A Risk Education Program Decreases Leftover Prescription Opioid Retention: An RCT

Abstract

Introduction:

Retaining leftover prescription opioids poses the risks of diversion, misuse, overdose, and death for youth and other family members. This study examined whether a new educational program would enhance risk perceptions and disposal intentions among parents and decrease their retention of leftover prescription opioids.

Study Design:

This study is an RCT (NCT03287622).

Setting/Participants:

A total of 648 parents whose children were prescribed opioid analgesics were recruited from a Midwestern, academic pediatric hospital between 2017 and 2019. Parents were randomized to receive routine information (control) with or without Scenario-Tailored Opioid Messaging Program intervention.

Intervention:

The intervention provided opioid risk and mitigation advice using interactive decisional feedback.

Main Outcome Measures:

The main outcome measures were parents’ perceptions of the riskiness of keeping/sharing opioids and child misuse measured at baseline, Days 3 and 14, their intention to dispose of leftover opioids, and their final retention decisions after the child’s use (at or around Day 14).

Results:

Perceived riskiness of child misuse and keeping/sharing opioids increased from baseline through Day 14 only for parents in the intervention group (p≤0.006). However, there were no significant differences in risk perceptions between groups and no intervention effect on disposal intentions at either follow-up. Despite these findings, the intervention reduced the likelihood of parents’ opioid retention when adjusted for important parent and child covariates (AOR=0.48; 95% CI=0.25, 0.93; p=0.028). Parents who reported past opioid misuse also showed higher retention behavior (AOR=4.78; 95% CI=2.05, 11.10; p<0.001).

Conclusions:

A scenario-specific educational intervention emphasizing the potential risks that leftover opioids pose to children and that provided risk mitigation advice decreased parents’ retention of their child’s leftover opioid medication. Removing leftover prescription drugs from homes with children may be an important step to reducing diversion, accidental poisoning, and misuse among youth.

Trial Registration:

This study is registered at www.clinicaltrials.gov NCT03287622.

INTRODUCTION

The presence of leftover prescription opioids in the home has contributed to widespread misuse, diversion, and poisonings among children, adolescents, and young adults in the U.S.^1,2 Indeed, a large majority of teens who have reported prescription opioid misuse obtained the drug from their own past prescription or from that of a family member or friend.^3–5 Three of 4 teen-aged children report unsupervised access to risky medications,⁵ and most unintentional pediatric opioid poisonings occur at home from exposure to readily available prescriptions.¹ Easy access is particularly concerning because a majority of prescribed opioid doses are left over after acute pain treatment,^6–10 and many adults admit to keeping their child’s or their own leftover opioids for later use.^11,12 Of concern, very few parents with opioids in the home report their storage in locked, inaccessible cabinets, leaving many children vulnerable to accidental or intentional ingestion.^13,14

Given the risks posed to children, families, and communities, primary prevention strategies that encourage prompt and safe removal of risky leftover medications from homes are needed. Community drug disposal programs such as take-back events as well as the provision of disposal kits and disposal information have led to improved disposal rates.^7,15,16 Yet, many people continue to keep leftover opioids and other controlled medications. Retention of high-risk medications may be, in part, related to the perception that these medications pose little risk to family members. Therefore, an educational intervention (i.e., Scenario-Tailored Opioid Messaging Program [STOMP]) was designed to enhance the perceived riskiness of leftover opioid retention and to encourage prompt disposal. The intervention was grounded in the tenets of Fuzzy Trace Theory, which supports the notion that people rely on gist representations (i.e., essential or bottom-line meaning) when making decisions–even when more precise or verbatim risk data (e.g., numerical rates) are provided.^17,18 Gist-based reasoning has been associated with better judgment, risk avoidance, and risk reduction behavior than details-based reasoning.¹⁷ Methods that emphasize bottom-line or gist meaning have been found to improve risk understanding.¹⁷ The intent of gist-based messaging is not meant to exaggerate risks. However, fuzzy trace posits that when the absolute rates of risk are low, such as in the case of opioid-related misuse, overdose, and death, verbatim or details-based reasoning can inadvertently promote risky behavior.

The goal of this randomized, controlled clinical trial was to examine the impact of the new educational intervention on the proximal outcomes, risk perception and cognitive intentions, and subsequent opioid retention. The aims were to

1. examine whether the STOMP intervention changes participants’ perceived riskiness of prescription opioid retention and misuse from preintervention to Days 3 and 14;

2. determine whether the STOMP intervention results in greater-risk perceptions or larger changes in risk perceptions after surgery than routine prescription instruction alone (i.e., controls);

3. examine whether the STOMP intervention, compared with routine instruction alone, impacts parents’ intentions to dispose of their child’s leftover prescription opioids on Days 3 and 14; and

4. examine whether the STOMP intervention reduces parents’ retention of their child’s leftover opioids after short-term use.

METHODS

The STOMP intervention uses scenario-specific feedback designed to alter risk perceptions and promote corrective action.¹⁹ The brief interactive program incorporates a series of clinically relevant pain and risk scenarios for which participants are asked to make intentional, analgesic use and handling decisions. Once the participant makes their decision, they are given immediate feedback that (1) uses categorically framed gist statements to explain the risks involved and (2) provides behavioral advice about the necessary steps to reduce risk. Similar use of feedback strategies has been shown to enhance health behaviors and outcomes compared with provider interactions alone.²⁰ The scenario-tailored messages were meant to reveal the bottom-line gist that “keeping leftover opioids poses the risk of harm to children.” The average time to completion of all scenarios was 8 minutes.

The focus of this study was a scenario where parents were asked to imagine that their child had recovered from surgery, had minimal pain, and was back to routine behavior. They were then asked to indicate what they would do with their child’s leftover prescription pain reliever in this situation: for example, Keep it in the medicine cabinet until we need it again…Put it in a locked cupboard…Throw it in the garbage…Take it to a police station…Regardless of their decision, all parents then received feedback describing the potential risk that leftover opioids pose to children and teens–that is, intentional and unintentional misuse, poisoning, and death. In addition to the risk messages, the intervention provided concrete advice about how to remove or mitigate the risk (e.g., getting rid of leftovers is the best way to reduce risk…here’s how to safely dispose of left-over opioids...).

Parents in the control group received only routine medication instruction. Throughout the time of this study, such education consisted of a review by a recovery room nurse and a standard, computer-generated, postoperative discharge form that, in addition to postoperative care instructions, included the child’s prescription details (e.g., hydrocodone 5 mg [1 tablet] every 6 hours as needed for pain). A link to a website describing how to dispose of leftover medications was provided at the end of the multipage form. Information about the risks that leftover medications pose to children was not part of the written discharge information at the time of this study.

Study Sample

This prospective, RCT was approved by the IRB at the University of Michigan and was registered with ClinicalTrials.gov hosted by the U.S. National Library of Medicine (IRBMED HUM00127009; NCT03287622). From October 2017 through May 2019, participants were recruited from preoperative clinics and waiting rooms at a large, Midwestern, tertiary care pediatric setting in the U.S. Parents of all children aged 5–17 years who were scheduled to undergo ambulatory or short-stay elective surgery and expected to receive a prescription opioid were approached. Written parental consent and child assent (where applicable) were obtained before inclusion.

All parents used a tablet computer to complete electronic surveys that assessed their risk perceptions at baseline–before the child’s surgery, prescription opioid dispensing, and discharge medication instruction. On the basis of computer-generated randomization (1:1 allocation), parents were preassigned to receive routine medication education (described earlier) with or without the STOMP intervention (administered by tablet computer). Parents’ risk perceptions were reassessed 3 and 14 days after the child’s hospital discharge by electronic survey sent through e-mail or text message links. These follow-up times were chosen to provide short- (i.e., early in the child’s postoperative recovery when analgesic use would be ongoing) and longer-term assessments (i. e., when the child’s pain has resolved and analgesic use has ended). In this manner, the potential impact of the intervention on the proximal outcomes, risk perceptions, and disposal intentions could be examined while controlling for other important factors (e.g., pain). A Day-14 assessment of final disposal outcomes was chosen on the basis of clinical experience that nearly all children complete their course of prescribed opioids within a week to 10 days after ambulatory surgery. It was expected that parents could make a final decision about retention by Day 14. Surveys contained no identifiable information but included unique identifiers that parents entered at the start of each survey. In this manner, surveys could be linked while maintaining parents’ and their children’s privacy. Parents received $15 for completion of each follow-up survey.

Measures

Prescription opioid misuse risk perceptions were measured at baseline and on Days 3 and 14 using a survey adapted from previous studies on prescription opioid storage perceptions.¹³ Survey items assessed the degree to which parents agreed or disagreed with risky behavior statements (e.g., It would be okay for me to share my own prescribed opioid with another family member; It would be okay for my child to take an opioid analgesic prescribed to a sibling; Dangerous things can happen when people keep leftover prescribed opioids in their home). An exploratory principal component factor analysis of the survey yielded 3 risk factors that explained 65% of the variance in overall risk perceptions (Kaiser criterion=0.6; p<0.001). These risk perception factors were labeled on the basis of commonality among factor items: (1) keeping/sharing opioids (among adults), (2) child misuse (using/sharing without supervision), and (3) access concern (the degree of concern/worry about child access to leftover medications). Possible scores for each of these factor scales ranged from ‒4 (low perceived riskiness/concern) to +4 (high perceived risk/concern). These scales were treated as continuous.

Prescription opioid retention decisions were assessed using survey items such as those used to ascertain medication disposal behavior in similar studies.^13,21 First, researchers assessed parents’ intentions to dispose at baseline and Days 3 and 14 using our hypothetical scenario describing the child’s resolving pain and opioid discontinuation. Parents were asked to indicate whether, under the described scenario, they would dispose of the leftover prescribed opioid. Decisions were collapsed into a dichotomous variable to indicate plan to keep (e.g., Keep it in the medicine cabinet until we need it again…Put it in a locked cupboard…) versus intend to dispose (e.g., Throw it in the garbage…Take it to a police station…). Second, after the child had completed their use of analgesics (before or around Day 14), we assessed actual opioid retention by asking parents, What have you done or will you do with your child’s leftover opioid medication? Parent responses were coded as the dichotomous outcome as disposed or will dispose versus will retain. Parents also submitted through e-mail or text message an image of their disposal process to confirm their response.

Similar to previous studies, past prescription opioid use was assessed by asking parents at baseline the question, Please indicate whether you, your child or another family member have used the following analgesics…This was followed by a list of generic and brand-name medications, including prescribed opioid pain relievers (e.g., Vicodin, Norco, hydrocodone, Percocet, hydromor-phone, codeine, Dilaudid, fentanyl). Parents’ use of any of these opioids was coded as a dichotomized variable, that is, yes, previous use of any prescribed opioid versus no use.

Prescription opioid kept in the home was assessed at baseline by asking parents to indicate Which of the following analgesics are currently available in your home…Parents used a drop-down menu to indicate whether they had any of the following drugs in the home: Vicodin, Norco, hydrocodone, Percocet, hydromor-phone, fentanyl. A dichotomous variable (yes versus no) was created to indicate the presence/absence of any prescribed opioid.

Parents also indicated whether they had acetaminophen, ibuprofen, or naproxen in the home. These items yielded the dichotomized variable yes (in the home) versus no (not in the home).

Similar to items used in national surveys,^4,22 parents were asked, Have you ever, even once taken a prescribed opioid in larger doses or more frequently than advised, Have you ever even once taken an opioid that was prescribed for someone else, and Have you ever shared their own prescribed opioid with a family member or friend. Responses were coded into a dichotomous variable to indicate past misuse versus no misuse. Parents’ motivations were assessed using dropdown and free text responses.

Pain interference (i.e., the degree to which pain interfered with the child’s functioning) was assessed with the Parents’ Postoperative Pain Measure-Short Form²³ on Days 3 and 14. This survey asks parents to indicate which behaviors the child has exhibited since surgery. Specifically, parents respond yes or no to items, including Cried more than usual, whined more than usual, Groaned or moaned more than usual, Play less than usual, Want to be close to you more than usual, Hold the sore part of his/her body, Act more worried, Act more quiet. This survey has excellent internal consistency (Cronbach’s α=0.85)²³ and agreement with self-reported pain intensity in children.²⁴ Scores range from 0 to 10, with 10 indicating high interference with functioning and 0 indicating no pain interference, and the scale is treated as continuous.

The Health Literacy-Brief tool assessed parents’ health literacy at baseline.²⁵ The 4-item instrument asks parents to indicate how often (never to always) they need help reading, understanding, or completing medical forms. The measure yields a continuous score from 0 to 16, with lower scores indicating low literacy (scores were flipped from the original version).

Demographics of the parent (i.e., sex; race/ethnicity; and education categorized as less than high school diploma, high school or equivalent diploma, some college or trade school, 4-year college degree, higher than 4-year degree) and child (i.e., sex, age) were also collected at baseline. Race/ethnicity was coded as White versus non-White, and education was coded as high school diploma or lower versus higher than high school diploma for the group comparisons and modeling.

Statistical Analysis

All data were analyzed using Stata (Release 14; StataCorp LP, College Station, TX). No missing data were imputed; however, partial data were included in the repeated measures-mixed models, controlling for the random effect of subject. Descriptive analyses were used to depict the sample characteristics and to examine the distribution of data. Univariate comparisons between study groups were made with chi-square (e.g., sex) or unpaired t-tests (e.g., health literacy). Data are presented as n (%), mean ± SD, or median (IQR) where appropriate. Beta coefficients (β), ORs or their adjusted values (e.g., AOR), and 95% CIs are reported. Significance was accepted at p<0.05. It was determined, a priori, that a sample of 650 (325 per group) would be more than sufficient to detect a significant difference in opioid retention rates between groups (expected rates of 30% vs 20%; sample needed 293 per group; β=0.20; α=0.05) while allowing for an expected 10% attrition rate.

The authors used generalized linear mixed models with repeated measures per subject (assuming constant variance between repeated measures within subject) to test the first hypothesis, controlling for covariates. Models estimated the association between risk perception factors and the fixed factors of group (STOMP versus control), time (Days 3 and 14 versus Day 0, assessed separately for each group), parent characteristics (sex, education level, race/ethnicity, health literacy, past misuse, and opioid storage behavior), and child sex and age. All models accounted for the random effect of the subject. Coefficient estimates were compared with determine significant differences between groups on Days 3 and 14.

The authors used a mixed-effect logistic regression model (with random effect of subject) to test the effect of STOMP on the outcome and disposal intentions on Days 3 and 14, controlling for the fixed effects of risk perceptions and other covariates (sex, child age, race/ethnicity, education, health literacy, past prescription opioid use, opioid at home, past prescription opioid misuse, pain interference).

Finally, a binary logistic regression tested the hypothesized effect of STOMP on parents’ final retention decision, adjusted for the effects of parent and child covariates.

RESULTS

The study recruitment and completion diagram are provided in Figure 1. A total of 648 parents were randomized, and 94% of these completed at least 1 follow-up assessment, and 89% completed both assessments. Attrition was significantly higher overall and for non-White participants in the STOMP group (p≤0.001). Table 1 describes the baseline characteristics of the parents, which were similar between groups except for higher past opioid use among STOMP parents. The children of the parents underwent similar procedures (61% orthopedic in the control vs 59% in the STOMP group; p=not significant) with similar prescribing patterns.

Figure 1.

CONSORT diagram depicting study enrollment and analytical sample.

*p<0.01 vs control.

LTFU, loss to follow-up; STOMP, Scenario-Tailored Opioid Messaging Program.

Table 1.

Baseline Characteristics of the Study Groups

Characteristics	Control (n=323)	STOMP (n=325)
Parent demographics
Female sex	254 (78.6)	273 (83.7)
Education
High school diploma or lower	32 (9.9)	42 (12.9)
Associate’s degree, trade, or some college	119 (36.9)	136 (41.8)
4-year bachelor’s degree or higher	172 (53.2)	147 (45.2)
Race/ethnicity
White/non-Hispanic	272 (84.2)	276 (84.9)
Black	22 (6.8)	26 (8)
Hispanic	11 (3.4)	11 (3.4)
Asian	7 (2.2)	7 (2.1)
Other	5 (<2.0)	5 (<2.0)
Health literacy	12.08±1.17	12.06±1.28
Past behaviors
Treated pain in past	305 (94.4)	303 (93.2)
Past prescription opioid use	147 (45.5)	179 (55.1)
Over-the-counter analgesic in home	294 (91.0)	287 (88.3)
Prescription opioid in home	70 (21.7)	63 (19.4)
Past prescription opioid misuse	37 (11.5)	37 (11.4)
Took higher dose	16 (5.0)	20 (6.2)
Shared with others	21 (6.5)	14 (4.3)
Took others’ opioid	23 (7.1)	21 (6.5)

Note: Data are presented as n (%) or mean ± SD.

STOMP, Scenario-Tailored Opioid Messaging Program.

Repeated measures ANOVA supported the hypothesized increase in perceived riskiness of child misuse of and keeping/sharing of opioids from baseline through Day 14 for the STOMP group (F=7.3, p=0.001; and F=5.0, p=0.001, respectively) but not for the control group (F=2.67, p=0.07; and F=0.53, p=0.590). Mixed effect regression models supported the increase in the perceived risk of keeping/sharing opioids and child misuse of opioids for STOMP through Day 14, controlling for important covariates (Table 2). These models also show how both groups became less worried over time about their child’s access to leftover medication.

Table 2.

Factors Associated With Parents’ Risk Perceptions Over Time

Factors	Child misuse	Keep/share opioid	Access concern
Parent female (versus male)	0.42 (0.25, 0.58), <0.001	0.04 (−0.13, 0.21), 0.627	−0.28 (−0.48, −0.07), 0.010
Parent White (versus non-White)	0.20 (0.02, 0.38), 0.028	0.16 (−0.02, 0.34), 0.089	−0.04 (−0.27, 0.19), 0.741
Education	0.09 (0.04, 0.14), 0.011	0.07 (0.02, 0.11), 0.006	0.17 (0.11, 0.23), <0.001
Health literacy	0.10 (0.04, 0.15), <0.001	0.06 (0.0041, 0.11), 0.036	−0.11 (−0.17, −0.04), 0.001
Past prescription opioid misuse	−0.67 (−0.89, −0.46), <0.001	−0.50 (−0.72, −0.29), <0.001	0.33 (0.06, 0.60), 0.016
Past prescription opioid use	0.04 (−0.10, 0.18), 0.552	−0.06 (−0.20, 0.08), 0.395	−0.17 (−0.35, 0.001), 0.051
Prescription opioid in home	−0.46 (−0.62, −0.30), <0.001	−0.39 (−0.55, −0.23), <0.001	0.08 (−0.12, 0.28), 0.446
Child female (versus male)	−0.02 (−0.15, 0.11), 0.763	−0.02 (−0.15, 0.12), 0.822	−0.27 (−0.44, −0.11), 0.001
Child age	−0.01 (−0.03, 0.01), 0.314	−0.01 (−0.03, 0.01), 0.183	0.07 (0.04, 0.09), <0.001
STOMP (versus Control)	−0.20 (−0.41, 0.02), 0.068	−0.15 (−0.37, 0.06), 0.160	−0.14 (−0.40, 0.13), 0.315
Day 3 X controla	0.25 (0.04, 0.46), <0.023	0.10 (−0.12, 0.31), 0.371	−0.55 (−0.82, −0.28), <0.001
Day 14 X controla	0.14 (−0.08, 0.35), 0.219	0.11 (−0.11, 0.32), 0.334	−0.74 (−1.01, −0.46), <0.001
Day 3 X STOMPa	0.37 (0.16, 0.59), <0.001	0.34 (0.13, 0.56), 0.002	−0.63 (−0.90, −0.36), <0.001
Day 14 X STOMPa	0.37 (0.15, 0.59), 0.001	0.28 (0.06, 0.50), 0.012	−0.73 (−1.00, −0.45), <0.001
Model statistics	F=12.52 (df14,1799) p<0.001	F=6.71 (df 14,1,799); p<0.001	F=12.72 (df 14,1,798) p<0.001

Note: Boldface indicates statistical significance (p<0.05).

Shown are results of the generalized linear mixed models accounting for the repeated measures of risk perceptions within the subject; data are presented as β (95% CI); p<0.001.

^aReference=baseline assessment.

STOMP, Scenario-Tailored Opioid Messaging Program.

There were no significant between-group differences in any of the risk perceptions on Days 3 and 14 (all post hoc comparisons p>0.05). Univariate comparisons of the resultant coefficients from Table 2 showed no significant differences between groups for changes in perceived risk of child misuse of opioids on Days 3 and 14 (i.e., chi-square [df1]=0.99, p=0.320; and chi-square [df1]=2.62, p=0.106, respectively), in perceived risk of keeping/sharing opioids (chi-square [df1]=3.70, p=0.055; chi-square [df1]=1.52, p=0.217), or access worry (chi-square [df1]=0.30, p=0.586; chi-square [df1]=0.00, p=0.955).

Most parents in the STOMP and control groups reported a baseline intention to dispose of leftovers (73.5% and 67.3%, respectively). This baseline difference was significant as shown in Appendix Table 1 (available online) (AOR=2.29; 95% CI=1.15, 4.56; p=0.018) when adjusted for fixed and random effects of other factors. However, the model revealed no significant effect of STOMP on Days 3 or 14. Parent risk perceptions and their past opioid misuse behavior were associated with decreased intentions to dispose of leftover opioids, whereas female sex of the child was associated with increased disposal intention.

At the final assessment, 50% fewer parents in the STOMP group than in the controls retained their child’s leftover opioid medication (6.2% vs 12.1%; OR=0.47; 95% CI=0.26, 0.88; p<0.018). Parents’ motivations for retention included potential future child need (n=43 [86%]), family need (n=7 [14%]), drug paid for (n=7 [14%]), and inconvenience or lack of knowledge regarding disposal (n=6 [12%]). Table 3 depicts the results of the logistic regression, which supported the hypothesis that STOMP decreased the odds of parents’ retention of leftover opioids (OR=0.48; 95% CI=0.25, 0.93; p=0.028), controlling for parent and child covariates.

Table 3.

Effect of STOMP on Parents’ Leftover Prescription Opioid Disposal After Child’s Short-Term Use

Factors	AOR	95% CI	p-value
STOMP (versus control)	0.48	0.25, 0.93	0.028
Parent female (versus male)	0.99	0.46, 2.10	0.970
Parent White/non-Hispanic (versus others)	0.96	0.38, 2.45	0.939
Education level	0.86	0.69, 1.09	0.211
Health literacy	1.02	0.79, 1.31	0.895
Past prescription opioid use (versus no)	0.82	0.39, 1.73	0.601
Prescription opioid in home (versus no)	1.91	0.97, 3.75	0.060
Past prescription opioid misuse (versus no)	4.78	2.05, 11.10	<0.001
Child female (versus male)	0.62	0.32, 1.18	0.146
Child age	1.05	0.96, 1.15	0.276
Child pain interference	1.01	0.85, 1.21	0.870

Note: Boldface indicates statistical significance (p<0.05).

Logistic regression model Wald chi-square (df11)=31.51; p<0.001; pseudo R²=0.095; p<0.001.

STOMP, Scenario-Tailored Opioid Messaging Program.

DISCUSSION

This study showed that an educational intervention providing bottom-line risk messages with an advice component can motivate risk mitigation action in the context of prescription opioid disposal decisions. Despite no between-group differences in risk perceptions, significantly fewer parents in the STOMP group retained their children’s leftover opioid medication in the final assessment. These findings suggest, perhaps, a scaffolding effect of the gist-based risk messages and behavioral advice component of STOMP (i.e., how to dispose).

Previous studies have described how heightened risk perceptions are fundamental toward reducing risky decision making.^26–29 However, higher risk perceptions may be insufficient to motivate preventive behavior, particularly if personal control over the threat is in doubt.³⁰ Interventions that combine risk or threat information with skill-building messages have been found to be most effective toward changing perceptions and behavioral intentions around diet and smoking behaviors.^31,32 Findings in this study similarly suggest that combining a gist-based approach with risk and action information may enhance the perceived threat and the sense of control over the threat posed by risky leftover medications.

In this study, 1 in 10 parents, overall, reported final retention of their children’s leftover prescription opioids after what was intended to be short-term use for acute pain. Although parents’ primary motivation for retention appeared to be therapeutic intent (i.e., future pain relief needs of the child and family), this behavior suggests a potential lack of perceived family vulnerability to the risks posed by retention and misuse. Indeed, lower perceptions of personal susceptibility have been shown to play a central role in risky health behavior.³³

Notably, parents who reported past misuse (i.e., sharing their own or borrowing a friend’s prescription opioid) were 70% less likely to indicate disposal intentions and more than twice as likely to retain their child’s leftover medication. Parents who previously misused opioids and who retained their child’s leftovers may perceive themselves and family members to be invulnerable to bad outcomes related to drug misuse. Previous studies suggest that personal risk-taking behavior often reduces perceptions about the riskiness of that behavior, particularly when the outcomes were beneficial or at least not harmful.^34,35 Parents who misuse a prescribed opioid and experience pain relief without noticeable ill effect may thereafter develop lowered risk perceptions. Indeed, these data show that parents’ past misuse was associated with lower-risk perceptions as well as subsequent drug retention. Risk perceptions are thus not only precursors to behavior but are also formed or influenced by behavior.^36,37 Although past prescription opioid use was not associated with opioid retention, this survey item included family use as well as personal use, limiting interpretation.

It is important to note that the rate of leftover prescription opioid retention in this sample of parents (both groups) was just under 10% compared with that of previous reports suggesting rates as high as 30%–60% among parents and adult patients in U.S. samples.^9,13,38 In addition, baseline disposal intentions were high in both study groups. Widespread media attention to the opioid crisis across the U.S. and Michigan may have increased the awareness and importance of prescription opioid risks. The increased attention to opioid use disorder and overdose may be particularly salient to parents who want to protect their children. Notably, in both groups, parents’ baseline risk perceptions were relatively high, whereas their worry about children’s access was low–even though 1 in 5 reported storing a prescription opioid in the home at the time of the survey. Low concern about access to these medications conveys a common parental underestimation of children’s vulnerability to household hazards. Earlier studies showed that parents are well aware of household hazards such as poisons or guns, but perceive their own child’s vulnerability to such hazards to be low.³⁹

Importantly, several parent and child characteristics influenced parents’ risk perceptions and decision making. For instance, higher health literacy was associated with higher risk perceptions, which has important implications for ensuring comprehensible patient education. This finding should be interpreted with caution and should be confirmed using a more comprehensive objective measure of health literacy. Also noteworthy is that parents’ concern over child access was largely influenced by their children’s sex and age, with greater concern for boys and older children. This finding suggests that parents may underestimate the ability of young children and girls to improperly access stored medications. Despite data suggesting variable poisoning rate differences between genders and age groups, it remains important that all parents be made aware of the potential risks posed by leftover medications.

Limitations

The findings from this study are strengthened by the randomized, controlled nature of this study, the large sample size, and low loss to follow-up. However, it is important to note that results may have been influenced by the repeated measure of risk and disposal intentions and the baseline differences between groups. It is possible that repeated assessments led to a satisficing bias among parents, thus reducing group differences. In addition, clinical conversations between parents and practitioners could have influenced the perceptions and behaviors of participants in unknown ways. The ability to generalize findings beyond the study setting is limited by the sample characteristics, including a majority of White, relatively well-educated, female parents recruited from a single institution. It is important to note that the effect of STOMP on decisions may have been overestimated or underestimated given the higher attrition rate of non-White participants in the intervention group, despite controlling for these characteristics in the hypotheses tests. Finally, this sample was drawn from 1 large, tertiary care pediatric hospital in the Midwest where awareness of prescription opioids risk may be relatively high. Further study of risk communication interventions in a more diverse, national sample is warranted to determine the broader efficacy.

CONCLUSIONS

In this RCT, a scenario-tailored educational intervention was associated with parents’ decreased retention of their child’s leftover opioids despite nonsignificant differences in risk perceptions and intentions from the controls. Providing gist-based risk messages combined with information about risk mitigation actions in a one-time educational session at the time of prescribing may therefore be a cost-effective way to motivate medication disposal. Given data suggesting up to 50% opioid retention rates,^9,40,41 intervening to reduce risky medication retention remains an important endeavor. Finally, adaptation of this or similar educational interventions to address the risks of other controlled substances (e.g., stimulants, anxiolytics) may have the potential to reduce inadvertent exposures among children who are prescribed these medications.