An important goal of anesthetic care is to minimize postoperative pain through the appropriate management of analgesia. Nonetheless, a substantial fraction of postoperative patients have severe pain after surgery,1 which is associated with reduced patient satisfaction2 and may lead to the development of chronic pain,3 cardiac complications,4 and other postoperative morbidity.5,6 Adequate pain control must be balanced against complications of excessive analgesia (e.g., hypoxemia due to respiratory depression), and an anesthesiologist’s ability to achieve this balance is one potential measure of skill. Rank-ordering of supervising anesthesiologists by their patients’ pain scores on admission to the postanesthesia care unit (PACU) might allow for identification of high and low performers, presenting an opportunity for improvement in analgesia management.
The PACU pain score has been suggested as a metric for quality of care7 and was incorporated into the anesthesiologists’ quality metrics at our institution. However, we were concerned about the validity of this approach because raw pain scores are significantly influenced by race,8–10 gender, age,11–13 chronic opioid use, preoperative pain, and surgical procedure.11–14 Additionally, nonuniform application of the pain score measurement in clinical settings (in contrast to their use in research studies) may also potentially contribute to confounding. Finally, multiple pain scales have been used in the postoperative period, including the visual analog scale15 and numeric rating scale (NRS), which each present specific methodological challenges.16 Although the visual analogue scale is a continuous scale that may assess the change in pain intensity more closely, the NRS is commonly used in the PACU setting because many postoperative patients require corrective eyewear for near vision (typically not available to them in the PACU) and might be unable to accurately mark the visual analog scale.
We hypothesized that a retrospective comparison of initial postoperative NRS pain scores could be used to develop a valid methodology for rank-ordering anesthesiologists by their patients’ pain scores on admission to the PACU, after adjusting for patient, supervisor, and procedural factors. This rank-ordering might provide the ability to identify highly effective practitioners, and thus allow sharing of their pain management practices with providers whose analgesic management is not as effective. Secondarily, we report the results from the multivariable model used to evaluate provider performance. We provide estimates of associations between covariates and the NRS as well as a measure of variable importance for predicting the NRS.
This study received approval by the Vanderbilt University IRB (Nashville, Tennessee) without requirement for written consent. We included patients 16 years of age and older at Vanderbilt University Medical Center who were admitted postoperatively to the 2 main adult PACUs between April 29, 2011, and March 14, 2013. During the study period, there were no systematic efforts to alter clinical practice regarding postoperative analgesia.
At our institution, all anesthetics are supervised by anesthesiologists, each of whom is responsible for a maximum of 2 anesthesia residents, 4 certified registered nurse anesthetists, or 2 student registered nurse anesthetists. As part of the anesthesiologist’s responsibility to formulate the anesthetic plan and periodically monitor the state of the patient during the anesthetic, he or she determines the modality for postoperative analgesia (e.g., use of regional nerve blocks, addition of adjuvant medications) and is ultimately responsible for the analgesic state of the patient on arrival to the PACU. After the patient is admitted to the PACU, pain management is provided by multiple individuals under the supervision of a separate anesthesiologist assigned to cover the PACU. Furthermore, the duration of the PACU stay varies. Thus, for development of our ranking metric, we focused on assessing the supervisory performance of operating room anesthesiologists by their patients’ initial PACU pain assessments.
Anesthesiologist and PACU Nurse Assignments
Individual anesthesiologists are typically assigned to perform cases for a subset of surgical services during regular working hours, although they often work outside of their regular assignments at other times. This assignment pattern leads to a nonuniform distribution of cases performed by surgical services among anesthesiologists. After surgery, patients not being admitted directly to an intensive care unit are assigned to one of the 2 PACUs, typically based on their primary surgeon’s surgical service but varying, dependent on bed availability. Nurses may work in either PACU, and assignments are interchangeable, although many nurses have preferential PACU assignments. Patients admitted to the PACU are assigned to available nurses on a rotation basis.
Nurses at our institution ask patients to rank their current pain from 0 to 10 using the NRS. Surgical patients’ pain is assessed at each surgical clinic visit and preoperatively in the holding area; thus, patients are familiar with the NRS before PACU admission. In the postoperative period, the receiving PACU nurse elicits the patient’s admission NRS pain score after receiving report on the patient from the transporting anesthesia provider and performing a baseline physiologic assessment. The NRS score is subsequently entered electronically using the Vanderbilt Perioperative Information Management System. Although the NRS documentation screen provides anchors of 0 through 10 (“no pain” to “worst possible pain”), the precise wording supplied for the upper anchor varies among nurses. The authors from Vanderbilt (JPW, JME) have observed inconsistent use by the PACU nurses of upper anchors, such as “worst pain imaginable,” “worst possible pain,” “most pain you could experience,” “the worst pain ever,” and assessments for which no anchor was provided.
Data were retrieved electronically from the perioperative data repository (SQL Server, Microsoft, Redmond, WA). We identified the first documented NRS value for each patient after arrival in the PACU. Patients for whom no NRS was recorded were excluded. Patient age, gender, self-identified race, surgical procedure, anesthesiologist, surgeon, PACU nurse, American Society of Anesthesiologists (ASA) physical status classification, presence of opioids in the preoperative medication list, emergency status, outpatient status, and initial postoperative pain score were collected with deidentification of the anesthesiologist, surgeon, and PACU nurse. To have a sufficient sample size, we only analyzed data for mixed effect analysis for which the anesthesiologist and the PACU nurse each had provided care for at least 100 patients.
Demographic, clinical, and procedural variables were summarized with median and interquartile range for continuous variables and with percentages for categorical variables. Two constructs for NRS pain scores were used for analyses. The primary analysis comprised a 4-category ordered outcome with levels of no pain (NRS = 0), mild pain (NRS = 1 to 3), moderate pain (NRS = 4 to 6), and severe pain (NRS = 7 to 10). We also analyzed a binary outcome model with levels of no pain (NRS = 0) and at least some pain (NRS > 0).17 Recategorizing the 11-category NRS scale into a 4-category scale was done primarily to more closely satisfy the proportional odds assumption because, particularly at the upper end of the NRS scale (high scores), we expected a measurement error resulting from the challenge of anchoring, as mentioned previously. The binary response analysis was added because although there is inconsistency in the description of the upper anchor, “no pain” is uniformly the description for an NRS score of 0. The Appendix describes this analysis.
A proportional odds mixed effects model was used to examine risk factors associated with the postoperative pain score for the ordered outcome, and a mixed effects logistic regression model was used for the binary outcome. The prespecified risk factor set for both models included patient age, gender, race, presence of opioids in the preoperative medication list, ASA physical status, emergency procedure (yes versus no), laparoscopic surgery (yes versus no), inpatient or outpatient status, the anesthesiologist identifier, the PACU nurse identifier, and the presence of multiple anesthesiologists during the procedure (yes versus no). Cases with multiple anesthesiologists were attributed to the last anesthesiologist who supervised care. Because the number of cases was large, age was modeled using restricted cubic splines with 4 knots (at quantiles 0.05, 0.35, 0.65, and 0.95) to permit analysis of nonlinear associations of age on the pain score. Anesthesiologist (69 identifiers) and PACU nurse (66 identifiers) associations were of interest and were modeled with person-specific, fixed effects.
Clinical Classification Software (CCS) codesa were assigned to each case based on the primary Current Procedural Terminology code associated with the case when it was scheduled. To acknowledge the correlation in pain scores within CCS codes (157 codes) and surgeons (348 identifiers), random effects for CCS codes and surgeons were included in the models. For all fixed effects, odds ratios (ORs) (with 95% confidence intervals [CIs]) were calculated to quantify the association of each variable on the pain score. In the proportional odds models used for analyses, each OR should be interpreted as an n-fold change in the odds (i.e., OR) of a higher level of pain associated with a unit change in a covariate (e.g., gender) for equal values of all other covariates and within surgeon and CCS code. For example, a patient whose care was supervised by an anesthesiologist associated with a fixed effect size estimate of 2 would have a 2-fold increase in the odds of having a higher level of pain compared with the median anesthesiologist. To evaluate the explanatory value of each variable (i.e., variable importance) on postoperative pain score while acknowledging that several variables required many df, we used the likelihood ratio χ2 statistic minus the df.18
The goal of this analysis was to rank anesthesiologists’ relative performance in managing pain. Since anesthesiologist comparisons were captured with fixed effects, we were able to conduct hypothesis tests to compare all pairs of anesthesiologists with Wald tests. We summarized the performance of each anesthesiologist with the proportions of other anesthesiologists who he or she performed significantly worse than and better than. Although the NRS is designed to allow the patient to relate his or her subjective pain level without being influenced by the individual recording the data or the instrument itself, we were concerned that there might be a strong “nurse factor” because of differential styles of obtaining NRS pain scores, as described previously for anchoring the value of “10.” Thus, we examined the PACU nurse association in the same way we evaluated the anesthesiologist association (i.e., by including 66 nurse-specific indicator variables).
All analyses were implemented using R 3.0.1 (R Foundation for Statistical Computing, Vienna, Austria). A significance level of 0.05 was used for statistical inference.
A total of 32,539 patients admitted to the PACU were identified; 1659 patients (5.1%) were excluded because no NRS values were recorded while the patient was in the PACU. After excluding observations of which the anesthesiologist and the PACU nurse each had not provided care for at least 100 patients, 26,680 cases were included (Table 1).
In unadjusted analyses, we observed that older patients tended to report less pain than younger patients: patients with severe postoperative pain had a median age of 46 years, compared with a median age of 56 years in those with no pain. Those with severe pain more frequently had opioids in their preoperative medication list (53% of those in severe pain versus 38% of those with no pain) and less frequently were outpatients (16% vs 40%).
The majority of cases were supervised by a single anesthesiologist (89%). The distribution of admission pain scores demonstrated a high frequency of patients with no pain (57.8%), followed by severe (17.4%), moderate (14.1%), and mild pain (10.7%). Although the descriptive results above are unadjusted for other factors, all results that follow are based on results from the covariate adjusted, mixed effects, proportional odds model.
Variation by Anesthesiologist
To evaluate their performance in managing pain, individual anesthesiologists were rank-ordered by conducting the pairwise comparisons based on the proportional odds mixed effect model with 4-category pain outcome (Fig. 1). The anesthesiologist OR panel (Fig. 1) shows the OR of a higher pain score comparing each anesthesiologist to the reference person whose model-based estimate was at the median. The anesthesiologist percentage panel (Fig. 1) shows all adjusted pairwise (between anesthesiologists) comparisons. For each individual, it displays the percentage of other anesthesiologists of whom he or she received significantly higher or lower scores than, and those for which the comparison was indeterminate (i.e., not significant at the 0.05 level). The ORs of a higher pain score had small variation and ranged from 0.60 (95% CI 0.37 to 0.99) to 1.44 (95% CI 0.98 to 2.11). The one anesthesiologist who had the largest OR of a higher pain was significantly different (higher) than 50% of the other anesthesiologists, whereas the one anesthesiologist who had the smallest OR of a higher pain was significantly different (lower) than 66%. The majority of pairwise comparisons were in a “gray zone,” wherein there was insufficient evidence to suggest differential performance, with only 6.4% of anesthesiologists different from each other.
Variation by Patient and Procedural Factors
Multivariable analysis demonstrated that patient age had a significant nonlinear relationship with increasing age associated with decreased odds of more pain (Fig. 2). Females were more likely to have higher pain scores than males (OR 1.21, 95% CI 1.15 to 1.28), and African American patients had lower pain scores than Caucasians (OR 0.82, 95% CI 0.76 to 0.89). ASA physical status had a nonlinear relationship for which ASA 2 and ASA 3 patients had higher pain scores than ASA 1, but ASA 4 and 5 patients had the lowest pain scores. Outpatients had significantly lower pain scores (OR 0.63, 95% CI 0.59 to 0.68), and patients with opioids in their preoperative medication list had significantly higher pain scores (OR 1.44, 95% CI 1.37 to 1.52). Lower pain scores were associated with laparoscopic versus open surgical approaches (OR 0.76, 95% CI 0.60 to 0.98). There was no significant effect from having single versus multiple anesthesiologists for a case (OR 1.00, 95% CI 0.92 to 1.08) or from emergency surgery versus elective surgery (OR 0.89, 95% CI 0.78 to 1.01). These patient and procedural factors were nearly identical in effect size between the 4-category model (Fig. 2) and the binary model (Appendix).
Variation by PACU Nurse
As with the anesthesiologists, individual PACU nurses were rank-ordered using pairwise comparisons to determine how many PACU nurses were more likely to elicit worse, indeterminate or better pain scores (Fig. 1, PACU nurse OR panel). Compared with the nurse who was ranked at the median, the observed associations of the PACU nurses ranged from an OR of more pain equal to 0.16 (95% CI 0.11 to 0.24) up to 2.95 (95% CI 2.43 to 3.59). Under this analysis, 16 of 66 PACU nurses were identified as eliciting significantly lower pain scores than the median, and 33 PACU nurses were distinguished as eliciting significantly higher pain scores than the median. The PACU nurse percentage panel (Fig. 1) shows that we observed far more significant differences in pain scores between pairs of nurses, with 61% of pairs observed to be nonindeterminate.
Comparison of Factor Effect Sizes
To compare the relative explanatory value of the factors included in the model, specifically the anesthesiologist, PACU nurse, patient and procedural values, we compared those factors’ likelihood ratio χ2 statistic minus the degrees of freedom used in the model for those factors (Fig. 3). This comparison demonstrated that the PACU nurse who elicited the initial pain score was the largest determinant of the initial PACU pain score. In contrast, the supervising anesthesiologist was a relatively unimportant determinant of patient-reported pain.
Anesthesiologists working in the Vanderbilt main operating rooms could not be evaluated reliably based on their patients’ NRS pain scores on arrival to the PACU after adjusting for confounding factors. Although such a metric is intuitively appealing, such ranking had poor utility after controlling for confounding factors, most notably the PACU nurse who elicited the admission pain score. Thus, there is no basis upon which to recommend providing such feedback to anesthesiologists based on their patients’ admission PACU NRS pain scores.
Our data confirm previous associations13–17 of age, gender, race, presence of opioids in the preoperative medication list, and surgical procedure as significant factors affecting reported pain scores within the mixed effects models. This provides face validity for our data analysis.
Confirming our concern related to the nonuniform manner in which the NRS score of 10 was anchored, we observed a large association with the initial PACU NRS pain scores attributable to the PACU nurse obtaining the measurement. This effect appears to overwhelm all other factors. The apparent lack of utility of ranking anesthesiologists according to their patients’ PACU admission NRS may be related to the strength of this effect.
The NRS was designed to be administered during research studies by providing 2 anchor points (“no pain” and “worst pain imaginable”) and asking the patient to assign his or her level of pain relative to these anchors. As at other institutions and described elsewhere, our instrument uses these anchor labels “no pain” for 0, and “the worst possible pain” for 10. However, in our clinical setting, additional information or nonstandard language is often added to the upper anchor. For example, one of the authors (JPW) observed a nurse evaluating a patient who appeared comfortable but initially reported a pain rating of 10. The nurse explained to the patient, “No, 10 is the worst pain imaginable, as in childbirth,” at which point the patient revised the pain rating down to 5. Many nurses appear to have a different pattern through which they elicit the pain score, which may have resulted in the large nurse factor seen in our data. Upper anchor differences in pain assessment by PACU nurses have been observed at other institutions, including Thomas Jefferson University Hospital (RHE, personal communication, 2013) and Massachusetts General Hospital (Karim Ladha, personal communication, 2013). We did not assess the NRS anchor provided by PACU nurses in nonacademic hospitals, but there is no reason to believe that they would be applied uniformly (i.e., as would be done in a research study).
A systematic review of the use of pain scales found 24 distinct anchoring phrases to describe the maximum NRS value among 54 included studies.19 This review suggested providing standardization for anchor labels, which our institutional interface for documenting postoperative pain provides. However, despite this, our ad hoc observations demonstrate that anchoring phrases vary, which likely represents typical clinical practice outside of a research setting.
At Vanderbilt University Medical Center, before performing this study, our Department of Anesthesiology Quality Improvement Committee had created a composite anesthesia care score that incorporated as a metric reported NRS pain scores above a threshold of 7 for supervising anesthesiologists. On the basis of the results from this analysis, we have removed this metric from our composite anesthesia care score. Additional education in pain assessment for PACU nurses might reduce the large confounding association they contributed to the recorded pain scores, allowing a valid method of rank-ordering anesthesiologists according to their patients’ admission PACU NRS pain scores. However, this would require repeating this study after collecting many years of additional data.
Our study has several limitations. The findings do not apply to anesthesiologists mostly supervising anesthetics involving children younger than 16 years of age because such patients were excluded from our analysis. We included only patients who were capable of providing a verbal response, which eliminated nonverbal patients, patients who were unable to communicate effectively because of language barriers, and those who were intubated. However, this affected only 5% of the sample size, and thus is unlikely to affect our results. Additionally, the PACU pain score was attributed to the last anesthesiologist on cases with multiple anesthesiologists, which may not be an accurate reflection of the burden of responsibility for the patients, for example, if the pain management was largely influenced by the plan of the anesthesiologist who initially supervised the case. However, this situation affected only 11% of the cases. The results also cannot be applied to settings in which anesthesiologists personally provide patient care (i.e., outside the anesthesia care team model) because we did not study this type of practice. Additionally, to the extent that we perceived to be feasible, we controlled for most potential risk factors for PACU pain. However, we recognize that there are a number of other potential confounders that were not included (e.g., past opioid use), potential confounders that could not be represented fully because of data availability (e.g., home medication administration data for preoperative oral opioids prescribed on an as needed basis) and considering those who are highly specialized (i.e., only do a single kind of surgery). It is possible that complete control for confounding may not be plausible. We did not correct for multiple comparisons, which would have resulted in wider CIs and would have made it more difficult to identify differences between anesthesiologists. Finally, this was a single-center study of an academic anesthesia practice; and thus, these results may not generalize to other academic anesthesia practices. Other institutions may have greater heterogeneity in supervising anesthesiologist performance and greater consistency in the way that nurses record pain. However, the background and training of the anesthesiologists working at Vanderbilt are diverse and similar to those found at many academic institutions.
In summary, we performed an analysis of clinically measured NRS pain scores at the time of PACU admission and did not find evidence to suggest that supervising anesthesiologists could be discriminated one from another after controlling for significant sources of variation. Given that the largest source of variability is the PACU nurse who elicited the initial pain score, we do not recommend using the admission pain score metric as a quality measure of anesthesia care by supervising anesthesiologists.
Analysis Based on the Presence or Absence of Pain on PACU Admission
In addition to the primary analysis provided in the results using the proportional odds model, we explored the potential utility of ranking anesthesiologists by the percentage of their patients who arrive with a pain score of 0. We selected this criterion as potentially useful because the NRS score of 0 is consistently anchored by PACU nurses as “no pain,” and 58.6% of patients reported no pain on PACU admission. Thus, although they use different terminology for the upper anchors, this would not have affected patients reporting 0 as their pain score. Spearman’s rho for anesthesiologist ranks was estimated to be 0.92 between the proportional odds model results and the results of this binary model, and an analysis of rank-orderings in Fig. 4 for the binary model is comparable to the results in Fig. 1 for the proportional odds model. Similarly, effect sizes as shown in Fig. 5 for the binary model are comparable to Fig. 2. Finally, the explanatory value of each variable for the binary model as shown in Fig. 6 is similar to those from the proportional odds model as shown in Fig. 3.
Name: Jonathan P. Wanderer, MD, MPhil.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: Jonathan P. Wanderer has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is one of the authors responsible for archiving the study files.
Name: Yaping Shi, MS.
Contribution: This author helped analyze the data and write the manuscript.
Attestation: Yaping Shi has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is one of the authors responsible for archiving the study files.
Name: Jonathan S. Schildcrout, PhD.
Contribution: This author helped analyze the data and write the manuscript.
Attestation: Jonathan S. Schildcrout has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Name: Jesse M. Ehrenfeld, MD, MPH.
Contribution: This author helped design the study and conduct the study.
Attestation: Jesse M. Ehrenfeld has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Name: Richard H. Epstein, MD, CPHIMS.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: Richard H. Epstein has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
This manuscript was handled by: Franklin Dexter, MD, PhD.
a Healthcare Cost and Utilization Project, Clinical Classification Software for ICD-9-CM. Available at: http://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp. Accessed December 6, 2013.
1. Apfelbaum JL, Chen C, Mehta SS, Gan TJ. Postoperative pain experience: results from a national survey suggest postoperative pain continues to be undermanaged. Anesth Analg. 2003;97:534–40
2. Myles PS, Williams DL, Hendrata M, Anderson H, Weeks AM. Patient satisfaction after anaesthesia and surgery: results of a prospective survey of 10,811 patients. Br J Anaesth. 2000;84:6–10
3. Kehlet H, Jensen TS, Woolf CJ. Persistent postsurgical pain: risk factors and prevention. Lancet. 2006;367:1618–25
4. Airaksinen KE. Autonomic mechanisms and sudden death after abrupt coronary occlusion. Ann Med. 1999;31:240–5
5. Beattie WS, Badner NH, Choi PT. Meta-analysis demonstrates statistically significant reduction in postoperative myocardial infarction with the use of thoracic epidural analgesia. Anesth Analg. 2003;97:919–20
6. Rodgers A, Walker N, Schug S, McKee A, Kehlet H, van Zundert A, Sage D, Futter M, Saville G, Clark T, MacMahon S. Reduction of postoperative mortality and morbidity with epidural or spinal anaesthesia: results from overview of randomised trials. BMJ. 2000;321:1493
7. Usichenko TI, Röttenbacher I, Kohlmann T, Jülich A, Lange J, Mustea A, Engel G, Wendt M. Implementation of the quality management system improves postoperative pain treatment: a prospective pre-/post-interventional questionnaire study. Br J Anaesth. 2013;110:87–95
8. Lee E, Teeple M, Bagrodia N, Hannallah J, Yazzie NP, Adamas-Rappaport WJ. Postoperative pain assessment and analgesic administration in Native American patients undergoing laparoscopic cholecystectomy. JAMA Surg. 2013;148:91–3
9. Sadhasivam S, Krekels EH, Chidambaran V, Esslinger HR, Ngamprasertwong P, Zhang K, Fukuda T, Vinks AA. Morphine clearance in children: does race or genetics matter? J Opioid Manag. 2012;8:217–26
10. Faucett J, Gordon N, Levine J. Differences in postoperative pain severity among four ethnic groups. J Pain Symptom Manage. 1994;9:383–9
11. Kalkman CJ, Visser K, Moen J, Bonsel GJ, Grobbee DE, Moons KG. Preoperative prediction of severe postoperative pain. Pain. 2003;105:415–23
12. Ip HY, Abrishami A, Peng PW, Wong J, Chung F. Predictors of postoperative pain and analgesic consumption: a qualitative systematic review. Anesthesiology. 2009;111:657–77
13. Gramke HF, de Rijke JM, van Kleef M, Kessels AG, Peters ML, Sommer M, Marcus MA. Predictive factors of postoperative pain after day-case surgery. Clin J Pain. 2009;25:455–60
14. Gerbershagen HJ, Aduckathil S, van Wijck AJ, Peelen LM, Kalkman CJ, Meissner W. Pain intensity on the first day after surgery: a prospective cohort study comparing 179 surgical procedures. Anesthesiology. 2013;118:934–44
15. DeLoach LJ, Higgins MS, Caplan AB, Stiff JL. The visual analog scale in the immediate postoperative period: intrasubject variability and correlation with a numeric scale. Anesth Analg. 1998;86:102–6
16. Ohnhaus EE, Adler R. Methodological problems in the measurement of pain: a comparison between the verbal rating scale and the visual analogue scale. Pain. 1975;1:379–84
17. McCaffery M, Beebe A Pain: Clinical Manual for Nursing Practice. 1989 Maryland Heights, Missouri: Mosby-Year Book
18. Harrell FE Regression Modeling Strategies: With Applications to Linear Models, Logistic Regression, and Survival Analysis. 2001 New York Springer
19. Hjermstad MJ, Fayers PM, Haugen DF, Caraceni A, Hanks GW, Loge JH, Fainsinger R, Aass N, Kaasa SEuropean Palliative Care Research Collaborative (EPCRC). . Studies comparing Numerical Rating Scales, Verbal Rating Scales, and Visual Analogue Scales for assessment of pain intensity in adults: a systematic literature review. J Pain Symptom Manage. 2011;41:1073–93