Risk Factors for Chronic Pain after Hysterectomy: A Nationwide Questionnaire and Database Study

The study was approved by the steering group of DHD and registered according to the Danish law of Data Protection. Studies based on questionnaires or registers are not to be notified to the Danish National Committee on Biomedical Research Ethics.

All women registered in DHD between October 1, 2003 and April 1, 2004 were identified (n = 1,299). A pain questionnaire with a prestamped return envelope was mailed to 1,299 women between February 15, 2005 and March 15, 2005 (range, 12.3–15.2 months after the operation). If not returned within 5 weeks, a reminder was sent. The questionnaire was in Danish, and it included the McGill short-form questionnaire and additional questions specified for posthysterectomy pain (for an English translation of the questionnaire, see  appendix). Chronic pain was defined as pelvic pain within the last 3 months, corresponding to “yes” in question 15: Have you had pain in the pelvic region within the last 3 months? (Both pain that was present before the operation and pain that has come after the operation. Pain related to sexual intercourse is not included in this question, but comes later in the questionnaire.)

Other questions included intensity, frequency, and location of pain both before hysterectomy and at the time of the survey, together with use of analgesics and others treatments, as, for example, physiotherapy and acupuncture. General questions were related to previous abdominal surgery, pain problems elsewhere, and occupation.

Returned questionnaires were identified by a patient number and responses were manually entered into a database. For the subsequent analysis, data from the DHD were linked to the questionnaire data. The statistical software was Intercooled Stata version 9 (StataCorp LP, College Station, TX). Descriptive statistics using median (range) and mean (SD) were used where appropriate. The chi-square test was used to compare categorical variables. Eight main risk factors were chosen for a multivariate logistic regression model. The choice of risk factors was based on the current knowledge about risk factors for chronic postsurgical pain and clinical experience. The dependent variable was chronic pain (question 15). The independent variables were (1) preoperative pelvic pain, (2) previous cesarean delivery, (3) pain problems elsewhere, (4) indication for hysterectomy, (5) type of hysterectomy, (6) spinal anesthesia, (7) epidural anesthesia during surgery, and (8) postoperative epidural analgesia. Variables 4 and 5 were categorical based on the categories registered in the DHD, and the reference variables were leiomyoma and total abdominal hysterectomy, respectively. The other variables were all dichotomous. We chose cesarean delivery as a risk factor instead of previous pelvic surgery for two reasons: Previous studies have found chronic pain after cesarean delivery, and women’s recall of a previous cesarean delivery is likely to be more accurate than the recall of other surgical procedures.11,12Postoperative complications were not included in the regression model because details about severity were not provided in the DHD. All P  values less than 0.05 were considered to be statistically significant.

Inclusion in the final analysis required complete data with respect to question 1: Did you have pain in the pelvic region before the operation? (yes/no/do not remember); question 15: Have you had pain in the pelvic region within the last 3 months? (yes/no); and the eight variables included in the regression model.

The questionnaire was returned by 1,173 women (90.3%). Three women had died, and 123 women did not return the questionnaire. Thirty-eight of the 1,173 respondents were excluded from the final analysis (20 returned a blank questionnaire, 6 returned a very incomplete questionnaire, and 12 had missing data with regard to variables included in the regression analysis). This left 1,135 questionnaires (87.4%) for evaluation (fig. 1). Basic demographic data and primary indication are listed in table 1. Type of hysterectomy and method of anesthesia are listed in table 2.

Overall, the included questionnaires had few missing data. The median response rate to each question was 98.6% (range, 78.6–100%). The lowest response rates were observed in question 23c (78.6%): Have you sought other treatments for the pain?; question 13b (85.3%): If you were initially pain free, how long after the operation did you acquire pain?; and question 17 (89.8%): Where is your pain located?” The remaining questions were all answered by more than 94.5% of women. All results refer to complete data for all 1,135 women except where otherwise mentioned.

We compared the 1,135 included women with the 164 not included (3 died, 123 nonresponders, and 38 excluded). The two groups were similar with regard to age, duration of surgery, and body mass index (Wilcoxon rank sum test, P > 0.1). The type of incision and the type of anesthesia was not different between the groups (chi-square test, P > 0.1). The surgical methods used were similar except for vaginal hysterectomy that was more frequent among the not-included women (38.4% vs.  29.3%; chi-square test, P = 0.0183). Prolapse was more frequently the main indication for surgery among the not-included women (20.1% vs.  12.9%; chi-square test, P = 0.0117).

The multiple regression model was significantly better than chance (P < 0.000), and with regard to sensitivity and specificity, the area under the receiver operator curve was 0.7471.

Hysterectomy improved pain in most women (tables 2 and 3). However, 362 women (31.9%) had pain in the pelvic area 1 yr after surgery. The most common pain location was in the middle of the pelvic region, but 70 women reported pain located in the abdominal scar. The most common McGill pain descriptors were “shooting” (51.9%) and “sharp” (25.1%), and the average pain intensity was 4 (range, 0–10), whereas the worst pain intensity was 6 (range, 1–10) on a 0–10 numeric rating scale. Analgesics were used because of pelvic pain by 161 of the pain patients (44.5%), and 139 of these used acetaminophen and/or nonsteroid antiinflammatory drugs. Eighty-five women (23.5%) used other treatments for their pain (e.g. , physiotherapy, acupuncture, massage).

Fifty-four (14.9%) of the women with chronic pain did not recall having pain before the operation. The surgical procedures in these women were abdominal hysterectomy (66.7%), vaginal hysterectomy (31.5%), and laparoscopically assisted vaginal hysterectomy (1.9%). Average pain intensity was 4 (range, 1–10), worst pain intensity was 5 (range, 1–10), and pain affected daily living “a lot” or “very much” in 16.7% (n = 9). Pain was most frequently located in the middle of the pelvic region (n = 15) and in the abdominal scar (n = 15).

The primary indications for hysterectomy are listed in table 1. In the multivariate logistic regression, we included the primary indication for hysterectomy as a categorical variable, and leiomyoma was chosen as the reference value (table 4). A recall of pain before hysterectomy was also included. There was an increased odds ratio for pain at follow-up if pain was the primary indication for hysterectomy (odds ratio [OR], 2.98; 95% confidence interval [CI], 1.54–5.77).

The different surgical procedures are listed in table 2. Multivariate logistic regression including all the surgical procedures showed no difference in chronic pain between total abdominal hysterectomy and vaginal hysterectomy (OR, 0.70; CI, 0.46–1.06), and likewise, no difference was found for total versus  subtotal abdominal hysterectomy (OR, 1.20; CI, 0.77–1.86) (table 4). The type of abdominal incision (Joel-Cohen or Pfannenstiel) did not significantly influence chronic postsurgical pain, but there was a trend toward less pain after Pfannenstiel incision (chi-square test, OR, 0.71; CI, 0.47–1.07). A subsequent analysis of the 70 patients who reported scar pain showed no differences between incision location. Few patients had a laparoscopic hysterectomy (n = 14), not allowing for statistical analysis.

Previous pelvic or abdominal surgery was reported by 552 women (48.6%; table 2), and procedures were cesarean delivery (n = 156), appendectomy (n = 107), sterilization (n = 104), myomectomy (n = 30), extrauterine pregnancy (n = 41), surgery related to adnexa (n = 119), and other abdominal/pelvic procedures (n = 176). Previous cesarean delivery was associated with an increased risk of having chronic postsurgical pain (multiple logistic regression, OR, 1.54; CI, 1.06–2.26). Previous pelvic or abdominal surgery in general was tested post hoc  in the regression model by replacing cesarean delivery, and a similar association with chronic pain was found (OR, 1.44; CI, 1.09–1.90).

Types of anesthesia are listed in table 2. Overall, 334 women had epidural anesthesia during surgery, most often in combination with general anesthesia (n = 325). In the multiple logistic regression model, spinal anesthesia was associated with a reduced OR for chronic pain (OR, 0.42; CI, 0.21–0.85). Epidural anesthesia during surgery and epidural anesthesia in the postoperative period did not have this association (OR, 0.76; CI, 0.50–1.17 and OR, 1.02; CI, 0.66–1.59, respectively).

One hundred and thirteen complications during primary hospital stay were registered in 94 patients (8.3%), and the most frequent complications were bleeding/hemorrhage, pain, and infection (table 2). The complications were not included in the regression analysis, but 41 women with a primary complication reported chronic pain, and a chi-square test showed chronic pain to be more frequent in women with complications (chi-square, OR, 1.74; CI, 1.10–2.72).

A pain problem elsewhere was defined as pain not related to the pelvic region or the operation in the woman’s own opinion. At the time of completing the questionnaire, 66.3% of all women had experienced pain problems elsewhere during the past 3 months, and 25.0% had pain daily or constantly. By far, the most common sites of pain were the head, neck, shoulder, or lower back region (83.9%). Among the women with pain problems elsewhere, 75.2% used analgesics, and 17.1% took pain medication daily. Women with pain problems elsewhere had an increased risk of having chronic postsurgical pain (OR, 3.19; CI, 2.29–4.44).

In this study, we found a pain prevalence of 31.9% 1 yr after hysterectomy, and the risk factors identified by multiple logistic regression were preoperative pelvic pain, pain as an indication for surgery, previous cesarean delivery, and pain problems elsewhere. In addition, we found that spinal anesthesia was associated with a reduced risk of chronic pain 1 yr after hysterectomy.

Thirty-two percent of women had pelvic pain 1 yr after hysterectomy, and 260 (22.9%) had pain that affected their daily living. Other studies of hysterectomy have found pain frequencies that vary from 4.7% to 26.2% in long-term follow-up, but details are lacking regarding pain frequency, intensity, and location.4–10Pelvic pain is a frequent symptom in fertile women, and posthysterectomy pain may reflect pain in a normal population.13The 3-month prevalence of pain among women aged 18–49 yr is found to be 14.7% in the United States and 24.0% in the United Kingdom.4,5,7,9,14,15However, we found a pain prevalence of 40.4% among women aged up to 49 yr, indicating a pain prevalence above the average population.

Significant risk factors identified in this study were primarily related to preoperative status. In the questionnaire, a recall bias could cause a higher report of preoperative pelvic pain in women with chronic postsurgical pain, but pain as the primary indication was registered preoperatively in the DHD, suggesting that preoperative pelvic pain is a true risk factor. Previous pelvic surgery, including cesarean delivery, was associated with an increased risk of chronic postsurgical pain, a relation also described previously. A questionnaire study 1 yr after cesarean delivery described daily or almost daily scar pain in 5.8% of women, and another study found previous cesarean delivery to be associated with chronic pelvic pain.11,12The mechanisms by which preoperative pelvic pain or previous surgery leads to chronic pain may be both physiologic and psychosocial. From human and animal studies, it is known that peripheral nerve trauma may induce neuroplastic changes in the central nervous system (central sensitization), leading to abnormal processing of sensory input from the site of injury, even after healing of the wound.3,16,17If sensitization of the central nervous system is present preoperatively, and this may be the case for both preoperative pelvic pain and previous surgery, postsurgical neuroplastic changes may be enhanced and lead to chronic pain. Psychosocial factors are of major importance in chronic pelvic pain, but the relative contribution in the case of posthysterectomy pain is still unknown. Future prospective studies, including preoperative and postoperative quantitative sensory testing, may clarify the exact contribution of preoperative sensitization to postsurgical pain. Recent studies have found suprathreshold heat pain stimuli and pressure pain tolerance to be predictors of acute postoperative pain.18–20It has, however, never been examined in hysterectomy patients whether preoperative sensory changes can predict pain on long-term follow-up.

Surprisingly, the type of surgical procedure did not affect chronic pain frequency despite the obvious difference between an abdominal and a vaginal approach, but it is possible that the segmental noxious input to the central nervous system is comparable. There was a tendency, though, toward vaginal hysterectomy being associated with less chronic pain compared with total abdominal hysterectomy (OR, 0.70; CI, 0.46–1.06; P = 0.091). Also, the type of incision in abdominal hysterectomy, e.g. , Pfannenstiel, Joel-Cohen, or vertical incision, did not relate to chronic postsurgical pain (table 2). Pfannenstiel incision did, however, have a tendency toward less chronic pain compared with the Joel-Cohen incision (chi-square test, OR, 0.71; CI, 0.47–1.07; P = 0.0985). The Joel-Cohen incision is considered to be less traumatic because the tissue is pulled apart rather than cut, but this may also damage the sensory nerves and contribute to central sensitization.

Importantly, spinal anesthesia was associated with a lower frequency of chronic pelvic pain, whereas epidural anesthesia did not influence chronic pain significantly. Epidural anesthesia was previously thought to have a protective role against central sensitization during surgery, but this theory has later been rejected.21–23It is possible that the stronger blockade of central impulse traffic in spinal anesthesia may have a protective effect for the development of chronic pain in some patients. This notion is supported by a study of cesarean delivery in which general compared with spinal anesthesia was associated with a higher frequency of chronic pain after 1 yr.11Other factors may be involved in cesarean delivery, because general anesthesia usually is restricted to urgent surgery. In contrast to the present findings, a recent, randomized study of spinal versus  general anesthesia in 89 women scheduled to undergo vaginal hysterectomy found no difference in pain at 12 weeks’ follow-up.24However, that study had some limitations. The mean pain intensities, recorded on an 11-point verbal numeric rating scale, were 0.0 (SD ±0.2) versus  0.1 (SD ±0.3) after 12 weeks (spinal vs.  general anesthesia), and accordingly, intergroup differences will be difficult to demonstrate. Also, the exclusion rate before randomization was high, and treatment assignment was not blinded to either patient or study personnel.

In conclusion, preoperative status is important for the development of chronic pain, and this supports the hypothesis that previous noxious input to the central nervous system may play the most important role, perhaps together with genetic and psychosocial factors. It is unknown whether central sensitization is present preoperatively in some patients (e.g. , with preoperative pelvic pain), and also, the natural course of such changes after hysterectomy and their correlation to chronic postsurgical pain are unknown.16Future studies should aim at describing the preoperative status in detail, including pain characteristics, quantitative sensory testing, and psychological profile, with postoperative follow-up of all tests. The current observation that spinal anesthesia, and hence a more complete block of afferent noxious activity, has a protective role for development of chronic pain is interesting and calls for a future, randomized study to address this issue.

The authors thank Helle Obenhausen Andersen (Secretary, Danish Pain Research Center, Aarhus University Hospital, Aarhus, Denmark) for excellent secretarial assistance through all the phases of this study.