Journal Logo

REVIEW PAPER

A Systematic Review of the Relative Frequency and Risk Factors for Prolonged Opioid Prescription Following Surgery and Trauma Among Adults

Pagé, M. Gabrielle PhD∗,†; Kudrina, Irina MD, CM; Zomahoun, Hervé Tchala Vignon PhD§,¶; Croteau, Jordie MSc; Ziegler, Daniela MSi||; Ngangue, Patrice PhD∗,∗∗; Martin, Elisabeth MSc; Fortier, Maude BSc; Boisvert, Esthelle Ewusi BA; Beaulieu, Pierre MD, PhD††; Charbonneau, Céline‡‡; Cogan, Jennifer MD, MSc†,§§; Daoust, Raoul MD, MSc¶¶,∗∗∗∗; Martel, Marc O. PhD||||; Néron, Andrée BPharm, FOPQ∗∗∗; Richebé, Philippe MD, PhD†††; Clarke, Hance MD, PhD‡‡‡,§§§

Author Information
doi: 10.1097/SLA.0000000000003403

In the United States, 80% and 50% of patients are treated with opioids after surgery1 or trauma,2 respectively. The incidence of prolonged opioid prescription, an ongoing prescription of opioids after the body's healing is well established (see below for details), can reach close to 50% of patients.3 These rates are alarming considering that long-term opioid use is associated with multiple adverse effects,4 including risk for misuse, opioid-induced hyperalgesia, mood disorders, and endocrine disturbances, to name but a few.5 The primary rationale for opioid prescription postoperatively is to control acute pain. A comparative study showed, however, that the North American hospitals prescribe postsurgical opioids to 98.3% of their patients compared with 70.2% of patients in European institutions; yet, the mean worse pain scores are lower among European patients (5.4/10) compared with North American patients (7.4/10).6 In a study describing successful postoperative nonopioid pain management among Dutch patients with ankle fracture, results suggested that pain and treatment satisfaction could be culturally mediated.7 In addition, data suggest that psychosocial factors play an important role, perhaps even more important than pain intensity, in driving opioid prescriptions.8 It is therefore important to optimise nonpharmacological pain management and to identify risk factors for the unwarranted opioid use.

Understanding of the underlying risks factors for prolonged opioid prescription could contribute to the improvement in the preoperative patient preparation process as well as postoperative monitoring of recovery and analgesia. These early interventions could potentially mitigate the risks and/or prevent the transition to unwarranted long-term opioid use. However, no knowledge on what constitutes the most common risk, protective and preventive factors for prolonged opioid prescriptions in different populations has been synthesized systematically.9 Only one narrative review conducted on the use of opioid treatment9 addressed alternatives to opioids during the postsurgical recovery period. The authors, however, omitted several relevant articles,10,11 and did not integrate results across studies, leading to potential reporting biases.12

Considering the limited data synthesis on the relative frequency and risk factors for prolonged opioid therapy following surgery and trauma among adult patients, the objectives of this systematic review are

  • (1) to examine the relative frequency of opioid prescriptions 3 to 6 months (Rx_3–6) and >6 months (Rx_>6) postsurgery or trauma, and
  • (2) to identify the risk factors for Rx_3–6 and Rx_>6 among patients undergoing surgery or trauma.

METHODS

The published protocol13 for this systematic review was registered with the PROSPERO database (registration number CRD42018089907). This review was edited to comply with the PRISMA14 and MOOSE15 reporting guidelines.

Operationalization of Key Constructs

Surgery referred to in-hospital surgical interventions requiring any length of in-hospital stay, including day surgeries. Trauma was defined as an injury to the living tissue that is caused by an external agent or other physical stressor, which requires hospitalization. These populations could be combined for analysis due to the comparable severity of postevent homeostatic disturbances, stress response, and endo-opioid system modulatory reactions, and leading to similarities of postevent opioid prescribing.16

For analysis, we categorized postevent opioid prescription based on its length. Rx_3–6 was defined as ≥1 opioid prescription of any length between 3 and 6 months postevent. The 90-day benchmark is, in many regards, similar to the definition put forward by the CONsortium to Study Opioid Risks and Therapeutics (CONSORT).17 Rx_≥6 was defined as ≥1 opioid prescription of any length >6 months posthospital discharge following surgery/trauma in a patient treated with RX_3–6. Patients with prolonged opioid prescription pre-event were those with opioid prescription ≥1 month before the event. Patients with no/short-term opioid prescription pre-event were those not treated with opioids in the months before surgery (except for the 30 d immediately before the indexed event).18 A minimum of a 3-month period must have been examined to categorize patients.

Eligibility Criteria

Both prospective and retrospective studies were considered for inclusion if they used an experimental, quasiexperimental, observational, or a mixed design, and examined adult populations (aged 18 y or older) undergoing either in-hospital surgery (including day surgery) or treated for trauma requiring hospitalization. The studies had to contain information regarding the operationalization of the postevent opioid prescription definition so that the opioid prescription could be linked back to the surgery or trauma (eg, continuous prescription, time to discontinuation, multiple opioid prescriptions within a specified timeframe). In addition, the studies had to account for the pre-event opioid prescription status. All opioid agonists or partial agonists administered using pure or mixed formulations, and all systemic routes of administration were considered.

We excluded studies with methadone or buprenorphine maintenance therapy, those with unclear time of initiation or therapy purpose, with active cancer-related pain or end-of-life care, when another trauma or surgical procedure during a follow-up period were reported, with <3-month follow-up time after the indexed event or an unknown pre-event opioid prescription status. Studies with mixed populations based on one or more of the inclusion criteria were retained if >75% of patients met each criterion, as was used in a previous systematic review on long-term effectiveness of opioid therapy.19

Data Sources and Search Strategy

We conducted a systematic search of 6 databases (MEDLINE, Pubmed, CINAHL, PsycINFO, EMB Review, and EMBASE) between January 1998 and April 2018. This time frame corresponds to the most significant shift in opioid prescribing practice since the first consensus statement issued by the American Academy of Pain Medicine and the American Pain Society.20 An additional manual search included references listed in already identified studies, relevant review articles, and journals’ tables of content (British Journal of Anaesthesia, PAIN, Anesthesiology, Anesthesia & Analgesia, Trauma and Acute Care Surgery, Annals of Emergency Medicine, Emergency Medicine International, and European Journal of Trauma and Emergency surgery). We also searched gray literature (ie, Google Scholar, Pro Quest Dissertation and Theses, and published reports; May 2018). Via Google Scholar, we searched relevant studies with distinct populations (ie, elderly, women, young adults, indigenous people, and individuals suffering from mental illness). We used a snowballing technique to identify other potentially relevant articles.

Research librarian (DZ) and field experts (GP, IK) developed the search strategy that used database-specific terms and terminology derived for the following conceptual groups: (1) opioid; (2) surgery or trauma; (3) prolonged/long-term/frequency/prevalence/incidence or risk factors. For each database, both controlled vocabulary and free-text searching were used. Terms were combined with limits for English or French-language publications (see Supplementary Table 1, http://links.lww.com/SLA/B673 for a list of potentially relevant articles that were excluded). The search strategy is presented in the Supplementary Table 2, http://links.lww.com/SLA/B673. The authors of studies of interest that lacked data to determine eligibility or other measures needed for this review were contacted to obtain the necessary information.

Study Selection and Data Extraction

Management of the references was done in the EndNote X8 (Clarivate Analytics, Philadelphia, PA). The identification of duplicate studies, study selection, and data extraction were completed using the DistillerSR software (DistillerSR, Evidence Partners, Ottawa, Canada). Five individuals (MGP, PN, EM, EEB, and MF) participated in the article selection and data extraction processes. Two independent reviewers assessed each record at every screening level. In case of disagreements, a third reviewer (IK) was involved. Inter-rater reliability (kappa statistic) was calculated for study selection. Data from selected articles were extracted using an adapted version of the Cochrane EPOC Data Abstraction Form and Data Extraction Instructions.21 Recorded study information concerned population and study characteristics, exposure and outcome characteristics, results on association of measures of interest, and results applicability. The data extraction was reviewed by HZ and JC.

Risk of Bias Assessment

The risk of bias was assessed independently by 2 reviewers. Disagreements were resolved through discussion and consensus, and with help of the third reviewer if necessary. Quality assessment was performed using the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies.22

Data Analysis and Synthesis

Definition and Categorization of Variables

Given the heterogeneity in the types and definitions of the risk factors, variables were renamed and categorized a posteriori by consulting clinicians and researchers from the field of interest, and a biostatistician. The 2 main dichotomous outcomes, RX_3–6 and RX_>6, were analyzed separately with respect to the potential risk factors reported in selected studies and based on pre-event opioid prescription status (no/short-term opioid prescription pre-event vs prolonged opioid prescription pre-event). We examined potential risk factors based on several categories: sociodemographic, health behaviors, baseline medication use, musculoskeletal disorders, neurological disorders, and psychiatric disorders. We also examined type and severity of procedure/trauma, peri-event medication use and procedures, comorbidities, micro- or macrovascular disorders, endocrine disorders, gastrointestinal disorders, respiratory diseases, and systemic/rheumatological disorders.

Analyses

We included only results from the study samples with ≥75% of patients with no/short-term opioid prescription pre-event or ≥ 75% of patients with prolonged opioid prescription pre-event.

Data were analyzed using a narrative approach. Descriptive statistics are reported in the form of median (range) when possible. All outcomes were analyzed and reported in a dichotomous form, effect size reported were odds ratio (OR), risk ratio (RR) and hazard ratio (HR) with 95% confidence interval (CI). Due to the heterogeneity in identification and measurement of risk factors and outcome definitions, meta-analysis was not performed.

RESULTS

Study Selection and Study Characteristics

The search identified 10,003 unique records. Thirty-five studies3,10,11,18,23–53 were included in the systematic review (see Fig. 1). Cohen's kappa coefficient (inter-rater agreement on full-text eligibility) was equal to 0.73. One32 of 3 authors contacted to determine inclusion in the review and 418,32,47,48 of 11 authors contacted during data extraction to provide details of measured risk factors responded.

FIGURE 1
FIGURE 1:
PRISMA flow diagram. Adapted from: Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med. 2009;6:e1000097.

Study characteristics are presented in Supplementary Table 3 (Supplemental Digital Content, http://links.lww.com/SLA/C142). The 35 studies included 1,398,182 patients (median=6,695 patients; range=6–641,941) who underwent surgery (29 studies; 96.9%) or suffered trauma (6 studies; 3.1%). The most frequently studied surgical populations were patients who underwent the knee, hip, or shoulder arthroplasty.10,25–27,32,33,35,37,38,43,44,52 All included studies were observational, and the majority (26/35 articles) had a retrospective longitudinal design using medicoadministrative databases. Two studies used a prospective study design,39,42 1 study was cross-sectional,40 and 6 studies were retrospective data from medical records.10,37,46,51–53

Relative Frequency of Rx_3–6 and Rx_>6

The relative frequency of Rx_3–6 and Rx_>6 was 12.45 and 22.20 times greater among patients with prolonged opioid prescription pre-event [Rx_3–633–35,37,45 median=50.93% (range: 42.6%–72.0%); Rx_>631,32,39,42,51 median=57.93% (range: 31.3%–88.2%)] compared to patients with no/short-term opioid prescription pre-event [Rx_3–61,12,29,30,32,34,38,40,52 median=4.1% (range: 0.1% = 45.0%); Rx_>611,25,27,30,32,38,47,48 median=2.61% (0.3%–29.0%)], respectively (see Table 1).

TABLE 1
TABLE 1:
Percentage of Patients on Opioid Prescription 3 to 6 months and >6 months Following Surgery/Trauma.

Details of frequency of Rx_3–6 and Rx_>6 according to pre-event opioid prescription status are presented in Figure 2 and definitions of Rx_3–6 and Rx_>6 for each study are presented in Supplementary Table 4, http://links.lww.com/SLA/B673. Few studies examined opioid prescription at multiple time points; generally, their results suggested a decrease in frequency of opioid prescription over time.3,23–25,27,29,47

FIGURE 2
FIGURE 2:
Percentage of patients with postevent opioid prescription based on their pre-event opioid prescription status. These studies only considered percentage of patients who received opioid prescription 3 to 6 months or >6 months postevent among those who initially received an opioid prescription after trauma. Study examining chronic pancreatitis. Three included studies24,43,49 are not shown in Figure 2 because of missing information in the article (eg, overall number of postevent opioid treated patients but not according to pre-event opioid prescription status). Bedard (2017b)27; Bedard (2017)28. Relative risk is classified under its shortest duration (6 months or more is shown as 6 months).

Risk of Bias

There were 3 categories of risk of bias (sample size justification or description of power, level of exposure dose or length of initial opioid prescription, and adjustment for confounding variables) (comorbidities, age, sex, pre-event opioid prescription status) that we assessed for (see Fig. 3) that revealed a great degree of bias. Overall, many studies had only 3/14 criteria unmet,18,27–30,32–34,36,38,41–43,45 whereas others had bias in more than half of the criteria examined46,51 (see Supplementary Table 5, http://links.lww.com/SLA/B673).

FIGURE 3
FIGURE 3:
Evaluation of risk of bias. Criteria do not apply to included studies and as such rated as “‘unclear/not applicable.”

Risk Factors for Rx_3–6 and Rx_>6

Individual risk factors for Rx_3–6 and Rx_>6 are shown in Supplementary Table 6 (Supplemental Digital Content; http://links.lww.com/SLA/C142) for patients with no/short-term opioid prescription pre-event and in Supplementary Table 7 (Supplemental Digital Content; http://links.lww.com/SLA/C142) for patients with prolonged opioid prescription pre-event. The results below highlight key findings.

Sociodemographic Risk Factors

Sex was generally not identified as a risk factor for Rx_3–6/Rx_>6 among patients with no/short-term opioid prescription pre-event [8/10 studies3,11,23,28,29,34,36,48 were negative; range of effect sizes (OR, RR, and HR) are shown in Supplementary Table 6; Supplemental Digital Content; http://links.lww.com/SLA/C142] or patients with prolonged opioid prescription pre-event (all 4 studies were negative32,33,37,45; results are shown in Supplementary Table 7 (Supplemental Digital Content; http://links.lww.com/SLA/C142).

Age was inconsistently associated with Rx_3–6/Rx_>6 among both patients with no/short-term opioid prescription,3,23,28,29,36,41,48,49 and those with prolonged opioid prescription pre-event32,33,35,37 [range of effect sizes shown in Supplementary Table 6; Supplemental Digital Content; http://links.lww.com/SLA/C142—multiple types (OR, RR, HR)]. Most studies,3,23,28,29,32,33,36,37,41,48,49 however, used arbitrary age groups that rendered comparisons across studies difficult.

Race was not associated (with the exception of a comparison between white and other/unknown race)41 with Rx_3–6/Rx_>6 among patients with no/short-term opioid prescription pre-event (range of OR: 0.73–1.13; median=0.96).28,41 The only study that examined race among patients with prolonged opioid prescription pre-event did not identify race as a risk factor for Rx_>6.32

Among patients with no/short-term opioid prescription pre-event, results showed that higher income (neighborhood29 or household36) decreased the odds of Rx_3–6 (range of OR: 0.7–0.8).

Health Behaviors

Tobacco dependence was identified as a risk factor for Rx_3–6 (ORs: 1.35, 1.6) among patients with no/short-term opioid prescription pre-event and in one45 out of two37,45 studies among patients with prolonged opioid prescription pre-event (OR: 1.96, 6.94). Results mostly converged to show that alcohol and/or drug dependence is possibly a risk factor for Rx_3–6/Rx_>6 among patients with no/short-term opioid prescription pre-event (significant association in three28,31,48 of the 4 studies; range of OR: 0.90–3.15; median=1.32).28,31,36,48 This does not seem to be the case for patients with prolonged opioid prescription pre-event (range of OR: 0.90–0.95).32,45

Medication Use

Three large studies29,41,48 examined the role of prescribed medications at baseline and the risk of Rx_3–6/Rx_>6 among patients with no/short-term opioid prescription pre-event; only one study examined the role of baseline prescribed medications among patients with prolonged opioid prescription pre-event (with the exception of hypnotic use35).

Current use of anti-depressants was associated with increased risk of Rx_3–6/Rx_>6 among patients with no/short-term opioid prescription pre-event (range of ORs: 1.26–1.65; median = 1.41)29,41,48 and with prolonged opioid prescription pre-event (RR [95% CI] = 1.27 [1.02–1.59])32; past use of anti-depressants was only examined in one study and was not identified as a risk factor for Rx_3–6/Rx_>6 (RR [95% CI] = 1.10 (0.97–1.24))32.

Results generally converged to show that current use of benzodiazepine was a risk factor for Rx_3–6 among patients with no/short-term opioid prescription pre-event (range of ORs: 1.26–1.82; median = 1.56)29,41,48; however, past use of benzodiazepine (used before last year before surgery) is possibly protective against Rx_3–6/Rx_>6 (OR [95% CI] = 0.84 [0.72–0.98]).41 Current or past use of prescribed benzodiazepine was not associated with Rx_>6 among patients with prolonged opioid prescription pre-event (RRs: 1.06–1.09).32

The current or past use of muscle relaxants might be a risk factor for Rx_3–6 among patients with no/short-term opioid prescription pre-event (ORs: 1.17–1.69)41 but not for patients with prolonged opioid prescription pre-event (RRs: 0.90–1.10),32 although this risk factor was only examined in one study in each of the pre-event opioid prescription status categories.

Other medications were examined among patients with prolonged opioid prescription pre-event. Results showed that past (RR [95% CI] =1.13 [1.02–1.26]) but not current (RR [95% CI] = 0.89 [0.59–1.34]) use of antiepileptics32 or NSAID (OR [95% CI] = 1.00 [0.70–1.43])33 were associated with the slightly increased odds of Rx_>6. Hypnotic use appears a risk factor for Rx_3–635 (OR [95% CI] = 2.52 [1.48–4.3]) but not for Rx_>632 (current use: RR [95% CI] = 0.80 [0.55–1.17]; past use: RR [95% CI] = 1.03 [0.91–1.17]).

Preoperative Opioid Prescription

Presence or number of opioid prescription pre-event were associated in all28,31,35,45 but one34 study with increased risk of Rx_3–6/Rx_>6 (range of ORs: 1.66–11.4; median = 3.52). Interestingly, there does not seem to be a dose–effect; 2 studies35,37 examined the oral morphine equivalent incremental doses of pre-event opioid prescriptions and did not find a significant association between increased opioid doses and risk of Rx_3–6 among patients with prolonged opioid prescription pre-event (range of ORs: 1.06–5.97).

Musculoskeletal Conditions

Across several studies, presence of another pain diagnosis affecting musculoskeletal system was inconsistently associated with the risk of Rx_3–6/Rx_>6 independent of the status of pre-event opioid prescription.11,28,32,33,35,36,41,45 There was a substantial heterogeneity in the examination of these covariates, with one study11 examining specific pain diagnoses associated with the surgical procedure (range of HRs: 0.73–1.29; median HR: 1.07), whereas others33,35,45 looking at the presence of specific chronic pain comorbidities (range of ORs: 0.57–2.12; median OR: 1.99), or the overall presence of chronic pain conditions28,32,36,41 (range of ORs: 1.0–1.39; median: 1.18; RR [95% CI] =1.07 [0.96–1.2]).

Neurological Conditions

Migraine was the only neurological conditions examined, and results were inconsistent regarding an association between the presence of migraines and Rx_3–6 among patients with prolonged opioid prescription pre-event (range of ORs: 1.42–1.76; median: 1.59).35,45

Psychiatric Disorders

The presence of a depressive disorder was by far the most studied psychiatric disorder; and results were inconsistent across studies for both, patients with no/short-term opioid prescription pre-event and those with prolonged opioid prescription pre-event. Five studies found that having a diagnosis of depression is associated with an increased risk of Rx_3–6/Rx_>6,11,28,31,45,48 whereas 2 others35,41 did not find such association (range of ORs: 0.92–1.63; median: 1.15; HR [95% CI] = 0.84 [0.77–0.9]).

A diagnosis of anxiety was also examined in 4 studies among patients with no/short-term opioid prescription pre-event; three11,31,41 out of 4 studies11,28,31,41 concluded that the presence of an anxiety disorder did not increase one's risk of Rx_3–6/Rx_>6 (range of ORs: 1.07–1.25; median: 1.12; HR [95% CI] = 0.85 [0.67–1.06]).

Type and Severity of Procedure/Trauma

Type of surgery or trauma was only examined among patients with no/short-term opioid prescription pre-event. Data are very heterogeneous with multiple different types of surgical procedures or trauma studied. None of the factors examined were consistently associated with risks of Rx_3–6/Rx_>6 across studies.

Peri-event Medication Use and Procedures

Results converge to suggest that a greater opioid dose and/or a longer duration of initial postevent prescription could be associated with an increased odds of Rx_3–6 or Rx_>6 (OR=0.96–3.05; median=1.23) among patients with no/short-term opioid prescription pre-event.28,31,41 These characteristics of the initial opioid prescription at hospital discharge were not examined among patients with prolonged opioid prescription pre-event.

Length of hospitalization was examined in 3 studies and results suggest that this factor is not associated with an increased risk of Rx_>6.11,32,33

Comorbidities (Micro-/Macrovascular Disorders, Endocrine Disorders, Gastrointestinal Disorders, Respiratory Diseases, and Systemic/Rheumatological Disorders)

Severity of comorbidities was assessed heterogeneously across studies with some using standard indexes such as the Charlson Comorbidity Index11,28,33 or the Elixhauser Index,32,36 whereas others simply using a comorbidity count.38 The 2 studies that examined overall comorbidity severity among patients with prolonged opioid prescription pre-event failed to show any association of this factor with Rx_>6.32,33 Presence of multiple comorbidities might be a risk factor for Rx_3–6 or Rx_>6 among patients with no/short-term opioid prescription pre-event (statistically significant association in three28,36,38 of 4 studies11,28,36,38 (OR=1.1–2.2; median=1.6; HR range: 0.94–1.06).

Some studies have examined an association between specific comorbidities (endocrine disorders,29,32,35,45 gastrointestinal disorders,35 systemic/rheumatologic disorders,35,45 respiratory diseases,29 and vascular disorders29,35,37) and Rx_3–6/Rx_>6. Results did not suggest that any specific condition could be a risk factor for Rx_3–6 or Rx_>6.

DISCUSSION

Results showed that the median relative frequency of Rx_3–6 and Rx_>6 is much higher among patients with prolonged opioid prescription pre-event (50.9% and 58.5%, respectively) compared with patients with no/short-term opioid prescription pre-event (4.1% and 2.6%, respectively). Quantitative examination of risk factors for Rx_3–6/Rx_>6 was not possible due to substantial heterogeneity across studies. Risk of bias was present in most studies; data were primarily retrospective from the medicoadministrative databases. Results seemed to converge to suggest that the income levels, tobacco dependence, use of anti-depressants, as well as the presence and number of pre-event opioid prescriptions (but not dose) are associated with increased risk of Rx_3–6/Rx_>6. Intuitive, the use of benzodiazepine (current use) or muscle relaxants, and presence of alcohol/drug dependence were found to be potential risk factors for Rx_3–6/Rx_>6 among patients with no/short-term opioid prescription pre-event. Results also converged overall to suggest that sex, race, anxiety, other psychiatric disorders, and length of hospitalization might not be associated with Rx_3–6/Rx_>6. Finally, results were inconsistent across studies regarding the associations between age, presence of chronic pain conditions, depressive disorder and other comorbidities and risks of Rx_3–6/Rx_>6.

Relative Frequency of Rx_3–6 and Rx_>6

Although the definition of Rx_>6 implies that there also be Rx_3–6, not all studies looking at the shorter duration (Rx_3–6), examined opioid use beyond this period. In addition, there was only a partial overlap in studies reporting on populations using Rx_3–6 and Rx_>6. This might explain why the relative median frequency of Rx_>6 is greater.

Although this review's results provide important information regarding the prevalence of Rx_3–6 and Rx_>6 after surgery and trauma, it was not possible to put this information in the context of chronic pain treatment. Indeed, none of the examined studies provided rates of chronic pain in their studied sample at the time that Rx_3–6/Rx_>6 was measured. It is theorized that with time, opioid use might contribute to the process of “pain chronification”54 and opioid induced hyperalgesia,55 thus maintaining or worsening patient pain experience. On the contrary, opioid prescriptions might become the means of managing psychological distress associated with pain.56 The continuous opioid prescribing despite physiological resolution of postoperative or posttrauma injury, which, in turn, might be indicative of pain chronification, psychosocial distress, opioid misuse or use disorder, was not described in the included studies. Our review did not address the appropriateness of continuous opioid prescribing postevent or the development of opioid dependence or other adverse effects associated with long-term opioid use. This would be a necessary next step to the elaboration of preventative and interventional strategies.

Risk Factors for Rx_3–6 and Rx_>6

There was heterogeneity across studies in the types of risk factors examined and how they were measured. Risk factors discussed here are only those that were examined in 2 or more studies given that the purpose of this review is to identify risk factors consistent across studies and different postoperative and posttrauma populations.

Common Risk Factors Among Patients With No/Minimal and Those With Prolonged Opioid Prescription Pre-event.

Tobacco dependence is an interesting and underappreciated risk factor for Rx_3–6, particularly among patients with no/short-term opioid prescription pre-event. This might be linked to higher postoperative pain scores and analgesic requirements in the days following surgery reported by smokers compared to non- or past smokers.57,58 In general, smoking is associated with lower opioid bioavailability59 and longer bone healing,60 both leading to higher doses required for analgesia. Another association could be lower socio-economic status (SES) among smokers, which was not possible to examine. However, lower SES emerged as an independent risk for Rx_3–6. Addressing tobacco use within a biopsychosocial context (eg, smokers typically report greater pain interference and higher rates of unemployment) might prove helpful.61

Antidepressants

The use of antidepressants before surgery was a risk factor for Rx_3–6.29,41 Although some antidepressants can be prescribed for analgesic purposes in neuropathic pain, this is not the case of all antidepressants (eg, SSRIs).62 As such, the use of antidepressants as examined in this review could also be a mood disorder treatment-related or an adjunct preoperative pain management, or even both, which was not distinguished by the authors. In addition, the review did not consistently identify depressive disorder as a risk factor for Rx_3–6/Rx_>6. Perhaps, the observed heterogeneity across studies and populations (type of surgical procedure/trauma, time to a diagnosis of depression) could partially explain this result.

There is an emerging literature on the bidirectional mutual depression-opioid use influences. For several years, Sullivan et al63 have published on a higher prevalence of opioid use among pain patients with mood disorders. Several authors54,64 have published extensively on opioids effects on the hypothalamo–pituitary–adrenal axis through the modulatory actions of the endogenous opioid system, thus affecting brain plasticity and promoting chemical and behavioral dependence on the mood-modifying properties of these drugs. Opioid use could lead to an increased risk of significant dysphoric effects, which could improve temporarily with each opioid administration, thus driving a new onset or relapse in depression and other related mood disorders. This results in a vicious cycle of dose escalation and difficulty stopping the therapy. Long-term opioid therapy is linked to the emergence of treatment-resistant depression,64 where time seems to play a more significant role than the cumulative dose of opioids. Therefore, opioid-sustained or opioid-induced mood symptoms could be difficult to treat in these patients.

Unmatched or Inconsistent Risk Factors Based on Pre-event Opioid Prescription Status

Alcohol/Drug dependence

Most studies examining alcohol or drug dependence identified this variable as a risk factor for Rx_3–6 and Rx_>6 among patients with no/short-term opioid prescription pre-event. This might be due to dysregulations in reward pathways (eg, mesolimbic dopaminergic system65) associated with the use of alcohol, drugs, or opioids. Evidence indicates that substance use problems may, over time, lead to enduring changes in the central nervous system (CNS). In turn, these CNS changes and plasticity may increase susceptibility to the development (ie, onset) of new substance use problems.66,67 Biological predispositions could thus represent one potential explanation as for why patients with a substance use history are more likely to transition to long-term opioid treatment after surgery. The persistence of pain symptoms after surgery could be a contributing factor in long-term opioid therapy, but the literature seems to suggest that a history of any substance use disorder is a stronger predictor of long-term opioid therapy than the pain-related variables.28 It would thus be important to carefully evaluate the risks and benefits of opioid prescribing at discharge in this specific population and consider optimizing all nonopioid pain management options first. Other risk factors identified in this review are the dose and/or length of initial opioid prescription.41 Prescribing high doses of opioids, and/or for many consecutive days at a time might further increase the risk of long-term opioid treatment possibly due to a higher risk of concurrent opioid-induced hyperalgesia and suppression of the neurohormonal analgesic influences.

Importantly, neither the presence nor the absence of alcohol/drug dependence among patients with prolonged opioid prescription pre-event is associated with the risk of Rx_3–6. The risk factors for the development of long-term opioid therapy might thus be different from the risk factors that contribute to the continuation of the opioid therapy.

Benzodiazepine

The situation with the use of benzodiazepine and muscle relaxants is similar. These variables are associated with an increased risk for Rx_3–6/Rx_>6 among patients with no/short-term opioid prescription pre-event but not among those with prolonged opioid prescription pre-event. The concurrent use of benzodiazepines and opioids can lead to serious adverse events. Indeed, the rates of overdose and misuse increase with the concurrent use of both medications compared with either one alone.68 These 2 drugs are frequently coabused, although the physiological mechanisms are unclear due to the lack of empirical evidence.69 It is possible that both have modulatory effects on respective pathways as, for example, benzodiazepines could be altering the pharmacokinetics of opioids.69 This would render patients vulnerable to the persistent postoperative or posttrauma opioid use.

Chronic Pain

The presence of pre-event chronic pain did not seem to be a risk factor for Rx_3–6/Rx_>6. This could be due to the heterogeneity in assessing chronic pain (overall presence of pain vs pain-specific conditions). It is also possible that opioid prescription status pre-event is a stronger risk factor compared to chronic pain. In addition, this association could further be confounded by the severity of pain (interference, intensity), an aspect that was not taken into account in any of the included studies.

Negative Findings

Sociodemographic Characteristics

Most sociodemographic variables examined were overall not or inconsistently associated with Rx_3–6/Rx_>6, including sex and race. This is in contrast with the limited literature on sex differences in opioid prescribing, showing that female patients are more likely to have ongoing opioid therapeutic use compared with men.70 With regard to the racial differences, some studies suggest that white patients are more likely to be using opioids despite lower pain scores as compared with black patients71; however, the issue is also confounded by prescribers’ practices. A recent study conducted in the United States showed racial disparities in discontinuation of persistent opioid therapy as well as in the use urine of testing for opioid misuse.72

Anxiety

Decades of research in this field showed that demographic factors (age, female sex), psychological factors (depression, anxiety, catastrophizing), preoperative pain, and type of surgery are risk the factors for persistent postsurgical pain.73 A surprising result in this review is that anxiety was not identified as a risk factor for Rx_3–6 or Rx_>6.11,28,31,41 Of note, no study examined pain-related anxiety as their data originated mostly from the medicoadministrative databases. Few studies that examined anxiety used only diagnostic codes. Because this methodology identifies only those who were officially diagnosed by their treating teams, it leads to an underrepresentation of anxiety symptoms, and might affect the review results.

Identified gaps

There was a great degree of heterogeneity across studies. Most were limited by the use of medicoadministrative databases, which precluded the examination of specific factors known to be associated with postoperative outcomes (eg, catastrophizing, pain status). There is also a clear lack of a uniform definition of postoperative/posttrauma opioid status. An expert consensus on how to best operationalize postsurgical opioid prescription (eg, define persistent therapy based on time duration, minimum dose, and/or number of prescriptions filled) is warranted. The lack of data on the pain status of patients on Rx_3–6 and Rx_>6 is concerning as this prevents the interpretation of the observed opioid prescription patterns and the development of adequate alternative treatment options for these patients.

Limitations

Bias assessment of the literature revealed a selection bias (retrospective studies) and a misclassification bias (medicoadministrative databases). Minor modifications were made to the study protocol following its publication (see Supplementary Table 8, http://links.lww.com/SLA/B673). Given the medicoadministrative nature of the data, patients or prescribers were not interviewed to confirm that opioids were prescribed for the management of postevent pain; therefore, the Rx_3–6 and Rx_>6 opioid statuses were inferred. Although the treatment and outcome definitions in most studies made this attribution likely, it is possible that in some cases opioids were prescribed after surgery or trauma for an unrelated condition.

Implications

The values observed for the prevalence of Rx_3–6 and Rx_>6 show that it is a serious behavioral health problem and it is imperative to identify relevant risk factors. The definitions of risk factors and outcomes were so heterogeneous that this precluded the conduct of meta-analysis. Nonetheless, this review identified several modifiable risk factors (tobacco dependence and baseline antidepressant use), thus opening doors for the initial preventative interventions targeting populations at risk.

This review found an important gap in our knowledge related to the lack of data on who are these community-dwelling patients and why they would continue opioid use postoperatively and posttrauma, ultimately leading to Rx_3–6 and Rx_>6 (eg, lack of postevent follow-up; access to physical, psychological, and alternative pain management approaches). With the continued advances in the field, we could already envision the development of a set of effective strategies and prevention programs in the community.

In conclusion, the high prevalence of Rx_3–6 and Rx_>6 among patients with prolonged opioid prescription pre-event indicates that more efforts must be deployed to decrease this statistic. This review identified several modifiable risk factors, thus opening doors for preventative interventions targeting populations at risk. The correlational nature of the data, however, does not allow the inference of causality and requires more focused research efforts.

Because of the complexity and multidimensional nature of the identified risk factors, a multidisciplinary team approach, such as a novel transitional pain service,74 should be considered. This model offers a promising venue to manage the risks of long-term opioid therapy. More targeted resource allocation to refine and implement similar models will thousands of patients during the challenging period of transition back to the functional and productive lives.

REFERENCES

1. Wunsch H, Wijeysundera DN, Passarella MA, et al. Opioids prescribed after low-risk surgical procedures in the United States, 2004–2012. JAMA 2016; 315:1654–1657.
2. Chaudhary MA, Schoenfeld AJ, Harlow AF, et al. Incidence and predictors of opioid prescription at discharge after traumatic injury. JAMA Surg 2017; 152:930–936.
3. Al Dabbagh Z, Jansson KA, Stiller CO, et al. Long-term pattern of opioid prescriptions after femoral shaft fractures. Acta Anaesthesiol Scand 2016; 60:634–641.
4. Benyamin R, Trescot AM, Datta S, et al. Opioid complications and side effects. Pain Physician 2008; 11: (2 Suppl.): S105–S120.
5. Henry SG, Wilsey BL, Melnikow J, et al. Dose escalation during the first year of long-term opioid therapy for chronic pain. Pain Med 2015; 16:733–744.
6. Chapman CR, Stevens DA, Lipman AG. Quality of postoperative pain management in American versus European institutions. J Pain Palliat Care Pharmacother 2013; 27:350–358.
7. Helmerhorst GT, Lindenhovius AL, Vrahas M, et al. Satisfaction with pain relief after operative treatment of an ankle fracture. Injury 2012; 43:1958–1961.
8. Darnall BD, Stacey BR, Chou R. Medical and psychological risks and consequences of long-term opioid therapy in women. Pain Med 2012; 13:1181–1211.
9. Hah JM, Bateman BT, Ratliff J, et al. Chronic opioid use after surgery: implications for perioperative management in the face of the opioid epidemic. Anesth Analg 2017; 125:1733–1740.
10. Rozet I, Nishio I, Robbertze R, et al. Prolonged opioid use after knee arthroscopy in military veterans. Anesth Analg 2014; 119:454–459.
11. Schoenfeld AJ, Nwosu K, Jiang W, et al. Risk factors for prolonged opioid use following spine surgery, and the association with surgical intensity, among opioid-naive patients. J Bone Joint Surg Am 2017; 99:1247–1252.
12. Mulrow CD. Rationale for systematic reviews. BMJ 1994; 309:597–599.
13. Page MG, Kudrina I, Zomahoun HTV, et al. Relative frequency and risk factors for long-term opioid therapy following surgery and trauma among adults: a systematic review protocol. Syst Rev 2018; 7:97.
14. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med 2009; 6:e1000097.
15. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000; 283:2008–2012.
16. Molina PE. Opioids and opiates: analgesia with cardiovascular, haemodynamic and immune implications in critical illness. J Intern Med 2006; 259:138–154.
17. Von Korff M, Saunders K, Thomas Ray G, et al. De facto long-term opioid therapy for noncancer pain. Clin J Pain 2008; 24:521–527.
18. Deyo RA, Hallvik SE, Hildebran C, et al. Use of prescription opioids before and after an operation for chronic pain (lumbar fusion surgery). Pain 2018; 159:1147–1154.
19. Chou R, Fanciullo G, Fine P, et al. Clinical guidelines for the use of chronic opioid therapy in chronic noncancer pain. J Pain 2009; 10:113–130.
20. Haddox JD, Joranson DE, Angarola RT, et al. The use of opioids for the treatment of chronic pain: a consensus statement from the American Academy of Pain Medicine and the American Pain Society. Clin J Pain 1997; 13:6–8.
21. Effective Practice and Organisatio nof Care (EPOC). EPOC resources for review authors. 2017. Available at: http://epoc.cochrane.org/epoc-specific-resources-review-authors. Accessed Novembre 13, 2017.
22. National Heart LaBI. Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies. 2014. Available at: http://mixedmethodsappraisaltolpublic.pbworks.com. Accessed November 30, 2017.
23. Al Dabbagh Z, Jansson KT, Stiller CO, et al. No signs of dose escalations of potent opioids prescribed after tibial shaft fractures: a study of Swedish National Registries. BMC Anesthesiol 2014; 14:4s.
24. Anthony CA, Westermann RW, Bedard N, et al. Opioid demand before and after anterior cruciate ligament reconstruction. Am J Sports Med 2017; 45:3098–3103.
25. Bedard NA, DeMik DE, Dowdle SB, et al. Trends and risk factors for prolonged opioid use after unicompartmental knee arthroplasty. Bone Joint J 2018; 100-B: (1 Suppl. A): 62–67.
26. Bedard NA, Pugely AJ, Dowdle SB, et al. Opioid use following total hip arthroplasty: trends and risk factors for prolonged use. J Arthroplasty 2017; 32:3675–3679.
27. Bedard NA, Pugely AJ, Westermann RW, et al. Opioid use after total knee arthroplasty: trends and risk factors for prolonged use. J Arthroplasty 2017; 32:2390–2394.
28. Brummett CM, Waljee JF, Goesling J, et al. New persistent opioid use after minor and major surgical procedures in US adults. JAMA Surg 2017; 152:e170504.
29. Clarke H, Soneji N, Ko DT, et al. Rates and risk factors for prolonged opioid use after major surgery: population based cohort study. BMJ 2014; 348:g1251.
30. Connolly J rd, J, Z, R, M, A, C, W, K, Y, F, B. Predictors of long-term opioid use following lumbar fusion, surgery. Spine 2017; 42:1405–1411.
31. Daoust R, Paquet J, Moore L, et al. Incidence and risk factors of long-term opioid use in elderly trauma patients. Ann Surg 2018; 268:985–991.
32. Hadlandsmyth K, Vander Weg MW, McCoy KD, et al. Risk for prolonged opioid use following total knee arthroplasty in veterans. J Arthroplasty 2018; 33:119–123.
33. Hansen CA, Inacio MCS, Pratt NL, et al. Chronic use of opioids before and after total knee arthroplasty: a retrospective cohort study. J Arthroplasty 2017; 32:811–817.
34. Holman JE, Stoddard GJ, Higgins TF. Rates of prescription opiate use before and after injury in patients with orthopaedic trauma and the risk factors for prolonged opiate use. J Bone Joint Surg Am 2013; 95:1075–1080.
35. Inacio MC, Hansen C, Pratt NL, et al. Risk factors for persistent and new chronic opioid use in patients undergoing total hip arthroplasty: a retrospective cohort study. BMJ Open 2016; 6:e010664.
36. Johnson SP, Chung KC, Zhong L, et al. Risk of prolonged opioid use among opioid-naive patients following common hand surgery procedures. J Hand Surg Am 2016; 41:947–957.
37. Kim KY, Anoushiravani AA, Chen KK, et al. Preoperative chronic opioid users in total knee arthroplasty—which patients persistently abuse opiates following surgery? J Arthroplasty 2018; 33:107–112.
38. Kim SC, Choudhry N, Franklin JM, et al. Patterns and predictors of persistent opioid use following hip or knee arthroplasty. Osteoarthritis Cartilage 2017; 25:1399–1406.
39. Lawrence JT, London N, Bohlman HH, et al. Preoperative narcotic use as a predictor of clinical outcome: results following anterior cervical arthrodesis. Spine 2008; 33:2074–2078.
40. Mohanty S, Lee JS, Ross RA, et al. New persistent opioid use after bariatric surgery. J Am Coll Surg 2017; 225:S123–S1123.
41. Mosher HJ, Hofmeyer BA, Hadlandsmyth K, et al. Predictors of long-term opioid use after opioid initiation at discharge from medical and surgical hospitalizations. J Hosp Med 2018; 13:243–248.
42. Moyer HR, Yezhelyev M, Ingram WL. Bilayer dermal matrix for the treatment of painful burn scars. Wounds 2011; 23:236–242.
43. Mueller KG, Memtsoudis SG, Mariano ER, et al. Lack of association between the use of nerve blockade and the risk of persistent opioid use among patients undergoing shoulder arthroplasty: evidence from the marketscan database. Anesth Analg 2017; 125:1014–1020.
44. Politzer CS, Kildow BJ, Goltz DE, et al. Trends in opioid utilization before and after total knee arthroplasty. J Arthroplasty 2017; 14:14.
45. Qureshi R, Werner B, Puvanesarajah V, et al. Factors affecting long-term postoperative narcotic use in discectomy patients. World Neurosurg 2018; 112:e640–e644.
46. Rios GA, Adams DB, Yeoh KG, et al. Outcome of lateral pancreaticojejunostomy in the management of chronic pancreatitis with nondilated pancreatic ducts. J Gastrointest Surg 1998; 2:223–229.
47. Soneji N, Clarke HA, Ko DT, et al. Risks of developing persistent opioid use after major surgery. JAMA Surg 2016; 151:1083–1084.
48. Sun EC, Darnall BD, Baker LC, et al. Incidence of and risk factors for chronic opioid use among opioid-naive patients in the postoperative period. JAMA Intern Med 2016; 176:1286–1293.
49. Weiss RJ, Montgomery SM, Stiller C-O, et al. Long-term follow-up of opioid use in patients with acetabular fractures. Injury Extra 2012; 43:49–53.
50. Westermann RW, Anthony CA, Bedard N, et al. Opioid consumption after rotator cuff repair. Arthroscopy 2017; 33:1467–1472.
51. Yang CJ, Bliss LA, Freedman SD, et al. Surgery for chronic pancreatitis: the role of early surgery in pain management. Pancreas 2015; 44:819–823.
52. Zarling BJ, Yokhana SS, Herzog DT, et al. Preoperative and postoperative opiate use by the arthroplasty patient. J Arthroplasty 2016; 31:2081–2084.
53. Zwisler ST, Hallas J, Larsen MS, et al. Opioid prescriptions before and after high-energy trauma. J Opioid Manag 2015; 11:313–318.
54. Rivat C, Ballantyne J. The dark side of opioids in pain management: basic science explains clinical observation. Pain Rep 2016; 1:e570.
55. Lee M, Silverman SM, Hansen H, et al. A comprehensive review of opioid-induced hyperalgesia. Pain Physician 2011; 14:145–161.
56. Helmerhorst GT, Vranceanu AM, Vrahas M, et al. Risk factors for continued opioid use one to two months after surgery for musculoskeletal trauma. J Bone Joint Surg Am 2014; 96:495–499.
57. Chiang HL, Chia YY, Lin HS, et al. The implications of tobacco smoking on acute postoperative pain: a prospective observational Study. Pain Res Manag 2016; 2016:9432493.
58. Kim DH, Park JY, Karm MH, et al. Smoking may increase postoperative opioid consumption in patients who underwent distal gastrectomy with gastroduodenostomy for early stomach cancer: a retrospective analysis. Clin J Pain 2017; 33:905–911.
59. Zevin S, Benowitz NL. Drug interactions with tobacco smoking. An update. Clin Pharmacokinet 1999; 36:425–438.
60. Shukla R, Jain N, Jain R, et al. Effects of smoking on healing of distal femur intra-articular fractures, treated with distal femur locking compression plate. J Orthop Traumatol Rehabil 2018; 10:54–56.
61. Shi Y, Weingarten TN, Mantilla CB, et al. Smoking and pain: pathophysiology and clinical implications. Anesthesiology 2010; 113:977–992.
62. Obata H. Analgesic mechanisms of antidepressants for neuropathic pain. Int J Mol Sci 2017; 18: pii: E2483.
63. Sullivan MD. Depression effects on long-term prescription opioid use, abuse, and addiction. Clin J Pain 2018; 34:878–884.
64. Scherrer JF, Salas J, Sullivan MD, et al. The influence of prescription opioid use duration and dose on development of treatment resistant depression. Prev Med 2016; 91:110–116.
65. Kim J, Ham S, Hong H, et al. brain reward circuits in morphine addiction. Mol Cells 2016; 39:645–653.
66. Volkow ND, Morales M. The brain on drugs: from reward to addiction. Cell 2015; 162:712–725.
67. Koob GF, Volkow ND. Neurocircuitry of addiction. Neuropsychopharmacology 2010; 35:217–238.
68. Hernandez I, He M, Brooks MM, et al. Exposure-response association between concurrent opioid and benzodiazepine use and risk of opioid-related overdose in medicare part D beneficiaries. JAMA Netw Open 2018; 1:e180919.
69. Jones JD, Mogali S, Comer SD. Polydrug abuse: a review of opioid and benzodiazepine combination use. Drug Alcohol Depend 2012; 125:8–18.
70. Serdarevic M, Striley CW, Cottler LB. Sex differences in prescription opioid use. Curr Opin Psychiatry 2017; 30:238–246.
71. Chen I, Kurz J, Pasanen M, et al. Racial differences in opioid use for chronic nonmalignant pain. J Gen Intern Med 2005; 20:593–598.
72. Gaither JR, Gordon K, Crystal S, et al. Racial disparities in discontinuation of long-term opioid therapy following illicit drug use among black and white patients. Drug Alcohol Depend 2018; 192:371–376.
73. Schug SA, Bruce J. Risk stratification for the development of chronic postsurgical pain. Pain Rep 2017; 2:e627.
74. Katz J, Weinrib A, Fashler SR, et al. The Toronto General Hospital Transitional Pain Service: development and implementation of a multidisciplinary program to prevent chronic postsurgical pain. J Pain Res 2015; 8:695–702.
Keywords:

frequency; long-term opioid prescription; opioids; postoperative; risk factors; surgery; systematic review; trauma

Supplemental Digital Content

Copyright © 2019 The Author(s). Published by Wolters Kluwer Health, Inc.