Abstract
Objective
Tapering of chronic opioids has increased, with subsequent reports of exacerbated pain among patients who tapered. We aimed to evaluate the association between opioid dose tapering and subsequent pain-related healthcare utilization (emergency department [ED] visits, hospitalizations and primary care visits).
Design, Setting, and Subjects
We conducted a retrospective cohort study from years 2015–2019 using data from the Optum Labs Data Warehouse that contains de-identified retrospective administrative claims data for commercial and Medicare Advantage enrollees in the United States. Adults aged ≥18 years who were prescribed stable doses of opioids, ≥50 morphine milligram equivalents (MME)/day, during a 12-month baseline period.
Methods
Tapering was defined as ≥15% relative reduction in mean daily opioid dose during one of 6 overlapping 60-day periods. Tapered patient-periods were subclassified as tapered-and-continued (MME > 0) vs tapered-and-discontinued (MME = 0). We modeled monthly counts of visits for pain diagnoses up to 12 months after cohort entry using negative binomial regression as a function of tapering, baseline utilization, and patient level-covariates.
Results
Among 47 033 patients, 13 793 patients tapered. Compared to no taper, any taper was associated with more ED visits for pain (adjusted incidence rate ratio [aIRR] 1.21, 95% confidence interval [CI]: 1.11–1.30), tapered then continued status was associated with more ED visits (aIRR 1.23, CI: 1.14–1.32) and hospitalizations (aIRR 1.14, CI: 1.03–1.27) f-or pain, and tapered-and-discontinued was associated with fewer primary care visits for pain (aIRR 0.68, CI: 0.61–0.76).
Conclusions
These associations suggest that opioid tapering may lead to increased emergency and hospital utilization for acute pain and possibly a decreased perceived need for primary care for those whose opioids were discontinued.
Keywords: tapering, opioid, primary care, emergency visits, healthcare utilization
Introduction
Since 1999, US drug overdose death rates have quintupled.1 In response to the rising drug overdose deaths, regulatory and policy responses have emphasized restricting the provision of prescription opioids, including both initial and ongoing opioid prescriptions for chronic pain. The 2016 Centers for Disease Control and Prevention (CDC) opioid prescribing guideline identified daily morphine milligram equivalent (MME) thresholds associated with higher overdose risk,2 which triggered an increase in opioid deprescribing, or tapering, among patients prescribed long-term opioid therapy (LTOT).3 More recent guidelines underscore weighing the risks and benefits of opioid dose continuation with the risks and benefits in terms of pain relief and functional improvement.4–6
Opioid tapering has led concerns that tapering may confer risks of use of illicit opioids, overdose, precipitated withdrawal, depression, anxiety, suicide, and worsened comorbid chronic condition care.7–12 Worsening pain after taper has been described as a risk of tapering,6 whereas the few empiric studies on non-cancer chronic pain control suggest that pain can be as well controlled with non-opioids and similar or better pain scale ratings has been observed in small samples of tapering patients within structured, multidisciplinary pain programs.13–16 As these and other studies elucidate concerns for both increased pain after tapering opioids and overuse of opioids when non-opioids would be sufficient for pain control, a greater understanding of pain and healthcare for pain after opioid tapering is needed.
If opioid tapering is associated with pain crises, this increased, uncontrolled pain might be seen with increased utilization of emergency and hospital services for pain. In a previous study examining all-cause health care utilization, we found an increase in all-cause emergency department (ED) visits after taper, many of which were unexplained by overdose or mental health crises and could be caused by newly uncontrolled pain.12 At the same time, if opioid tapering disrupts or removes pain care from primary care relationships, tapering might be associated with fewer primary care visits for pain.17 To expand our understanding of how opioid tapering may specifically impact pain-related health care utilization, we conducted a retrospective cohort study among a large sample of patients prescribed LTOT in the United States. We hypothesized that tapering would be associated with increased rates of pain-related emergency visits and pain-related hospitalizations with reduced rates of pain-related primary care visits.
Methods
Data source
This study used de-identified administrative claims data from the Optum Labs Data Warehouse® (OLDW) from years 2015–2019. The database contains longitudinal health information on enrollees and patients, representing a diverse mixture of ages and geographic regions across the United States. The claims data in OLDW includes medical and pharmacy claims, laboratory results and enrollment records for commercial and Medicare Advantage enrollees.18 The de-identified data was used in compliance with the Health Insurance Portability and Accountability Act Privacy Rule, and the study did not require IRB review.
Design and participants
The study had a retrospective cohort design with patient-periods sampled based on stable opioid dosing during each month of a 12-month baseline period occurring from July 1, 2015, to December 31, 2018. Patients were adults (age ≥ 18 years on the first day of follow-up) who were prescribed opioids with a mean dose of ≥50 morphine milliequivalents (MME)/day for each 30-day period of the prior 12 months and a stable dose across the period (based on monthly dose fluctuations of less than 10% compared to the mean dose across the entire year).10,12 We restricted the study period to time after the transition to International Classification of Diseases, 10th revision (ICD-10). Patients were required to have at least 14 months of continuous health plan enrollment to allow for 12 months of baseline data and at least 2 months of follow-up.10 Using claims from the baseline year, we excluded patients with buprenorphine prescriptions (any formulation, as a potential marker of opioid use disorder, as we could not differentiate prescriptions for pain vs opioid use disorder), a non-skin cancer diagnosis (as cancer pain is often treated differently from other chronic pain), any use of hospice or palliative care, or prolonged skilled nursing care (≥90 days). After the 12-month baseline period, patient data were assessed for at least 2 months and up to 12 months for outcomes (Figure 1). During follow-up, patients were censored if they died, disenrolled from the health plan, or met one of the exclusion criteria above, including transitioning to buprenorphine. Patients could contribute more than one person-period if they met eligibility criteria after 12 months of follow-up ended in a prior person-period. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.
Figure 1.
Timeline for baseline, taper ascertainment, and follow up periods with example of patient who undergoes opioid tapering in Period 4. All patients had at least 12-months of stable opioid dosing. After the 12-month baseline, they entered the cohort and were assessed for opioid dose tapering during 6 overlapping periods of 60 days. If the average daily MME in a follow up period was 15% of the baseline period’s MME, the patient was considered as tapered at the end of that period. Among patients who tapered, their opioid dose was calculated for each follow up period to determine if they continued on any opioid dose (MME > 0) or if they discontinued (MME = 0). In this example, the patient had a dose reduction to 15% of their baseline dose during Period # 4 and thus was defined to be tapered at the end of Period 4. During the remaining follow up periods, their dose was assessed for continuation or discontinuation. Abbreviation: MME = morphine milligram equivalent.
Specification of tapering
As in prior studies,3,10,12 we defined an opioid dose taper as a ≥ 15% relative reduction in the mean daily opioid dose compared to baseline dosing during any one of the first 6 overlapping 60-day periods in the follow up period. This definition was selected as it incorporates gradual tapers (∼2.5% per month over six months), as recommended by guidelines,2,19 but also more rapid dose reductions. Tapering status during follow-up was defined as a time-varying binary variable. Patients were classified as tapered in all subsequent study months after the 60-day period when tapering was identified as these patients all experienced an opioid taper.
To examine potentially distinct effects of opioid dose reduction vs complete discontinuation of opioids on pain outcomes, follow-up periods from tapered patients were subclassified based on opioid dose during the follow up period in which tapering was identified. If a tapered patient’s opioid dose was non-zero during a 60-day follow-up period, the subsequent period was classified as “tapered-and-continued.” If a tapered patient’s opioid dose dropped to zero during any 60-day period, all subsequent periods were classified as “tapered-and-discontinued.” Patients who had dose increases post-index date were censored to compare between dose tapered and dose-maintained populations.
Outcomes
Outcomes included: (1) pain-related ED visits that did not result in hospitalization, (2) pain-related hospitalizations (that may have been preceded by an ED visit), and (3) pain-related primary care visits (defined as outpatient encounters with family medicine and general internal medicine clinicians where clinicians were physicians, advance-practice nurses and physician associates).20 For each outcome, we identified counts of events during each 30-day period of up to 12 months of follow-up using pain diagnosis codes as described below to classify encounters as “pain-related.” We also created monthly counts of non-pain related visits, excluding all visits for the listed pain diagnoses.
As there is no standard definition of a pain-related visit, we created a list of pain diagnoses by using the Healthcare Cost and Utilization Project Clinical Classification Software Refined (CCSR)21 to identify diagnosis categories that included pain codes for chronic conditions likely to be treated with LTOT (eg, musculoskeletal pain or regional pain syndromes) but that excluded pain codes for acute injuries or other emergent conditions (eg, fracture, infection, post-operative injuries). We required these codes to be in the primary position for ED visits and hospitalizations and in any position for primary care visits. We also excluded diagnoses that might be due to emergent or non-chronic pain conditions, such as unspecific abdominal pain or headache. Because our initial approach did not allow inclusion of neck-related diagnosis codes (due to inclusion of neck infection within the same CCSR category), we used a list of neck pain ICD-10 diagnosis codes developed for a separate cohort study using the OLDW data.22 We also included non-specific chronic pain codes but did not include as pain visits pain-related codes such as “long-term use of opioid analgesic.” The final pain diagnosis last overlaps substantially with pain diagnoses used in other studies.23 Supplement A lists pain-related diagnosis codes.
Covariates
We included as covariates sociodemographic, clinical and healthcare utilization variables in the baseline year. The sociodemographic variables included: Age (categorical), education (based on median adult household education level in patient’s U.S. residential Census block; missingness was approximately 6% and its own category), rurality (dichotomized as metropolitan/micropolitan vs small town/rural [using Rural-Urban Commuting Area codes 1–6 vs 7–10]; missingness was ∼0.2% and grouped with small town/rural, the smaller category), and insurance (commercial or Medicare Advantage). We did not have a patient-reported race/ethnicity variable in the data set.
We also included clinical variables that might contribute to health care use or status in patients prescribed LTOT: Comorbidities (27 indicator variables for non-cancer conditions in the Elixhauser comorbidity index),24 ED, hospital and outpatient visits for depression/anxiety or suicidality, benzodiazepine co-prescription on the date of cohort entry, presence of ≥1 overdose events in the 90 days preceding cohort entry, and LTOT dose (average daily MME during the baseline year; categorized as 50–89, 90–149, 150–299, and ≥300 MME). To reflect baseline propensity to use emergency, hospital, and primary care services, we computed the mean number encounters during the baseline year in the following six categories: Pain-related ED visits, ED visits unrelated to pain, pain-related hospitalizations, hospitalizations unrelated to pain, pain-related primary care visits and primary care visits unrelated to pain. Lastly, we included the year of cohort entry year to account for secular factors.
Analyses
We conducted our analyses using Stata MP, v.15.1 (StataCorp). We performed descriptive analyses to characterize the entire cohort and by whether patients never tapered, tapered-and-continued, or tapered-and-discontinued at the end of each 60-day follow up period. We used negative binomial regression to model monthly counts of each of the three types of pain-related visits as a function of tapering and the covariates above, accounting for varying person-period lengths by using the natural log of the duration as an offset. These models estimated adjusted incidence rate ratios (aIRRs) of pain-related encounters among patients in the tapered vs non-tapered state during follow-up. We used post-estimation commands to compute adjusted rate differences associated with tapering for each outcome. To elucidate findings from initial models, we then fit analogous models with the tapering variable operationalized as a three-level categorical variable: Non-tapered, tapered-and-continued LTOT, and tapered-and-discontinued LTOT.
In all analyses, we set the significance level (α) to 0.05 and used robust standard errors to account for clustering of person-periods (the units of analysis) within patients and potential slight departures from modeling assumptions.
Results
The overall cohort had 47 033 patients who were observed for tapering and utilization outcomes events after 67 784 baseline periods of stable opioid dosing. The mean number of person-periods per patient was 1.4 (median 1.0). Across person-periods, the mean follow-up was 11.1 months (SD 2.3). After the 67 784 baseline periods of stable opioid dosing, tapering occurred during follow-up in 14 923 (22.0%) patient-periods, including 11 690 periods (17.2%) when patients tapered-and-continued opioids and 3233 periods (4.8%) when patients tapered-and-discontinued opioids (Figure 2).
Figure 2.
Flowchart showing patient cohorts for patients prescribed stable opioid therapy ≥50 morphine milliequivalents (MME)/day. Patient cohort was derived from a chronic opioid cohort defined for prior studies and then limited to baseline years such that the index date for cohort entry was on or after 10/1/2016 to include only ICD-10 diagnosis codes for pain-visits. After cohort entry, patients were assessed for taper or no taper, and among patients who tapered, they were assessed for taper then dose continuation (MME>0) or dose discontinuation (MME=0). Patients were censored from the cohort if their dose increased after index date or if they met cohort exclusion criteria.
As shown in Table 1, during their most recent person-period, patients had a mean age of 60.8 (SD = 10.9), were 54% female, with 8.7% residing in a small-town or rural area and 77.9% with Medicare Advantage (vs 22.1% with commercial insurance). As compared to the overall cohort, patients who tapered during the most recent observation period were prescribed higher baseline opioid doses, were more likely to be co-prescribed a benzodiazepine, and had a higher frequency of any drug overdose during the baseline period.
Table 1.
Baseline characteristics of patients prescribed stable LTOT ≥ 50 MME/day for at least 12 months prior to cohort entry.
| Baseline patient characteristicsa | Overall cohort | Patients who tapered and continued LTOT during observation | Patients who tapered and discontinued LTOT during observation | Patients who did not taper during observation |
|---|---|---|---|---|
| 47 033 patients | 10 579 patients | 3214 patients | 33 240 patients | |
| Age in years, n (%) | ||||
| 18≤35 | 541 (1.1) | 133 (1.3) | 55 (1.7) | 353 (1.1) |
| 35≤50 | 6195 (13.2) | 1490 (14.1) | 533 (16.6) | 4172 (12.6) |
| 50≤65 | 23 302 (49.5) | 5540 (52.4) | 1634 (50.8) | 16 128 (48.5) |
| ≥65 | 16 995 (36.1) | 3416 (32.3) | 992 (30.9) | 12 587 (37.9) |
| Age, average in years, SD | 60.8 (10.9) | 60.0 (10.7) | 59.3 (11.3) | 61.3 (10.9) |
| Sex, n (%) | ||||
| Male | 21 656 (46.0) | 4 579 (43.3) | 1642 (51.1) | 15 435 (46.4) |
| Female | 25 377 (54.0) | 6000 (56.7) | 1572 (48.9) | 17 805 (53.6) |
| Education b , n (%) | ||||
| HS or less | 19 434 (41.3) | 4207 (39.8) | 1355 (42.2) | 13 872 (41.7) |
| >HS | 22 722 (48.3) | 5178 (49.0) | 1478 (46.0) | 16 066 (48.3) |
| Unknown/missing | 4877 (10.4) | 1194 (11.3) | 381 (11.8) | 3302 (9.9) |
| Rurality c , n (%) | ||||
| Metro+micro | 42 917 (91.3) | 9670 (91.4) | 2890 (89.9) | 30 357 (91.3) |
| Small town+rural | 4116 (8.7) | 909 (8.6) | 324 (10.1) | 2883 (8.7) |
| Insurance, n (%) | ||||
| Medicare Advantage | 36 662 (77.9) | 7981 (75.4) | 2434 (75.7) | 26 247 (79.0) |
| Commercial | 10 371 (22.1) | 2598 (24.6) | 780 (24.3) | 6993 (21.0) |
| Opioid dose in MME, n (%) | ||||
| 50 ≤90 | 19 524 (41.5) | 2900 (27.4) | 1326 (41.3) | 15 298 (46.0) |
| 90 ≤150 | 13 282 (28.2) | 2960 (28.0) | 951 (29.6) | 9371 (28.2) |
| 150 ≤300 | 10 689 (22.7) | 3298 (31.2) | 720 (22.4) | 6671 (20.1) |
| ≥300 | 3538 (7.5) | 1421 (13.4) | 217 (6.7) | 1900 (5.7) |
| Benzodiazepine co-prescription d , n (%) | 12 787 (27.2) | 3073 (29.1) | 893 (27.8) | 8821 (26.5) |
| Drug overdose e , n (%) | 1190 (2.5) | 312 (3.0) | 122 (3.8) | 756 (2.3) |
| Comorbidities f , n (%) | ||||
| 0 | 19 786 (42.1) | 4501 (42.6) | 1427 (44.4) | 13 858 (41.7) |
| 1 | 12 420 (26.4) | 2814 (26.6) | 805 (25.1) | 8801 (26.5) |
| 2 | 6273 (13.3) | 1416 (13.4) | 393 1(2.2) | 4464 (13.4) |
| ≥3 | 8554 (18.2) | 1848 (17.5) | 589 (18.3) | 6117 (18.4) |
| Mood or suicide visit | 23 954 (50.9) | 5645 (53.4) | 1679 (52.2) | 16 630 (50.0) |
| Healthcare utilization, mean during year (SD) | ||||
| Primary care visit not for pain | 5.9 (5.9) | 5.9 (5.5) | 5.6 (6.1) | 5.9 (6.0) |
| Primary care visit for pain | 1.1 (2.4) | 1.1 (2.4) | 1.1 (2.5) | 1.1 (2.4) |
| ED visit not for pain | 0.9 (2.1) | 0.9 (1.8) | 0.9 (1.8) | 0.9 (2.3) |
| ED visit for pain | 0.1 (0.5) | 0.1 (0.5) | 0.1 (0.5) | 0.1 (0.5) |
| Hospitalization not for pain | 0.2 (0.7) | 0.2 (0.7) | 0.3 (0.7) | 0.2 (0.7) |
| Hospitalization for pain | 0.03 (0.2) | 0.04 (0.2) | 0.03 (0.2) | 0.03 (0.2) |
| Mean follow-up months (SD) | 11.1 (2.3) | 11.4 (1.7) | 11.6 (1.4) | 11.0 (2.5) |
Abbreviations: ED = emergency department; HS = high school; LTOT = long-term opioid therapy; MME = morphine milligram equivalent; SD = standard deviation.
Values are at the person-level for the most recent baseline period (if patient was eligible after more than one baseline period). The overall cohort had 47 033 people with 67 784 baseline periods. There were 13 793 people who had any taper with 14 923 baseline periods. There were 33 240 people who did not taper with 47 700 baseline periods. There were 10 579 people who tapered then continued LTOT with 11 690 baseline periods. There were 3214 people who tapered and discontinued with 3233 baseline periods.
Education estimated based on median household education level for patient’s US census block.
Rurality derived from Rural Urban Commuting Area (RUCA) codes. Missing was added to the small town+rural (small category), and missingness was 56 (0.1%), 21 (0.2%), 5 (0.2%), and 30 (0.1%) for each patient group above, respectively.
Benzodiazepine co-prescription, based on pharmacy claims, on date of cohort entry.
≥1 drug overdose in the 90 days prior to cohort entry (see methods for determination of drug overdose).
Elixhauser comorbidities; 27 non-cancer Elixhauser comorbidities were measured.
As compared to patients in the non-tapered state, patients in the tapered state had a higher adjusted rate of pain-related ED visits (adjusted incidence rate ratio [aIRR] 1.21, 95% confidence interval [CI]: 1.11–1.30) (Table 2) but did not differ significantly in the rate of pain-related hospitalization (aIRR 1.08, CI: 0.97–1.20). Patients in the tapered (vs the non-tapered) state had a slightly lower rate of pain-related primary care visits during follow-up (aIRR 0.96, CI: 0.92–1.00).
Table 2.
Healthcare visits for pain diagnoses among patients previously prescribed stable LTOT ≥50 MME, comparing those who tapered opioid doses to those who did not, and among those who tapered those who either tapered-and-continued or tapered-and-discontinued.
| ED visits for paind |
Hospitalizations for painf |
Outpatient visits for paine |
|||||||
|---|---|---|---|---|---|---|---|---|---|
| Opioid dose taperb | Event rate per 1000-person years | Adjusted rate difference per 1000-person years (CI) | aIRRa (CI) | Event rate per 1000-person years | Adjusted rate difference per 1000-person years (CI) | aIRRa (CI) | Event rate per 1000-person years | Adjusted rate difference per 1000-person years (CI) | aIRRa (CI) |
| Non-tapered | 132 | ref | ref | 52 | ref | ref | 1091 | ref | ref |
| All tapers b | 162 | 19 (11 to 27) | 1.21 (1.11 to 1.30) | 61 | 3.5 (1.4 to 8.4) | 1.08 (0.97 to 1.20) | 1044 | −32 (−62 to −0.3) | 0.96 (0.92 to 1.00) |
| Tapered-and-continuedc | 164 | 20.7 (12 to 29) | 1.23 (1.14 to 1.32) | 64 | 6.3 (1.0 to 12) | 1.14 (1.03 to 1.27) | 1097 | 11 (−23 to 45) | 1.01 (0.97 to 1.06) |
| Tapered-and-discontinuedc | 149 | 10.6 (−6.2 to 27) | 1.12 (0.95 to 1.31) | 40 | −11 (−22 to −0.8) | 0.74 (0.54 to 1.02) | 759 | −252 (−314 to −190) | 0.68 (0.61 to 0.76) |
Abbreviations: AIRR, adjusted incidence rate ratios; CI = confidence interval; ED = emergency department; LTOT = long-term opioid therapy; MME = morphine milligram equivalent.
Analyses adjusted for: Age, sex, education, RUCA, insurance, baseline opioid dose (morphine milligram equivalents), baseline benzodiazepine prescription (at time of cohort entry), baseline drug overdose (in 90 days prior to index date), comorbidity (Elixhauser and depression/anxiety/suicidality), baseline primary care visits (for pain and not for pain), baseline ED visits (for pain and not for pain), baseline hospitalizations (for pain and not for pain), and year.
Predictor variable for this model is all tapers ≥15% (includes Tapered-and-Continued and Tapered-and-Discontinued). The unit of analysis is the person-period. The overall cohort had 47 033 people with 67 784 baseline periods. There were 13 793 people who had any taper with 14 923 baseline periods. There were 33 240 people who did not taper with 47 700 baseline periods.
In a separate model, the predictor taper variable had three levels: Non-tapered, tapered ≥15% compared to baseline with continued prescriptions for MME > 0 (Tapered-and-Continued), and tapered ≥15% compared to baseline and then discontinued opioids (MME = 0) at any time after taper (Tapered-and-Discontinued). There were 10 579 people who tapered then continued LTOT with 11 690 baseline periods. There were 3214 people who tapered and discontinued with 3233 baseline periods.
ED visits for pain were defined as ED visits without hospitalization for ICD-10 pain codes in any position; see Supplement X for list of codes.
Outpatient visits for pain were defined as billed visits to an ambulatory facility with a primary care provider (family medicine or general internal medicine), ICD-10 pain codes in any position; see Supplement for list of pain codes.
Hospitalizations for pain were defined as an inpatient visit to a hospital facility for ICD-10 pain codes in any position; see Supplement for list of codes.
When taper status was subcategorized based on continuation vs. discontinuation of opioids after taper, tapered-and-continued status (compared to non-tapered) remained associated with a 23% higher rate of pain-related ED visits (aIRR 1.23, CI: 1.14–1.32) and tapered-and-discontinued (compared to non-tapered) status was associated with a similar rate of pain-related ED visits (aIRR 1.12, CI: 0.95–1.31) (Table 2). With regard to hospitalizations, patients in the tapered-and-continued state (vs. non-tapered) had significantly higher rates of pain-related hospitalizations (aIRR 1.14, CI 1.03–1.27), while patients in a tapered-and-discontinued state had a lower, though not statistically significant, rate (vs non-tapered) (aIRR 0.74, CI 0.54–1.02).
Meanwhile, as compared to the non-tapered state, the tapered-and-discontinued state was associated with a 32% relative reduction in pain-related primary care visits (aIRR 0.68, CI: 0.61–0.76), while pain-related primary care visit rates were similar in tapered-and-continued and non-tapered states (aIRR 1.01, CI: 0.97–1.06).
Discussion
Among patients prescribed stable LTOT, opioid tapering during follow-up was associated with an increase in pain-related ED visits and a small decrease in pain-related primary care visits. When tapered patients were subclassified based on whether opioids were continued or discontinued after initiating taper, the small decrease in pain-related primary care visits overall among tapered patients was explained by a large decline in pain-related primary care visits among the minority of tapered patients who tapered-and-discontinued opioids. Meanwhile, among patients who tapered-and-continued opioids during follow-up, there was a significant increase in pain-related ED visits and hospitalizations.
Several factors may account for the observed shifts from primary care to ED or hospital settings for pain-related visits after opioid tapering. We consider the below two most relevant to our results.
First, opioid tapering has also been associated with disruption in primary care relationships25 and tapered patients may have difficulty finding new accepting primary care clinicians.17 Such patients may be more likely to seek care for pain-related diagnoses in EDs rather than from primary care after tapering than prior to tapering. Our results show a large reduction in pain-related primary care visits among those who tapered-and-discontinued, compared to those who tapered-and-continued, suggests that many discontinued patients either cease or substantially reduce visits to primary care for pain-related care after opioid discontinuation. This decrease in primary care visits could reflect a reduction in patients’ need for fewer pain-related primary care visits after opioid discontinuation, as could be expected if pain was stable or improved after taper, or could be due to a disruption in the patient-provider relationship for pain control26; our data cannot inform the extent of either contribution to the noted primary care visit reductions.
Second, opioid tapering, especially when rapid or after discontinuation, can precipitate withdrawal or exacerbated pain, leading to patients who seek urgent attention for this pain at emergency rooms and potentially leading to hospital admission for prolonged pain control. After subcategorization of tapered patients as continued vs discontinued, patients who tapered-and-continued opioids had a significantly increased rate of ED visits and hospitalizations for pain-related diagnoses while those who tapered-and-discontinued opioids had a significant decrease in primary care visits for pain. This suggests increased uncontrolled pain for which the patient seeks ED treatment, not primary care, either due to the severity of the pain or a disruption in the primary care patient-provider relationship. We encourage cautious interpretation of the association with hospitalization, as we observed no significant association between tapering overall and pain-related hospitalization.
Among a similarly defined cohort, we previously found associations between opioid tapering and increases in all-cause ED visits and hospitalizations with a small decline in primary care visits.12 The current study elucidates the extent to which the tapering-associated differences in those all-cause ED and hospital utilization may be explained by changes in pain-related care-seeking. Overall, the adjusted rate difference associated with tapering (vs non-tapering) was 19 pain-related ED visits per 1000 person-years, which is ∼15% of the adjusted rate difference associated with tapering in all-cause ED visits observed in our prior study.12 Similarly, while we observed a decline of 144 all-cause primary care visits per 1000 person-years associated with tapering in our prior study,12 the current study suggests that tapering was associated with a decline of 32 pain-related primary care visits per 1000 person-years (∼22% of the associated decline in primary care visits). Together, these findings suggest that previously observed decreases in overall primary care visits and increases in ED and hospital utilizations associated with tapering are partly attributable to pain-related utilization; tapering pain medication is associated with changes in healthcare utilization for pain, consistent with concerns that tapering can lead to uncontrolled pain.
Our study has limitations. We used claims records and therefore were unable to determine who initiated the taper and why, including for new diagnosis of opioid use disorder, or the medical appropriateness of changes in health care utilization. We examined pain-related healthcare utilization, and although visits can be considered a proxy for patient pain experience, we did not have the data to assess patient-reported pain (ie, pain scales). Pain-related visits were determined by diagnostic codes and therefore we are limited in the meaning of these codes to represent the content of the visit and are not able to capture when pain was discussed but not coded with one of our included pain codes. The amount of time spent discussing pain and its severity and stability cannot be ascertained from diagnostic codes. We also likely missed visits that discussed pain but did not include that as a diagnostic code or included a pain-related code that was not in our pain code list or pain visits that we did not include as we could not differentiate uncontrolled chronic pain vs acute crises requiring ED visits (eg, sickle cell disease without crisis vs sickle cell disease with acute chest syndrome or Crohn’s disease without complications vs Crohn’s disease with intestinal obstruction). In this case we would still be able to measure changes in pain-coded visits but the results would be tempered. We also were unable to capture non-visit healthcare interactions, such as telephone calls or patient portal messages, and we did not include urgent care visits. The maximum follow-up time was 1 year, and this study does not assess long-term healthcare utilization after taper. We were not able to include outpatient pain specialist visits as these could not be differentiated in our data. We aimed to study outcomes after tapering from stable long-term opioid doses and these results do not include the experiences for patients with unstable baseline opioid doses. Finally, although we conducted this study with a diverse, US national sample, the data are from years 2015–2019 and were limited to patients with stable commercial insurance or Medicare Advantage without Medicaid, and the data set lacked a self-report measure of patient race/ethnicity.26,27
Conclusions
In this observational study of pain-related visits after LTOT tapering, tapering then continuing opioids was associated with more ED visits and hospitalizations for pain-related diagnoses, and tapering and discontinuation of opioids was associated with fewer outpatient primary care visits for pain-related diagnoses. These associations suggest that opioid tapering may lead to increased pain for patients whose doses were reduced, and to a decreased perceived need for primary care for pain control those who opioids were discontinued. Clinicians and policymakers should consider potential changes in pain control and pain-related health care utilization when making recommendations for opioid tapering.
Supplementary Material
Contributor Information
Elizabeth Magnan, Department of Family and Community Medicine, University of California, Davis, Sacramento, CA 95817, United States; Center for Healthcare Policy and Research, University of California, Davis, Sacramento, CA 95817, United States.
Daniel J Tancredi, Center for Healthcare Policy and Research, University of California, Davis, Sacramento, CA 95817, United States; Department of Pediatrics, University of California, Davis, Sacramento, CA 95817, United States.
Guibo Xing, Center for Healthcare Policy and Research, University of California, Davis, Sacramento, CA 95817, United States.
Alicia Agnoli, Department of Family and Community Medicine, University of California, Davis, Sacramento, CA 95817, United States; Center for Healthcare Policy and Research, University of California, Davis, Sacramento, CA 95817, United States.
I E Tseregounis, Center for Healthcare Policy and Research, University of California, Davis, Sacramento, CA 95817, United States; Department of Internal Medicine, University of California, Davis, Sacramento, CA 95817, United States.
Joshua J Fenton, Department of Family and Community Medicine, University of California, Davis, Sacramento, CA 95817, United States; Center for Healthcare Policy and Research, University of California, Davis, Sacramento, CA 95817, United States.
Supplementary material
Supplementary material is available at Pain Medicine online.
Funding
This study was supported by a University of California–Optum Labs Research Credit through Dr Fenton and the Department of Family and Community Medicine, University of California, Davis. This project was supported by the University of California Davis Center for Healthcare Policy and Research (CHPR). Dr Agnoli was supported by the University of California, Davis School of Medicine Dean’s Office (Dean’s Scholarship in Women’s Health Research) and NICHD K12 HD051958 (Building Interdisciplinary Research Careers in Women’s Health at UC Davis).
Conflicts of interest: The authors have no conflicts of interest to declare.
References
- 1. Hedegaard H, Miniño AM, Spencer MR, Warner M. Drug Overdose Deaths in the United States, 1999–2020. National Center for Health Statistics; 2021. NCHS Data Brief, no 428. [PubMed]
- 2. 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]
- 3. Fenton JJ, Agnoli AL, Xing G, et al. Trends and rapidity of dose tapering among patients prescribed long-term opioid therapy, 2008-2017. JAMA Netw Open. 2019;2(11):e1916271. 10.1001/jamanetworkopen.2019.16271 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. U.S. Department of Health and Human Services. HHS guide for clinicians on the appropriate dosage reduction or discontinuation of long-term opioid analgesics; October 2019. Accessed December 5, 2024. https://www.hhs.gov/system/files/Dosage_Reduction_Discontinuation.pdf
- 5. Dowell D, Ragan KR, Jones CM, Baldwin GT, Chou R. CDC clinical practice guideline for prescribing opioids for pain—United States, 2022. MMWR Recomm Rep. 2022;71(3):1-95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Larochelle M, Lagisetty PA, Bohnert AS. Opioid tapering practices—time for reconsideration? JAMA. 2021;326(5):388-389. [DOI] [PubMed] [Google Scholar]
- 7. Scherrer JF, Salas J, Copeland LA, et al. Prescription opioid duration, dose, and increased risk of depression in 3 large patient populations. Ann Fam Med. 2016;14(1):54-62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Mark TL, Parish W. Opioid medication discontinuation and risk of adverse opioid-related health care events. J Subst Abuse Treat. 2019;103:58-63. [DOI] [PubMed] [Google Scholar]
- 9. Oliva EM, Bowe T, Manhapra A, et al. Associations between stopping prescriptions for opioids, length of opioid treatment, and overdose or suicide deaths in US veterans: observational evaluation. BMJ. 2020;368:m283. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Agnoli A, Xing G, Tancredi DJ, Magnan E, Jerant A, Fenton JJ. Association of dose tapering with overdose or mental health crisis among patients prescribed long-term opioids. JAMA. 2021;326(5):411-419. 10.1001/jama.2021.11013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Binswanger IA, Glanz JM, Faul M, et al. The association between opioid discontinuation and heroin use: a nested case-control study. Drug Alcohol Depend. 2020;217:108248. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Magnan EM, Tancredi DJ, Xing G, Agnoli A, Jerant A, Fenton JJ. Association between opioid tapering and subsequent health care use, medication adherence, and chronic condition control. JAMA Netw Open. 2023;6(2):e2255101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Mackey K, Anderson J, Bourne D, Chen E, Peterson K. Benefits and harms of long-term opioid dose reduction or discontinuation in patients with chronic pain: a rapid review. J Gen Intern Med. 2020;35(suppl 3):935-944. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Alenezi A, Yahyouche A, Paudyal V. Interventions to optimize prescribed medicines and reduce their misuse in chronic non-malignant pain: a systematic review. Eur J Clin Pharmacol. 2021;77(4):467-490. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. McDonagh MS, Selph SS, Buckley DI, et al. Nonopioid Pharmacologic Treatments for Chronic Pain. Agency for Healthcare Research and Quality (US); 2020. Comparative Effectiveness Review, No. 228. Report No.: 20-EHC010.
- 16. Skelly AC, Chou R, Dettori JR, et al. Noninvasive Nonpharmacological Treatment for Chronic Pain: A Systematic Review Update. Agency for Healthcare Research and Quality (US); 2020. Report No.: 20-EHC009.
- 17. Lagisetty PA, Healy N, Garpestad C, Jannausch M, Tipirneni R, Bohnert AS. Access to primary care clinics for patients with chronic pain receiving opioids. JAMA Netw Open. 2019;2(7):e196928. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. OptumLabs. Data from: OptumLabs and OptumLabs Data Warehouse (OLDW) descriptions and citation. Eden Prairie, MN; 2020. Accessed December 5, 2024. https://www.optumlabs.com
- 19. Centers for Disease Control and Prevention. Pocket Guide: Tapering Opioids for Chronic Pain. US Department of Human and Health Services; 2018. [Google Scholar]
- 20. Venkatesh AK, Mei H, Kocher KE, et al. Identification of emergency department visits in medicare administrative claims: approaches and implications. Acad Emerg Med. 2017;24(4):422-431. 10.1111/acem.13140 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Healthcare Cost and Utilization Project (HCUP). Clinical Classifications Software Refined (CCSR). Agency for Healthcare Research and Quality. 2023. Accessed September 11, 2023. www.hcup-us.ahrq.gov/toolssoftware/ccsr/ccs_refined.jsp
- 22. Fenton JJ, Fang S-Y, Ray M, et al. Longitudinal care patterns and utilization among patients with new-onset neck pain by initial provider specialty. Spine (Phila Pa 1976). 2023;48(20):1409-1418. [DOI] [PubMed] [Google Scholar]
- 23. Wei Y-JJ, Chen C, Lewis MO, Schmidt SO, Winterstein AG. Trajectories of prescription opioid dose and risk of opioid-related adverse events among older Medicare beneficiaries in the United States: a nested case–control study. PLoS Med. 2022;19(3):e1003947. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8-27. 10.1097/00005650-199801000-00004 [DOI] [PubMed] [Google Scholar]
- 25. Perez HR, Buonora M, Cunningham CO, Heo M, Starrels JL. Opioid taper is associated with subsequent termination of care: a retrospective cohort study. J Gen Intern Med. 2020;35(1):36-42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Shoemaker K, Smith CP. The impact of patient-physician alliance on trust following an adverse event. Patient Educ Couns. 2019;102(7):1342-1349. [DOI] [PubMed] [Google Scholar]
- 27. Darnall BD. The national imperative to align practice and policy with the actual CDC opioid guideline. Pain Med. 2020;21(2):229-231. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.