As patients recover from surgery, they face an ongoing choice either to continue taking prescribed opioids or to stop opioids and undertake non-opioid pain treatment (e.g., acetaminophen and nonsteroidal antiinflammatory drugs). Very little, however, is known about the factors that determine this patient choice. Despite the increase in prescription opioids for both pain1 and recreational use,2 previous research has not focused on the predictors of when or if patients will discontinue prescribed opioids. Because opioids are prescribed to reduce pain intensity, it would be reassuring if pain intensity was the principal driver of continuing opioid use after surgery or acute injuries. However, studies of patients with chronic pain suggest that this may not be the case.3,4
Among patients with chronic pain, higher levels of psychological distress and substance abuse, rather than pain intensity, best predict treatment with opioids rather than non-opioid analgesics.3,4 Psychological distress and substance abuse have also been associated with greater opioid use in the immediate postoperative period. More specifically, preoperative depressive and anxiety symptoms (distinct from an actual established mood disorder diagnosis) correlate with higher consumption of opioid analgesics in the first 24 to 72 hours after surgery.5,6 Similarly, preoperative substance abuse has been associated with increased opioid use immediately and remotely after surgery.7,8
One potential association between early postinjury opioid consumption and subsequent remote risk of chronic opioid use is a decreased rate of opioid discontinuation. Previous studies of postoperative opioid use have not reported how substance abuse history and psychological state, rather than injury and pain, might predict a patient's ongoing choice to stop rather than continue opioid therapy.
Therefore, we undertook this study to determine whether preoperative substance abuse history and psychological state predicted the postoperative rate of opioid discontinuation after routine surgical procedures. We specifically hypothesized that substance abuse and psychological distress (anxiety, depression, and posttraumatic stress disorder [PTSD]) were associated with a decreased rate of opioid discontinuation after surgery.
METHODS
Participants
We conducted a prospective, longitudinal observational study of subjects' opioid use and pain after surgery. Subjects were identified for recruitment by participating surgeons between January 2007 and April 2009. All patients scheduled to undergo total hip replacement, total knee replacement, thoracotomy, mastectomy, and lumpectomy with any of the participating surgeons were considered eligible. The only exclusion criterion was an inability to complete the study forms because of either mental incapacity or a language barrier. All aspects of the protocol were approved by the Stanford University IRB and all subjects provided informed consent in writing.
Preoperative Assessment
Before surgery, subjects completed the following baseline questionnaires.
Measures of Pain and Opioid Use
Preoperatively, we determined baseline opioid use and pain using the Brief Pain Inventory (BPI).9 Subjects were specifically asked to complete the BPI twice: first, with respect to pain at the surgery site; and second, with respect to pain not at the surgery site (e.g., back pain in those undergoing a total knee replacement). Preoperative opioid use was classified as “legitimate preoperative opioid use,” in which subjects reported taking an opioid prescribed to them for pain; “illicit preoperative opioid use,” in which subjects used an opioid preoperatively that was not prescribed to them (even if it was taken ostensibly for pain); and “any preoperative opioid use,” which included all subjects from either of these groups.
Measures of Risk of Developing Problematic Opioid Use
To assess previously identified risk factors for developing problematic opioid use behaviors, we administered the Screener and Opioid Assessment for Patients with Pain (SOAPP 24).10 Subjects were also given an author-generated measure of Self-Perceived Susceptibility to Addiction. This was intended to quantify how subjects' baseline fears of prescribed opioids influenced their subsequent use of opioids. Subjects were asked, “How likely do you think it is that you will develop an addiction problem from pain medication you take after surgery?” and chose from 1 of 4 answers: 1: “not at all”; 2: “unlikely”; 3: “somewhat likely”; or 4: “very likely.” Past substance abuse behaviors were measured with the Addiction Severity Index Drug and Alcohol Use section,11,12 which assesses lifetime and past 30-day use of illicit substances and alcohol. These questionnaires were given in the following consistent order: SOAPP 24, Self-Perceived Susceptibility to Addiction, and then the Addiction Severity Index Drug and Alcohol Use section.
Measures of Mood, PTSD, Anxiety-Sensitivity, and Fear
We assessed depressive symptoms with the Beck Depression Inventory-II (BDI-II)9; PTSD symptoms with the Primary Care Posttraumatic Stress Disorder Screen13; somatic fear and anxiety with the Anxiety-Sensitivity Index14; and fear of pain with the Fear of Pain Questionnaire.15
Preoperative Instructions on Opioid Use
Preoperatively (and again postoperatively), a research team member gave each subject the following standardized verbal and written instructions on how to take their opioid medication:
“Following your surgery you are going to have a certain amount of pain for a short period of time. Your doctor will either prescribe pain medication or instruct you to take over-the-counter pain medication. You should take these pain medications only when you are in pain. You should stop taking the medications when you no longer have pain. If you do not require the entire amount of medication prescribed, you should dispose of the remainder. It is alright for you not to finish all the medication you are given.”
Postoperative
Starting on postoperative day 1 and each day thereafter, we determined continuing opioid use and pain using the BPI.9 Subjects were specifically instructed to answer the BPI with reference to pain at the surgical site. We continued to collect daily BPI measurements by telephone until the subject reported 5 consecutive days of zero prescription opioid use and 5 consecutive days of zero average pain. All reported medications were reviewed by a study physician to identify opioid use. The time from surgery until the first of 5 consecutive days of zero opioid use was defined as postoperative “time to opioid cessation,” and this was the primary end point. Subjects taking preoperative opioids were considered to have reached their primary end point when they had stopped taking any new postoperative opioids (i.e., they had returned to their preoperative regimen). The time from surgery until the first of 5 consecutive days of zero average pain was defined as postoperative “pain duration.” Daily assessment generally lasted between 2 and 3 minutes per contact. Major changes in subjects' health status were sought on a monthly basis (e.g., cancer recurrence, or chemotherapy) to identify possible confounds to the primary self-reported pain data.
Statistical Analysis
SAS version 9.2 (SAS Institute, Cary, NC) was used for all analyses. A Pearson correlation coefficient was calculated for the correlation between opioid use duration and pain duration. Time to prescription opioid cessation was analyzed using Kaplan-Meier analysis and Cox proportional hazards regression. Survival curves were generated for variables that significantly predicted time to opioid cessation in univariate Cox proportional hazard analysis. All Cox proportional hazard analyses included the surgery type in the model as either an explanatory variable or stratification variable as appropriate. The proportional hazards assumption for Cox regression was met.
Multivariate Model Selection and Validation
Although each preoperative characteristic was considered a plausible determinant of postoperative opioid use, the absence of previous data in the literature precluded a rationally designed a priori multivariate model of preselected variables. Therefore, multivariate model selection was accomplished using an automated stepwise algorithm for the selection of variables for the model. All variables that were significant in the univariate analysis were considered for inclusion in the multivariate model. There can potentially be errors in model building that are as important as sampling error in the variables analyzed. To analyze the sensitivity of our results to the specific model-building algorithm or order effects, the model was also constructed using both a backward and forward variable-selection algorithm. To address the sensitivity of our model to factors unique to any one specific type of surgery, we reanalyzed the Cox proportional hazard analysis using the variables from our final multivariate model while sequentially excluding each kind of surgery from the analysis. Similarly, we reconstructed the Cox proportional hazard model using the opioid-naïve population only as part of our sensitivity analysis.
After identification of the multivariate model using the procedure described above, pain duration, baseline pain at the surgical site, baseline pain other than at the surgical site, and pain intensity at the time of opioid cessation were each sequentially added back into the model to determine whether the inclusion of pain improved or reduced model fit.
RESULTS
Patient characteristics are shown in Table 1. One hundred nine of 134 patients approached for inclusion in this study agreed to participate.
;)
Table 1: Patient Characteristics (n = 109)
Median time to opioid discontinuation by surgery type is presented in Table 1. Overall, 6% of patients continued taking new opioids 150 days after surgery.
Factors Influencing Time to Opioid Cessation in Univariate Analysis
Pain duration was correlated with time to opioid cessation with a Pearson correlation coefficient of 0.69 (95% confidence interval [CI] 0.55–0.78), accounting for 48% of variance (Fig. 1). However, neither preoperative pain intensity nor the pain intensity at the time of opioid cessation significantly predicted the decision to continue or discontinue opioids. Factors influencing time to opioid cessation in univariate analysis are presented in Table 2.
Figure 1: Duration of opioid use is correlated with pain duration after surgery: Pearson correlation coefficient 0.69 (95% confidence interval 0.55–0.78), P < 0.0001, R 2 = 0.48. Patients using preoperative opioids take postoperative opioids longer than opioid-naïve patients, often until their pain completely resolves. Red circles = used opioids preoperatively; blue circles = did not use opioids preoperatively.
Table 2: Univariate Analysis of Preoperative Determinants of Prolonged Opioid Use After Surgery
The following preoperative variables predicted reduced rates of opioid discontinuation in univariate analysis: BDI-II score, preoperative legitimate opioid use, any preoperative opioid use, Self-Perceived Risk of Addiction score, SOAPP 24 score (10), preoperative illicit opioid use, positive PTSD symptoms, Anxiety-Sensitivity Index score, and Fear of Pain score.
Factors Influencing Time to Opioid Cessation in Multivariate Analysis
Results from the stepwise multivariate model building are displayed in Table 3. Preoperative legitimate prescribed opioid use, self-perceived risk of addiction, and depressive symptoms each independently predicted more prolonged opioid use after surgery. Each of these factors was a better predictor of prolonged opioid use than postoperative pain duration or severity.
Table 3: Multivariate Analysis of Preoperative Determinants of Prolonged Opioid Use After Surgery
Preoperative Opioid Use
Twenty-one of 103 respondents (20%) reported taking legitimately prescribed opioids before surgery. This preoperative legitimate opioid use was associated with a 73% (95% CI 41%–87%) reduction in the rate per unit time of opioid cessation after surgery (P = 0.0009). Figure 1 illustrates that, for any given duration of pain (e.g., 90 days), patients who used legitimate opioids preoperatively were more likely to use their postoperative opioids until their pain had almost completely abated.
Six of 109 subjects declined to provide data on their possible use of legitimate preoperative opioids, and 3 declined to provide data on their possible use of illicit opioids. Eight of the 21 (38%) subjects taking prescribed preoperative opioids also reported taking illicit opioids. In contrast, only 6 of the 82 (7%) subjects not taking prescribed opioids reported taking illicit opioids. Thus, subjects taking prescribed opioids before surgery were much more likely to also use illicit opioids than subjects not taking prescribed opioids before surgery: relative risk = 5.2 (95% CI 2.0–13.4; P = 0.001).
Self-Perceived Susceptibility to Addiction
Preoperative self-perceived susceptibility to addiction was also independently associated with marked increases in the duration of opioid use postoperatively. In response to the question, “How likely do you think it is that you will develop an addiction problem from pain medication you take after surgery?” every 1-point increase (on a 4-point categorical scale) was independently associated with a 53% (95% CI 23%–71%; P = 0.003) reduction in the opioid discontinuation rate per unit time (Table 3). In contrast, traditionally recognized risk factors for addiction, including a family history of substance use disorder, history of illicit substance abuse, and tobacco or alcohol use did not predict prolonged duration of prescription opioid use as strongly as self-perceived susceptibility to addiction.
Preoperative Depressive Symptoms
Each 10-point increase in the 63-point BDI-II score was independently associated with a 42% (95% CI 18%–58%; P = 0.002) reduction in the opioid discontinuation rate (Table 2). To illustrate how preoperative depressive symptoms predicted subsequent duration of opioid use, Kaplan-Meier survival curves (Fig. 2) were generated with subjects stratified according to whether their preoperative BDI-II score was more than or less than 14 (the cutoff for minimal versus mild depression). As Figure 2 illustrates, subjects with elevated levels of preoperative depressive symptoms have lower rates of opioid cessation after surgery and are more likely to continue taking opioids 6 months after their operation (log-rank P = 0.0004).
Figure 2: Elevated levels of preoperative (pre-op) depressive symptoms predict more persistent opioid use after surgery. Beck Depression Inventory-II (BDI-II) ≥14 correlates with “mild depression” or greater (log-rank P = 0.0004).
Pain Duration and Severity in Predicting Time to Opioid Cessation
Variance in pain duration predicts no more than 48% of the variance in opioid use duration (Fig. 1) in univariate analysis. However, pain duration was not selected for the multivariate model, suggesting other variables more completely explain the variance in time to opioid cessation. To evaluate whether pain independently predicted time to opioid cessation when already accounting for preoperative opioid use, self-perceived risk of addiction, and depressive symptoms, we individually and sequentially added to this model the following measures of pain: (1) pain duration, (2) pain severity at the time of opioid cessation, (3) preoperative baseline pain severity at the surgical site, and (4) preoperative baseline pain severity other than at the surgical site (e.g., chronic back pain in a patient undergoing total knee replacement) (Table 4). In each case, the model's ability to predict time to opioid cessation was reduced by the addition of pain as a predictor variable, as assessed by an increasing Akaike Information Criteria. Furthermore, when these measures of pain are included in the model, pain duration and severity demonstrated only small effects on time to opioid cessation that were not statistically significant. In contrast, the metrics of preoperative opioid use, self-perceived risk of addiction, and depressive symptoms all showed only minor changes to their hazard ratios, and each remained a significant predictor of time to opioid cessation in each new multivariate model. In totality, these results suggest that the variance in time to opioid cessation explained by pain duration in univariate analysis (Fig. 1) is actually better predicted and explained by the variance in preoperative opioid use, self-perceived risk of addiction, and depressive symptoms.
Table 4: Sensitivity Analysis of the Multivariate Model
Model Validation
To address the sensitivity of the model to any one specific surgery, we reran the Cox proportional hazard analysis using the variables from our final multivariate model while sequentially excluding each kind of surgery from the analysis. Detailed results are presented in Table 4. In each model, hazard ratios were changed for each variable only minimally, and P values remain significant for each variable in all models with only one exception (depressive symptoms when excluding subjects undergoing total knee replacement, P = 0.09).
To analyze our multivariate model for sensitivity to the model-building algorithm, or order effects, the model was also reconstructed using both backward and forward variable selection algorithms. Depressive symptoms, self-perceived risk of addiction, and use of preoperative opioids were again significant in the models constructed by both the backward and forward model selection algorithms. The model presented in Table 3 demonstrated the best model fit as assessed by Akaike Information Criteria.
DISCUSSION
Overall, 6% of patients continued on new opioids 6 months after surgery. It is not clear at what point such continuing postoperative opioid use should be considered new “chronic” opioid use. Between 1999 and 2005, an average of 17.6 million surgical patients were discharged per year from hospitals across the country.16 If 6% of 17.6 million surgical procedures result in new prolonged opioid use, surgery would contribute 1.1 million new users of opioids to the population each year. If used to effectively treat chronic postsurgical pain that had been previously ignored, this new prolonged opioid use may be a positive development.
However, the more prolonged opioid use predicted by preoperative opioid use, self-perceived risk of addiction, and depressive symptoms (rather than coexisting pain) may be undesirable for the following 4 reasons. First, there is a disconnect between the primary purpose for which these opioids are ostensibly prescribed and consumed (pain) and the factors that actually predict their continuing use (preoperative opioid use, self-perceived risk of addiction, and depressive symptoms). Second, newly emerging data suggest that patients treated for chronic noncancer pain with opioids, rather than nonsteroidal antiinflammatory drugs or cyclooxygenase-2 inhibitors, may be at markedly increased risk of falls, cardiovascular events, and all-cause mortality.17 Third, when these patients present for subsequent further surgical procedures, they are at increased risk of postoperative oversedation, poorly controlled pain, prolonged hospital stays, and may even be at higher risk of revision surgery.18,19 Finally, individuals with continuing legitimate, pain-relieving, opioid consumption constitute a significant pool of people at risk of subsequently developing prescription opioid addiction. The potential importance of iatrogenic opioid exposure in the development of prescription opioid addiction is highlighted by studies showing that 31% to 84% of prescription opioid addicts seeking inpatient treatment report that the prescription opioids they later abused had first been legitimately prescribed by a physician for the treatment of pain.20–22 It should be noted that in this study we did not attempt to make assessments of whether patients who continued taking postoperative opioids had developed opioid abuse, dependency, or addiction.
Pain and the Duration of Opioid Use
Pain duration and pain intensity at the time of opioid cessation did not explain the duration of opioid use as well as legitimate preoperative opioid use, self-perceived risk of addiction, or depressive symptoms. No patient reported taking opioid medications after pain had resolved, and most postsurgical patients stopped opioid use before their pain had resolved (Fig. 1). Thus, the variance in opioid use duration occurred only in the setting of persistent pain. In other words, if 2 patients both had pain for 90 days after knee surgery, one may have stopped opioids on day 10 whereas the other continued opioids until day 80. This variance is explained by our multivariate model, suggesting that the variance between these 2 patients is best explained by their preoperative differences in opioid use, self-perceived risk of addiction, and depressive symptoms.
These findings extend growing research in patients with chronic pain suggesting that many (and perhaps most) patients with pain discontinue prescribed opioids,23,24 but those patients with pain and with higher levels of affective distress are more likely to continue to use prescribed opioids.3,4,25
Preoperative Opioid Use
Patients using legitimate preoperative opioids discontinued opioids more slowly after surgery, even though opioid discontinuation was defined for them as simply returning to their preoperative opioid regimen. These data confirm the clinical impression that preoperative opioid use is a strong predictor of continuing postoperative opioid use. Our results also indicate that subjects taking prescribed opioids before surgery were much more likely to also use illicit opioids than subjects not taking prescribed opioids before surgery: relative risk = 5.2 (95% CI 2.0–13.4; P = 0.001). This relationship has not been reported previously. Of note, if we had relied exclusively on urine drug testing rather than patient self-report we would not have been able to identify that 38% of patients taking legitimate prescribed opioids were also taking illicit unprescribed opioids. Conversely, if we had used urine drug screens, we may have identified a higher rate of opioid use (as well as illicit drug use) preoperatively among patients who denied legitimate or illicit preoperative opioid use. The high concordance of illicit and legitimate preoperative opioid use among our patients helps to explain why both illicit and legitimate preoperative opioid use are strong predictors of time to opioid cessation in univariate models, yet only legitimate preoperative opioid use remains a strong predictor in the final multivariate model. Further studies with more patients who only use illicit opioids (and not simultaneous legitimate opioids) will be needed to determine whether illicit opioid use also independently predicts a reduced rate of opioid discontinuation.
Self-Perceived Risk of Addiction
Subjects who reported an increased sense of vulnerability to opioid addiction actually continued postoperative prescription opioids significantly longer than those subjects who did not perceive themselves at risk. Self-perceived risk of addiction remained a strong predictor of delayed opioid cessation, even when all the patients who had been taking preoperative opioids were excluded from the analysis (Table 4). Thus, the predictive value of self-perceived susceptibility to addiction is not merely a function of fear among subjects currently experiencing continuing preoperative opioid use. Alternatively, higher ratings of self-perceived risk of addiction may reflect patients' experiences of positive emotional responses such as euphoria during previous short-term opioid exposures (e.g., after tooth extraction).
It is likely that most patients characterizing their own risk of addiction do not understand the difference between physical dependence, tolerance, abuse, and addiction. However, we present strong evidence that the patients' self-perceived risk of addiction (as understood by these laymen patients) strongly predicts their future opioid-consuming behavior. Patients identifying themselves as having a higher self-perceived risk of addiction may benefit from a more specific, physician-guided opioid taper, rather than being left to titrate off opioids on their own. This area will need to be further explored to fully understand the connection between a preoperative, self-perceived risk of addiction and a more prolonged postsurgical opioid use that does not conform to the medical definition of addiction.
Depressive Symptoms Contribute to Chronicity of Opioid Use
Even minor levels of depressive symptoms, reflected by BDI-II scores correlating with minimal depression (Fig. 2), independently and strongly predicted more prolonged opioid use. This supports and extends previous findings that preoperative depressive symptoms predict greater opioid use in the first 48 hours postoperatively.5,6,26 The work presented herein also extends findings from patients with chronic pain demonstrating that dysthymia and depression strongly predict subsequent chronic opioid use,4 even when pain intensity does not.3 Our data suggest that one mechanism of increased prevalence of opioid use in those chronic pain patients with diminished mood is reduced rates of opioid discontinuation. A better understanding of the connection between a patient's mood and their decision to stop or continue opioid therapy may help formulate pain management strategies that appropriately use opioids to minimize pain while simultaneously minimizing nonanalgesic opioid use.
Model Validation
We evaluated different model-building algorithms, the addition of different measures of pain to the multivariate model, and the exclusion of different patient subsets from the model. In each model, hazard ratios changed minimally for preoperative legitimate opioid use, self-perceived risk of addiction, and depressive symptoms, whereas P values remained significant for each variable in all models with only one exception (depressive symptoms when excluding subjects undergoing total knee replacement, P = 0.09). Furthermore, the identical variables were identified for the final model when surgery type was used as a stratification variable, rather than an independent variable, in the model. En bloc, these results support the conclusion that preoperative legitimate opioid use, self-perceived risk of addiction, and depressive symptoms predict subsequent time to opioid cessation across divergent surgical procedures and their associated divergent patient populations.
CONCLUSIONS
This pilot study measured factors leading to more prolonged opioid use after surgery; however, these factors' relation to opioid abuse and addiction remain to be elucidated. Future work will need to elucidate the relationship between early, objective postsurgical opioid-use measures such as time to opioid cessation and the important subsequent measures of opioid misuse or addiction. Furthermore, future studies should aim to include more patients undergoing each type of surgery, to determine whether factors that determine time to opioid cessation after major surgical procedures, such as thoracotomy, are identical to the factors that determine time to opioid cessation after minor procedures such as lumpectomy.
STUDY FUNDING
Dr. Carroll gratefully acknowledges funding from the Foundation for Anesthesia Education and Research (FAER) in the form of a mentored research training grant. Subsequent funding was provided by K23 grant 1K23DA025152 from the National Institute on Drug Abuse. Dr. Mackey was supported by NIH NINDS NS053961, John and Dodie Rosekrans Pain Research Endowment, and Chris Redlich Pain Research Fund. Any views expressed herein are not necessarily those of the US Government. The above sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.
DISCLOSURES
Name: Ian Carroll, MD, MS.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: Ian Carroll has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.
Name: Peter Barelka, MD.
Contribution: This author helped conduct the study and write the manuscript.
Name: Charlie Kiat Meng Wang, BS.
Contribution: This author helped conduct the study and write the manuscript.
Name: Bing Mei Wang, BS.
Contribution: This author helped conduct the study and write the manuscript.
Name: Matthew John Gillespie, BS.
Contribution: This author helped conduct the study and write the manuscript.
Attestation: Matthew John Gillespie approved the final manuscript.
Name: Rebecca McCue, BA.
Contribution: This author helped conduct the study and write the manuscript.
Attestation: Rebecca McCue has seen the original study data and reviewed the analysis of the data.
Name: Jarred W. Younger, PhD.
Contribution: This author helped design the study and write the manuscript.
Attestation: Jarred W. Younger approved the final manuscript.
Name: Jodie Trafton, PhD.
Contribution: This author helped design the study and write the manuscript.
Name: Keith Humphreys, PhD.
Contribution: This author helped design the study and write the manuscript.
Attestation: Keith Humphreys approved the final manuscript.
Name: Stuart B. Goodman, MD, PhD.
Contribution: This author helped conduct the study and write the manuscript.
Name: Fredrick Dirbas, MD.
Contribution: This author helped conduct the study and write the manuscript.
Name: Richard I. Whyte, MD, MBA.
Contribution: This author helped conduct the study and write the manuscript.
Name: Jessica S. Donington, MD.
Contribution: This author helped conduct the study and write the manuscript.
Name: Walter B. Cannon, MD.
Contribution: This author helped conduct the study and write the manuscript.
Name: Sean Charles Mackey, MD, PhD.
Contribution: This author helped design the study, conduct the study, and write the manuscript.
This manuscript was handled by: Spencer S. Liu, MD.
REFERENCES
1. Kuehn BM. Opioid prescriptions soar: increase in legitimate use as well as abuse. JAMA 2007;297:249–51
2. Substance Abuse and Mental Health Services Administration, Results from the 2010 National Survey on Drug Use and Health: Summary of National Findings, NSDUH Series H-41, HHS Publication No. (SMA) 11-4658. Rockville, MD: Substance Abuse and Mental Health Services Administration, 2011
3. Breckenridge J, Clark JD. Patient characteristics associated with opioid versus nonsteroidal anti-inflammatory drug management of chronic low back pain. J Pain 2003;4:344–50
4. Sullivan MD, Edlund MJ, Zhang L, Unutzer J, Wells KB. Association between mental health disorders, problem drug use, and regular prescription opioid use. Arch Intern Med 2006;166:2087–93
5. De Cosmo G, Congedo E, Lai C, Primieri P, Dottarelli A, Aceto P. Preoperative psychologic and demographic predictors of pain perception and tramadol consumption using intravenous patient-controlled analgesia. Clin J Pain 2008;24:399–405
6. ÖZalp G, Sarioglu R, Tuncel G, Aslan K, Kadiogullari N. Preoperative emotional states in patients with breast cancer and postoperative pain. Acta Anaesthesiol Scand 2003;47:26–9
7. Massey GM, Dodds HN, Roberts CS, Servoss TJ, Blondell RD. Toxicology screening in orthopedic trauma patients predicting duration of prescription opioid use. J Addict Dis 2005;24:31–41
8. Singh J, Lewallen D. Predictors of pain and use of pain medications following primary total hip arthroplasty (THA): 5,707 THAs at 2-years and 3,289 THAs at 5-years. BMC Musculoskelet Disord 2010;11:90
9. Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measuring depression. Arch Gen Psychiatry 1961;4:561–71
10. Butler S, Budman S, Fernandez K, Jamison R. Validation of a screener and opioid assessment measure for patients with chronic pain. Pain 2004;112:65–75
11. McLellan AT, Luborsky L, Woody GE, O'Brien CP. An improved diagnostic evaluation instrument for substance abuse patients: the Addiction Severity Index. J Nerv Ment Dis 1980;168:26–33
12. Makela K. Studies of the reliability and validity of the Addiction Severity Index. Addiction 2004;99:398–410
13. Kimerling R, Ouimette P, Prins A, Nisco P, Lawler C, Cronkite R, Moos RH. Brief report: utility of a short screening scale for DSM-IV PTSD in primary care. J Gen Intern Med 2006;21:65–7
14. Reiss S, Peterson RA, Gursky DM, McNally RJ. Anxiety sensitivity, anxiety frequency and the prediction of fearfulness. Behav Res Ther 1986;24:1–8
15. McNeil DW, Rainwater AJ III. Development of the Fear of Pain Questionnaire—III. J Behav Med 1998;21:389–410
16. Li G, Warner M, Lang BH, Huang L, Sun LS. Epidemiology of anesthesia-related mortality in the United States, 1999–2005. Anesthesiology 2009;110:759–65
17. Solomon DH, Rassen JA, Glynn RJ, Lee J, Levin R, Schneeweiss S. The comparative safety of analgesics in older adults with arthritis. Arch Intern Med 2010;170:1968–78
18. Rapp SE, Ready LB, Nessly ML. Acute pain management in patients with prior opioid consumption: a case-controlled retrospective review. Pain 1995;61:195–201
19. Zywiel M, Stroh DA, Lee S, Bonutti P, Mont MA. Chronic opioid use prior to total knee arthroplasty. J Bone Joint Surg Am 2011;93:1988–93
20. Passik SD, Hays L, Eisner N, Kirsh KL. Psychiatric and pain characteristics of prescription drug abusers entering drug rehabilitation. J Pain Palliat Care Pharmacother 2006;20:5–13
21. Potter JS, Hennessy G, Borrow JA, Greenfield SF, Weiss RD. Substance use histories in patients seeking treatment for controlled-release oxycodone dependence. Drug Alcohol Depend 2004;76:213–5
22. Hays L, Kirsh KL, Passik SD. Seeking drug treatment for OxyContin abuse: a chart review of consecutive admissions to a substance abuse treatment facility in Kentucky. J Natl Compr Canc Netw 2003;1:423–8
23. Furlan AD, Sandoval JA, Mailis-Gagnon A, Tunks E. Opioids for chronic noncancer pain: a meta-analysis of effectiveness and side effects. CMAJ 2006;174:1589–94
24. Kalso E, Edwards JE, Moore RA, McQuay HJ. Opioids in chronic non-cancer pain: systematic review of efficacy and safety. Pain 2004;112:372–80
25. Sullivan MD, Edlund MJ, Steffick D, Unützer J. Regular use of prescribed opioids: association with common psychiatric disorders. Pain 2005;119:95–103
26. Fukuda K, Hayashida M, Ikeda K, Koukita Y, Ichinohe T, Kaneko Y. Diversity of opioid requirements for postoperative pain control following oral surgery: is it affected by polymorphism of the mu-opioid receptor? Anesth Prog 2010;57:145–9
