Abstract
Objective
To identify common opioid tapering trajectories among patients commencing opioid taper from long-term opioid therapy for chronic non-cancer pain and to examine patient-level characteristics associated with these different trajectories.
Design
A retrospective cohort study.
Setting
Australian primary care.
Subjects
Patients prescribed opioid analgesics between 2015 and 2020.
Methods
Group-based trajectory modeling and multinomial logistic regression analysis were conducted to determine tapering trajectories and to examine demographic and clinical factors associated with the different trajectories.
Results
A total of 3369 patients commenced a taper from long-term opioid therapy. Six distinct opioid tapering trajectories were identified: low dose / completed taper (12.9%), medium dose / faster taper (12.2%), medium dose / gradual taper (6.5%), low dose / noncompleted taper (21.3%), medium dose / noncompleted taper (30.4%), and high dose / noncompleted taper (16.7%). A completed tapering trajectory from a high opioid dose was not identified. Among patients prescribed medium opioid doses, those who completed their taper were more likely to have higher geographically derived socioeconomic status (relative risk ratio [RRR], 1.067; 95% confidence interval [CI], 1.001–1.137) and less likely to have sleep disorders (RRR, 0.661; 95% CI, 0.463–0.945) than were those who didn’t complete their taper. Patients who didn’t complete their taper were more likely to be prescribed strong opioids (eg, morphine, oxycodone), regardless of whether they were tapered from low (RRR, 1.444; 95% CI, 1.138–1.831) or high (RRR, 1.344; 95% CI, 1.027–1.760) doses.
Conclusions
Those prescribed strong opioids and high doses appear to be less likely to complete tapering. Further studies are needed to evaluate the clinical outcomes associated with the identified trajectories.
Keywords: opioids, opioid taper, chronic non-cancer pain, long-term opioid therapy, general practice, group-based trajectory models
Introduction
During the past 3 decades, there has been a substantial increase in the use of prescription opioids globally and in Australia.1–3 In Australia, almost 15 million opioid prescriptions were dispensed in 2015.3 Two million Australians were found to initiate prescription opioids every year.4 Parallel to the increase in prescribing of opioids, a rise in opioid-related harms in Australia has also been reported. In 2017, opioids accounted for more than half of all unintended deaths, 56% of which involved prescription opioids.5
Escalating opioid use has been attributed to greater use of stronger, longer-acting opioids for the treatment of chronic non-cancer pain.6 Despite limited evidence of efficacy, which is compounded by risks of misuse, addiction, and overdose, opioids continue to be prescribed for chronic non-cancer pain.7–9 In efforts to reduce harms from opioids, there has been a considerable focus on deprescribing of opioids in Australia and internationally.10–12 Deprescribing is a key clinical strategy to reduce opioid-related harms and improve the quality of life.13–15 However, rapid tapering of doses and abrupt discontinuation from long-term opioid therapy were found to be associated with unintended consequences, including withdrawal symptoms (eg, uncontrolled pain, psychological distress), illicit opioid use, and suicide attempts.16,17 Given the complexity of pain management and the undesirable effects of opioids (eg, physical dependence, tolerance), experts emphasize the importance of individualized tapering regimes which may take into account factors such as starting dosages, required tapering speeds, combinations of opioids and non-opioids, and other patient-level factors (eg, age, concurrent medication use, and comorbidities including substance dependence and mental health disorders).10,18,19
Opioid deprescribing can be a challenging process and is poorly described in the current Australian and international clinical guidelines.11,20,21 The 2022 Evidence-Based Clinical Practice Guideline for Deprescribing Opioid Analgesics, commissioned by the Australian Government National Health and Medical Research Council (NHMRC), highlighted the “limited evidence to inform a preferred protocol for opioid deprescribing,” referring clinicians to existing local guidelines, which are based mostly on expert consensus.20,22 Moreover, the current guidelines do not provide specific guidance for any key clinical subpopulations (eg, those with mental health or substance use disorders), despite the increased risks of adverse opioid-related outcomes in these populations.18,23 A recently published systematic review of international opioid deprescribing guidelines highlighted a need for high-quality clinical evidence to strengthen guideline recommendations.24
In a recent priority-setting study, opioid deprescribing in primary care has been identified as a top priority for research.25 At present, there are few studies worldwide and none from Australia that provide a detailed picture of the different trajectory pathways for patients who commence an opioid taper. Studies that describe the common trajectories of opioid taper and patient characteristics associated with different taper trajectories could help to address this gap and to provide a more detailed understanding of individual patients’ pathways as they commence tapering their opioids. Identifying characteristics of patients and trajectories that are associated with taper completion could help understand how to achieve better taper outcomes and identify those in need of further support. Therefore, the aim of the present study was to identify common opioid tapering trajectories among patients commencing opioid taper from long-term opioid therapy for chronic non-cancer pain and to examine patient-level characteristics associated with these different trajectories.
Methods
Data source
The study used the Population Level Analysis and Reporting (POLAR) dataset, a primary care dataset from 464 Victorian general practices across the eastern half of Australia, consisting of almost 700 000 de-identified patient records in which opioid analgesics were prescribed.26 We accessed clinical records including prescriptions from eligible practices between January 2015 and December 2020. The data included age in 5-year brackets to protect patient confidentiality. Further information can be found in the Opioid Prescribing, Policy Impacts, and Clinical Outcomes (OPPICO) study protocol.27
Study design
We undertook a retrospective cohort study of patients prescribed opioid analgesics between January 2015 and December 2020. To identify changes in opioid dose after a baseline period of long-term opioid therapy, we examined opioid doses over a 12-month follow-up period from the index date, defined as the first day of a reduction in opioid dose. All opioid formulations indicated for chronic non-cancer pain were examined, as outlined in the OPPICO study protocol. Opioid doses were converted to oral morphine equivalents (OME) with the use of published conversion factors.28 We accounted for repeat prescriptions by incorporating the OME of the prescription every 28 days for transdermal and slow-release formulations and every 14 days for all other formulations as observed in our data. This study was reported according to the RECORD-PE Checklist.29
Eligibility criteria
We defined our cohort on the basis of the following criteria: Patients (1) were ≥19 years of age at cohort entry; (2) had received at least 6 months of continuous opioid prescriptions (operationalized as having a gap of fewer than 60 days between subsequent prescriptions, based on the standard supply of opioids in Australia26) (Figure 1A); (3) had received a stable dosage level during the 6-month baseline period (operationalized as the average daily opioid dose during any single month during the 6-month baseline period that did not vary by more than 10% from the average baseline dose, defined as the average of the total dose prescribed during the 6-month baseline period) (Figure 2); (4) had an absence of terminal illness or cancer diagnoses (other than non-melanoma skin cancer) in the 12 months before and 12 months after the index date; (5) had an absence of opioid prescriptions for opioid replacement therapy (such as high-dose sublingual buprenorphine and buprenorphine-naloxone, methadone liquid) in the 12 months before and 12 months after the index date; and (6) had at least one clinic activity (eg, general practitioner visit or telehealth) or medication data recorded within 90 days after the date of the last opioid prescription (Figure 1B). This latter criterion examining activity within a 90-day window after an opioid taper was deemed necessary to distinguish between discontinuation of opioid therapy and disenrollment from the medical practices.
Figure 1.
(A) Visualization of cohort entry and exposure and covariate assessment windows. (B) Illustration of the study time frame. # Screening window for disengagement from practice (90 days after the last opioid script). ORT= opioid replacement therapy. Figure adapted from Schneeweiss S, Rassen JA, Brown JS, Rothman KJ, Happe L, Arlett P, et al. Graphical depiction of longitudinal study designs in health care databases. Ann Intern Med. 2019;170(6):398-406.
Figure 2.
Timeline illustrating baseline and follow-up periods. Average baseline dose was determined by calculating the average of the total dose prescribed during the baseline period (B1–B6). Opioid doses in each of the baseline months did not vary by more than 10% from the average baseline dose. Index date (day = 0) was defined as commencement of opioid taper. To be considered a taper, a minimum 10% reduction in the average daily opioid dose must have been observed over a 90-day period (period 1) compared with the average baseline dose. Opioid doses were assessed for 12 months after the index date in 10 overlapping 90-day periods (periods 1–10). Figure adapted from Fenton et al.31
Definition of taper
To determine the commencement of a taper, we defined the index date of a taper episode as the point at which the following criteria were fulfilled: (1) the index date immediately following a 6-month period of stable baseline opioid dose based on prescription data, as described in the eligibility criteria (Figure 2), and (2) greater than 10% reduction in the average daily opioid dose of the first 90 days after the index date (ie average daily dose during period 1; Figure 1) compared with the average baseline dose. This definition was based on previously used definitions30,31 and Australian guideline recommendations of tapering by a 10%–25% reduction every 4 weeks for patients taking >3 months of opioids.20 Patients were followed up for 12 months from the index date, with a 90-day rolling average dose used to smooth out dose fluctuations due to minor variations in prescription dates. Completion of a taper was defined as the average daily OME over a rolling 90-day period being approximately equal to 0 mg (allowing for occasional pro re nata use). Values of the average daily OME in the 90-day periods were capped to a maximum value of 250 mg, equivalent to 3 standard deviations from the mean. To ensure a 12-month lookback for covariates and a full 12 months of follow-up plus a 90-day screening window for discontinuation, the index date was censored between January 1, 2016, and September 30, 2019, meaning that those who commenced a taper before January 2016 or after September 2019 were excluded from our cohort. Given that only a small proportion (72 out of 3369 patients, 2.1%) of the overall sample tapered more than once, we chose to keep the first chronological taper episode, consistent with other published literature.32,33
Covariates
A series of covariates that were established in prior studies to be relevant for tapering outcomes include: (1) demographic characteristics, including age, sex, concessional status (which enables access to government-subsidized health services and medicines), remoteness, and geographically derived socioeconomic status (SES) (measured with the Socio-Economic Indexes for Areas [SEIFA] score), as detailed in the OPPICO study protocol27; (2) characteristics of opioids prescribed during the baseline period, including type of opioid (weak [eg, codeine, tramadol] vs strong [eg, morphine, oxycodone] and short vs long acting [Table S1]), route of administration, and number of types of opioid prescribed; (3) diagnosis of mental health conditions (Table S2); (4) concurrent prescriptions of other analgesics during the baseline period, including nonsteroidal anti-inflammatory drugs (NSAIDs), paracetamol, and gabapentinoids, as outlined in the OPPICO protocol; and (5) referral to other health care professionals and services, including allied health services, medical and surgical specialists, and mental health and addiction services (Table S3). For categorical covariates, category cutoffs were considered to ensure sufficient sample sizes within the trajectory groups.
Statistical analyses
Group-based trajectory modeling is a statistical method that is used to identify clusters of individuals following similar trajectories over time.34 We performed group-based trajectory modeling to determine the common trajectories of opioid tapering, using the rolling monthly average of daily OME as a dependent variable over 10 consecutive periods, equivalent to a 12-month follow-up. This means that the dose groups were determined through the group-based trajectory modeling analysis, as opposed to dose cutoffs being used to define groups. Because of the wide variation in the daily OME, which made the models unstable, we applied the log transformation. The model used censored normal distributions (cnorm) and included fifth-order polynomial, the most flexible function form. The Bayesian information criterion is a score calculated from the model’s maximized likelihood, sample size, and the number of parameters in the model. In general, the model with the highest Bayesian information criterion score is selected as the best-fitting model.34 However, the Bayesian information criterion score of our model continued to increase with the addition of more groups, which is known to occur in some circumstances (Table S4). Therefore, we based our model selection on clinical context and model parsimony (ie, the simplest model that illustrates the distinct features of opioid tapering trajectories), as recommended in the literature.34 Nagin’s criteria were used to assess for model adequacy (Table S5). The trajectories commencing tapers from opioid doses of <50 mg, 50–100 mg, and >100 mg were labeled as tapers from low, medium, and high doses, respectively, on the basis of the definitions used in local guidelines and the Victorian prescription drug monitoring program.35 Multinomial logistic regression and descriptive statistics were conducted to determine the differences in demographic and clinical characteristics between the patients within the opioid trajectories. We chose the group with the largest sample size and the most clinically meaningful comparator as a reference group to enable comparisons with various tapering groups through the multinomial logistic regression. Additional dichotomous logistic regression analysis was performed to further enable comparison across different tapering trajectories from comparable doses. To ensure the validity of the regression model, we tested for multicollinearity and ensured that the variance inflation factor was less than 5 (the highest variance inflation factor was 1.30).36 Baseline daily OME was not examined in the regression analysis, as the tapering trajectories were differentiated by the probabilities of group membership based on the baseline daily OME. All analyses were conducted in Stata Statistical Software, Release 17 (StataCorp, 2021). Graphical illustrations were created with GraphPad Prism version 9.5.0 (San Diego, California, USA) and Microsoft PowerPoint Version 16.70.
Data approval and ethics review
The study was approved by the Monash University Human Research Ethics Committee (Approval number: 24139).
Results
Sample characteristics
Of a total of 676 966 patients who were prescribed at least one opioid during our study period, 8643 patients (1.3%) had at least 6 months of continuous, stable baseline opioid dose, as determined by analysis of opioid prescriptions (Figure 3). Of those, 3369 patients (39.0%) commenced a taper. A total of 887 (26%), 840 (25%), 930 (28%), and 712 (21%) patients commenced tapers in 2016, 2017, 2018, and 2019, respectively. At cohort entry, most patients were between the ages of 39–58 years (31.7%) and 59–78 years (39.5%), and 61.6% were female (Table 1). A greater proportion of the total sample were concession beneficiaries (76.9%), compared with 38.3% for the larger cohort of patients prescribed at least one opioid. Almost three-quarters of those who commenced a taper resided in major cities (74.6%), similar to those in the larger cohort (83.7%). The mean SEIFA score was 6.2 (SD: 2.8) out of 10 (with a score of 10 indicating the least disadvantaged) for those who commenced a taper, lower than that of the larger cohort (6.9 [SD: 2.6]).
Figure 3.
Derivation of study sample.
Table 1.
Key demographic and clinical characteristics of patients by opioid tapering trajectory groups.a
| Discontinuation opioid trajectories |
Noncompleted opioid tapering trajectories |
||||||
|---|---|---|---|---|---|---|---|
| Low dose / completed taper (435 (12.9%) | Medium dose / faster taper (411 (12.2%) | Medium dose / gradual taper (221 (6.5%) | Low dose / noncompleted taper (716 (21.3%) | Medium dose / noncompleted taper (1026 (30.4.%) | High dose / noncompleted taper (560 (16.7%) | Total (n = 3369) | |
| Demographic characteristics | |||||||
| Ageb | |||||||
| 19–38 years | 51 (11.7%) | 28 (6.8%) | 19 (8.6%) | 63 (8.8%) | 73 (7.1%) | 48 (8.6%) | 282 (8.4%) |
| 39–58 years | 119 (27.4%) | 110 (26.8%) | 70 (31.7%) | 186 (26.0%) | 324 (31.6%) | 260 (46.4%) | 1069 (31.7%) |
| 59–78 years | 162 (37.2%) | 156 (38.0%) | 79 (35.7%) | 282 (39.4%) | 436 (42.5%) | 216 (38.6%) | 1331 (39.5%) |
| ≥79 years | 103 (23.7%) | 117 (28.5%) | 53 (24.0%) | 185 (25.8%) | 193 (18.8%) | 36 (6.4%) | 687 (20.4%) |
| Sexc | |||||||
| Femalei | 270 (62.1%) | 260 (63.3%) | 137 (62.0%) | 469 (65.5%) | 617 (60.1%) | 321 (57.3%) | 2074 (61.6%) |
| Concessional beneficiary statusd | |||||||
| Beneficiaryi | 307 (70.6%) | 320 (77.9%) | 179 (81.0%) | 544 (76.0%) | 809 (78.8%) | 433 (77.3%) | 2592 (76.9%) |
| Remotenesse | |||||||
| Major cities of Australiai | 348 (80.0%) | 300 (73.0%) | 165 (74.7%) | 568 (79.3%) | 725 (70.7%) | 407 (72.7%) | 2513 (74.6%) |
| Geographically derived socioeconomic status | |||||||
| Mean SEIFAf score (SD) | 6.3 (2.8) | 6.1 (2.7) | 6.5 (2.6) | 6.4 (2.7) | 6.0 (2.8) | 6.0 (2.7) | 6.2 (2.8) |
| Opioid characteristics | |||||||
| Type of opioid | |||||||
| Strong opioidsg,i | 223 (51.3%) | 297 (72.3%) | 156 (70.6%) | 479 (66.9%) | 712 (69.4%) | 447 (79.8%) | 2314 (68.7%) |
| Long-acting opioidsh,i | 223 (51.3%) | 381 (92.7%) | 201 (91.0%) | 529 (73.9%) | 951 (92.7%) | 541 (96.6%) | 2826 (83.9%) |
| ≥2 types of opioidsi | 26 (6.0%) | 95 (23.1%) | 62 (28.1%) | 101 (14.1%) | 315 (30.7%) | 311 (55.5%) | 910 (27.0%) |
| Baseline average daily OME in mg (SD) | 14.90 (23.03) | 67.36 (67.67) | 68.09 (93.87) | 22.93 (22.42) | 57.27 (36.28) | 162.34 (108.55) | 63.91 (77.19) |
| Mental health diagnosis | |||||||
| Depression and/or anxietyi | 270 (62.1%) | 269 (65.5%) | 146 (66.1%) | 420 (58.7%) | 647 (63.1%) | 430 (76.8%) | 2182 (64.8%) |
| Substance use disorderi | 29 (6.7%) | 21 (5.1%) | 16 (7.2%) | 26 (3.6%) | 55 (5.4%) | 37 (6.6%) | 184 (5.5%) |
| Sleep disorderi | 127 (29.2%) | 95 (23.1%) | 47 (21.3%) | 178 (24.9%) | 285 (27.8%) | 182 (32.5%) | 914 (27.1%) |
| Concurrent analgesic prescriptions | |||||||
| NSAIDsi | 86 (19.8%) | 81 (19.7%) | 39 (17.6%) | 159 (22.2%) | 226 (22.0%) | 145 (25.9%) | 736 (21.8%) |
| Paracetamoli | 45 (10.3%) | 47 (11.4%) | 30 (13.6%) | 87 (12.2%) | 119 (11.6%) | 60 (10.7%) | 388 (11.5%) |
| Gabapentinoidsi | 35 (8.0%) | 51 (12.4%) | 28 (12.7%) | 83 (11.6%) | 127 (12.4%) | 91 (16.3%) | 415 (12.3%) |
| Referrals to other health care services | |||||||
| Allied healthi | 71 (16.3%) | 67 (16.3%) | 45 (20.4%) | 126 (17.6%) | 180 (17.5%) | 76 (13.6%) | 565 (16.8%) |
| Medical specialistsi | 134 (30.8%) | 125 (30.4%) | 63 (28.5%) | 224 (31.3%) | 329 (32.1%) | 189 (33.8%) | 1064 (31.6%) |
| Surgical specialistsi | 85 (19.5%) | 91 (22.1%) | 55 (24.9%) | 168 (23.5%) | 263 (25.6%) | 122 (21.8%) | 784 (23.3%) |
| Mental health and addictioni | 34 (7.8%) | 32 (7.8%) | 25 (11.3%) | 54 (7.5%) | 89 (8.7%) | 54 (9.6%) | 288 (8.6%) |
Abbreviations: OME = oral morphine equivalents; NSAIDs = nonsteroidal anti-inflammatory drugs.
Groups were labeled according to low (average daily OME <50 mg), medium (between 50 and 100 mg), and high dose (>100 mg).
Age at cohort entry. Age was provided in 5-year brackets.
There were missing data in one patient record.
Beneficiary includes those who hold one of the following: (1) Commonwealth Seniors Health Card, (2) Department of Veterans’ Affairs Card, (3) Healthcare Card, or (4) Pensioner Concession Card. Where no beneficiary status was recorded, patients were assumed to be non-beneficiary.
Remoteness was measured as a dichotomous variable of patients in “major cities of Australia” and in “regional and remote Australia.”
Socio-Economic Indexes for Areas deciles (SEIFA) is a geographically derived socioeconomic index based on patient’s postcode of residence. SEIFA is ranked on a scale of 1 to 10, with 1 indicating the most disadvantaged.
Prescription of strong opioids with or without concurrent prescription of weak opioids.
Prescription of long-acting opioids with or without concurrent prescription of short-acting opioids.
Single level was reported for binary variables.
During the 6-month baseline period, 68.7% were prescribed strong opioids (with or without concurrent prescription of weak opioids), and 83.9% were prescribed long-acting opioids (with or without concurrent prescription of short-acting opioids). Three-quarters were prescribed one type of opioid (73.0%) (eg, buprenorphine only or oxycodone only). Most were prescribed oral formulations (82.5%), and more than one-quarter of the patients were prescribed transdermal opioid formulations (25.7%). Almost 1 in 5 patients (17.3%) had an average daily OME of >100 mg during the 6 months before commencing a taper.
Approximately two-thirds had a diagnosis of depression or anxiety (64.8%) in the year before commencing an opioid taper. The most frequent referral category was medical specialists (31.6%), which included anesthetists and pain specialists, and the least frequent was referral to mental health and addiction services (8.6%).
Trajectories of opioid tapering
Six distinct opioid tapering trajectories were identified. Three completed tapering trajectories emerged: (1) low dose / completed taper (12.9%), (2) medium dose / faster taper (12.2%), and (3) medium dose / gradual taper (6.5%). Three persistent opioid trajectories emerged: (1) low dose / noncompleted taper (21.3%), (2) medium dose / noncompleted taper (30.4%), and (3) high dose / noncompleted taper (16.7%) (Figure 4). Means of the average daily OME during the 6-month baseline period for each trajectory were 14.90 mg and 22.93 mg for the low dose / completed and low dose / noncompleted taper groups, respectively; 67.36 mg, 68.09 mg, and 57.27 mg for the medium dose / faster, medium dose / gradual, and medium dose / noncompleted taper groups, respectively; and 162.34 mg for the high dose / noncompleted taper group. Of those who commenced a taper, approximately one-third completed the taper (31.7%). The taper was completed in 37.8% and 38.1% of those commencing tapers from a low and medium dose, respectively. The model did not identify trajectories with completed tapers from a high opioid dose.
Figure 4.
Trajectories of opioid dose in the 6 months before and 12 months after the commencement of a taper. The average daily oral morphine equivalents (OME) of individuals in each trajectory were averaged and graphed on a non-logarithmic scale. Periods 1–6 indicate overlapping 90-day periods during the 6-month baseline period, and periods 7–16 indicate that during the 12-month follow-up period. Index date was defined as the point at which there was >10% reduction in the opioid dose. As consecutive periods are overlapping, the index date falls between periods 5 and 7 (shaded in grey).
Characteristics associated with opioid tapering trajectories
Low dose / completed taper
Patients who completed a taper from a low dose were less likely to be concession beneficiaries (relative risk ratio [RRR], 0.685; 95% confidence interval [CI], 0.509–0.923), less likely to be prescribed long-acting opioids (RRR, 0.086; 95% CI, 0.062–0.121), less likely to be prescribed ≥2 types of opioids (RRR, 0.223; 95% CI, 0.144–0.347), and less likely to be referred to surgical specialists (RRR, 0.736; 95% CI, 0.543–0.988) than were those in the medium dose / noncompleted tapering trajectory (reference group) (Table 2, Figure S1).
Table 2.
Demographic and clinical characteristics of patients associated with opioid tapering trajectories.
| Characteristics | Low dose / completed taper vs Medium dose / noncompleted taper (ref) | Medium dose / faster taper vs Medium dose / noncompleted taper (ref) | Medium dose / gradual taper vs Medium dose / noncompleted taper (ref) | Low dose / noncompleted taper vs Medium dose / noncompleted taper (ref) | High dose / noncompleted taper vs Medium dose / noncompleted taper (ref) |
|---|---|---|---|---|---|
| RRRa (95% CI) | RRRa (95% CI) | RRRa (95% CI) | RRRa (95% CI) | RRRa (95% CI) | |
| Age | |||||
| 19–38 years | Reference | Reference | Reference | Reference | Reference |
| 39–58 years | 0.639 (0.398–1.024) | 0.880 (0.538–1.438) | 0.861 (0.485–1.527) | 0.722 (0.480–1.084) | 1.282 (0.850–1.932) |
| 59–78 years | 0.961 (0.597–1.546) | 0.986 (0.602–1.617) | 0.705 (0.392–1.269) | 0.902 (0.600–1.355) | 0.845 (0.553–1.291) |
| ≥79 years | 1.695 (0.998–2.878) | 1.603 (0.940–2.735) | 0.983 (0.515–1.874) | 1.248 (0.797–1.953) | 0.346 (0.200–0.597) |
| Sex | |||||
| Male | Reference | Reference | Reference | Reference | Reference |
| Femalec | 1.152 (0.890–1.492) | 1.067 (0.836–1.362) | 1.032 (0.758–1.405) | 1.248 (1.010–1.541) | 0.922 (0.740–1.150) |
| Concessional beneficiary status | |||||
| Non-beneficiary | Reference | Reference | Reference | Reference | Reference |
| Beneficiaryc | 0.685 (0.509–0.923) | 0.864 (0.643–1.160) | 1.203 (0.815–1.777) | 0.848 (0.659–1.091) | 1.161 (0.885–1.524) |
| Remoteness | |||||
| Regional and remote Australia | Reference | Reference | Reference | Reference | Reference |
| Major cities of Australiac | 1.365 (0.979–1.903) | 1.088 (0.811–1.460) | 1.013 (0.690–1.486) | 1.409 (1.083–1.835) | 1.120 (0.857–1.463) |
| SEIFA b | 1.003 (0.953–1.055) | 0.987 (0.940–1.035) | 1.067 (1.001–1.137) | 1.009 (0.968–1.052) | 1.009 (0.965–1.054) |
| Opioid characteristics | |||||
| Weak | Reference | Reference | Reference | Reference | Reference |
| Strongc | 1.034 (0.779–1.372) | 1.122 (0.850–1.481) | 1.007 (0.709–1.429) | 1.444 (1.138–1.831) | 1.344 (1.027–1.760) |
| Short-acting | Reference | Reference | Reference | Reference | Reference |
| Long-actingc | 0.086 (0.062–0.121) | 0.929 (0.586–1.473) | 0.762 (0.442–1.314) | 0.195 (0.142–0.266) | 1.891 (1.111–3.217) |
| 1 type of opioid | Reference | Reference | Reference | Reference | Reference |
| ≥2 types of opioidsc | 0.223 (0.144–0.347) | 0.679 (0.513–0.900) | 0.934 (0.662–1.319) | 0.434 (0.333–0.566) | 2.154 (1.711–2.712) |
| Mental health diagnosis | |||||
| Depression and anxietyc | 1.116 (0.859–1.449) | 1.275 (0.988–1.644) | 1.190 (0.858–1.650) | 0.920 (0.744–1.139) | 1.353 (1.051–1.743) |
| Substance use disorderc | 1.280 (0.749–2.185) | 1.088 (0.641–1.848) | 1.334 (0.736–2.417) | 0.701 (0.423–1.161) | 0.988 (0.630–1.549) |
| Sleep disorderc | 0.964 (0.733–1.270) | 0.781 (0.596–1.024) | 0.661 (0.463–0.945) | 0.831 (0.661–1.044) | 1.137 (0.899–1.436) |
| Concurrent analgesics use | |||||
| NSAIDsc | 0.979 (0.717–1.336) | 0.966 (0.720–1.294) | 0.766 (0.521–1.127) | 1.156 (0.905–1.477) | 1.057 (0.821–1.361) |
| Paracetamolc | 0.962 (0.648–1.428) | 0.894 (0.619–1.291) | 1.170 (0.753–1.818) | 1.035 (0.759–1.411) | 0.987 (0.698–1.395) |
| Gabapentinoidsc | 0.799 (0.524–1.218) | 0.992 (0.696–1.414) | 0.980 (0.627–1.532) | 1.013 (0.744–1.379) | 1.180 (0.869–1.601) |
| Referral to specialists | |||||
| Allied healthc | 0.956 (0.686–1.333) | 0.915 (0.666–1.256) | 1.222 (0.838–1.782) | 0.986 (0.756–1.286) | 0.840 (0.618–1.143) |
| Medical specialistsc | 1.036 (0.791–1.355) | 0.976 (0.756–1.260) | 0.847 (0.608–1.178) | 1.021 (0.821–1.270) | 1.182 (0.936–1.492) |
| Surgical specialistsc | 0.736 (0.543–0.998) | 0.857 (0.647–1.136) | 1.000 (0.706–1.418) | 0.945 (0.745–1.198) | 0.850 (0.654–1.104) |
| Mental health and addictionc | 0.920 (0.582–1.456) | 0.957 (0.621–1.475) | 1.316 (0.811–2.138) | 0.910 (0.627–1.323) | 0.958 (0.658–1.393) |
Abbreviations: Ref = reference group; RRR = relative risk ratio; CI = confidence interval; NSAIDs = nonsteroidal anti-inflammatory drugs.
Bold values indicate statistical significance at P < .05.
Adjusted for all covariates listed in this table.
Socio-Economic Indexes for Areas deciles (SEIFA) was included as a continuous variable.
Covariates included as a binary variable.
Medium dose / faster taper and medium dose / gradual taper
Patients who completed a faster taper from a medium dose were less likely to be prescribed ≥2 types of opioids (RRR, 0.679; 95% CI, 0.513–0.900) than were those in the medium dose / noncompleted tapering trajectory (reference group) (Table 2, Figure S1). Patients who completed a gradual taper from a medium dose were more likely to have higher SEIFA scores (indicating higher geographically derived SES; RRR, 1.067; 95% CI, 1.001–1.137) and less likely to have sleep disorders (RRR, 0.661; 95% CI, 0.463–0.945) than were those in the reference group. Additional dichotomous logistic regression analysis comparing those in the gradual and faster tapering trajectories commencing tapers from medium doses showed that those who completed gradual tapers were more likely to have higher SEIFA scores (odds ratio, 1.086; 95% CI, 1.008–1.171) than were those who completed faster tapers from comparable doses (Table 3, Figure S2).
Table 3.
Demographic and clinical differences between patients completing gradual and faster opioid taper.
| Characteristics | Medium dose / gradual taper vs Medium dose / faster taper (ref) |
|---|---|
| Odds ratioa (95% CI) | |
| Age | |
| 19–38 years | Reference |
| 39–58 years | 1.061 (0.531–2.120) |
| 59–78 years | 0.749 (0.370–1.517) |
| ≥79 years | 0.646 (0.306–1.364) |
| Sex | |
| Male | Reference |
| Femalec | 0.950 (0.668–1.351) |
| Concessional beneficiary status | |
| Non-beneficiary | Reference |
| Beneficiaryc | 1.440 (0.920–2.254) |
| Remoteness | |
| Regional and remote Australia | Reference |
| Major cities of Australiac | 0.956 (0.615–1.487) |
| SEIFA b | 1.086 (1.008–1.171) |
| Opioid characteristics | |
| Weak opioid | Reference |
| Strong opioidc | 0.879 (0.590–1.309) |
| Short-acting opioid | Reference |
| Long-acting opioidc | 0.793 (0.424–1.481) |
| 1 type of opioid | Reference |
| ≥2 types of opioidsc | 1.404 (0.935–2.109) |
| Mental health diagnosis | |
| Depression and anxietyc | 0.934 (0.639–1.365) |
| Substance use disorderc | 1.256 (0.608–2.593) |
| Sleep disorderc | 0.866 (0.575–1.304) |
| Concurrent analgesics use | |
| NSAIDsc | 0.774 (0.494–1.213) |
| Paracetamolc | 1.342 (0.800–2.251) |
| Gabapentinoidsc | 1.019 (0.609–1.705) |
| Referral to specialists | |
| Allied healthc | 1.358 (0.865–2.133) |
| Medical specialistsc | 0.841 (0.573–1.235) |
| Surgical specialistsc | 1.118 (0.742–1.685) |
| Mental health and addictionc | 1.407 (0.782–2.532) |
Abbreviations: Ref = reference group; RRR = relative risk ratio; CI = confidence interval; NSAIDs = nonsteroidal anti-inflammatory drugs.
Bold values indicate statistical significance at P < .05.
Adjusted for all covariates listed in this table.
Socio-Economic Indexes for Areas deciles (SEIFA) was included as a continuous variable.
Covariates included as a binary variable.
Low dose / noncompleted taper
Patients who commenced a taper from a low dose but didn’t completed the taper were more likely to be female (RRR, 1.248; 95% CI, 1.010–1.541) and more likely to reside in major cities (RRR, 1.409; 95% CI, 1.083–1.835) than were those in the medium dose / noncompleted tapering trajectory (reference group) (Table 2, Figure S1). Patients in the low dose / noncompleted tapering trajectory were more likely to be prescribed strong opioids (RRR, 1.444; 95% CI, 1.138–1.831), less likely to be prescribed long-acting opioids (RRR, 0.195; 95% CI, 0.142–0.266), and less likely to prescribed ≥2 types of opioids (RRR, 0.434; 95% CI, 0.333–0.566) than were those in the reference group.
High dose / noncompleted taper
Patients who commenced a taper from a high dose but didn’t complete the taper were less likely to be ≥79 years of age (RRR, 0.346; 95% CI, 0.200–0.597) than were those in the medium dose / noncompleted tapering trajectory (reference group) (Table 2, Figure S1). Patients in the high dose / noncompleted tapering trajectory were more likely to be prescribed strong opioids (RRR, 1.344; 95% CI, 1.027–1.760) and long-acting opioids (RRR, 1.891; 95% CI, 1.111–3.217) than were those in the reference group. Patients in high dose / noncompleted tapering trajectory were the only ones who were more likely to be prescribed ≥2 types of opioids (RRR, 2.154; 95% CI, 1.711–2.712) and more likely to have depression or anxiety (RRR, 1.353; 95% CI, 1.051–1.743) than were patients in the reference group.
Discussion
Six distinct tapering trajectories were identified and characterized according to dosage level and taper completion. Patients completing gradual tapers from medium doses were more likely to have higher geographically derived SES than were those who completed faster tapers and those who didn’t complete tapers from medium doses. Patients in both the low dose / noncompleted and high dose / noncompleted trajectories were more likely to be prescribed strong opioids than were those who completed tapers from medium doses, but only those in the high dose / noncompleted trajectory were more likely to be prescribed long-acting opioids and ≥2 types of opioids and more likely to have a diagnosis of depression or anxiety than were those who completed tapers from medium doses.
Our study found that in primary care settings, twice as many patients discontinued opioids at a faster rate, compared with tapering at a gradual rate from a similar dosage level, with few differences between these groups, other than SES, that might explain the difference in tapering trajectories. This novel finding raises critical questions that can drive future research on opioid deprescribing. It is still unknown whether those tapers were conducted at a faster rate for justifiable reasons, such as patient preference to reduce the duration of a taper for reasons of patient discomfort, or clinician concerns about opioid use disorder.37 It is possible that faster tapers were in response to the recent implementation of a number of national and state-based opioid-related interventions to curb opioid prescribing in Australia.38–39 After the publication of 2016 Centers for Disease Control and Prevention Guideline for Prescribing Opioids for Chronic Pain in the United States, there have been reports of inflexible application of dosage and duration thresholds, abrupt cessation of opioids, and clinician dismissal of patient care.40 Further studies are needed to better understand what drives a faster taper, as well as short-term and long-term health outcomes associated with the differing rates of tapering. This would be especially important given the highest-level recommendation for “gradual tapering of opioids” in the new 2022 Evidence-Based Clinical Practice Guideline for Deprescribing Opioid Analgesics, highlighting the risks of rapidly ceasing opioids, especially for those on long-term opioid therapy.20
Our study identified some key patient characteristics that could influence the rate of opioid tapering. First, those completing tapers at a gradual rate were more likely to have higher geographically derived SES than were those who completed faster tapers and those who didn’t complete tapers from a similar dosage level. Prior studies found clear links between socioeconomic disadvantage and increased rates of opioid prescribing and overdose.41 Opioid tapering is a difficult and complex process that requires greater clinician engagement, provision of non-opioid alternatives, and coordination of multidisciplinary pain management strategies in most circumstances.10 Prior studies also highlighted a need for empathetic chronic pain management and a strong clinician–patient relationship to provide individualized and gradual tapering regimes to prevent withdrawal symptoms.42 We can speculate about whether higher health resource allocation and better clinician engagement and social support correlated with higher SES might facilitate a more gradual taper. Moreover, patients with higher SES are less likely to experience prescribing biases (eg, inadequate pain treatment, not being referred to pain specialists).43 This could also explain our result demonstrating higher geographically derived SES in patients completing gradual tapers. Second, our study also found that those who completed a gradual taper were less likely to have a diagnosis of sleep disorders in the year before taper commencement than were those who didn’t complete tapers. This finding seems logical, given that poor sleep can exacerbate pain and pain can interfere with sleep.44,45 Moreover, sleep disorders are frequently associated with increased risk of psychiatric morbidity and underlying substance use problems,46 which could increase the risk of opioid dependence and reduce the likelihood of completing a taper.
The absence of a completed tapering trajectory from a high dose suggests that patients commencing tapers from high doses less commonly discontinue opioid therapy within 12 months of commencing the taper. Previous literature confirmed that patients who persist on opioid therapy have higher average OME and a greater number of months on opioid therapy.47,48 The common trajectory for these patients was that after the initial drop in their opioid doses during commencement of the taper, their doses rebounded to near pre-taper levels within 3 months of commencing the taper (Figure 4). Further studies are needed to examine whether this trend reflects the difficulty in maintaining a taper for those on higher doses or a greater opioid requirement for pain management. Prescription of strong opioids and multiple types of opioids and diagnosis of depression or anxiety in the months before taper commencement were associated with noncompleted tapers. Prior research also showed that comorbid mental health conditions predict noncompleted tapers.48,49 Moreover, high opioid doses and mental health disorders are risk factors for problematic opioid use.18 Targeted policy and strategic interventions could be required to provide further support for this high-risk group during tapering. These could include further investigation of the medical and psychosocial backgrounds these patients might have to help maximize non-opioid (eg, topical or oral NSAIDs) and nonpharmacological (eg, physiotherapy) therapies and other strategies, such as transition to buprenorphine, a partial agonist opioid.12
Our study found high rates of strong opioid prescriptions across all trajectory groups, with more than half of the patients prescribed strong opioids. Prescription of strong opioids was positively associated with noncompleted taper groups for both low and high doses. That almost two-thirds of those on lower doses didn’t complete tapers demonstrates that tapering might be a challenging process even for patients on lower doses, and especially for those prescribed strong opioids. Further studies examining whether any opioid type (eg, oxycodone, morphine) is associated with tapering outcomes could help to understand this finding. In addition, patients who might have transitioned to another opioid type without a dose change would not have been captured in our trajectory model. Whether transition to a different opioid type is associated with different tapering outcomes could be explored through future research. Our finding that the use of long-acting opioids might be associated with noncompleted tapers could contrast with guidelines that recommend tapering on a single modified-release opioid to ensure delivery of a consistent level of analgesic effects and reduce inter-dose withdrawal symptoms.50 Expert opinions also suggest that short-acting opioids might intensify psychological dependence, as these formulations are required to be taken more frequently.50 Future research could help clarify these relationships, but our findings suggest that the characteristics of opioids prescribed in the 6 months before commencement of a taper could influence the tapering outcome.
Our study used health care data containing one of the largest primary care cohorts of opioid prescribing in Australia and is one of the most detailed analyses of opioid tapering to date in Australia. The dataset enabled examination of all opioid prescriptions, irrespective of subsidy, and assessment of key clinical factors through the rich longitudinal patient-level information available through the dataset.
Our study has some inherent limitations. First, we cannot determine the actual opioid fills or consumption from prescribing data. In addition, some medicines (such as paracetamol or NSAIDs) that are available both by prescription and over the counter might not be fully captured in our data, so their use could be underestimated here. Second, it is possible that diagnoses and referral data are underrepresented because of incomplete data. Furthermore, although our data capture patients’ referrals to other health care specialists or services, we cannot confirm whether patients ultimately received these services or sought help without a referral. For this reason, referral to allied health specialists might be more underestimated than are medical or surgical referrals. Third, although 10% dose reductions were examined over a 90-day period to differentiate between dose reductions due to tapering and those attributed to gaps in prescriptions, there is a possibility that a small proportion of tapers that did not truly represent attempted dose reductions were included in our analysis. However, this limitation applies to all trajectory groups and therefore is not likely to affect our conclusions. Fourth, group-based trajectory modeling classifies all patients into a trajectory, regardless of fit. This means that some patients might be forced into trajectories that do not optimally represent their opioid therapy patterns. However, given high odds of correct classification and average posterior probabilities >0.96 for all groups, the impact of these assumptions might be minimal.
Conclusion
Six distinct opioid tapering trajectories were identified; these included 2 tapering trajectories from low doses (completed and noncompleted taper), 3 trajectories from medium doses (gradual taper, faster taper, and noncompleted taper), and 1 trajectory from high doses (noncompleted taper). A completed tapering trajectory from high doses was not identified. Among those completing tapers from medium doses, twice as many patients tapered at a faster rate than those who tapered at a gradual rate. Demographic and clinical characteristics, such as geographically derived SES and mental health diagnoses, differed among the tapering trajectories. These trajectory models add to our understanding of factors that could contribute to completing opioid tapers. Further studies are needed to examine whether being prescribed any specific opioid (ie, buprenorphine, oxycodone) influences tapering outcome, as well as the clinical outcomes associated with the different tapering trajectories.
Supplementary Material
Acknowledgments
The authors acknowledge Outcome Health and the participating Primary Health Networks (Eastern Melbourne, South Eastern Melbourne, and Gippsland Primary Health Networks) for provision of the data.
All authors contributed to the study conception and design. M.J. cleaned and analyzed the data with assistance from T.X. and S.N. M.J. produced the first draft of the manuscript. All authors reviewed, provided feedback, and approved the final version of the manuscript.
Contributor Information
Monica Jung, Monash Addiction Research Centre, Eastern Health Clinical School, Monash University, Melbourne, VIC 3199, Australia; Centre for Medicine Use and Safety, Faculty of Pharmacy and Pharmaceutical Sciences, Monash University, Melbourne, VIC 3052, Australia.
Ting Xia, Monash Addiction Research Centre, Eastern Health Clinical School, Monash University, Melbourne, VIC 3199, Australia.
Jenni Ilomäki, Centre for Medicine Use and Safety, Faculty of Pharmacy and Pharmaceutical Sciences, Monash University, Melbourne, VIC 3052, Australia; Department of Epidemiology and Preventive Medicine, School of Public Health and Preventive Medicine, Monash University, Melbourne, VIC 3004, Australia.
Christopher Pearce, Melbourne East General Practice Network (trading as Outcome Health), Surrey Hills, VIC 3127, Australia; Department of General Practice, School of Primary and Allied Health Care, Monash University, Melbourne, VIC 3168, Australia.
Suzanne Nielsen, Monash Addiction Research Centre, Eastern Health Clinical School, Monash University, Melbourne, VIC 3199, Australia.
Supplementary material
Supplementary material is available at Pain Medicine online.
Funding
The project is funded by the Australian National Health and Medical Research Council (#2002193). M.J. is a recipient of PhD Scholarship from the Monash Addiction Research Centre.
Conflicts of interest: S.N. has received unrelated untied research funding from Seqirus to examine pharmaceutical opioid-related harms and is a named investigator on a research grant from Indivior for an implementation trial of buprenorphine depot (no funding received). J.I. has grants from Amgen and AstraZenica unrelated to this research topic.
References
- 1. Ju C, Wei L, Man KKC, et al. Global, regional, and national trends in opioid analgesic consumption from 2015 to 2019: a longitudinal study. Lancet Public Health. 2022;7(4):e335-e46. [DOI] [PubMed] [Google Scholar]
- 2. Degenhardt L, Grebely J, Stone J, et al. Global patterns of opioid use and dependence: harms to populations, interventions, and future action. Lancet. 2019;394(10208):1560-1579. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Australian Institute of Health and Welfare (AIHW). Opioid Harm in Australia and Comparisons between Australia and Canada. Cat. No. HSE 210. AIHW; 2018.
- 4. Lalic S, Ilomaki J, Bell JS, Korhonen MJ, Gisev N.. Prevalence and incidence of prescription opioid analgesic use in Australia. Br J Clin Pharmacol. 2019;85(1):202-215. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Penington Institute. Australia’s Annual Overdose Report 2019. Penington Institute; 2019. [Google Scholar]
- 6. Karanges EA, Blanch B, Buckley NA, Pearson SA.. Twenty‐five years of prescription opioid use in Australia: A whole‐of‐population analysis using pharmaceutical claims. Br J Clin Pharmacol. 2016;82(1):255-267. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Busse JW, Wang L, Kamaleldin M, et al. Opioids for chronic noncancer pain: a systematic review and meta-analysis. JAMA. 2018;320(23):2448-2460. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Krebs EE, Gravely A, Nugent S, et al. Effect of opioid vs nonopioid medications on pain-related function in patients with chronic back pain or hip or knee osteoarthritis pain: the SPACE randomized clinical trial. JAMA. 2018;319(9):872-882. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Chou R, Turner JA, Devine EB, et al. The effectiveness and risks of long-term opioid therapy for chronic pain: a systematic review for a National Institutes of Health Pathways to Prevention Workshop. Ann Intern Med. 2015;162(4):276-286. [DOI] [PubMed] [Google Scholar]
- 10. Glare P, Ashton-James C, Han E, Nicholas M.. Deprescribing long-term opioid therapy in patients with chronic pain. Intern Med J. 2020;50(10):1185-1191. [DOI] [PubMed] [Google Scholar]
- 11. Busse JW, Craigie S, Juurlink DN, et al. Guideline for opioid therapy and chronic noncancer pain. CMAJ. 2017;189(18):E659-E666. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. 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]
- 13. Tardif H, Hayes C, Allingham SF.. Opioid cessation is associated with reduced pain and improved function in people attending specialist chronic pain services. Med J Aust. 2021;214(9):430-432. [DOI] [PubMed] [Google Scholar]
- 14. Dowell D, Compton WM, Giroir BP.. Patient-centered reduction or discontinuation of long-term opioid analgesics: the HHS guide for clinicians. JAMA. 2019;322(19):1855-1856. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Frank JW, Lovejoy TI, Becker WC, et al. Patient outcomes in dose reduction or discontinuation of long-term opioid therapy: a systematic review. Ann Intern Med. 2017;167(3):181-191. [DOI] [PubMed] [Google Scholar]
- 16. 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. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Coffin PO, Rowe C, Oman N, et al. Illicit opioid use following changes in opioids prescribed for chronic non-cancer pain. PLoS One. 2020;15(5):e0232538. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Campbell G, Noghrehchi F, Nielsen S, et al. Risk factors for indicators of opioid-related harms amongst people living with chronic non-cancer pain: findings from a 5-year prospective cohort study. EClinicalMedicine. 2020;28:100592. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Kalkman GA, Kramers C, van Dongen RT, van den Brink W, Schellekens A.. Tackling rising numbers of opioid prescriptions users. Lancet Public Health. 2020;5(3):e139. [DOI] [PubMed] [Google Scholar]
- 20. Langford AS, Lin CWC, Bero L, et al. Evidence-Based Clinical Practice Guideline for Deprescribing Opioid Analgesics. The University of Sydney; 2022. [Google Scholar]
- 21. NHS. Opioid deprescribing for persistent non-cancer pain. V1.0. 2020. Accessed April 12, 2023. https://www.nottsapc.nhs.uk/media/1564/opioid_deprescribing.pdf
- 22. eTG Complete. The role of analgesics in chronic noncancer pain [Internet]. Therapeutic Guidelines Limited. 2021. Accessed April 12, 2023. Https://tgldcdp.tg.org.au.acs.hcn.com.au/viewTopic?etgAccess=true&guidelinePage=Pain%20and%20Analgesia&topicfile=role-analgesics-chronic-noncancer-pain
- 23. Edlund MJ, Martin BC, Devries A, Fan MY, Braden JB, Sullivan MD.. Trends in use of opioids for chronic noncancer pain among individuals with mental health and substance use disorders: the TROUP study. Clin J Pain. 2010;26(1):1-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Hamilton M, Kwok WS, Hsu A, et al. Opioid deprescribing in patients with chronic noncancer pain: a systematic review of international guidelines. Pain. 2023;164(3):485-493. [DOI] [PubMed] [Google Scholar]
- 25. Jung M, Cangadis‐Douglass H, Nielsen S, et al. What are the research priorities for optimising the safe and effective use of opioids in Australian general practice? Drug Alcohol Rev. 2023;42(3):604-613. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Nielsen S, Buchbinder R, Pearce C, et al. Cohort profile: using primary care data to understand Opioid Prescribing, Policy Impacts and Clinical Outcomes (OPPICO) in Victoria, Australia. BMJ Open. 2023;13(5):e067746. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Cangadis-Douglass H, Jung M, Xia T, et al. Using primary care data to understand opioid prescribing, policy impacts and clinical outcomes: a protocol for the OPPICO study. Res Social Adm Pharm. 2022;18(12):4129-4137. [DOI] [PubMed] [Google Scholar]
- 28. Nielsen S, Degenhardt L, Hoban B, Gisev N.. A synthesis of oral morphine equivalents (OME) for opioid utilisation studies. Pharmacoepidemiol Drug Saf. 2016;25(6):733-737. [DOI] [PubMed] [Google Scholar]
- 29. Langan SM, Schmidt SA, Wing K, et al. The reporting of studies conducted using observational routinely collected health data statement for pharmacoepidemiology (RECORD-PE). BMJ. 2018;363: K 3532. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Nataraj N, Strahan AE, Guy GP Jr., Losby JL, Dowell D.. Dose tapering, increases, and discontinuity among patients on long-term high-dose opioid therapy in the United States, 2017-2019. Drug Alcohol Depend. 2022;234:109392. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. 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. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Hartung DM, Markwardt S, Johnston K, et al. Association between treatment setting and outcomes among Oregon Medicaid patients with opioid use disorder: a retrospective cohort study. Addict Sci Clin Pract. 2022;17(1):45-41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Zheng D, Brett J, Daniels B, Buckley N, Pearson SA, Schaffer AL.. Potentially inappropriate benzodiazepine use in Australian adults: a population-based study (2014-2017). Drug Alcohol Rev. 2020;39(5):575-582. [DOI] [PubMed] [Google Scholar]
- 34. Nagin DS, Nagin D.. Group-Based Modeling of Development. Harvard University Press; 2005. [Google Scholar]
- 35. Safescript. Patient Alerts. Department of Health, State of Victoria, Australia; 2019. Accessed April 12, 2023. https://help.safescript.vic.gov.au/help-topics/patients-alerts.htm
- 36. Shrestha N. Detecting multicollinearity in regression analysis. AJAMS. 2020;8(2):39-42. [Google Scholar]
- 37. Bedford T, Kisaalita N, Haycock NR, et al. Attitudes toward a pre-authorized concealed opioid taper: a qualitative analysis of patient and clinician perspectives. Front Psychiatry. 2022;13:820357. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Prathivadi P, Luckett T, Barton C, Holliday S, Mazza D.. General practitioner attitudes towards systems-level opioid prescribing interventions: a pooled secondary qualitative analysis. Aust J Gen Pract. 2021;50(5):309-316. [DOI] [PubMed] [Google Scholar]
- 39. Wilson H, Armstrong F, Thong I.. Opioid prescribing: what do the 2020 PBS changes mean for GPs. Med Today. 2020;21(9):21-28. [Google Scholar]
- 40. Dowell D, Haegerich T, Chou R.. No shortcuts to safer opioid prescribing. N Engl J Med. 2019;380(24):2285-2287. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Adewumi AD, Maravilla JC, Alati R, et al. Pharmaceutical opioids utilisation by dose, formulation, and socioeconomic status in Queensland, Australia: a population study over 22 years. Int J Clin Pharm. 2021;43(2):328-339. [DOI] [PubMed] [Google Scholar]
- 42. Altekruse SF, Cosgrove CM, Altekruse WC, Jenkins RA, Blanco C.. Socioeconomic risk factors for fatal opioid overdoses in the United States: findings from the Mortality Disparities in American Communities Study (MDAC). PLoS One. 2020;15(1):e0227966. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Atkins N, Mukhida K.. The relationship between patients’ income and education and their access to pharmacological chronic pain management: a scoping review. Can J Pain. 2022;16(1):142-170. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Haack M, Simpson N, Sethna N, Kaur S, Mullington J.. Sleep deficiency and chronic pain: potential underlying mechanisms and clinical implications. Neuropsychopharmacology. 2020;45(1):205-216. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Cheatle MD, Foster S, Pinkett A, Lesneski M, Qu D, Dhingra L.. Assessing and managing sleep disturbance in patients with chronic pain. Sleep Med Clin. 2016;11(4):531-541. [DOI] [PubMed] [Google Scholar]
- 46. Stein MD, Herman DS, Bishop S, et al. Sleep disturbances among methadone maintained patients. J Subst Abuse Treat. 2004;26(3):175-180. [DOI] [PubMed] [Google Scholar]
- 47. Minegishi T, Garrido MM, Stein M, Oliva EM, Frakt AB.. Opioid discontinuation among patients receiving high-dose long-term opioid therapy in the Veterans Health Administration. J Gen Intern Med. 2020;35(suppl 3):903-909. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48. Lalic S, Gisev N, Bell JS, Korhonen MJ, Ilomaki J.. Predictors of persistent prescription opioid analgesic use among people without cancer in Australia. Brit J Clinical Pharma. 2018;84(6):1267-1278. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49. Smith AY, Kirk JK, Smith PF, Wells BJ.. Mental health and benzodiazepine use among patients on chronic opioid therapy. J Am Board Fam Med. 2021;34(1):99-104. [DOI] [PubMed] [Google Scholar]
- 50. Murphy L, Babaei-Rad R, Buna D, et al. Guidance on opioid tapering in the context of chronic pain: evidence, practical advice and frequently asked questions. Can Pharm J (Ott). 2018;151(2):114-120. [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.