Chronic Opioid use in Adult Survivors of Pediatric ICU Admissions in Nova Scotia: A Retrospective Cohort Study

Introduction

Opioids are a mainstay of sedation and analgesia management for children in pediatric intensive care units (PICUs). ¹ While necessary for comfort and safety, opioids may cause adverse outcomes such as withdrawal, delirium, tolerance, and dependence. ² Preliminary evidence in both pediatric and adult populations indicates that exposure to opioids is a risk for future use and abuse^3,4; high school students demonstrated a 33% increased risk of opioid abuse after prescription. ⁵ The relationship between opioid administration for sedation and analgesia in PICU and later opioid prescribing and abuse in adult survivors has not been explored.

Chronic opioid use was demonstrated up to 2 years after adult ICU admission in one population-based study ⁶ though this was not reproduced in another, smaller study. ⁷ However, early exposure of the developing brain to neuroactive agents may impact long-term outcomes and children may have a different response to opioids at critical periods of neurodevelopment. Therefore, we performed this exploratory, hypothesis-generating study to examine the relationship between PICU-related opioid exposure and chronic prescription opioid use in adult survivors of PICU admissions. We hypothesized that the prevalence of chronic opioid use in our study population in adulthood is similar to other adults in our region (6.2%, N = 2595). ⁷

Materials and Methods

Participant Identification and Data Collection

We identified potential participants through IWK's internally-maintained, electronic PICU database, which identifies patients by both health care and hospital chart numbers to allow linkage between admissions, and includes dates of admission, basic demographic data, presence of invasive ventilation, reason for admission, and mortality. We then examined the paper medical records of potential participants to confirm eligibility and, once confirmed, to collect PICU admission-related data including: demographics; major surgical procedure(s); reason for each admission; presence of chronic underlying medical conditions (eg congenital heart disease, asthma, cerebral palsy); number and length of PICU admission(s); total ventilated days in childhood; mean and maximum Therapeutic Intervention Scoring System (TISS) score (a validated measure of illness severity based on the intensity of ICU interventions)^9,10 for each admission; and opioids received for each PICU admission. Data that retained the provincial HCN but was otherwise de-identified was sent to NSPMP. The NSPMP representative removed the data file for potential participants who did not have an active NS HCN at any point from 2015–2019, inserted opioid prescriptions that had been filled for the study cohort, removed the HCN, and returned the anonymized data. See Figure 1 for description of data collection and linking flow. Potential participants for whom medical record or NSPMP records were not available were removed from the study.

OPEN IN VIEWER

Figure 1.

Data linking process.

Results

We identified 313 eligible participants through the PICU database based on inclusion and exclusion criteria, excluded 155 through further chart review, and lost 11 to follow-up (Figure 2). The NSPMP excluded 28 for no active Nova Scotia HCN from 2015–2019. We included 119 patients in the final cohort. Demographic and baseline data are presented in Table 1.

OPEN IN VIEWER

Figure 2.

Patient inclusion and flow.

OPEN IN VIEWER

Table 1.

Admission and Demographic Data for Cohort and by Chronic Prescription Opioid Status. (Note, cell reporting has been censored where the number of participants is <5).

	Cohort (N = 119)	No Chronic Rx Opioid N = 113	Chronic Rx Opioid N = 6	p
Demographic Data
Sex = Female, n (%)	49 (42%)		–	0.69
Received open heart surgery in childhood, n (%)	42 (35%)		–	1
Chronic Medical Condition (prior to first admission), n (%)	69 (58%)	64 (57%)	5 (83%)	0.40
Chronic medical condition in childhood, n (%)	76 (64%)	70 (62%)	6 (100%)	0.086
Age at first admission (months, median[IQR])	74.2 (3.9, 152.6)	73.2 (4.1, 152.6)	97.25 (2.0, 120.0)	0.91
Weight at first admission (kg, median[IQR})	20 (5.1, 50)	20 (5.1, 45)	24.5 (5.5, 54.8)	0.84
Admission Data
# of PICU admissions in childhood (median[IQR])	1 (1, 2)	1 (1, 2)	3 (1, 6)	0.040
# of planned admissions in childhood (median[IQR])	0 (0, 1)	0 (0, 1)	1.5 (0, 4)	0.16
# PICU admissions associated with surgery (median[IQR])	1 (0, 2)	1 (0, 1)	1.5 (0, 3)	0.27
Cumulative length of ICU admission in childhood (days, median[IQR]))	8 (5, 15)	8 (5, 15)	12 (5, 14)	0.66
# of PICU admissions requiring mechanical ventilation in childhood (median[IQR])	1 (1, 2)	1 (1, 1)	1 (1, 2)	0.56
Cumulative length of mechanical ventilation in childhood (days, median[IQR])	5 (3, 10)	6 (3, 10)	4 (4, 5)	0.39
Mean TISS score (mean [SD])	35.58 (7.51)	35.91 (7.43)	29.34 (6.62)	0.036
Max TISS score (mean[SD])	51.07 (10.31)	51.48 (10.17)	43.33 (10.73)	0.059

*TISS = Therapeutic Intervention Scoring System, IQR = Interquartile Range, Rx = Prescription.

Six participants (5%) met criteria for chronic opioid use as an adult. This was similar (Fisher's Exact probability 0.84) to the prevalence of chronic opioid use previously reported in a local population of adults prior to ICU admission. ⁷ Five participants (83%) who developed chronic opioid use in adulthood had chronic medical conditions in childhood (eg, cancer, congenital heart disease). Participants with chronic prescription opioid use had a median (IQR) of 3(1-6) PICU admissions in childhood, significantly more than those who did not have chronic prescription opioid use (1[1-2], [p = 0.04]), though they had a lower mean (SD) TISS than those without chronic use (29.3[6.6] vs 35.9[7.4] respectively, p = 0.036).

The range of PICU-based opioid exposures during childhood in the entire cohort was 0.04–100.6 mg/kg morphine equivalents. There was no difference in median (IQR) cumulative PICU opioid exposure between those with (8.4 [3.2-11.4] mg/kg morphine equivalents) and without (7.3 [2.8-17.0] mg/kg morphine equivalents) chronic prescription opioid use (p = 0.49) (Table 1). We did not identify a correlation between total PICU morphine equivalents (mg/kg) and either percent of adult days receiving opioids (Spearman's rho 0.003, p = 1.0) or morphine milligram equivalents prescribed per day as an adult (Spearman's rho 0.01, p = 0.9).

Discussion

In this exploratory study, the prevalence of chronic opioid use in adults who survived childhood PICU admission was 5% and similar to that in Nova Scotians prior to their adult ICU admission (6.2%). ⁷ There was no demonstrable relationship between the amount of opioid received during PICU admissions and opioid prescription requirement as an adult. This suggests that opioid exposure during a PICU admission may not be a risk factor for chronic opioid use in adults.

While extensive literature addresses neurodevelopmental, psychological, and chronic disease burden of neonatal and pediatric critical care-related opioid exposure, we are not aware of longitudinal data linking childhood ICU exposure to either adult or chronic opioid use. Prescription in children and youth during healthcare encounters, including during cancer therapy ¹¹ and surgery, ¹² has been associated with subsequent opioid use and, in some cohorts, longer term use into adulhood.^4,13 However, these relationships are likely confounded by underlying illness and chronic pain.

We noted that participants with multiple PICU admissions, lower nursing care requirements, and those with chronic medical conditions were statistically more likely to have chronic opioid use as an adult. One might hypothesize that multiple admissions to PICU with recurrent childhood exposure to opioids places these children at higher risk for opioid use in adulthood than would be seen with higher doses of opioids associated with a single, high severity of illness admission. We also noted non-significant trends to increased opioid use in adult survivors who had more frequent planned PICU admissions, underwent more surgeries, and had pre-existing chronic medical conditions in childhood. Chronic disease starting in childhood and requiring PICU admission or co-morbidities acquired because of PICU admission may continue into adulthood, increasing risk for mental illness and pain. This, rather than the PICU admissions themselves, may account for chronic opioid use in adulthood, particularly given clear associations between pain, mental illness, and chronic disease.^4,13,14

For similar reasons, it should be noted that the comparison group – adult patients prior to ICU admission – likely over-represents chronic opioid use compared with the general population. Adult ICU admission rates are higher for individuals who receive opioids; a Dutch study reported 1-year adult ICU admission rates of 5.9 per 1000 person-years in individuals who received any opioid prescription in the preceding 3 months, versus 2.2 per 1000 person-years for those who had not, though individuals who had chronic opioid use had lower admission rates than those who received a single prescription. ¹⁵ The population-based prevalence in our region has not been reported so, while our results are comparable to adult studies in Nova Scotia that used similar methodology, ⁷ it remains possible that the prevalence in our sample is higher than for the general population. Additionally, it may be important that a significant portion of the PICU population have severe neurologic impairment that limit communication and mobility ¹⁶ and may preclude manifestation of opioid dependence.

While this study represents the novel use of data linkage between two robust databases that have been maintained for several decades, there are several limitations that restrict generalizability. Our historic cohort size was small with significant loss to follow-up, primarily from migration or health record loss. Repeating the study in 10–15 years would provide a significantly larger number of adult participants, though studies using this methodology will not allow attribution of adult prescription opioid use to pediatric exposure. As a single-centered study, practice variations related to opioid management may have impacted opioid sensitization. Change in PICU populations and practice over eras may limit generalizability to current practice, as pre-PICU liberation opioid prescribing practices were likely more liberal. ¹⁷ It is also likely that our findings under-estimate the true prevalence of chronic opioid use in adults as we could only examine prescription, rather than non-prescribed, opioid use.

Conclusions

In this hypothesis generating, exploratory study, pediatric critical illness necessitating PICU admission and opioid prescription showed no correlation with chronic opioid prescription use in adulthood in our cohort. Future exploration, particularly multicentered, of the relationship between chronic opioid use and repeated PICU admission, planned PICU admissions, and chronic medical conditions may be warranted.

Supplemental Material

Supplemental Material