year of the end of therapy.
Among those aged 0-11 years, 367 (140 survivors, 227 noncancer peers) were dropped for the outcome “Any SUD” because of perfect prediction in fully adjusted model.
P values were calculated from logistic regression models and reflected 2-sided test of statistical significance.
Figure 1.
Adjusted percentage of any opioid prescription. The predicted percentages represented the model-adjusted likelihoods of specific outcomes among survivors vs noncancer peers, with other model covariates held at their observed values, estimated with logistic regressions. Partially adjusted models were regression models adjusting for sociodemographic factors and calendar year of the end of therapy. Fully adjusted models were regression models adjusting for sociodemographic factors, health status, and calendar year of the end of therapy. P values were calculated from logistic regression models and reflected 2-sided test of statistical significance.
Indicators of Potential Misuse
Bivariate analyses showed a higher proportion of survivors than noncancer peers had at least 1 indicator for potential misuse of prescription opioids across age groups (children: 1.1% vs 0.04%; adolescents: 3.6% vs 0.5%; YAs: 5.8% vs 1.1%; P < .001; Table 2; Supplementary Table 5, available online). These differences between survivors and noncancer peers in potential misuse persisted in multivariable regression models (P < .001; Table 3 and Figure 2). For example, in models that controlled for both sociodemographic factors and health status, the adjusted likelihood of any indicator for potential misuse among adolescent and YA survivors was 2.7 times higher (ME = 1.6 ppt, 95% CI = 1.1 to 2.1; P < .001) and 2.0 times higher (ME = 2.6 ppt, 95% CI = 2.0 to 3.2; P < .001), respectively, than among noncancer peers.
Figure 2.
Adjusted percentages of (a) any indicator of potential misuse, (b) any substance use disorder (SUD), and (c) any indicator of potential misuse or SUD. The predicted percentages represent the model-adjusted likelihoods of specific outcomes among survivors vs noncancer peers, with other model covariates held at their observed values, estimated with logistic regressions. Partially adjusted models were regression models adjusting for sociodemographic factors and calendar year of the end of therapy. Fully adjusted models were regression models adjusting for sociodemographic factors, health status, and calendar year of the end of therapy. P values were calculated from logistic regression models and reflected 2-sided test of statistical significance.
Substance Abuse Disorders
In bivariate analyses, the likelihood of experiencing a SUD during the first year posttherapy was higher in survivors than noncancer peers for children (0.2% vs 0.02%; P < .001) and YAs (4.9% vs 3.3%; P < .001) but not for adolescents (1.1% vs 0.9%; P = .26; Table 2). These findings persisted after adjusting for sociodemographic factors (Table 3). For example, YA survivors were 48.5% (ME = 1.6 ppt, 95% CI = 0.8 to 2.4; P < .001) more likely than noncancer peers to experience a SUD. After further controlling for health status, the difference between survivors and noncancer peers in SUD became nonstatistically significant for all ages.
Indicators of Potential Misuse or SUD
Bivariate analyses examining the prevalence of potential misuse or SUD showed that 1.4% of child, 4.7% of adolescent, and 9.4% of young adult survivors fulfilled at least 1 criterion; this rate was higher than noncancer peers (0.1%, 1.4%, and 4.3%, respectively; Table 2). These differences between survivors and noncancer peers persisted in all multivariable regression models (P < .001; Table 3).
Factors Associated With Opioid Prescription, Potential Misuse, and SUD Among Survivors
In multivariable analyses among survivors, a clear age gradient was observed in all outcomes, with YA survivors facing the highest likelihoods of a filled opioid prescription, potential misuse, and SUD diagnosis (Figures 1 and 2 and Table 4). Additionally, survivors with painful conditions were more likely to fill opioid prescriptions (ME = 14.1 ppt, 95% CI = 12.3 to 16.0; P < .001), experience potential misuse (ME = 5.3 ppt, 95% CI = 4.4 to 6.2; P < .001), and have SUD (ME = 1.7 ppt, 95% CI = 1.0 to 2.4; P < .001). Similarly, the presence of anxiety disorder was associated with higher likelihoods of opioid prescriptions (ME = 6.0 ppt, 95% CI = 2.1 to 10.0; P = .003), potential misuse (ME = 2.0 ppt, 95% CI = 0.3 to 3.7; P = .02), and SUD (ME = 1.6 ppt, 95% CI = 0.3 to 3.0; P = .02). Compared with receiving surgery only, receiving HSCT was associated with a higher likelihood of filling opioid prescriptions (ME = 6.2 ppt, 95% CI = 2.3 to 10.2; P = .002). Compared with hematologic cancer survivors, bone cancer survivors (ME = 3.0 ppt, 95% CI = 1.5 to 4.4; P < .001) and those with multiple cancer types (ME = 2.3 ppt, 95% CI = 0.4 to 4.3; P = .02) were more likely to experience potential misuse. Moreover, EOT year was an important predictor of opioid prescriptions and potential misuse, with persistent reductions in rates observed since 2016.
Table 4.
Factors associated with any opioid prescription, any indicator of potential misuse, and SUD among pediatric cancer survivors (n = 8969)
| Any opioid prescription |
Any indicator of potential misuse |
Any SUD |
||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Unadjusted percentage |
Adjusted percentage point difference |
Unadjusted percentage |
Adjusted percentage point difference |
Unadjusted percentage |
Adjusted percentage point difference |
|||||||
| Characteristics | %a | P b | Estimate | P d | %a | P b | Estimate | P d | %a | P b | Estimate | P d |
| (95% CI)c | (95% CI)c | (95% CI)c | ||||||||||
| Age at end of therapy, y | ||||||||||||
| 0-11 | 12.7 (Referent) | Referent | 1.1 (Referent) | Referent | 0.3 (Referent) | Referent | ||||||
| 12-17 | 22.9 | <.001 | 8.4 (6.5 to 10.3) | <.001 | 3.6 | <.001 | 1.7 (1.0 to 2.5) | <.001 | 1.1 | <.001 | 0.7 (0.3 to 1.2) | .001 |
| ≥18 | 26.0 | <.001 | 13.0 (11.0 to 15.0) | <.001 | 5.8 | <.001 | 4.3 (3.3 to 5.2) | <.001 | 4.9 | <.001 | 4.0 (3.3 to 4.7) | <.001 |
| Sex | ||||||||||||
| Male | 19.8 (Referent) | Referent | 3.5 (Referent) | Referent | 2.7 (Referent) | Referent | ||||||
| Female | 21.0 | .17 | 0.9 (−0.8 to 2.5) | .30 | 3.5 | .97 | −0.4 (−1.1 to 0.3) | .30 | 1.4 | <.001 | −1.3 (−1.9 to −0.7) | <.001 |
| Region | ||||||||||||
| Northeast | 14.8 (Referent) | Referent | 3.2 (Referent) | Referent | 2.2 (Referent) | Referent | ||||||
| North Central | 20.1 | <.001 | 5.5 (3.2 to 7.8) | <.001 | 3.1 | .93 | 0.1 (−1.0 to 1.1) | .92 | 2.7 | .33 | 0.6 (−0.3 to 1.5) | .19 |
| South | 24.1 | <.001 | 9.7 (7.6 to 11.8) | <.001 | 4.0 | .13 | 1.0 (−0.04 to 2.0) | .06 | 1.8 | .28 | −0.1 (−0.9 to 0.7) | .84 |
| West | 19.5 | <.001 | 4.6 (2.2 to 6.9) | <.001 | 3.4 | .67 | 0.3 (−0.8 to 1.4) | .65 | 2.0 | .54 | −0.05 (−0.9 to 0.8) | .92 |
| Rurality of residence | ||||||||||||
| Rural | 22.8 (Referent) | Referent | 3.8 (Referent) | Referent | 2.6 (Referent) | Referent | ||||||
| Urban | 19.9 | .03 | −1.4 (−3.9 to 1.0) | .26 | 3.4 | .54 | −0.03 (−1.1 to 1.1) | .96 | 2.1 | .33 | −0.4 (−1.4 to 0.5) | .37 |
| Health plan type | ||||||||||||
| Health maintenance organization alone | 21.6 (Referent) | Referent | 2.9 (Referent) | Referent | 1.8 (Referent) | Referent | ||||||
| High-deductible plan alone | 22.5 | .60 | 0.2 (−3.3 to 3.7) | .92 | 4.0 | .15 | −1.0 (−2.7 to 0.7) | .27 | 2.2 | .49 | −0.7 (−1.9 to 0.5) | .25 |
| Preferred provider organization alone | 19.9 | .22 | −3.7 (−6.3 to −1.0) | .01 | 3.6 | .18 | −0.8 (−2.2 to 0.5) | .24 | 2.2 | .32 | −0.3 (−1.3 to 0.8) | .61 |
| Other or multiple plan types | 19.1 | .11 | −3.7 (−6.8 to −0.6) | .02 | 3.1 | .78 | −1.3 (−2.8 to 0.2) | .09 | 2.1 | .51 | −0.2 (−1.4 to 1.0) | .71 |
| Painful conditions | ||||||||||||
| No | 15.1 (Referent) | Referent | 1.3 (Referent) | Referent | 1.4 (Referent) | Referent | ||||||
| Yes | 30.8 | <.001 | 14.1 (12.3 to 16.0) | <.001 | 7.8 | <.001 | 5.3 (4.4 to 6.2) | <.001 | 3.6 | <.001 | 1.7 (1.0 to 2.4) | <.001 |
| Mood disorder | ||||||||||||
| No | 19.6 (Referent) | Referent | 2.9 (Referent) | Referent | 1.6 (Referent) | Referent | ||||||
| Yes | 33.6 | <.001 | 3.5 (−0.3 to 7.3) | .08 | 13.6 | <.001 | 3.9 (1.9 to 5.9) | <.001 | 12.1 | <.001 | 4.2 (2.4 to 6.1) | <.001 |
| Anxiety disorder | ||||||||||||
| No | 19.6 (Referent) | Referent | 3.1 (Referent) | Referent | 1.8 (Referent) | Referent | ||||||
| Yes | 31.9 | <.001 | 6.0 (2.1 to 10.0) | .003 | 10.0 | <.001 | 2.0 (0.3 to 3.7) | .02 | 8.1 | <.001 | 1.6 (0.3 to 3.0) | .02 |
| Other mental health disorder | ||||||||||||
| No | 19.8 (Referent) | Referent | 3.3 (Referent) | Referent | 1.9 (Referent) | Referent | ||||||
| Yes | 25.9 | <.001 | 3.0 (−0.1 to 6.1) | .06 | 5.7 | .01 | 0.3 (−0.9 to 1.5) | .66 | 4.8 | <.001 | 1.0 (−0.2 to 2.2) | .09 |
| Site and type of cancer | ||||||||||||
| Hematologic | 21.9 (Referent) | Referent | 3.0 (Referent) | Referent | 1.9 (Referent) | Referent | ||||||
| Bone or connective tissue | 25.7 | .01 | 1.6 (−1.2 to 4.4) | .27 | 7.7 | <.001 | 3.0 (1.5 to 4.4) | <.001 | 2.6 | .18 | 0.3 (−0.7 to 1.3) | .57 |
| Central nervous system | 13.3 | <.001 | −7.6 (−9.7 to −5.4) | <.001 | 2.1 | .03 | −0.8 (−1.8 to 0.1) | .07 | 1.5 | .22 | −0.2 (−1.1 to 0.6) | .56 |
| Gonadal | 20.0 | .19 | −2.9 (−5.7 to −0.1) | .04 | 3.1 | .79 | −0.3 (−1.5 to 0.9) | .62 | 4.4 | <.001 | 0.4 (−0.5 to 1.4) | .36 |
| Multiple types of cancer | 26.2 | .03 | 4.2 (0.3 to 8.1) | .04 | 5.9 | .01 | 2.3 (0.4 to 4.3) | .02 | 1.6 | .58 | −0.1 (−1.6 to 1.3) | .84 |
| Cancer treatment | ||||||||||||
| Surgery only | 16.4 (Referent) | Referent | 3.0 (Referent) | Referent | 2.3 (Referent) | Referent | ||||||
| Any HSCT | 22.1 | .002 | 6.2 (2.3 to 10.2) | .002 | 4.0 | .24 | 1.9 (−0.2 to 3.9) | .07 | 1.5 | .16 | −0.4 (−1.8 to 1.0) | .55 |
| Any radiation therapy (none HSCT) | 22.4 | <.001 | 3.6 (1.2 to 6.0) | .003 | 4.8 | .002 | 1.2 (0.1 to 2.2) | .04 | 2.0 | .41 | −0.4 (−1.4 to 0.5) | .35 |
| Any chemotherapy (none radiation or HSCT) | 21.2 | <.001 | 3.4 (1.2 to 5.6) | .002 | 3.1 | .78 | 0.4 (−0.6 to 1.3) | .45 | 2.2 | .84 | −0.1 (−0.9 to 0.8) | .90 |
| Year of end of therapy | ||||||||||||
| 2009 | 23.1 (Referent) | Referent | 3.5 (Referent) | Referent | 2.3 (Referent) | Referent | ||||||
| 2010 | 21.4 | .37 | −2.7 (−6.4 to 0.9) | .14 | 2.9 | .43 | −0.7 (−2.5 to 1.0) | .40 | 1.8 | .41 | −0.3 (−1.7 to 1.0) | .62 |
| 2011 | 20.3 | .11 | −4.4 (−7.9 to −1.0) | .01 | 3.4 | .94 | −0.5 (−2.2 to 1.1) | .53 | 1.8 | .43 | −0.6 (−1.8 to 0.6) | .36 |
| 2012 | 19.4 | .04 | −5.9 (−9.5 to −2.4) | .001 | 4.3 | .37 | 0.1 (−1.7 to 1.8) | .95 | 1.7 | .33 | −0.8 (−2.1 to 0.4) | .18 |
| 2013 | 21.0 | .26 | −4.1 (−7.7 to −0.4) | .03 | 4.1 | .45 | 0.1 (−1.7 to 1.8) | .94 | 3.3 | .15 | 0.8 (−0.6 to 2.3) | .24 |
| 2014 | 22.0 | .58 | −4.3 (−8.2 to −0.4) | .03 | 4.4 | .32 | 0.03 (−1.9 to 1.9) | .97 | 2.9 | .43 | 0.3 (−1.2 to 1.7) | .73 |
| 2015 | 24.1 | .63 | −3.4 (−7.4 to 0.6) | .10 | 4.9 | .16 | 0.2 (−1.7 to 2.1) | .84 | 2.3 | .99 | −0.4 (−1.8 to 1.0) | .56 |
| 2016 | 20.0 | .13 | −7.7 (−11.6 to −3.8) | <.001 | 2.7 | .37 | −1.9 (−3.6 to −0.2) | .03 | 2.3 | .96 | −0.4 (−1.8 to 0.9) | .53 |
| 2017 | 15.8 | <.001 | −11.8 (−15.8 to −7.8) | <.001 | 1.9 | .05 | −2.7 (−4.4 to −1.1) | .001 | 1.7 | .41 | −1.0 (−2.3 to 0.4) | .16 |
| 2018 | 11.7 | <.001 | −15.5 (−19.1 to −11.8) | <.001 | 1.4 | .005 | −2.9 (−4.6 to −1.3) | <.001 | 1.4 | .17 | −1.1 (−2.5 to 0.2) | .10 |
Row percentage reported. CI = confidence interval; HSCT = hematopoietic stem cell transplantation; SUD = substance use disorder.
P values were calculated from Wald tests; all statistical tests were 2-sided.
Adjusted percentage point difference was based on results from multiple logistic regression models that controlled for all covariates in the table.
P values were calculated from logistic regression models and reflected 2-sided test of statistical significance.
Sensitivity Analyses
We reconstructed outcome measures excluding the initial 3 months after EOT, a period when survivors may experience persistent complications and/or residual pain from their treatment. Results from this analysis remained qualitatively similar in direction and significance, with slight decreases in magnitude (Supplementary Tables 6 and 7, available online). In another analysis examining opioid prescription rates quarterly, survivors’ rates were the highest immediately following the EOT and decreased gradually over time (Supplementary Figure 1, available online). For example, 13.1% of YA survivors filled an opioid prescription during the first quarter posttherapy (vs 3.7% among noncancer peers), followed by 9.6%, 8.2%, and 7.8% (vs 3.5%, 3.5%, and 3.3% among noncancer peers) during the subsequent quarters posttherapy. Notably, opioid prescriptions remained increased among survivors as compared with noncancer peers across the full year posttherapy (P < .001).
Discussion
This is the first assessment of early posttherapy opioid prescriptions, potential misuse, and SUD among pediatric cancer survivors. In this nationwide sample of privately insured individuals, pediatric cancer survivors were 74.4%-404.8% more likely to fill opioid prescriptions than noncancer peers across age groups. Importantly, the absolute percentage with an indicator for potential misuse and/or a SUD diagnosis was generally low, with 1.4% of child, 4.7% of adolescent, and 9.4% of YA survivors fulfilling at least 1 of these criteria; nevertheless, these percentages were statistically significantly higher among survivors than noncancer peers. Whereas the difference between survivors and noncancer peers in SUD was attenuated when adjusting for health status, differences in opioid prescription and potential misuse persisted in fully adjusted models. A clear age gradient was observed, with YA survivors exhibiting the highest likelihood of each of our outcomes.
There are several possible explanations for the higher rate of opioid prescriptions, potential misuse, and SUDs in survivors than noncancer peers. Survivors often experience persistent pain despite completion of therapy (5,6), resulting in potentially appropriate opioid prescriptions (9). In our sample, survivors were 1-4 times more likely than noncancer peers to have a painful condition (Table 1). Notably, survivors’ opioid prescription rate decreased over the first year posttherapy, indicating improvement in pain over time, although the prescription rate remained statistically significantly higher among survivors than noncancer peers (Supplementary Figure 1, available online). Our data and prior research have also revealed a higher prevalence of mental health conditions, such as anxiety and mood disorders, in survivors compared with the general population (12–14). Mental health disorders may contribute to continued opioid use and development of an SUD during cancer survivorship (36). Indeed, when our analyses adjusted for mental health comorbidities and painful conditions, the differences between survivors and noncancer peers reduced for all outcome measures, and the difference in SUD was fully mitigated.
However, even in fully adjusted models, the differences between survivors and noncancer peers in opioid prescription and potential misuse remained statistically significant, suggesting other factors may play a role. Some survivors may have dependency from use of opioids earlier in their disease trajectory requiring a continued prescription taper (37). Additionally, survivors may be more engaged than peers in the health-care system, where opioid prescriptions can be obtained. Importantly, increased access to opioids, combined with a lack of knowledge about safe use of opioids among young survivors and their caregivers (38,39), may raise the risk for misuse (10). Lastly, indicators for potential misuse may be related to the complex treatment regimens survivors received; some survivors may continue benzodiazepines for anxiety after therapy completion (40,41), raising the possibility of opioid and benzodiazepine overlap. Additional research is warranted to determine the causes of elevated rates of opioid prescribing and potential misuse among early survivors of pediatric cancer.
When comparing across age groups, we found that the rate of opioid prescribing in child survivors (12.7%) was considerably lower than in YA survivors (26.0%). The low rate of opioid prescribing in very young survivors may be suggestive of potential undertreatment of children’s chronic pain (42–44). Prior research has demonstrated several barriers to pain management for children, such as parents’ limited understanding of opioids and negative attitudes toward analgesic use (42,43). Additionally, unlike YAs, children may be too young to effectively express their symptoms (45), and health-care providers may have less knowledge about prescribing opioids to younger patients (46). More studies are needed to better understand the adequacy of pain management for young children surviving cancer.
Whereas potential misuse and SUD were rare (1.4%) in child survivors, these were more common in adolescent (4.7%) and YA (9.4%) survivors. This is consistent with previous self-reported data showing the potential for prescription opioid misuse in adolescent and YA survivors of cancer (13,47). This may be a result of the psychosocial effects of cancer experience in adolescents and YAs, including disruptions in social relationships and activities, fear of recurrence, and financial hardship (48,49). Cancer can also complicate the developmental trajectory experienced by adolescents and YAs, such as achieving independence, educational, career, and relationship transitions, which may add to their risk for aberrant drug-related behaviors (15,48,50).
These findings point to a clear need for efforts to prevent opioid misuse and SUDs, while maintaining adequate pain control, for cancer survivors. As survivors transition from active treatment to survivorship, providers involved in their care should monitor drug-related behaviors, assess risks of opioid misuse and/or abuse, and make referrals to addiction specialists as needed. Prior research has shown that pediatric oncology trainees have high levels of comfort with prescribing opioids and sympathy for cancer-related pain but have knowledge gaps in opioid dosing and conversion, which may raise the risk of overprescribing (46). Thus, it is crucial that future efforts in provider education include safe opioid prescribing. Importantly, as survivors see oncologists less frequently during survivorship, these efforts should also be made by primary care providers. Additionally, more efforts toward multidisciplinary care for adolescent and YA survivors are needed given the potential for mental health problems that may impact risk for opioid misuse and/or SUDs (42). Health-care providers should also consider safe pain management alternatives to opioids—such as acetaminophen, nonsteroidal anti-inflammatory drugs, gabapentinoids, serotonin-norepinephrine reuptake inhibitors or other antidepressants, physical therapy, psychotherapy, and acupuncture—to minimize risks for opioid misuse and addiction (51).
Our results should be interpreted with caution. First, multiple factors could lead to underestimated rates of opioid prescriptions, potential misuse, and SUDs. We had no data on self-paid or other insurer-paid prescriptions. SUDs are frequently underdiagnosed, and their true prevalence may be undercounted using claims data (52). Other indicators of potential misuse (eg, nonprescription use of opioids or other substances) could not be captured. However, there should not be differential bias among survivors and noncancer peers in our data. Additionally, we lack data on weight, and no weight-based misuse indicators exist; using the 100 MME per day cutoff may underestimate the percentage of young children receiving high doses of opioids. Furthermore, commercial claims data exclude those with Medicaid and the uninsured. Yet, research has shown that pain and symptom distress, opioid prescription, and SUD are more prevalent for individuals with public health insurance than those privately insured (53–56), suggesting that our estimates may be conservative among the entire population of pediatric cancer survivors.
Second, relying on diagnosis codes may not fully exclude controls with a cancer history, which would potentially underestimate the observed outcome differences between survivors and noncancer peers. Third, the MarketScan data are a convenience sample in which those insured through policies offered by large employers are overrepresented and, thereby, are not representative of the US population (57). However, the detailed and rigorously maintained claims in this database offer a unique opportunity to examine the real-world patterns of opioid prescribing in a large, geographically dispersed cohort of privately insured cancer survivors (57). Another limitation is that our study focused on opioid prescribing within 1 year posttherapy to minimize sample attrition because of discontinued enrollment over time. Because survivors’ pain experience immediately following therapy completion may not reflect their future trajectory of pain, future research should explore other data sources to understand longer-term use and misuse of prescription opioids among this population. Finally, we lacked data on the effect of race, ethnicity, stage at diagnosis, and cancer recurrence on opioid-related outcomes, an area that also merits future research.
Despite the limitations, this nationwide, claims data-based study serves as the initial step toward understanding opioid prescribing, misuse, and SUD during early pediatric cancer survivorship. Future studies should investigate the trajectories of opioid misuse and associated outcomes among long-term survivors of pediatric cancer, use of nonopioid pharmacologic and nonpharmacologic approaches for pain management, and barriers to accessing alternative pain treatments. These will be crucial steps to help pediatric cancer survivors receive adequate pain management and minimize the risk of opioid misuse to improve long-term outcomes.
Funding
This work was supported by a Junior Faculty Focused Award of Pediatric Research Alliance Pilot Grant Programs (XJ, ACM) and an Aflac Pilot Grant of Children’s Healthcare of Atlanta (XJ, KEB, ACM, KEE). The content is solely the responsibility of the authors and does not necessarily represent the official views of the Pediatric Research Alliance or Children’s Healthcare of Atlanta.
Notes
Role of the funders: The funders had no role in the design of the study; collection, analysis, and interpretation of the data; writing of the manuscript; and decision to submit the manuscript for publication.
Disclosures: The authors have no conflicts of interest to disclose.
Author contributions: XJ, KEE: Scientific design, visualization, and project administration. XH: Formal analysis. XJ: Writing—original draft. XJ, XH, KEB, ACM, JRC, KEE: Conceptualization, methodology, interpretation of results, and writing—review & editing. XJ, KEB, ACM, KEE: Funding acquisition.
Previous presentations: Abstract presentations at the 2021 American Society of Pediatric Hematology/Oncology (ASPHO) Conference and the AcademyHealth 2021 Annual Research Meeting.
Data Availability
Data for this analysis were made available to the authors through third-party license from IBM, a commercial data provider in the United States. As such, the authors cannot make these data publicly available because of a data use agreement. Other researchers can access the data by purchasing a license through IBM. Inclusion criteria specified in the Methods section would allow other researchers to identify the same cohort of patients we used for this analysis. Interested individuals may see https://marketscan.truvenhealth.com/marketscanportal/ for more information on accessing IBM MarketScan databases.
Supplementary Material
Contributor Information
Xu Ji, Department of Pediatrics, Emory University School of Medicine, Atlanta, GA, USA; Aflac Cancer & Blood Disorders Center, Children’s Healthcare of Atlanta, Atlanta, GA, USA.
Xin Hu, Department of Health Policy and Management, Emory University Rollins School of Public Health, Atlanta, GA, USA.
Katharine E Brock, Department of Pediatrics, Emory University School of Medicine, Atlanta, GA, USA; Aflac Cancer & Blood Disorders Center, Children’s Healthcare of Atlanta, Atlanta, GA, USA.
Ann C Mertens, Department of Pediatrics, Emory University School of Medicine, Atlanta, GA, USA; Aflac Cancer & Blood Disorders Center, Children’s Healthcare of Atlanta, Atlanta, GA, USA; Department of Epidemiology, Emory University Rollins School of Public Health, Atlanta, GA, USA.
Janet R Cummings, Department of Health Policy and Management, Emory University Rollins School of Public Health, Atlanta, GA, USA.
Karen E Effinger, Department of Pediatrics, Emory University School of Medicine, Atlanta, GA, USA; Aflac Cancer & Blood Disorders Center, Children’s Healthcare of Atlanta, Atlanta, GA, USA.
References
- 1. Robison LL, Hudson MM. Survivors of childhood and adolescent cancer: life-long risks and responsibilities. Nat Rev Cancer. 2014;14(1):61–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Howlader N, Noone AM, Krapcho M, et al. SEER Cancer Statistics Review, 1975-2016. National Cancer Institute Surveillance, Epidemiology, and End Results Program SEER 18 Regs Research Data, Nov 2018 Sub (1976-2016). National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch. https://seer.cancer.gov/archive/csr/1975_2016/. Published April 2019. Accessed January 10, 2022. [Google Scholar]
- 3. Oeffinger KC, Hudson MM. Long‐term complications following childhood and adolescent cancer: foundations for providing risk‐based health care for survivors. CA Cancer J Clin. 2004;54(4):208–236. [DOI] [PubMed] [Google Scholar]
- 4. Oeffinger KC, Mertens AC, Sklar CA, et al. ; for the Childhood Cancer Survivor Study. Chronic health conditions in adult survivors of childhood cancer. N Engl J Med. 2006;355(15):1572–1582. [DOI] [PubMed] [Google Scholar]
- 5. Alberts NM, Gagnon MM, Stinson JN. Chronic pain in survivors of childhood cancer: a developmental model of pain across the cancer trajectory. Pain. 2018;159(10):1916–1927. [DOI] [PubMed] [Google Scholar]
- 6. Schulte FS, Patton M, Alberts NM, et al. Pain in long‐term survivors of childhood cancer: a systematic review of the current state of knowledge and a call to action from the Children’s Oncology Group. Cancer. 2020;127(1):35–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Karlson CW, Alberts NM, Liu W, et al. Longitudinal pain and pain interference in long‐term survivors of childhood cancer: a report from the Childhood Cancer Survivor Study. Cancer. 2020;126(12):2915–2923. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Grundy R, Hulse RP. Chronic pain and childhood cancer survivorship. Curr Opin Physiol. 2019;11:58–61. [Google Scholar]
- 9.Substance Abuse and Mental Health Services Administration. Managing Chronic Pain in Adults With or in Recovery from Substance Use Disorders. Treatment Improvement Protocol (TIP) Series 54. HHS Publication No. (SMA) 12-4671; 2011. https://www.ncbi.nlm.nih.gov/books/NBK92048/pdf/Bookshelf_NBK92048.pdf. Accessed June 1, 2021. [PubMed]
- 10. Matson KL, Johnson PN, Tran V, Horton ER, Sterner-Allison J; for the Advocacy Committee on behalf of Pediatric Pharmacy Advocacy Group. Opioid use in children. J Pediatr Pharmacol Ther. 2019;24(1):72–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Miech R, Johnston L, O’Malley PM, Keyes KM, Heard K. Prescription opioids in adolescence and future opioid misuse. Pediatrics. 2015;136(5):e1169–e1177. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Ji X, Cummings J, Marchak J, Han X, Mertens A. Mental health among non-elderly adult cancer survivors–a national estimate. Cancer. 2020;126(16):3768–3776. doi: 10.1002/cncr.32988. [DOI] [PubMed] [Google Scholar]
- 13. Ji X, Cummings J, Mertens AC, Wen H, Effinger KE. Substance use, substance use disorders, and treatment in adolescent and young adult cancer survivors–results from a national survey. Cancer. 2021;127(17):3223–3231. doi: 10.1002/cncr.33634. [DOI] [PubMed] [Google Scholar]
- 14. Ji X, Marchak JG, Mertens AC, Curseen KA, Zarrabi AJ, Cummings JR. National estimates of mental health service use and unmet needs among adult cancer survivors. Cancer. 2021;127(22):4296–4305. doi: 10.1002/cncr.33827. [DOI] [PubMed] [Google Scholar]
- 15.Institute of Medicine. Identifying and Addressing the Needs of Adolescents and Young Adults with Cancer: Workshop Summary . Washington, DC: National Academies Press; 2013. https://www.nap.edu/download/18547. Accessed December 27, 2020. [PubMed] [Google Scholar]
- 16.Centers for Disease Control and Prevention. About CDC’s opioid prescribing guideline; 2021. https://www.cdc.gov/drugoverdose/prescribing/guideline.html. Accessed May 1, 2021. [Google Scholar]
- 17. Dowell D, Haegerich TM, Chou R. CDC guideline for prescribing opioids for chronic pain—United States, 2016. JAMA. 2016;315(15):1624–1645. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Centers for Disease Control and Prevention. MarketScan. commercial insurance claims at a glance; 2021. https://www.cdc.gov/visionhealth/vehss/data/claims/marketscan.html. Accessed December 27, 2019. [Google Scholar]
- 19.IBM Watson Health. IBM MarketScan research databases for life sciences researchers. https://www.ibm.com/us-en/marketplace/marketscan-research-databases. Accessed October 15, 2021.
- 20. Smitherman AB, Mohabir D, Wilkins TM, Blatt J, Nichols HB, Dusetzina SB. Early post-therapy prescription drug usage among childhood and adolescent cancer survivors. J Pediatr. 2018;195:161–168.e167. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Agency for Healthcare Research and Quality. Healthcare Cost and Utilization Project (HCUP). Clinical Classifications Software (CCS) for ICD-9-CM; 2017. https://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp. Accessed July 1, 2021.
- 22. Ward E, DeSantis C, Robbins A, Kohler B, Jemal A. Childhood and adolescent cancer statistics, 2014. CA Cancer J Clin. 2014;64(2):83–103. [DOI] [PubMed] [Google Scholar]
- 23.Agency for Healthcare Research and Quality. Summary data tables—condition categories. https://meps.ahrq.gov/data_stats/conditions04.shtml. Accessed December 5, 2021.
- 24.Agency for Healthcare Research and Quality. MEPS HC-110I: attachment 1. clinical classification code to ICD-9-CM code crosswalk. https://meps.ahrq.gov/data_stats/download_data/pufs/h110i/h110i_icd9codes.shtml. Access November 1, 2021.
- 25. Baxter NN, Goldwasser MA, Paszat LF, Saskin R, Urbach DR, Rabeneck L. Association of colonoscopy and death from colorectal cancer. Ann Intern Med. 2009;150(1):1–8. [DOI] [PubMed] [Google Scholar]
- 26. Cvancarova M, Samuelsen SO, Magelssen H, Fossa SD. Reproduction rates after cancer treatment: experience from the Norwegian radium hospital. J Clin Oncol. 2009;27(3):334–343. [DOI] [PubMed] [Google Scholar]
- 27.Centers for Disease Control and Prevention (CDC). Data resources. analyzing opioid prescription data and oral morphine milligram equivalents (MME); 2021. https://www.cdc.gov/opioids/data-resources/index.html. Accessed June 1, 2021.
- 28. Liu Y, Logan JE, Paulozzi LJ, Zhang K, Jones CM. Potential misuse and inappropriate prescription practices involving opioid analgesics. Am J Manag Care. 2013;19(8):648–658. [PubMed] [Google Scholar]
- 29.Chronic Conditions Data Warehouse (CCW). Other chronic health, mental health, and potentially disabling conditions. https://www2.ccwdata.org/web/guest/condition-categories. Accessed July 1, 2021.
- 30. Follman S, Arora VM, Lyttle C, Moore PQ, Pho MT. Naloxone prescriptions among commercially insured individuals at high risk of opioid overdose. JAMA Netw Open. 2019;2(5):e193209. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Khanna AD, Duca LM, Kay JD, Shore J, Kelly SL, Crume T. Prevalence of mental illness in adolescents and adults with congenital heart disease from the Colorado Congenital Heart Defect Surveillance System. Am J Cardiol. 2019;124(4):618–626. [DOI] [PubMed] [Google Scholar]
- 32. Ji X, Wilk AS, Druss BG, Lally C, Cummings JR. Discontinuity of Medicaid coverage: impact on cost and utilization among adult Medicaid beneficiaries with major depression. Med Care. 2017;55(8):735–743. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Majhail NS, Mau L-W, Denzen EM, Arneson TJ. Costs of autologous and allogeneic hematopoietic cell transplantation in the United States: a study using a large national private claims database. Bone Marrow Transplant. 2013;48(2):294–300. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Broder MS, Quock TP, Chang E, et al. The cost of hematopoietic stem-cell transplantation in the United States. Am Health Drug Benefits. 2017;10(7):366–374. [PMC free article] [PubMed] [Google Scholar]
- 35. Norton EC, Dowd BE, Maciejewski ML. Marginal effects—quantifying the effect of changes in risk factors in logistic regression models. JAMA. 2019;321(13):1304–1305. [DOI] [PubMed] [Google Scholar]
- 36. Sverdlichenko I, Hawke LD, Henderson J. Understanding the service needs of youth with opioid use: a descriptive study of demographics and co-occurring substance use and mental health concerns. J Subst Abuse Treat. 2022;132:108592. [DOI] [PubMed] [Google Scholar]
- 37. Sutradhar R, Lokku A, Barbera L. Cancer survivorship and opioid prescribing rates: a population‐based matched cohort study among individuals with and without a history of cancer. Cancer. 2017;123(21):4286–4293. [DOI] [PubMed] [Google Scholar]
- 38. Bruera E, Del EF. Pain management in the era of the opioid crisis. Am Soc Clin Oncol Educ Book. 2018;(38):807–812. [DOI] [PubMed] [Google Scholar]
- 39. Reddy A, de la Cruz M, Rodriguez EM, et al. Patterns of storage, use, and disposal of opioids among cancer outpatients. Oncologist. 2014;19(7):780–785. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Lohr L. Chemotherapy-induced nausea and vomiting. Cancer J. 2008;14(2):85–93. [DOI] [PubMed] [Google Scholar]
- 41. Adel N. Overview of chemotherapy-induced nausea and vomiting and evidence-based therapies. Am J Manag Care. 2017;23(suppl 14):S259–S265. [PubMed] [Google Scholar]
- 42. Fortier MA, Yang S, Phan MT, Tomaszewski DM, Jenkins BN, Kain ZN. Children’s cancer pain in a world of the opioid epidemic: challenges and opportunities. Pediatr Blood Cancer. 2020;67(4):e28124. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Fortier MA, Wahi A, Bruce C, Maurer EL, Stevenson R. Pain management at home in children with cancer: a daily diary study. Pediatr Blood Cancer. 2014;61(6):1029–1033. [DOI] [PubMed] [Google Scholar]
- 44. Hockenberry M. Symptom management research in children with cancer. J Pediatr Oncol Nurs. 2004;21(3):132–136. [DOI] [PubMed] [Google Scholar]
- 45. Mathews L. Pain in children: neglected, unaddressed and mismanaged. Indian J Palliat Care. 2011;17(suppl):S70–S73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. MacDonell‐Yilmaz RE, Murillo A, Panicker C, Welch JG. Evaluation of knowledge and comfort with opioid prescribing among pediatric hematology/oncology fellows. Pediatr Blood Cancer. 2021;68(3):e28786. [DOI] [PubMed] [Google Scholar]
- 47. Ehrentraut JH, Kern KD, Long SA, An AQ, Faughnan LG, Anghelescu DL. Opioid misuse behaviors in adolescents and young adults in a hematology/oncology setting. J Pediatr Psychol. 2014;39(10):1149–1160. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48. Park EM, Rosenstein DL. Depression in adolescents and young adults with cancer. Dialogues Clin Neurosci. 2015;17(2):171–180. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49. Daniel CL, Emmons KM, Fasciano K, Nevidjon B, Fuemmeler BF., Demark-Wahnefried W. Needs and lifestyle challenges of adolescents and young adults with cancer: summary of an Institute of Medicine and Livestrong Foundation Workshop. Clin J Oncol Nurs. 2015;19(6):675–681. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50. Schultz KAP, Ness KK, Whitton J, et al. Behavioral and social outcomes in adolescent survivors of childhood cancer: a report from the childhood cancer survivor study. J Clin Oncol. 2007;25(24):3649–3656. [DOI] [PubMed] [Google Scholar]
- 51. Volkow ND, Collins FS. The role of science in addressing the opioid crisis. N Engl J Med. 2017;377(4):391–394. [DOI] [PubMed] [Google Scholar]
- 52. O’Malley KJ, Cook KF, Price MD, Wildes KR, Hurdle JF, Ashton CM. Measuring diagnoses: ICD code accuracy. Health Serv Res. 2005;40(5 pt 2):1620–1639. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53. Mikosz CA, Zhang K, Haegerich T, et al. Indication-specific opioid prescribing for us patients with Medicaid or private insurance, 2017. JAMA Netw Open. 2020;3(5):e204514. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54. Orgera K, Tolbert J. The opioid epidemic and Medicaid’s role in facilitating access to treatment. San Francisco, CA: Kaiser Family Foundation; 2019. https://www.kff.org/medicaid/issue-brief/the-opioid-epidemic-and-medicaids-role-in-facilitating-access-to-treatment/. Accessed August 10, 2021. [Google Scholar]
- 55. Ilowite MF, Al‐Sayegh H, Ma C, et al. The relationship between household income and patient‐reported symptom distress and quality of life in children with advanced cancer: a report from the PediQUEST study. Cancer. 2018;124(19):3934–3941. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56. Lyon ME, Cheng YI, Needle J, et al. The intersectionality of gender and poverty on symptom suffering among adolescents with cancer. Pediatr Blood Cancer. 2021;68(8):e29144. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 57. Kulaylat AS, Schaefer EW, Messaris E, Hollenbeak CS. Truven health analytics MarketScan databases for clinical research in colon and rectal surgery. Clin Colon Rectal Surg. 2019;32(1):54–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data for this analysis were made available to the authors through third-party license from IBM, a commercial data provider in the United States. As such, the authors cannot make these data publicly available because of a data use agreement. Other researchers can access the data by purchasing a license through IBM. Inclusion criteria specified in the Methods section would allow other researchers to identify the same cohort of patients we used for this analysis. Interested individuals may see https://marketscan.truvenhealth.com/marketscanportal/ for more information on accessing IBM MarketScan databases.