Free
Meeting Abstracts  |   March 2006
Multifactorial Preoperative Predictors for Postcesarean Section Pain and Analgesic Requirement
Author Affiliations & Notes
  • Peter H. Pan, M.D.
    *
  • Robert Coghill, Ph.D.
  • Timothy T. Houle, Ph.D.
  • Melvin H. Seid, M.D.
    §
  • W Michael Lindel, M.D.
    §
  • R Lamar Parker, M.D.
    §
  • Scott A. Washburn, M.D.
    §
  • Lynne Harris, B.S.N.
  • James C. Eisenach, M.D.
    #
  • * Associate Professor, Department of Anesthesiology, Wake Forest University School of Medicine, Winston-Salem, North Carolina, and Forsyth Medical Center, Winston-Salem, North Carolina. † Associate Professor, ‡ Research Assistant Professor, Department of Anesthesiology, Wake Forest University School of Medicine. § Obstetrician and Gynecologist, Lyndhurst Gynecological Associates, Winston-Salem, North Carolina, and Forsyth Medical Center. ∥ Research Coordinator, Department of Anesthesiology, Wake Forest University School of Medicine, and Forsyth Medical Center. # FM James III Professor, Department of Anesthesiology, Wake Forest University School of Medicine, and Forsyth Medical Center.
Article Information
Meeting Abstracts   |   March 2006
Multifactorial Preoperative Predictors for Postcesarean Section Pain and Analgesic Requirement
Anesthesiology 3 2006, Vol.104, 417-425. doi:
Anesthesiology 3 2006, Vol.104, 417-425. doi:
INDIVIDUAL variability in severity of postoperative pain is influenced by multiple factors, including sensitivity to pain, psychological factors (e.g.  , state of anxiety and somatization), age, and genetics.1–4 Such factors are thought to drive both interpatient variability in the postoperative pain experience as well as requirements for analgesics after the same or similar surgical procedure.1–3 Despite advancement in postoperative pain management, postoperative pain relief and patient satisfaction are still inadequate in some patients.1–3 In addition, a high percentage of hospitalized patients continue to experience moderate to severe postoperative pain.5 Recently, the Joint Commission on Accreditation of Healthcare Organization established standards for pain assessment and treatment in healthcare facilities with the goal to generate uniformly low pain scores of no more than 3 of 10 both at rest and with movement. Furthermore, improved pain relief may improve postoperative outcomes, whereas severe postoperative pain may lead to development of chronic pain.6,7 The visual analog scale (VAS) has been validated for pain assessment, allowing a reliable and consistent measure and study of pain intensity and unpleasantness.8,9 Accurate prediction of postoperative pain and/or analgesic requirement in an individual patient could lead to tailored methods to reduce severe postoperative pain and improve acute and chronic outcomes.
Previous studies have observed weak but statistically significant correlations between degree of postoperative pain and preoperative conditions, such as thermal or pressure pain threshold and tolerance, anxiety, expectation, and age.1–4,10–12 Granot et al.  10 reported, for example, that postoperative VAS pain scores at rest and during activity were significantly correlated (r = 0.4–0.5) with preoperative pain scores to suprathreshold heat stimuli, whereas there was no such correlation with pain threshold to heat. Similarly, VAS pain on the first postoperative day is correlated with pressure pain tolerance (r =–0.5).12 In addition to response to experimental pain, anxiety, somatization, and catastrophizing are predictive single factors of postoperative pain.13 
Few previous studies have explored multiple psychological and physical responses to generate a multifactorial predictive model for postoperative pain. Similarly, prior experience, patient expectations, and VAS use rarely have been included together in such studies, and never in a comprehensive fashion. In addition, few previous studies have examined predictive factors for analgesic drug use, yet this is a particularly interesting outcome, because it is largely controlled by the patients in the postoperative period and because opioid drug dose correlates with bothersome side effects.14,15 Finally, chronic nociception can result in altered response to experimental pain stimuli in affected dermatomes, including low thoracic and upper lumbar dermatomes at term pregnancy,16 yet previous studies have performed experimental pain testing before cesarean section in the arm, distant from these dermatomes.10–12 The purpose of the current study was to determine factors that predict postoperative pain and analgesic drug use after cesarean section using a far-ranging assessment of pain threshold and response to suprathreshold thermal stimuli in both cervical and lumbar dermatomes as well as degree of somatization and anxiety.
Materials and Methods
Study Population
With approval from institutional review boards (Wake Forest University and Forsyth Medical Center, Winston-Salem, NC), written informed consent was obtained from 34 healthy American Society of Anesthesiology Class I –II parturients at normal term pregnancy and who were scheduled for elective cesarean section under subarachnoid anesthesia. Patients with a history of psychiatric disease, preeclamptic or hypertensive disease, hearing deficits, alcohol or drug abuse, chronic opioid or nonsteroidal antiinflammatory drug use, or weight over 250 lbs and non-English speaking parturients were excluded.
Preoperative Assessment Preparation
Preoperative testing was performed within 2 weeks before the cesarean section in a quiet room. Patients received a detailed explanation of each testing procedure and were trained in the use of the VAS to report pain intensity and unpleasantness (one end indicating no pain or not at all unpleasant and the other end indicating the most intense pain imaginable or the most unpleasant imaginable) using a radio analogy.
Assessment of Emotional and Psychological State
All patients completed the State Trait and Anxiety Inventory (STAI), a validated tool that measures trait (20 questions) and state (20 questions) anxiety. Patients also completed a questionnaire assessing their views about use of medications, preoperative anxiety, baseline pain, unpleasantness during pregnancy, their expectations of surgical and postsurgical pain, and their expected amount of postoperative pain medication needed.
Assessment of Audio Sensitivity
The patients rated the intensity (loudness) of a series of audio tones using the mechanical VAS. This trained the patients in using the scale to report intensity of stimuli and to evaluate the patient’s scale use ability.
Assessment of Thermal Pain Sensitivity and Threshold
Thermal stimuli were administered to the ventral surface of the dominant forearm and then to the lower back with a Thermal Sensory Analyzer (Model TSA II; Medoc Inc., Ramat Yishai, Israel) via  a 16 × 16-mm Peltier contact thermode. Each thermal testing trial was separated by approximately 2 min and at a separate site to avoid sensitization or adaptation.
Thermal Pain Threshold Temperature
Thermal pain threshold was determined by methods of limit. The thermode was applied to the ventral surface of the dominant forearm or the lower back and the temperature was increased at 1°C/s from 35 to 50°C. Patients were asked to indicate the transition point at which the nonpainful warm sensation changed into a painful heat sensation by pressing a button, and the simultaneous thermode temperature was recorded electronically. This test was repeated successively six times and the mean threshold temperature was calculated.
Suprathreshold Thermal Pain Intensity and Unpleasantness
Stimulus responses for noxious heat stimuli were performed by applying phasic heat stimuli at eight different temperatures (35, 43, 44, 45, 46, 47, 48, and 49°C). Each stimulus started at the level of 35°C and increased at a rate of 1°C/s. After reaching the destination temperature, the stimulus remained for 5 s at that temperature, and then it rapidly decreased back to baseline at a rate of 8°C/s. The interstimulus interval between successive stimuli was 30 s. The various temperature stimuli were administered first to the ventral surface of the dominant forearm and then the lower back each four times, with the first time in ascending temperature, the second and third times in random order of temperature, and the fourth time in descending order of temperature. Patients were asked to report the level of perceived pain intensity and unpleasantness, respectively, by means of the VAS.
Protocol for Cesarean Anesthetic and Postoperative Pain Management
On the day of surgery while in the preoperative holding area, vital signs, anxiety score using VAS (limits of “not anxious at all” and “the most anxious imaginable”), and fetal heart rate were obtained. After intravenous infusion of 1 l lactated Ringer’s solution, a 27- or 25-gauge Whitacre spinal needle was inserted at the L3–L4 or L4–L5 interspace with patients in sitting position, and 11.25 mg hyperbaric 0.75% bupivacaine and 20 μg fentanyl were administered intrathecally. Intravenous morphine or fentanyl was administered as needed in the case of intraoperative discomfort, but no anxiolytics were administered during surgery during the study period. After surgery, patients received either 50 mg rofecoxib or 800 mg ibuprofen orally shortly after arrival in the recovery room and intravenous morphine at 2- to 4-mg bolus increments as needed in the recovery room for verbal pain score of more than 3 of 10 and/or at patient request for pain. Ibuprofen 800 mg was administered every 8 h, whereas rofecoxib was administered once during the first postoperative 24 h. Patients were discharged from the recovery room to the ward with a verbal numeric pain score of 3 or less of 10. On arrival on the ward, intravenous patient-controlled analgesia (PCA) was initiated using a 3-mg morphine loading dose a 1-mg demand dose, with an 8-min lock out and a maximum limit of 8 mg/h. PCA settings could be increased to a maximum hourly limit of 16 mg if the initial settings were inadequate for postoperative pain control.
Cesarean Section and Postcesarean Section Pain Assessment
Sensory level to pin-prick testing was obtained just before skin incision, at the end of surgery, and 2 h after injection of subarachnoid anesthetic. Duration of surgery (from incision to end of surgery procedure), duration of anesthetic (time from subarachnoid anesthetic injection to time when sensory blockade level receded to a T10 or lower level), time to first request for supplement analgesia, amount of supplement analgesia required during surgery and after surgery in the recovery room, and in the first 6 and 18 to 24 postoperative hours were recorded. On the morning of the first postoperative day (20 to 24 h after surgery), patients rated the intensity of pain at the surgical wound by means of the VAS, both at rest and during activity (changing from supine to sitting up at bedside: evoked pain). Patients also rated the overall level of pain since surgery, effectiveness of PCA pain relief, overall postoperative pain control satisfaction, and overall satisfaction with the cesarean surgery experience, each using the VAS.
Statistical Analysis
Statistical analyses were conducted using SPSS software version 13.0 (SPSS Inc., Chicago, IL). Descriptive statistics were calculated for all variables (i.e.  , mean, SD, median, and range) as appropriate. Because the medication outcome variables were positively skewed, a nonparametric Spearman correlation was used to examine the relationship between predictive factors and outcome variables. For all analyses, α was set at 0.05 for statistical significance.
To examine if predictors could be reduced into meaningful subsets based on their relationship to each other, principal component factor analysis with varimax rotation was used. Factor analysis often is used as a data reduction technique or as a method to mathematically identify meaningful subgroups of items. Care was taken to create a solution of item groups (factors) that were very similar to each other, but relatively uncorrelated. Because factor analysis is typically conducted on much larger samples than that of the current exploratory study, a minimally acceptable factor loading of 0.80 or more was used to better ensure the stability of identified factors. Predictors were combined to form factor scores by summing the individual items in the factor. Reliability estimates of the factors were conducted using Cronbach α.
To examine if groups of predictors (factors) could be used in conjunction to improve prediction of each outcome variable (pain and medication use), multiple regression analyses were used. A model-building strategy was used that attempted to account for the greatest amount of variance in the outcome variable while considering only factors that were statistically significant (or likely to be statistically significant with a larger sample). Any variable that was independently correlated with the outcome variable was entered into the equation, but was removed if not statistically significant when considered with the other predictors. In essence, this strategy aims at maximizing the adjusted r2value, or variance accounted for by the model if applied to a different sample. This conservative strategy resulted in parsimonious regression models that are likely to be replicated.
Finally, to index the predictive ability of the regression models and for clinical applicability, a receiver operating characteristic curve was used to estimate sensitivity and specificity of the models’ ability to predict the upper twentieth percentile of individuals’ pain and medication use. The optimal cutoff value of the prediction equation was chosen such that sensitivity and specificity were as high as possible, while still remaining balanced.
Results
The 34 parturients enrolled in this study had a mean (SD) body weight of 90 (19) kg, height of 161 (8) cm, and a median (range) parity of 1 (1–3). The mean (SD) cesarean surgery duration was 45 (14) min. There was a large interindividual variability in the results of all preoperative thermal pain tests as well as the ranges of postoperative pain scores and analgesic requirement. Complete descriptive statistics for predictor and outcome variables are provided in table 1.
Table 1. Descriptive Statistics of Main Predictor Variables and Outcome Variables 
Image not available
Table 1. Descriptive Statistics of Main Predictor Variables and Outcome Variables 
×
Preoperative Assessment
The mean (SD) thermal pain threshold on the arm and back were 47 (3)°C and 47 (2)°C, respectively. The mean (SD) VAS pain intensity and unpleasantness to the 49°C stimulus were 33 (27) and 33 (29) for the arm, respectively, and 22 (24) and 25 (28) for the back, respectively. The mean (SD) pain threshold temperature for the arm and the back were 47.5 (3.1)°C and 47.3 (2.3)°C, respectively.
Postcesarean Section Pain Assessment and Analgesic Requirement
The mean (SD) VAS were 20 (20), 50 (28), and 46 (24) for postcesarean section resting pain, evoked pain, and overall pain, respectively. The mean (SD) analgesic requirements were 2.9 (5.4), 18.8 (15.0), 13.8 (7.2), and 36.1 (18.7) mg morphine equivalents for intraoperative, recovery room, first 6 h of intravenous PCA consumption on the ward, and total analgesic required for the three periods above combined, respectively.
Predictive Model Using Factor Analysis
Tables 2 and 3display the results of the principal component factor analysis. The analysis resulted in six relatively orthogonal (uncorrelated) newly formed predictive factors (groups) that accounted for 90% of the total observed variances in predictors. It is of note that the STAI was moderately to highly correlated with many of the factors and was removed from the factor solution and used as a separate predictor category (although not orthogonal to the others). The usefulness of these factors then was examined in regression models for predicting the outcome measures in postoperative pain and analgesic requirement. Each factor was given a name consistent with its several independent measured component variables and its clinical relevance as appropriate (table 2).
Table 2. The Six Predictive Factors (Groups) and Their Corresponding Component Predictor Variables 
Image not available
Table 2. The Six Predictive Factors (Groups) and Their Corresponding Component Predictor Variables 
×
Table 3. Final Principal Components Solution of Predictor Variables (with Varimax Rotation), with Reliability Estimates 
Image not available
Table 3. Final Principal Components Solution of Predictor Variables (with Varimax Rotation), with Reliability Estimates 
×
The reliability estimates (Cronbach α) of the individual components forming each predictor factor (group) are shown in table 3. An α of 0.8 or more was used as an acceptable reliability to include the component in the corresponding factor (group) for final regression analyses. Because two factors (thermal pain threshold, intraoperative factors) did not meet that criterion, the single item most highly related to those factors was considered independently for the purposes of regression analyses.
The outcome measures were:
  • Postoperative Pain. The outcome variables were the VAS (0–100) scores on postoperative resting pain, evoked pain, and overall pain rating and a calculated composite pain score, which was the sum of the three VAS scores for resting, evoked, and overall pain rating.

  • Analgesic Requirement. The outcome variables were the milligrams of morphine equivalents required for analgesia during intraoperative period, recovery room period, and the first 6 h of PCA in the ward and the total analgesia required that was the sum of analgesia consumed during the above three periods.

Spearman Correlations between Pain and Analgesic Outcome Variables and Individual Predictor Factors
The Spearman correlation between the pain outcome variables and each individual predictor factor (group) is shown in table 4. Table 5shows the Spearman correlation between each individual predictor factor (group) and the analgesic outcome variables. It is of interest to note that evoked pain and resting pain were best predicted by different predictor factors (groups). Similarly, intraoperative analgesia and recovery room analgesia also were best predicted by different predictor factors.
Table 4. Spearman Correlations between Pain Outcomes and Individual Predictor Factors (Groups) 
Image not available
Table 4. Spearman Correlations between Pain Outcomes and Individual Predictor Factors (Groups) 
×
Table 5. Spearman Correlations between Analgesic Outcome Variables and Individual Predictor Factors (Groups) 
Image not available
Table 5. Spearman Correlations between Analgesic Outcome Variables and Individual Predictor Factors (Groups) 
×
Final Predictive Model with Multiple Regression Analysis
Final predictive model with multiple regression analysis for pain outcomes and analgesic outcomes are shown in tables 6 and 7, respectively. Two predictive factors (pain and unpleasantness and expectation) provided the best predictive model for resting pain (r2= 0.26, P  < 0.01). Thermal pain threshold in the back area alone provided the best predictive model for evoked pain (r2= 0.20, P  < 0.009), whereas two factors (pain and unpleasantness and blood pressure) fitted the best predictive model for the composite pain score (r2= 0.28, P  < 0.008); fig. 1and table 6).
Table 6. Multiple Regression Analyses for Each Pain Outcome (Final Model) 
Image not available
Table 6. Multiple Regression Analyses for Each Pain Outcome (Final Model) 
×
Table 7. Multiple Regression Analyses for Each Analgesic Use Outcome (Final Model) 
Image not available
Table 7. Multiple Regression Analyses for Each Analgesic Use Outcome (Final Model) 
×
Fig. 1. Graphs showing the standardized predictive value of linear composite of predictors  versus  postoperative visual analog scale pain scores for (  A  ) resting, (  B  ) evoked, and (  C  ) composite pain. 
Fig. 1. Graphs showing the standardized predictive value of linear composite of predictors  versus  postoperative visual analog scale pain scores for (  A  ) resting, (  B  ) evoked, and (  C  ) composite pain. 
Fig. 1. Graphs showing the standardized predictive value of linear composite of predictors  versus  postoperative visual analog scale pain scores for (  A  ) resting, (  B  ) evoked, and (  C  ) composite pain. 
×
For predicting analgesic requirement, the factor of preexisting pain alone provided the best predictive model for intraoperative analgesic need (r2= 0.22, P  < 0.006), whereas two factors (thermal pain threshold and STAI) fitted best for recovery room analgesia (r2= 0.27, P  < 0.01) and STAI total score alone for the predictive model of total analgesic need (r2= 0.22, P  < 0.01; fig. 2and table 7).
Fig. 2. Graphs showing the standardized predictive value of the linear composite of predictors  versus  postoperative analgesia outcomes for (  A  ) recovery room analgesia and (  B  ) total analgesia. 
Fig. 2. Graphs showing the standardized predictive value of the linear composite of predictors  versus  postoperative analgesia outcomes for (  A  ) recovery room analgesia and (  B  ) total analgesia. 
Fig. 2. Graphs showing the standardized predictive value of the linear composite of predictors  versus  postoperative analgesia outcomes for (  A  ) recovery room analgesia and (  B  ) total analgesia. 
×
When dichotomizing the pain (composite pain score) outcome to predict the upper twentieth percentile of individuals’ pain using a receiver operating characteristic curve, this multifactorial predictive model yielded an optimal sensitivity of 0.71 and specificity of 0.76 in predicting the composite pain score outcome. Similarly for total analgesic requirement, the predictive model yielded a sensitivity of 0.80 and a specificity of 0.80.
Discussion
These results suggest that factor analysis with a thoughtful combination of predictors yields a model with significant improvements over single variables in predicting pain and analgesic requirement outcomes after cesarean section. Thermal pain sensitivity study is an accepted method for experimental pain assessment.17,18 Granot et al.  10 reported postoperative pain VAS score at rest and during activity to be significantly correlated (r = 0.4–0.5) with preoperative pain VAS score on the arm to thermal stimuli, especially at 48°C, but failure of thermal pain threshold to correlate with postoperative pain. Our results with thermal testing in the arm are consistent with those findings, but we enlarge on them by concurrent measurement of heat pain responses to the lower back. Bajaj et al.  16 reported hypoalgesia in late pregnancy to be generalized in women with pelvic pain, but localized to presumed referred pain sites in the lower back and sacrum in women without pelvic pain. The authors suggest that descending noxious inhibition may be activated in late pregnancy and is probably more intense and generally activated in women with pelvic pain but only segmentally activated in women without pain. Our data did not find preexisting pain during pregnancy to be a strong predictor factor for postoperative pain. However, the thermal pain threshold in the lower back near the dermatomes of the surgical wound interestingly was the most important predictive factor for evoked pain and for recovery room analgesia requirement after cesarean section, consistent with variability in homosegmental inhibition at term pregnancy as a determinant for postinjury pain.
In contrast to evoked pain after cesarean section, the degree of pain at rest was predicted by two other factors: pain and unpleasantness and patient expectation. This may suggest that the mechanisms of resting and evoked pain are different and that psychological aspects may play a significantly greater role in resting pain than evoked pain. It is conceivable that different treatments or combinations of treatments may be more appropriate to control resting pain as compared with evoked pain.
Surgery and postoperative recovery evoke significant physiologic and psychological reactions. Jamison et al.  19 found significant correlations between stress and postoperative analgesic usage, whereas Hansson et al.  20 did not. These conflicting results and those of others on stress and postoperative pain and analgesic drug use may reflect in part interstudy differences in patient gender, surgical procedures, and presence or absence of pregnancy.19–23 Cesarean section is a unique surgery, accompanied by the significant hormonal and emotional changes associated with pregnancy, arrival of the newborn with attendant care responsibilities, expectation to recover rapidly, and sleep deprivation resulting from maternal neonatal interactions. Patients’ expectations and their pain experience during pregnancy or with previous surgery also may impact postoperative pain experience. For example, preoperative expectation of pain is correlated with postoperative pain intensity in nonpregnant patients.22 Expectations powerfully modulate both subjective reports of acute pain and pain-induced brain activity through mechanisms likely involving the prefrontal cortex, anterior cingulated cortex, and the anterior insular cortex.24 In our study, regression analysis also identified preoperative expectation to be a key component to predict resting pain but not evoked pain.
Both pain and analgesic drug use vary widely after surgery, and these share a complex relationship. Aubrun et al.  ,14 in a study of 3,045 patients, reported the relationship between postoperative pain and morphine requirements was not linear and was best described as sigmoidal. Therefore, predictors for postoperative analgesic requirement may be different from those for postoperative pain intensity. Our results are consistent with this presumption. Furthermore, the key predictive factor for postoperative analgesia requirement may differ even for different stages of recovery or activity. For example, preexisting pain during pregnancy was the key predictive factor for intraoperative analgesic drug use in the current study, whereas thermal pain threshold and STAI were key factors predicting analgesic drug use in the recovery room, and STAI was the key predictor for total analgesic drug use. Hsu et al.  12 reported a significant correlation between postoperative analgesic requirements and the effect of a probe dose of fentanyl on experimental pressure pain tolerance. It is conceivable that such pharmacodynamic predictors could further improve our multifactorial model.
Postoperative systemic opioid analgesia consumption varied widely with a significant interindividual variability. Macintyre and Jaarvis4 reported a 10-fold difference in postoperative morphine consumption, which also is consistent with findings in our study. It is not known whether this large interindividual variability in response to systemic opioids also exists for other forms of analgesia or multimodal analgesia. Therefore, application of regression analysis with meaningful subsets of variables to predict postoperative pain and analgesic requirement may be useful in individualizing analgesic methods and identifying patients at higher risk for severe postoperative pain and high analgesic requirement who may then benefit from more intensive therapy (be it higher dose or multimodal analgesia) to reduce variability in postoperative pain and further improve comfort and quality of care.
There are a number of limitations in our exploratory study other than the small sample size. The unanticipated withdrawal of rofecoxib by Merck during our study period led us to switch to the use of 50 mg rofecoxib once to 800 mg ibuprofen every 8 h during the first 24 h after surgery. These two regimens generally are considered to have similar analgesic effects during the study period.25 We also allowed opioid analgesic supplement during surgery when needed or requested by patients for pain and discomfort. Even though the average amount of discretional supplemental opioid used during surgery was small (2.9 mg of morphine equivalents), this may make the recovery room analgesic need difficult to interpret. However, denial of opioid analgesic supplement when needed or when requested for pain and discomfort would not represent normal good practice and may make the study enrollment more difficult or may even be deemed unethical. Instead, in the design of our study, we included the total analgesic use (intraoperative, recovery room, and ward intravenous PCA use) as one of the main outcome measure that would have taken the intraoperative supplement analgesic use into consideration. The pain scores assessed on postoperative day 1 should also not be affected significantly by the small amount of intraoperative opioid supplement. Even with a relatively small sample size in this exploratory study, statistical significance in predictive power and correlations were observed, as well as clinical relevance. However, results of this study must be applied with caution within the limitations of the study, and also a significant part of the variability in the outcomes measured still cannot be predicted fully by our existing multifactorial predictive model. Further research is needed to determine a more complete multifactorial prediction model for postoperative pain and analgesic required.
In conclusion, our results suggest that a multifactorial predictive model with factor analysis, using a meaningful combination of patients’ responses to thermal sensitivity tests and to psychological questionnaires, provides a reliable method in predicting postoperative pain and analgesic requirement outcomes after cesarean section. Furthermore, different predictors may be useful for predicting different types of postoperative pain and analgesic requirement. These findings may have important clinical implications to help identify patients at risk for postoperative hyperalgesia who may then benefit from tailored treatment to improve outcomes and quality of care.
The authors thank Lori Bailey (Administrative Assistant, Forsyth Medical Center Winston-Salem, North Carolina) for her technical assistance in the preparation of the manuscript and Gina Curry, R.N. (Nurse Clinician III, Department of Anesthesiology, Wake Forest University School of Medicine, Winston-Salem, North Carolina), for her technical assistance in the study. The authors also thank Terry Bogard, M.D. (Associate Professor), Robert D’Angelo, M.D. (Associate Professor), Laura Dean, M.D. (Assistant Professor), David Hood, M.D. (Professor), Kenneth Nelson, M.D. (Assistant Professor), Medge Owen, M.D. (Associate Professor), Vernon Ross, M.D. (Assistant Professor), and John Thomas, M.D. (Assistant Professor, all from Department of Anesthesiology, Wake Forest University School of Medicine, and Forsyth Medical Center, Winston-Salem, North Carolina), for making this study possible.
References
Turk DC, Okifuji A: Assessment of patients’ reporting of pain: An integrated perspective. Lancet 1999; 352:1784–8Turk, DC Okifuji, A
Lynch EP, Lazor MA, Gellis JE, Orav J, Goldman L, Marcantonio ER: Patient experience of pain after elective non cardiac surgery. Anesth Analg 1997; 85:117–25Lynch, EP Lazor, MA Gellis, JE Orav, J Goldman, L Marcantonio, ER
Bisgaard T, Klarskov B, Rosenberg J, Kehlet H: Characteristics and prediction of early pain after laparoscopic cholecystectomy. Pain 2001; 90:261–9Bisgaard, T Klarskov, B Rosenberg, J Kehlet, H
Macintyre PE, Jaarvis DA: Age is the best predictor of postoperative morphine requirement. Pain 1996; 64:357–64Macintyre, PE Jaarvis, DA
Dolin ST, Cashman JN, Bland JM: Effectiveness of acute postoperative pain management: I. Evidence from published data. Br J Anaesth 2002; 89:409–23Dolin, ST Cashman, JN Bland, JM
Perkins FM, Kehlet H,: Chronic pain as an outcome of surgery: A review of predictive factors. Anesthesiology 2000; 95:1123–33Perkins, FM Kehlet, H
Callesen T, Bech K, Kehlet H: Prospective study of chronic pain after groin hernia repair. Br J Surg 1999; 86:1528–33Callesen, T Bech, K Kehlet, H
Price DD, McGrath PA, Rafii A, Buckingham B: The validation of visual analogue scales as ratio scale measures for chronic and experimental pain. Pain 1983; 17:45–56Price, DD McGrath, PA Rafii, A Buckingham, B
Rosier EM, Iadarola MJ, Coghill RC: Reproducibility of pain measurement and pain perception. Pain 2002; 98:205–16Rosier, EM Iadarola, MJ Coghill, RC
Granot M, Lowenstein AL, Yarnitsky D, Tamir A, Zimmer EZ: Postcesarean section pain prediction by preoperative experimental pain assessment. Anesthesiology 2003; 98:1422–6Granot, M Lowenstein, AL Yarnitsky, D Tamir, A Zimmer, EZ
Werner MU, Dunn P, Kehlet H: Prediction of postoperative pain by preoperative nociceptive responses to heat stimulation. Anesthesiology 2004; 100:115–9Werner, MU Dunn, P Kehlet, H
Hsu Y, Somma J, Hung Y, Tsai P, Yang C, Chen C: Predicting postoperative pain by preoperative pressure pain assessment. Anesthesiology 2005; 103:613–8Hsu, Y Somma, J Hung, Y Tsai, P Yang, C Chen, C
Robinson ME, Riley JL 3rd, Meyers CD: Psychosocial contributions to sex-related differences in pain responses, Sex, Gender and Pain. Edited by Fillingim RB. Seattle, IASP Press, 2000, pp 41–68Robinson, ME Riley, JL Meyers, CD Fillingim RB Seattle IASP Press
Aubrun F, Langeron O, Quesnel C, Coriat AP, Riou B: Relationships between measurement of pain using visual analog score and morphine requirements during postoperative intravenous morphine titration. Anesthesiology 2003; 98:1415–21Aubrun, F Langeron, O Quesnel, C Coriat, AP Riou, B
Berde CB, Brennan TJ, Raja SN: Opioids: More to learn, improvements to make. Anesthesiology 2003; 98:1309–12Berde, CB Brennan, TJ Raja, SN
Bajaj P, Bajaj P, Madsen H, Moller M, Arendt-Nielsen L: Antenatal women with or without pelvic pain can be characterized by generalized or segmental hypoalgesia in late pregnancy. J Pain 2002; 6:451–60Bajaj, P Bajaj, P Madsen, H Moller, M Arendt-Nielsen, L
Yarnitsky D, Sprecher E, Zaslansky R, Hemli JA: Multiple session experimental pain measurement. Pain 1996; 67:237–33Yarnitsky, D Sprecher, E Zaslansky, R Hemli, JA
Yarnitsky D, Ochoa JL: Studies of heat pain sensation in man: Perception thresholds, rate of stimulus rise and reaction time. Pain 1990; 40:85–91Yarnitsky, D Ochoa, JL
Jamison RN, Taft K, O’Hara JP, Ferrante FM: Psychosocial and pharmacologic predictors of satisfaction with intravenous patient-controlled analgesia. Anesth Analg 1993; 77:121–5Jamison, RN Taft, K O’Hara, JP Ferrante, FM
Hansson P, Ekblom A, Thomsson M, Fjellner B: Pain development and consumption of analgesics after oral surgery in relation to personality characteristics. Pain 1989; 37:271–7Hansson, P Ekblom, A Thomsson, M Fjellner, B
Taenzer P, Melzack R, Jeans ME: Influence of psychological factors on postoperative pain, mood and analgesic requirements. Pain 1986; 24:331–42Taenzer, P Melzack, R Jeans, ME
Svensson I, Sjostrom B, Haljamae H: Influence of expectations and actual pain experiences on satisfaction with postoperative pain management. Eur J Pain 2001; 5:125–33Svensson, I Sjostrom, B Haljamae, H
Lamberg L: Girls’ and boys’ differing response to pain starts early in their lives. JAMA 1998; 280:1035–6Lamberg, L
Koyma T, McHaffie JG, Laurienti PJ, Coghill RC: The subjective experience of pain: Where expectations become reality. Proc Natl Acad Sci U S A 2005; 102:12950–5Koyma, T McHaffie, JG Laurienti, PJ Coghill, RC
Gotzsche PC: Non-steroidal anti-inflammatory drugs. Clin Evid 2003; 9:1292–300Gotzsche, PC
Fig. 1. Graphs showing the standardized predictive value of linear composite of predictors  versus  postoperative visual analog scale pain scores for (  A  ) resting, (  B  ) evoked, and (  C  ) composite pain. 
Fig. 1. Graphs showing the standardized predictive value of linear composite of predictors  versus  postoperative visual analog scale pain scores for (  A  ) resting, (  B  ) evoked, and (  C  ) composite pain. 
Fig. 1. Graphs showing the standardized predictive value of linear composite of predictors  versus  postoperative visual analog scale pain scores for (  A  ) resting, (  B  ) evoked, and (  C  ) composite pain. 
×
Fig. 2. Graphs showing the standardized predictive value of the linear composite of predictors  versus  postoperative analgesia outcomes for (  A  ) recovery room analgesia and (  B  ) total analgesia. 
Fig. 2. Graphs showing the standardized predictive value of the linear composite of predictors  versus  postoperative analgesia outcomes for (  A  ) recovery room analgesia and (  B  ) total analgesia. 
Fig. 2. Graphs showing the standardized predictive value of the linear composite of predictors  versus  postoperative analgesia outcomes for (  A  ) recovery room analgesia and (  B  ) total analgesia. 
×
Table 1. Descriptive Statistics of Main Predictor Variables and Outcome Variables 
Image not available
Table 1. Descriptive Statistics of Main Predictor Variables and Outcome Variables 
×
Table 2. The Six Predictive Factors (Groups) and Their Corresponding Component Predictor Variables 
Image not available
Table 2. The Six Predictive Factors (Groups) and Their Corresponding Component Predictor Variables 
×
Table 3. Final Principal Components Solution of Predictor Variables (with Varimax Rotation), with Reliability Estimates 
Image not available
Table 3. Final Principal Components Solution of Predictor Variables (with Varimax Rotation), with Reliability Estimates 
×
Table 4. Spearman Correlations between Pain Outcomes and Individual Predictor Factors (Groups) 
Image not available
Table 4. Spearman Correlations between Pain Outcomes and Individual Predictor Factors (Groups) 
×
Table 5. Spearman Correlations between Analgesic Outcome Variables and Individual Predictor Factors (Groups) 
Image not available
Table 5. Spearman Correlations between Analgesic Outcome Variables and Individual Predictor Factors (Groups) 
×
Table 6. Multiple Regression Analyses for Each Pain Outcome (Final Model) 
Image not available
Table 6. Multiple Regression Analyses for Each Pain Outcome (Final Model) 
×
Table 7. Multiple Regression Analyses for Each Analgesic Use Outcome (Final Model) 
Image not available
Table 7. Multiple Regression Analyses for Each Analgesic Use Outcome (Final Model) 
×