Chronic pain after breast cancer surgery is widely recognized with a reported incidence range of between 20% and >50%.1–10 Because chronic pain can seriously affect a patient’s daily life, chronic pain after breast cancer surgery has become an important medical issue.1–3 Many investigators have suggested several risk factors associated with chronic pain after breast cancer surgery, including younger age,2,4,5 greater height and weight,3,6 low level of social support,5 genetic susceptibility,7 type of surgery,1,6,8 location of the breast cancer,2 axillary lymph node dissection,1,4,9 radiotherapy,1,4 chemotherapy,2 and acute pain after breast cancer surgery.1,7,10
The efforts of anesthesiologists to prevent chronic pain after surgery have to be focused on acute postoperative pain management. Evidence continues to accumulate concerning the role of sensitization in chronic pain after surgery. Meanwhile, preventive and multimodal analgesia have been introduced to reduce central sensitization and this has provided an overall benefit on reducing both acute and chronic postoperative pain.11,12 In addition, a variety of preventive and multimodal analgesic techniques have been used pre-, intra-, and postoperatively.13–15
Previously, Shin et al.16 concluded that remifentanil hyperalgesia was induced by high-dose remifentanil-based anesthesia during sevoflurane anesthesia, whereas that was not apparent during propofol anesthesia. Also, remifentanil hyperalgesia did not occur during low-dose remifentanil-based anesthesia. The mechanism of opioid-induced hyperalgesia is still unclear but it is believed that central sensitization, as with the mechanism of chronic pain, has an important role.17
If general anesthetics affect acute postoperative pain, and the intensity of acute pain after breast cancer surgery influences the development of chronic pain, the choice of anesthetics could affect chronic postoperative pain, because of a similar modulation of opioid-induced hyperalgesia. Thus, we hypothesized that anesthetics affect acute pain and the development of chronic pain after breast cancer surgery. Our primary goal was to determine the effects of propofol versus sevoflurane on the incidence, severity, and duration of chronic pain after breast cancer surgery. Our secondary objective was to identify risk factors associated with the incidence and severity of chronic pain after breast cancer surgery.
The study was approved by the IRB of the Pusan National University Hospital as an exempted observational study, and written informed consent was waived. However, we provided a study information sheet to the potential participants in the breast clinic. The study flow diagram is presented in Figure 1. We investigated 228 women (n = 111 in the propofol group and n = 117 in the sevoflurane group) aged 20 to 65 years with an ASA physical status of I or II and who had undergone breast cancer surgery at our institution between March 2007 and December 2008. Exclusion criteria were any antidepressant intake due to psychiatric disorder; emigration; ketamine use; discontinued patient-controlled analgesia (PCA) with morphine because of intractable nausea and vomiting; and cancer relapse, occurrence of metastasis, or death. An investigator who was blinded to the patients’ clinical data interviewed 189 women by telephone on July 2011 using a standard questionnaire. A total of 175 women (n = 86 in the propofol group and n = 89 in the sevoflurane group) responded.
Demographic Data and Clinical Characteristics
The demographic data and clinical characteristics were obtained from clinical records. The anthropometric variables included age, weight, height, and body mass index (BMI). We also obtained type of anesthetics; intraoperative remifentanil consumption; duration of surgery; type of surgery; axillary lymph node dissection; 24-hour postoperative visual analog scale (VAS) score and morphine consumption; history of radiotherapy, chemotherapy, and hormone therapy; and length of time since surgery. BMI (kg/m2) categories were selected according to the World Health Organization’s West Pacific Region definitions18: underweight (<18.5), normal (18.5–22.9), overweight (23.0–24.9), and obese (≥25.0). Because there were few patients in the underweight category (1%–2% in each), we combined the underweight and normal weight categories, resulting in 3 BMI categories: normal (<23.0), intermediate (23.0–24.9), and obese (≥25.0).
Intraoperative anesthetic management and postoperative pain control were standardized as described in the study by Shin et al.16 Postoperative pain was controlled by a PCA device with morphine and rated using a VAS (0 cm = no pain; 10 cm = worst pain imaginable). The PCA was programmed to deliver demand doses of morphine sulfate 1.0 mg with a 20-minute lockout interval and continuous infusion of 1.0 mg/h. The 4-hour limit of morphine sulfate was set to not exceed 20 mg. If patient requirement or VAS scale exceeded 5, the patient was administered morphine 4 mg IV as a rescue analgesic. The choice of anesthetics was randomized because most subjects in this study participated in the trial conducted by Shin et al.16 Type of surgery was classified by invasiveness into mastectomy with or without local flap and mastectomy with latissimus dorsi or external oblique musculocutaneous flap. Because all patients received chemotherapy, they were categorized as chemotherapy with or without paclitaxel, which causes neuropathic pain.19 Hormone therapy included tamoxifen, Femara, and Arimidex.
The first question was whether the patient had pain for more than 2 months after surgery. If the answer was positive, we asked for the length of time that the pain had lasted after surgery and what the worst pain intensity was using a numeric rating scale (NRS) (0 = no pain; 10 = worst pain imaginable). The NRS was assigned to 3 categories according to severity: mild (1–3), moderate (4–6), and severe (7–10) pain. Because it was a telephone questionnaire, an NRS was used instead of a VAS. It has been reported that NRS and VAS have significant correlation and provide similar pain information.20,21 There were questions based on EuroQol-5D.22 We asked whether there were problems with performing self-care (washing or dressing oneself) and usual activities (work, study, housework, or leisure activities) to determine the impact of pain on daily life. We also asked about impaired mobility of the ipsilateral arm and shoulder caused by pain, and their pain-treatment experiences in the hospital. The questionnaire is shown in Appendix 1.
First, we compared propofol and sevoflurane groups for confounding. Incidence, severity, and duration of chronic pain and quality of life between propofol and sevoflurane groups were also compared. We then determined the factors associated with the incidence and severity of chronic pain. For the analysis of chronic pain severity, we restricted patients to those who reported chronic pain. We performed power analysis for the incidence of chronic pain in propofol and sevoflurane groups after data collection. Power analysis was done using OpenEpi calculator (freely available at www.openepi.com) based on the sample size table.23
All continuous variables were checked for normal distributions with skewness, kurtosis, Kolmogorov-Smirnov test, and Shapiro-Wilk test. Age, weight, and height were normally distributed (all P > 0.069) and were compared with Student t test and analysis of variance. Other continuous variables were analyzed using the Mann-Whitney U test and the Kruskal-Wallis test. The post hoc Mann-Whitney U test with Bonferroni correction was used in the case of significance with the Kruskal-Wallis test. For the post hoc test for 3 pain severity categories, a P value <0.0167 (0.05 divided by 3) was considered statistically significant. The χ2 test or Fisher exact test was used to compare groups with categorical variables. Those variables that showed significance were applied to the subsequent univariate and multivariate regression models. Age, intraoperative remifentanil consumption, duration of surgery, axillary lymph node dissection, 24-hour postoperative morphine consumption, radiotherapy, and type of anesthetics were included in the univariate and multivariate logistic regression models for the incidence of chronic pain. As Dexter et al.24 described in their study, scheduled duration of surgery should be used instead of actual duration because this can lead to bias in the logistic regression model. However, it was reasonable to use actual duration in our model because it was not significant in the model. The variables of 24-hour postoperative VAS score and morphine consumption were included in the univariate and multivariate ordinal logistic regression models for the severity of chronic pain. The proportional odds assumption was not violated in any of the univariate or multivariate ordinal logistic regression models. Unadjusted and adjusted odds ratio (OR) and 95% confidence intervals (CIs) were calculated, and the Wald test was used to test the overall significance of each variable. Because incidence of chronic pain is common, we calculated the approximate relative risk (RR) based on the formula of Zhang and Yu25 if OR was <0.5 or >2.5. For all tests, a P value <0.05 was considered statistically significant. Data were analyzed using SPSS 18.0 software (SPSS Inc., Chicago, IL).
Demographic Data and Clinical Characteristics
Demographic data and clinical characteristics are shown in Table 1. There were no significant differences between the propofol and sevoflurane groups. An exception was significantly higher VAS scores in the sevoflurane group compared with the propofol group.
Table 2 shows the results of the questionnaire. The overall incidence of chronic pain was 56.0% (n = 98) after breast cancer surgery. The incidence of chronic pain was significantly higher in the sevoflurane group compared with the propofol group. For severity and duration of chronic pain, there was no significant difference between the propofol and sevoflurane groups. More patients who had sevoflurane anesthesia visited the hospital for pain treatment and had arm and shoulder disability. Patients who had sevoflurane anesthesia experienced a greater significant impact of pain in their daily lives compared with patients who had propofol anesthesia.
Predictive Factors for the Development of Chronic Pain After Breast Cancer Surgery
Risk factor variables for the development of chronic pain are compared in Table 3. Four significant risk factors were identified in Table 4: younger age (OR 0.949, 95% CI 0.907–0.992), axillary lymph node dissection (approximated RR 1.591, 95% CI 1.204–1.898), 24-hour postoperative morphine consumption (OR 1.058, 95% CI 1.004–1.116), and sevoflurane (approximated RR 1.514, 95% CI 1.146–1.809).
Predictive Factors for the Severity of Chronic Pain After Breast Cancer Surgery
As shown in Table 5, frequencies of hospital visits for pain treatment, arm and shoulder disability, and effects on daily life are significantly different among the 3 pain severity categories. Post hoc tests revealed that the mild pain category showed significantly lower incidences in all variables than the moderate and severe categories. The moderate pain category also showed significantly lower incidences of hospital visits and effects on daily life than the severe pain category. Significant differences were found among the 3 pain severity categories for 24-hour postoperative VAS score and morphine consumption. Post hoc tests detected that the mild pain category had lower VAS scores and less morphine consumption compared with the severe category. Table 6 shows that there was an association between severity of chronic pain and 24-hour postoperative morphine consumption. An ordinal multivariate analysis showed a proportional OR of 1.175 (95% CI 1.001–1.379) for a 1-mg increase in 24-hour postoperative morphine consumption.
We found an overall incidence of 56% of patients with chronic pain after breast cancer surgery, which was significantly higher for sevoflurane anesthesia (67.4%) compared with propofol anesthesia (44.2%). The sevoflurane group was 1.514 times more likely to develop chronic pain than the propofol group (95% CI 1.146–1.809). Frequencies of hospital visits for pain treatment, ipsilateral arm and shoulder disability, and effects on daily life were significantly different. However, among patients with chronic pain, neither the severity (95% CI 0.516–7.419) nor duration (95% CI 0.106–1.007) differed between patients receiving sevoflurane or propofol. Therefore, patients with sevoflurane anesthesia developed chronic pain after breast cancer surgery with greater frequency, but the pain was not more severe nor the duration longer compared with propofol anesthesia.
The overall incidence of chronic pain in our study was higher than previously reported.1–10 This could have been due to chemotherapy received by all our patients, because chemotherapy is a risk factor for chronic pain after breast cancer surgery.2 Another reason for the higher incidence is that we asked patients whether they had persistent pain related to surgery for longer than 2 months. As a result of this question, there was a wide range of responses.
There have been several studies to identify risk factors for chronic pain after breast cancer surgery.1–10 However, this is the first attempt to include anesthetics into logistic regression models of chronic pain after breast cancer surgery. Moreover, the effects of inhaled and IV anesthetics on chronic pain were rarely compared. As was mentioned earlier, we began this study because propofol reduced acute pain under opioid-induced hyperalgesia,16 and severe acute pain could affect the development of chronic pain.1,7,10 However, anesthetics may make a difference in chronic pain not only by affecting severity of acute pain but also by independent (or direct) mechanisms. When multivariate analyses were conducted to control for association among the variables, including 24-hour postoperative morphine consumption and anesthetics, sevoflurane showed a stronger association with the development of chronic pain after breast cancer surgery compared with propofol. Thus, it seems that propofol has its own analgesic effects by preventing sensitization mechanisms.26,27 Although sevoflurane prevents sensitization by inhibiting the N-methyl-D-aspartate receptor, this effect cannot sufficiently prevent chronic pain.28
Anesthetics did not show significant effects on the severity of chronic pain. We are unsure of the interpretation of this finding. Propofol may inhibit development of chronic pain through modulation of sensitization, but once it develops, propofol might not contribute to reducing the severity. The lack of difference in severity of chronic pain between propofol and sevoflurane groups can also be explained by limited statistical power. The latter interpretation is a more reasonable explanation. Although not statistically significant, more patients with severe pain were in the sevoflurane group than the propofol group. Moreover, the sevoflurane group had significantly higher frequencies of hospital visits, more arm and shoulder disability, and daily life was affected to a greater extent.
In addition to sevoflurane, we found risk factors for the development of chronic pain after breast cancer surgery, such as age, axillary lymph node dissection, and 24-hour postoperative morphine consumption. Younger aged patients tend to have more aggressive breast cancer requiring more aggressive management, including surgical treatment, chemotherapy, and radiotherapy.29 During axillary lymph node dissection, significant nerve injury can occur, which leads to central sensitization and hyperalgesia associated with chronic pain.30 Severe acute postoperative pain has been reported to be associated with the development of chronic pain after breast cancer surgery1,7,10 and other types of procedures, such as thoracotomy,31 hernia,32 and limb amputation.33
There have been several studies to identify risk factors for chronic pain severity after breast cancer surgery4,5 using 2 categories, but no studies have been conducted to quantify the degree of association between those risk factors and 3 pain severity categories. Three pain severity categories are more widely used than 2 pain severity categories, because their clinical outcomes are different and therefore their treatment strategy should be different. First, we compared the duration of chronic pain, frequencies of hospital visits for pain treatment, ipsilateral arm and shoulder disability, and effects on daily life. There were significant differences among the 3 pain severity categories except for duration of chronic pain. Post hoc findings revealed there were significant differences between each category except for frequency of disability of the arm and shoulder. This suggests that the pain severity categories consider different qualities of life and that the categorization we used was clinically meaningful.
Using multivariate ordinal logistic regression analyses, 24-hour postoperative morphine consumption significantly affected the severity of chronic pain. It is interesting that the variables predicting development of chronic pain after breast cancer surgery, except 24-hour postoperative morphine consumption, did not affect pain severity. The mechanisms that govern the development and intensity of chronic pain seem different; however, this may be partially explained by a limitation of our study. We recruited 175 patients but only 98 patients (56.0%) developed chronic pain after breast cancer surgery. Only 23 patients developed moderate pain and 27 patients developed severe pain, both of which were relatively small compared with the mild pain category. This limited statistical power may explain why variables other than 24-hour postoperative morphine consumption failed to predict chronic pain severity. Because morphine consumption showed a significant relationship to pain severity, it is reasonable to assume that severe acute pain is a strong predictor of chronic pain severity, as in the development of chronic pain.
Our study has several limitations. As we mentioned above, limited statistical power may be a reason that factors influencing the incidence of chronic pain failed to predict pain severity. However, power of incidence of chronic pain between propofol and sevoflurane was calculated after data collection and the analysis revealed 87.76%. Second, our study includes a cross-sectional study that provided a questionnaire at a single point, 2.5 to 4 years after breast cancer surgery. Because of the nature of our study, interpretations of the results must be made with caution. The information obtained from the patients does not provide definite cause-and-effect relationships, and the long time interval between surgery and answering the questionnaire may have caused recall bias. Third, we conducted this study in a single hospital and patients were treated by a single surgeon. All their treatments were well controlled by consistent guidelines. Thus, we do not know whether our model can be generalized to other settings because the development of chronic pain from surgical and medical treatments may be affected differently.
This study suggests that anesthetics during breast cancer surgery may affect the development of chronic pain. Propofol reduced the incidence of chronic pain compared with sevoflurane. Young age, axillary lymph node dissection, and more 24-hour postoperative morphine consumption increased the incidence of chronic pain after breast cancer surgery. However, the above factors did not increase the severity of chronic pain, and only higher 24-hour postoperative morphine consumption did. In conclusion, according to this study, propofol may have a contributing role in perioperative preventive and multimodal analgesic management, and further prospective studies are needed to confirm the validity of these provocative findings.
APPENDIX 1: QUESTIONNAIRE FORM
Questionnaire for Chronic Pain After Breast Cancer Surgery
1. Did you suffer from pain for more than 2 months after your surgery?□ Yes □ No
2. If your answer to 1 is positive, how long has the pain lasted after surgery?( ) months
3. If your answer to 1 is positive, what was the worst pain intensity? Score it using a numeric rating scale (0 = no pain; 10 = worst pain imaginable).( )
Questionnaire for Quality of Life
2. Have you had any problem in using your ipsilateral arm and shoulder?
3. □ Yes □ No
4. Self-care and usual activities
5. Have you had any problem washing or dressing yourself, or performing your usual activities (e.g., work, study, housework, family or leisure activities)?
6. □ Yes □ No
8. Have you visited a hospital to manage your pain?
9. □ Yes □ No
Name: Ah-Reum Cho, MD.
Contribution: This author helped analyze the data and write the manuscript.
Attestation: Ah-Reum Cho has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Name: Jae-Young Kwon, MD, PhD.
Contribution: This author helped design the study.
Attestation: Jae-Young Kwon has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Name: Kyung-Hoon Kim, MD, PhD.
Contribution: This author helped design the study.
Attestation: Kyung-Hoon Kim has seen the original study data and approved the final manuscript.
Name: Hyeon-Jeong Lee, MD, PhD.
Contribution: This author helped design the study.
Attestation: Hyeon-Jeong Lee approved the final manuscript.
Name: Hae-Kyu Kim, MD, PhD.
Contribution: This author helped analyze the data.
Attestation: Hae-Kyu Kim has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Name: Eun-Soo Kim, MD.
Contribution: This author helped conduct the study.
Attestation: Eun-Soo Kim has seen the original study data, approved the final manuscript, and is the author responsible for archiving the study files.
Name: Jung-Min Hong, MD.
Contribution: This author helped write the manuscript.
Attestation: Jung-Min Hong approved the final manuscript.
Name: Choongrak Kim, PhD.
Contribution: This author helped analyze the data.
Attestation: Choongrak Kim has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
This manuscript was handled by: Spencer S. Liu, MD.
1. Tasmuth T, Kataja M, Blomqvist C, von Smitten K, Kalso E. Treatment-related factors predisposing to chronic pain in patients with breast cancer: a multivariate approach. Acta Oncol. 1997;36:625–30
2. Vilholm OJ, Cold S, Rasmussen L, Sindrup SH. The postmastectomy pain syndrome: an epidemiological study on the prevalence of chronic pain after surgery for breast cancer. Br J Cancer. 2008;99:604–10
3. Macdonald L, Bruce J, Scott NW, Smith WC, Chambers WA. Long-term follow-up of breast cancer survivors with post-mastectomy pain syndrome. Br J Cancer. 2005;92:225–30
4. Gärtner R, Jensen MB, Nielsen J, Ewertz M, Kroman N, Kehlet H. Prevalence of and factors associated with persistent pain following breast cancer surgery. JAMA. 2009;302:1985–92
5. Poleshuck EL, Katz J, Andrus CH, Hogan LA, Jung BF, Kulick DI, Dworkin RH. Risk factors for chronic pain following breast cancer surgery: a prospective study. J Pain. 2006;7:626–34
6. Smith WC, Bourne D, Squair J, Phillips DO, Chambers WA. A retrospective cohort study of post mastectomy pain syndrome. Pain. 1999;83:91–5
7. Hickey OT, Nugent NF, Burke SM, Hafeez P, Mudrakouski AL, Shorten GD. Persistent pain after mastectomy with reconstruction. J Clin Anesth. 2011;23:482–8
8. Wallace MS, Wallace AM, Lee J, Dobke MK. Pain after breast surgery: a survey of 282 women. Pain. 1996;66:195–205
9. Steegers MA, Wolters B, Evers AW, Strobbe L, Wilder-Smith OH. Effect of axillary lymph node dissection on prevalence and intensity of chronic and phantom pain after breast cancer surgery. J Pain. 2008;9:813–22
10. Hickey OT, Burke SM, Hafeez P, Mudrakouski AL, Hayes ID, Shorten GD. Severity of acute pain after breast surgery is associated with the likelihood of subsequently developing persistent pain. Clin J Pain. 2010;26:556–60
11. Ong CK, Lirk P, Seymour RA, Jenkins BJ. The efficacy of preemptive analgesia for acute postoperative pain management: a meta-analysis. Anesth Analg. 2005;100:757–73
12. Karanikolas M, Aretha D, Tsolakis I, Monantera G, Kiekkas P, Papadoulas S, Swarm RA, Filos KS. Optimized perioperative analgesia reduces chronic phantom limb pain intensity, prevalence, and frequency: a prospective, randomized, clinical trial. Anesthesiology. 2011;114:1144–54
13. Sentürk M, Ozcan PE, Talu GK, Kiyan E, Camci E, Ozyalçin S, Dilege S, Pembeci K. The effects of three different analgesia techniques on long-term postthoracotomy pain. Anesth Analg. 2002;94:11–5
14. Remérand F, Le Tendre C, Baud A, Couvret C, Pourrat X, Favard L, Laffon M, Fusciardi J. The early and delayed analgesic effects of ketamine after total hip arthroplasty: a prospective, randomized, controlled, double-blind study. Anesth Analg. 2009;109:1963–71
15. Amr YM, Yousef AA. Evaluation of efficacy of the perioperative administration of venlafaxine or gabapentin on acute and chronic postmastectomy pain. Clin J Pain. 2010;26:381–5
16. Shin SW, Cho AR, Lee HJ, Kim HJ, Byeon GJ, Yoon JW, Kim KH, Kwon JY. Maintenance anaesthetics during remifentanil-based anaesthesia might affect postoperative pain control after breast cancer surgery. Br J Anaesth. 2010;105:661–7
17. Woolf CJ, Salter MW. Neuronal plasticity: increasing the gain in pain. Science. 2000;288:1765–9
18. WHO Western Pacific Region, IASO, IOTF. The Asia-Pacific perspective: redefining obesity and its treatment. 2000 Melbourne Health Communications Australia
19. Reyes-Gibby CC, Morrow PK, Buzdar A, Shete S. Chemotherapy-induced peripheral neuropathy as a predictor of neuropathic pain in breast cancer patients previously treated with paclitaxel. J Pain. 2009;10:1146–50
20. Bijur PE, Latimer CT, Gallagher EJ. Validation of a verbally administered numerical rating scale of acute pain for use in the emergency department. Acad Emerg Med. 2003;10:390–2
21. 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
22. EuroQol Group. . EuroQol: a new facility for the measurement of health-related quality of life. The EuroQol Group. Health Policy. 1990;16:199–208
23. Machin D, Campbell MJ, Fayers PM, Pinol APY Sample Size Tables for Clinical Studies. 19972nd ed Oxford Blackwell Science
24. Dexter F, Dexter EU, Ledolter J. Statistical grand rounds: importance of appropriately modeling procedure and duration in logistic regression studies of perioperative morbidity and mortality. Anesth Analg. 2011;113:1197–201
25. Zhang J, Yu KF. What’s the relative risk? A method of correcting the odds ratio in cohort studies of common outcomes. JAMA. 1998;280:1690–1
26. Bandschapp O, Filitz J, Ihmsen H, Berset A, Urwyler A, Koppert W, Ruppen W. Analgesic and antihyperalgesic properties of propofol in a human pain model. Anesthesiology. 2010;113:421–8
27. O’Connor TC, Abram SE. Inhibition of nociception-induced spinal sensitization by anesthetic agents. Anesthesiology. 1995;82:259–66
28. Richebé P, Rivalan B, Rivat C, Laulin JP, Janvier G, Maurette P, Simonnet G. Effects of sevoflurane on carrageenan- and fentanyl-induced pain hypersensitivity in Sprague-Dawley rats. Can J Anaesth. 2009;56:126–35
29. Kroman N, Jensen MB, Wohlfahrt J, Mouridsen HT, Andersen PK, Melbye M. Factors influencing the effect of age on prognosis in breast cancer: population based study. BMJ. 2000;320:474–8
30. Kehlet H, Jensen TS, Woolf CJ. Persistent postsurgical pain: risk factors and prevention. Lancet. 2006;367:1618–25
31. Katz J, Jackson M, Kavanagh BP, Sandler AN. Acute pain after thoracic surgery predicts long-term post-thoracotomy pain. Clin J Pain. 1996;12:50–5
32. Callesen T, Bech K, Kehlet H. Prospective study of chronic pain after groin hernia repair. Br J Surg. 1999; 86:1528–31
33. Nikolajsen L, Ilkjaer S, Krøner K, Christensen JH, Jensen TS. The influence of preamputation pain on postamputation stump and phantom pain. Pain. 1997;72:393–405