Association Between Opioids Prescribed to Medical Inpatients With Pain and Long-Term Opioid Use

Kaitlin E. Keenan, MD; Michael B. Rothberg, MD, MPH; Shoshana J. Herzig, MD, MPH; Simon Lam, PharmD, BCPS; Vicente Velez, MD; Kathryn A. Martinez, PhD, MPH


South Med J. 2021;114(10):623-629. 

In This Article


Of 18,949 admissions initially meeting inclusion criteria, 13,152 (69%) reported a pain score > 0. Of these, 10,181 (77%) were opioid experienced before admission, 77% of whom received opioids during their hospitalization. Our analytic sample consisted of the 2971 patients who reported pain and were opioid naïve at the time of admission, 64% of whom received opioids during their hospitalization.

Opioid use During Hospitalization

Table 1 presents sample characteristics of opioid-naïve patients with pain stratified by opioid receipt during hospitalization. More than half of the patients were older than 65 years (63%), 72% of patients were White, 86% were nonsmokers, 38% had a history of chronic pain, and 27% had a psychiatric history. Most (85%) had a single admission during the study period. The average LOS was 6.8 days (standard deviation 4.1). Of a maximum of 10, the mean maximum pain score was 7.4 (standard deviation 2.4). Thirty-six percent received opioids during the hospitalization but not at discharge, 28% received them at discharge, and 36% did not receive opioids.

Opioid receipt during the hospitalization was associated with younger age (P < 0.001), White race (P < 0.001), smoking (P < 0.001), longer LOS (P < 0.001), maximum pain score (P < 0.001), history of chronic pain (P < 0.001), history of psychiatric diagnosis (P < 0.001), and having only 1 admission during the study period (P < 0.001). Opioid receipt at discharge was associated similarly with the factors above, with the exception of history of psychiatric diagnosis, which was not significantly associated with opioids at discharge.

Table 2 presents the multivariable logistic regression models of odds of opioid receipt during hospitalization and at discharge. Opioid receipt during the hospitalization was associated with chronic pain history (adjusted odds ratio [aOR] 1.37, 95% confidence interval [CI] 1.14–1.65), maximum pain score (aOR per point increase 1.42, 95% CI 1.37–1.48), LOS (aOR/day 1.11, 95% CI 1.08–1.14), and smoking (aOR 1.34, 95% CI 1.02–1.76). Patients were significantly less likely to receive opioids if they were Black as opposed to White (aOR 0.54, 95% CI 0.43–0.69), 65 years and older versus 18 to 50 years (aOR 0.54, 95% CI 0.40–0.73), if they had a psychiatric diagnosis (aOR 0.65, 95% CI 0.54–0.79), or if they had multiple admissions during the study period (aOR 0.69, 95% CI 0.54–0.87). There was no difference by sex. Opioid receipt at discharge was associated with maximum pain score (aOR per point increase 1.28, 95% CI 1.24–1.33), chronic pain history (aOR 1.42, 95% CI 1.13–1.76), and > 1 admission during the study period (aOR 0.62, 95% CI 0.45–0.84). There were no other differences in odds of opioid receipt at discharge by patient factors.

Long-term Opioid use

Table 3 presents bivariate comparisons between sample characteristics and long-term use. Overall, 3% of patients received prescriptions for opioids between 6 and 12 months. Two percent of patients who did not receive opioids at all during their index hospitalization, 3% of patients who received them during the hospitalization but not at discharge, and 5% of patients who received opioids at discharge had met our criteria for long-term use (P < 0.001). The only other factor significantly associated with long-term use in the bivariate analysis was the last pain score recorded during the index hospitalization (P = 0.024).

Table 4 presents the adjusted mixed-effects logistic regression model of odds of long-term opioid use both including and excluding last recorded pain score. Excluding the last patient-reported pain score, opioid receipt during the hospitalization alone (ie, stopped on discharge) was not associated with long-term use (aOR 1.57, 95% CI 0.89–2.77), but being discharged with an opioid was (aOR 2.61, 95% CI 1.49–4.55). Inclusion of the last recorded pain score attenuated these effects. No other factors were significantly associated with long-term use.

Indication for Opioid use

In the review of 100 patient charts, only 13 contained a physician note documenting the reason for an opioid, most commonly back pain (3) and chronic pancreatitis (2). Eight other charts contained an indication in the nursing documentation—headache (2), pain (5), and chronic pancreatitis (1). Another 54 noted that an opioid was given without specifying a reason. Twenty-five charts contained no documentation regarding opioids.