Impact of stress coping capacity on recovery from abdominal hysterectomy in a fast-track programme: a prospective longitudinal study

Authors

  • P Kjølhede,

    1.  Division of Obstetrics and Gynaecology, Department of Clinical and Experimental Medicine, Faculty of Health Sciences, Linköping University, Linköping, Sweden
    2.  Department of Obstetrics and Gynaecology, County Council of Östergötland, Linköping, Sweden
    Search for more papers by this author
  • N Borendal Wodlin,

    1.  Division of Obstetrics and Gynaecology, Department of Clinical and Experimental Medicine, Faculty of Health Sciences, Linköping University, Linköping, Sweden
    2.  Department of Obstetrics and Gynaecology, County Council of Östergötland, Linköping, Sweden
    Search for more papers by this author
  • L Nilsson,

    1.  Division of Drug Research, Anaesthesiology and Intensive Care, Department of Medical and Health Sciences, Linköping University, Linköping, Sweden
    2.  Department of Anaesthesia and Intensive Care, County Council of Östergötland, Linköping, Sweden
    Search for more papers by this author
  • M Fredrikson,

    1.  Linköping Academic Research Centre, University Hospital, Linköping, Sweden
    2.  Division of Occupational and Environmental Medicine, Department of Experimental and Clinical Medicine, Faculty of Health Sciences, Linköping University, Linköping, Sweden
    Search for more papers by this author
  • K Wijma

    1.  Unit of Medical Psychology, Department of Clinical and Experimental Medicine, Faculty of Health Sciences, Linköping University, Linköping, Sweden
    Search for more papers by this author

Errata

This article is corrected by:

  1. Errata: Errata Volume 119, Issue 10, 1291, Article first published online: 13 August 2012

  • The study was registered in the ClinicalTrial.gov Protocol Registration System (NCT00527332), with initial release 7 September 2007.

Dr P Kjølhede, Department of Obstetrics and Gynaecology, University Hospital, S-581 85 Linköping, Sweden. Email preben.kjolhede@liu.se

Abstract

Please cite this paper as: Kjølhede P, Borendal Wodlin N, Nilsson L, Fredrikson M, Wijma K. Impact of stress coping capacity on recovery from abdominal hysterectomy in a fast-track programme: a prospective longitudinal study. BJOG 2012;119:998–1007.

Objective  To evaluate the effect of stress coping capacity in combination with mode of anaesthesia on postoperative recovery in fast-track abdominal hysterectomy.

Design  Prospective longitudinal study.

Setting  Five hospitals in the south-east of Sweden.

Population  A cohort of 162 women undergoing fast-track abdominal hysterectomy for benign conditions.

Methods  Self-administered questionnaires, the Stress Coping Inventory (SCI) and the Swedish Postoperative Symptom Questionnaire (SPSQ), and clinical information were collected prospectively. Stress coping capacity was categorised as high or low according to the summed score of the SCI. Comparisons of effect variables were adjusted using a propensity score-matching model.

Main outcome measures  Associations between stress coping capacity and hospital stay, sick leave, use of analgesic and self-reported postoperative symptoms.

Results  Women with high stress coping capacity had a significantly shorter sick leave, experienced postoperative symptoms significantly less often, and with lower intensity, than women with low stress coping capacity. With the exception of symptom intensity, these findings were related to having had the operation under spinal anaesthesia as opposed to general anaesthesia. Hospital stay, use of analgesics and abdominal pain were not related to stress coping capacity.

Conclusions  In patients for whom spinal anaesthesia was applied, high stress coping seems to be a quality that helps patients manage the burden of surgery. It is desirable for the individual, as well as for the healthcare system, to enhance recovery by using intervention programmes designed to improve or manage stress coping, particularly for individuals with low stress coping capacity. This recommendation merits further investigation.

Introduction

Abdominal hysterectomy is a common gynaecological operation,1 mostly carried out for benign conditions such as bleeding disturbances and/or uterine fibroids. The complex nature of hysterectomy suggests that multiple factors are likely to influence postoperative recovery. Psychological problems often decrease a woman’s capacity to cope with symptoms from surgery that would otherwise be tolerable, and are related to more subjective feelings of distress. This may interfere with treatment and delay recovery.2 The experience and intensity of postoperative symptoms are related to a capacity to cope with stressful situations: women with lower stress coping capacity have greater problems recovering after surgery than those with a higher stress coping capacity.3–5

During the past 10 years, fast-track surgery has gained much attention. Fast-track surgery is intended to enhance recovery.6 By giving the patient thorough and detailed information about the operation and the postoperative period, the fast-track programme prepares the individual for a faster recovery.7 Providing accurate information prior to gynaecological surgery and using regional anaesthesia reduces postoperative analgesic requirements, and results in a more rapid return to full health.7,8

The connection between stress coping capacity and postoperative recovery in fast-track programmes has not been evaluated, however. As the recovery time after fast-track surgery is shorter than after traditionally conducted surgery, it seems reasonable to assume that the physical burden for the patient will be lower. Therefore, one could theorise that a fast-track programme, by reducing the physical problems for the patient, could also be expected to place fewer demands on a patient’s stress coping capacity during postoperative recovery.

We conducted an open randomised multicentre trial comparing general anaesthesia and spinal anaesthesia in fast-track abdominal hysterectomy for benign conditions (the ‘GASPI study’), in order to analyse postoperative recovery. The results of the primary and secondary objectives of the trial previously presented showed a significantly better outcome for patients given spinal anaesthesia, with fewer perceived symptoms postoperatively, a faster recovery of quality of life and a shorter sick leave.9–11 There was, however, no significant difference between the two groups as concerns length of hospital stay. Assuming that the mode of anaesthesia and psychological characteristics were responsible for the benefits observed in postoperative recovery, it might have been expected that even the duration of hospital stay would have been shortened. The results from the GASPI study indicate that the length of hospital stay in a fast-track programme, as measured using standardised discharge criteria, was not significantly affected by mode of anaesthesia or by the psychological characteristics of the patients. We therefore wondered if the fast-track programme counteracted the predicted relationship between psychological traits and duration of hospital stay.

Based on these considerations, we hypothesised that in a fast-track regime the difference in recovery measures that otherwise could be anticipated after abdominal hysterectomy between women with low and high stress coping capacity would disappear.

This study of women who had fast-track abdominal hysterectomy is a secondary post hoc analysis of the GASPI study. The aim of this study was to examine the role of stress coping capacity, in combination with mode of anaesthesia, in affecting recovery from surgery. Recovery from surgery was operationalised by measuring the duration of hospital stay, sick leave, use of analgesic and self-reported postoperative symptoms, and by taking into account the mode of anaesthesia, either general or spinal anaesthesia.

Methods

The population in this study comprises a cohort of women who all had an abdominal hysterectomy for benign diseases in a fast-track programme. The women participated in an open randomised controlled multicentre study that compared general anaesthesia and spinal anaesthesia including intrathecal morphine.9 The trial was conducted in five hospitals in the south-east of Sweden during March 2007–December 2009. Ethical approval was obtained from the Regional Ethical Review Board in Linköping.

Women admitted to hospital for hysterectomy for benign conditions were assessed for eligibility in the study. Inclusion criteria were: aged 18–60 years; scheduled for abdominal total or subtotal hysterectomy; benign indications (cervical dysplasia included); at least one ovary preserved at the conclusion of surgery; and speak Swedish fluently. The exclusion criteria were: contraindication against spinal or general anaesthesia; American Society of Anesthesiologists (ASA) score ≥ 3; gynaecological malignancy present or in history; previous bilateral oophorectomy; operation expected to encompass more than hysterectomy, with or without salpingectomy and appendectomy; postmenopausal without hormone therapy (if patient was on hormone therapy, this therapy was to be maintained for 6 months postoperatively); physically disabled to a degree that might be expected to affect mobilisation postoperatively; and severe psychiatric disease or learning difficulties. The patients agreed to participate in the study by signing informed consent.

The enrolled women were randomised to surgery under general anaesthesia or spinal anaesthesia with intrathecal morphine. A total of 180 women were randomised and 162 completed the study (Figure 1).

Figure 1.

 Flowchart of the study participants.

The fast-track programme and the standardised modes of anaesthesia have been presented and described in detail previously.9,10 All women received similar information about care and advice for the perioperative period. The postoperative monitoring of the two groups was identical. The duration of hospital stay was defined as the time from the start of anaesthesia to the time of discharge from the gynaecology ward. Clinical data were collected consecutively during the hospital stay and at follow-up visits 5 weeks and 6 months postoperatively. After disge the women completed a diary reporting their use of analgesics until the 5 week postoperative follow-up visit. The total quantity of analgesics given during the hospital stay and after discharge until 5 weeks postoperatively was summarised.

Assessment of stress coping capacity

The capacity to cope with stress was measured approximately 1 week preoperatively by the Stress Coping Inventory (SCI).12,13 The respondent rates how often she thinks she is able to cope with each of the 41 stressful situations described in the SCI form.13 The answers were rated on a six-point Likert-type scale: 1, almost never; 2, rarely; 3, occasionally; 4, rather often; 5, very often; 6, almost always. The minimum score is 41 and the maximum score is 246 for the SCI. A higher score indicates a greater stress coping capacity. The score was calculated after the study period and was thus blinded for the participants and investigators during the study. A clinical categorisation of the SCI score has not been scientifically evaluated, but the value of the lower 25th percentile has been considered as a discriminatory level for low and high stress coping capability.14 In this study we used the lower 25th percentile (score ≤ 169) as the cut-off for low and high stress coping capacity.

In a previous study,13 the SCI showed the following correlations with four commonly used psychological questionnaires: Beck Depression Inventory, r = −0.53 (95% CI −0.65 to −0.39);15 State–Trait Anxiety Inventory, r = −0.71 (95% CI −0.79 to −0.60);16 Psychological Well Being Inventory, r = 0.57 (95% CI 0.43–0.68);17 and Women’s Health Questionnaire, r = −0.55 (95% CI −0.66 to −0.41).18 The internal consistency of the SCI has been found to be very good, with Cronbach’s alphas at of 0.94–0.95 and a split-half reliability of 0.93.5,12,14

Assessment of postoperative symptoms

Postoperative symptoms were assessed by the Swedish Postoperative Symptoms Questionnaire (SPSQ).19 The patient completed the form on a daily basis, preferably in the evening during the first 7 days postoperatively, starting in the evening after surgery (day 0), and thereafter once weekly until the 5-week postoperative visit. The questions were both open-ended and closed. The closed questions could be answered by choosing an answer from a set given on a Likert-type scale. The open-ended questions required written responses.

Initially the patient was asked if she, at the moment of completing the form, experienced a number of symptoms commonly reported after surgery (pain in the area of surgery, nausea, retching, headache, abdominal pain, tiredness, drowsiness and blurred vision), and how she rated the intensity of each of these symptoms. The answers were rated on a four-point scale: 0, none; 1, yes, a little; 2, yes, somewhat; 3, yes, a lot. The total score (SPSQ sum score) was calculated to estimate overall postoperative discomfort. Second, the patient was asked to report the intensity of pain in the surgical area, both when it was at its worst and how it felt on average on that particular day. The answers to these questions were rated on a seven-point scale: 0, none; 1, very mild; 2, mild; 3, moderate; 4, bad; 5, severe; 6, very severe. Third, the patient reported the occurrence and comprehensive intensity of eight specified troublesome symptoms [nausea/vomiting (PONV), headache, abdominal pain, drowsiness, fatigue, blurred vision, itching and other]. Similar assessments for the intensity of pain in the surgical area were made.

For the analyses in this study, we used the mean values of the repeated estimates carried out within the 5-week postoperative period (i.e. on 12 occasions): (1) the SPSQ sum score of the intensity of symptoms the patients experienced at the moment of completing the form; (2) the average intensity of pain in the surgical area and of troublesome symptoms; and (3) the occurrence of troublesome symptoms. In order to assess the impact of stress coping capacity on effect variables more generally, and not only at the moment or during a short time period, we selected only those troublesome symptoms that occurred with a frequency of more than 10% and were sustained for more than a week.

Sick leave

At discharge from the hospital the patient was granted sick leave for 14 days. The research nurse contacted the patient by telephone the day after discharge and then once weekly until the 5-week postoperative visit. The purpose of frequent contact was dual: to be supportive and to provide a reminder to fill in the study questionnaires. At these contacts the sick leave was prolonged by at most 7 days at a time on the basis of patient demand until the patient was able to return to work or felt sufficiently recovered. The duration of the sick leave was defined as the time from the day of surgery to the day of returning to work to the same extent as preoperatively. Women who were on sick leave for reasons other than the hysterectomy, who were unemployed or who had a disability pension were excluded from the analysis of sick leave.

Statistics

The estimation of sample size and the power calculation were based on the primary outcome measure of the trial, i.e. the duration of hospital stay, and have been described previously.9 No power calculations were performed a priori for the secondary outcomes. All analyses were performed according to intention-to-treat principles. Data are denoted as medians and ranges or numbers and percentages. Mann–Whitney U-tests were used for the analyses of effect variables on a continuous scale. Nominal data were compared by using Yates’ corrected χ2-tests or Fisher’s exact tests, as appropriate.

As the women with low and high stress coping capacity were not randomly assigned to the modes of anaesthesia, potential confounding effects might be present that could influence the results. In order to reduce or eliminate potential confounding effects a propensity score-matching analysis was conducted.20,21 The propensity scores were calculated with a logistic regression analysis with low or high stress coping capacity as outcome and age, body mass index (BMI), parity, previous anaesthesia, concomitant diseases, main indication of hysterectomy, ASA, mode of hysterectomy, mode of skin incision and mode of anaesthesia as independent factors. The women in the low stress coping capacity group were then matched to women from the high stress coping capacity group based on the propensity score in a 1:1 matching procedure. To match each woman with low stress coping capacity we used a woman with high stress coping capacity that had an identical propensity score. In case this was not found we used the woman with the closest propensity score. The data from these matched women were entered into a linear regression analysis to adjust for potential confounding (propensity score). To take the matching into account each matched pair had a unique number that was used as identity in a regression model for panel data (stata 12.0; StataCorp LP, College Station, TX, USA). As some of the dependent variables were not normally distributed, logarithmic transformations were used for these variables.

Results

Of the 162 women who completed the study, 42 women were categorised as having a low stress coping capacity, i.e. had SCI scores ≤169, constituting the low stress capacity group (LSC-G), and 120 women were categorised as having a high stress coping capacity (SCI score >169), constituting the high stress capacity group (HSC-G). The median SCI scores were 154 (107–169) and 194 (170–241) for the LSC-G and HSC-G groups, respectively. The SCI scores did not differ significantly between women who had the operation performed under general anaesthesia (GA) and those with spinal anaesthesia (SA): 183 (107–241) for GA and 187 (121–241) for SA (= 0.532). Demographic, descriptive and clinical data of the two stress coping capacity groups are presented in Table 1. The results for effect variables subdivided by categorisation of stress coping capacity are presented in Table 2, and the results for the effect variables subdivided by mode of anaesthesia and categorisation of stress coping capacity are presented in Table 3.

Table 1.   Demographic, descriptive and clinical data.
CharacteristicsLow stress capacity group (n = 42)High stress capacity group (n = 120)P
  1. Figures denote medians and (ranges) or numbers and (%); ASA, the American Society of Anesthesiologist classification of physical status CIN, cervical intraepithelial dysplasia.

  2. *A woman may have had more anaesthesia or concomitant diseases.

  3. **Mann–Whitney U-test.

  4. ***Yates’ corrected inline image test.

  5. ****Yates’ corrected inline image test.

  6. *****Fisher’s exact test.

Age (years)44.5 (33–58)46.0 (34–58)0.150**
Body mass index(kg/m2)26.4 (18.9–41.5)25.1 (19–39.2)0.238**
Parity2.0 (0–4)2.0 (0–8)0.752**
Smokers7 (16.7%)22 (18.3%)0.993***
Previous laparotomy(no. of women)17 (40.5%)41 (34.2%)0.584***
Previous anaesthesia*
General anaesthesia15 (35.7%)27 (22.5%)0.140***
Spinal/epidural anaesthesia5 (11.9%)17 (14.2%)0.915****
Concomitant diseases
Psychiatric8 (19.0%)6 (5.0%)0.014***
Musculoskeletal5 (11.9%)12 (10.0%)0.772****
Cardiovascular5 (11.9%)12 (10.0%)0.772****
Chronic pulmonary5 (11.9%)5 (4.2%)0.128****
Concomitant medication
Analgesics9 (21.4%)23 (19.2%)0.927***
Antidepressants7 (16.7%)6 (5.0%)0.040***
Main indication of hysterectomy
Bleeding disturbances30 (71.4%)62 (51.7%)0.088****
Mechanical symptoms11 (26.2%)45 (37.5%)
CIN/endometrial hyperplasia0 (0.0%)9 (7.5%)
Endometriosis/dysmenorrhea1 (2.4%)4 (3.3%)
ASA
Class I24 (57.1%)90 (75.0%)0.047***
Class II18 (42.9%)30 (25.0%)
Mode of hysterectomy
Total abdominal27 (64.3%)79 (65.8%)1.000***
Subtotal abdominal15 (35.7%)41 (34.2%)
Mode of skin incision
Midline3 (7.1%)10 (8.3%)1.000*****
Low transverse39 (92.9%)110 (91.7%)
Mode of anaesthesia
General anaesthesia20 (47.6)60 (50%)0.931***
Spinal anaesthesia22 (52.4)60 (50%)
Table 2.   Results associated with measures performed by the healthcare system and with estimates performed by the study participants in relation to stress coping capacity
 LSC-G
(n = 42)
HSC-G
(n = 120)
P*Adjusted P**
(n = 42 + 42)
  1. Data are given as medians and ranges.

  2. *Mann–Whitney U-test.

  3. **Regression analysis, adjusted for matched propensity score.

  4. ***Total consumption for 16 days postoperatively.

  5. ****Total consumption for 5 weeks postoperatively.

Measures performed by the healthcare system
Duration of hospital stay (hours)50.2 (24.4–124.6)47.2 (21.9–123.4)0.2270.893
Duration of sick leave (days)29.0 (14–81)23.0 (2–258)0.0010.003
Analgesics. Equivalent morphine (mg)***15.0 (0–130.0)15.0 (0–255.4)0.9690.631
Analgesics. Non-opioids (sum of RDD)****19.4 (1.8–58.0)17.2 (2.4–47.6)0.0370.062
Estimates performed by study participants
SPSQ score (0–24, min.–max.)5.1 (1.4–12.3)3.6 (0–8.6)0.0030.002
Intensity of pain in surgical area (0–6)2.3 (0.9–3.5)1.7 (0.1–5.0)0.0010.001
Intensity of troublesome symptoms (0–6)2.3 (0.5–3.6)1.8 (0.2–5.0)0.0020.004
Abdominal pain: occurrence (%)58 (0–100)42 (0–100)0.0620.090
Fatigue: occurrence (%)29 (0–100)17 (0–100)0.0090.036
Table 3.   Results associated with measures performed by the healthcare system, and with estimates performed by the study participants in relation to category of stress coping capacity, subdivided according to mode of anaesthesia
 Spinal anaesthesiaGeneral anaesthesia
LSC-G
(n = 22)
HSC-G
(n = 60)
P*Adjusted P** (n = 22 + 22)LSC-G
(n = 20)
HSC-G
(n = 60)
P*Adjusted P** (n = 20 + 20)
  1. Data given as medians and ranges.

  2. *Mann–Whitney U-test.

  3. **Regression analysis, adjusted for matched propensity score.

  4. ***Total consumption for 16 days, postoperatively.

  5. ****Total consumption for 5 weeks postoperatively.

Measures performed by the healthcare system
Duration of hospital stay (hours)51.4 (26.3–124.6)34.0 (21.9–123.4)0.0160.22049.1 (24.4–80.4)50.5 (24.5–100.3)0.3830.079
Duration of sick leave (days)29.0 (14–81)19.5 (2–258)0.0010.00130.5 (20–49)28.0 (7–81)0.2190.100
Analgesics.Equivalent morphine (mg)***4.5 (0–113.3)2.5 (0–255.4)0.3920.38020.5 (7.5–130.0)25.2 (5.0–243.5)0.4140.154
Analgesics. Non-opioids (sum of RDD)****19.4 (5.8–58.0)15.4 (3.2–46.5)0.0420.04219.2 (1.8–56.7)18.6 (2.4–47.6)0.2430.255
Estimates performed by the study participants
SPSQ score (0–24, min.–max.)5.1 (1.4–12.3)3.0 (0–8.5)0.0010.0074.8 (1.9–10.0)4.1 (1.3–8.6)0.3530.296
Intensity of pain in surgical area (0–6)2.2 (0.9–3.5)1.6 (0.1–3.3)0.0020.0082.4 (1.3–3.3)1.9 (0.3–5.0)0.0550.030
Intensity of troublesome symptoms (0–6)2.3 (0.6–3.4)1.8 (0.2–3.3)0.0050.0532.4 (0.5–3.6)1.9 (0.3–5.0)0.1010.011
Abdominal pain: occurrence (%)50 (0–100)42 (0–100)0.4030.87871 (17–100)54 (0–100)0.0760.007
Fatigue: occurrence (%)29 (0–100)8 (0–100)0.0190.00729 (0–83)25 (0–92)0.2770.915

Objective effect measures

Duration of hospital stay

Overall, and subdivided after mode of anaesthesia, the duration of hospital stay did not differ significantly between women in the two stress coping capacity groups.

Sick leave

Of the 162 women in the study, 151 received sick leave: 113 in the HSC-G and 38 in the LSC-G. Of the 11 women who did not receive sick leave, seven were on disability pension and four were on sick leave for reasons other than the hysterectomy. The duration of sick leave was strongly associated with the stress coping capacity, being significantly shorter for women in the HSC-G than in the LSC-G. This association was strongly connected to mode of anaesthesia. In the group of women who underwent hysterectomy in SA, the length of sick leave in the HSC-G was significantly shorter than for the women in the LSC-G, whereas for the women who had the operation under GA no such difference was observed.

Consumption of analgesics

The consumption of opioids and the use of non-opioids (sum of recommended daily doses, RDD) did not differ significantly between the two stress coping capacity groups overall, or in the SA and GA subgroups. Although the women who had the hysterectomy under GA consumed significantly more opioids postoperatively, and an equal quantity of non-opioids as those who had the operation with SA [equivalent morphine, 23.9 mg (5.0–243.5 mg) versus 3.0 mg (0.0–255.4 mg), < 0.0001; non-opioids (sum of RDD), 18.4 (1.8–56.7) versus 17.0 (3.2–58.0), = 0.196], no significant differences were observed in the consumption of opioids or non-opioids between the low and high stress capacity groups in either of the modes of anaesthesia.

Subjective effect measures

Postoperative symptoms

The LSC-G had a significantly higher SPSQ score, pain intensity, on average, and intensity of troublesome symptoms, on average, than the HSC-G. The adjusted analysis showed that this was the case regardless of the mode of anaesthesia, with the exception of the SPSQ score, which was only found to be significantly higher in women undergoing hysterectomy with SA.

Only the symptoms of abdominal pain and fatigue reached a frequency of 10% or more, and were sustained for more than a week. Overall, the mean frequencies of the troublesome symptom ‘abdominal pain’ and ‘fatigue’ did not differ significantly between the LSC-G and HSC-G (Table 2). The influences of mode of anaesthesia on the prevalence of the two troublesome symptoms ‘abdominal pain’ and ‘fatigue’ in relation to low and high stress coping capacity varied as listed in Table 3. Although there were no significant differences in the occurrence of abdominal pain between the women who had GA and SA [63% (0–100%) versus 42% (0–100%); = 0.014], the LSC-G women who had GA had a significantly higher occurrence of abdominal pain than the HSC-G. No such difference was observed among the women who had the operation performed in SA. In contrast, fatigue occurred significantly more often in the women who had GA compared with SA [25% (0–92%) versus 8% (0–100%); = 0.005]. By subdividing with mode of anaesthesia, the LSC-G women showed a significantly higher occurrence of fatigue than the HSC-G women in the group who received SA, whereas no such difference was seen in the GA group.

Discussion

The length of recovery after abdominal hysterectomy in a fast-track programme was found to be associated with stress coping capacity, but only in the SA group. As was shown for almost all variables for the patients for whom SA was used, high stress coping seemed to be a quality that facilitated the management of the burden of surgery. Only the use of analgesics and having abdominal pain were not related to stress coping capacity.

Many studies have explored the relationship between preoperative psychological factors and postoperative recovery, with ambiguous findings.3,5,22–25 Fast-track programmes had not been used in all the groups covered by these studies. It is not clear whether fast-track programmes have any impact on a possible correlation of pre-operative psychological characteristics and recovery.

Stress coping capacity did not appear to be of value for the recovery of women who had GA. The stress coping capacity of patients had significance only with SA. One explanation could be that the women who had the operation under general anaesthesia developed postoperative cognitive dysfunction (POCD) to a greater degree than those who had SA. This agrees with the findings of Rasmussen et al.26 POCD affects a wide variety of cognitive functions, such as attention span, memory executive function and speed of information processing.27 The aetiology remains elusive, but POCD has been connected with effects of general anaesthesia and the post-operative use of opioids. A multimodal approach within the fast-track principles of pain management with reduced opioids, combined with a reduction in inflammatory responses, with minimal invasive surgery and pharmacological anti-inflammatory intervention may therefore have the potential for improving postoperative cognitive function.28 The women in the GA group had significantly more symptoms postoperatively, and used between five and ten times more morphine-equivalent analgesics postoperatively, than those who had SA. This might indicate that the intensity of their suffering and POCD overruled their stress coping capacity. Another concurrent factor could be that the stress coping capacity of patients is of use when the patient is awake or sedated during spinal anaesthesia. A patient who is submitted to surgery has to cope with various emotions and cognition, and has to use her will to force herself to be compliant and committed. A patient who has a higher capacity to stand uncertainties and believes more in her own power to manage is better prepared for such an unsafe situation.

The length of hospital stay after surgery is mainly determined by clinical and medical criteria, and therefore is assumed to be less likely to be influenced by pre-operative psychological characteristics, in particular when the hospital stay is supposed to be short, as in fast-track programmes. Oxlad et al.29 reported that psychological factors like anxiety, depression and post-traumatic stress disorder constitute a risk for prolonged hospital stay in coronary bypass surgery. We found that hospital stay was not related to the stress coping capacity, independent of mode of anaesthesia. It was noted in the fast-track programme and in the information given repeatedly perioperatively that discharge was allowed when the standardised criteria for discharge were met. This was an attempt to encourage the women to leave the clinic as soon as the criteria were fulfilled. It seemed that, independent of stress coping capacity and mode of anaesthesia, the women complied and felt reassured about this.

The length of sick leave may be considered as a blunt measurement of recovery, as it is usually not strictly determined by medical decisions. Social welfare benefits, local traditions, personal reasons, workload and the attitudes of professionals in the healthcare system may all have an effect on the length of sick leave. In the present study the participants had weekly telephone contacts with the research nurses, and a prolonged sick leave was granted if the woman did not feel able to return to work. In such a way we believe that sick leave expressed the woman’s perceived experience of recovery. The symptoms of pain and fatigue were described as main contributors to prolonged postoperative convalescence.30 We found that sick leave and fatigue in a fast-track setting was associated with stress coping capacity, but only in the group who had the operation performed with SA. The HSC-G women who had SA had a far shorter sick leave than all the other women. This group of women also presented with the lowest occurrence of fatigue. The length of sick leave and occurrence of fatigue for the LSC-G with SA was quite similar to the length of sick leave and occurrence of fatigue for all women in GA, independent of stress coping capacity.

The postoperative consumption of analgesics has been shown to be associated with preoperative anxiety and coping strategies.3,23,24 In the present study we found no association between the consumption of analgesics and stress coping, independent of mode of anaesthesia. This may be because our multimodal analgesic therapy combined paracetamol, nonsteroidal anti-inflammatory drugs (NSAIDs) and opioids, whereas Kain et al.3 and Cohen et al.23 only used intravenous patient-controlled administration of morphine. All women in those studies had GA, and their dosage of morphine during the 24–48 hours postoperatively exceeded by far the dosage of our patients with GA. This could indicate that the multimodal analgesic therapy provided advantages by reducing the need of opioids, which is considered an advantage in the fast-track programme, even for women with low stress coping capacity.

The preoperative psychological preparation of patients for surgery has been common practice for many years. Patient education concerning postoperative care is an important element of the fast-track concept, and it is claimed to play a major role in modifying the individual’s response to the operative experience.7 To the best of our knowledge no studies have analysed the effect of stress coping capacity on recovery in fast-track surgery. However, our results need to be confirmed. Although some studies have shown associations between preoperative psychological characteristics and recovery, results are ambiguous.8,31–37 In our study it seemed that mode of anaesthesia played an important role in enabling stress coping capacity to aid recovery.

The study has methodological limitations. It constitutes a post hoc analysis, and the results should be considered as exploratory. The participants in the trial were randomised to mode of anaesthesia, and no stratification was performed according to stress coping capacity. The number of women in the two categories of stress coping capacity would, according to the definition of the categories, be uneven, and lowest in the LSC-G. Consequently, this group determined the statistical power of the study. In order to reduce or eliminate the effects of confounding we applied the propensity score matching model to estimate treatment effects. Propensity score methods have in recent years increasingly been advocated to reduce the impact of treatment-selection bias in the estimation of treatment effects in studies using observational data.38 It is a popular and effective way of dealing with confounding, and is probably most useful in small studies where it is not possible to include a large number of possible confounding factors in the analysis. In this study, we used the method of matching on the propensity score. This gave a group of women that should be relatively homogeneous with regard to the confounding factors, but concurrently unfortunately a much smaller number were used in the analysis. This might indicate a lower statistical power and consequently a possible reduced effect of the analysis, but at the same time gives a lower risk of overestimating the results.

The results of the subgroup analyses should be interpreted with caution because of the relatively small number of women in these subgroups. In the fast-track programme it was stated that the participants should receive thorough information about perioperative care. Although the information about perioperative care was given preoperatively, and during their hospital stay, no assessments were performed to ensure that all participants had received the complete information, or to what level the participant had understood it. This issue is always troublesome in clinical trials, and may constitute a sort of selection bias.

In conclusion, psychological stress coping capacity seems to be of importance in fast-track abdominal hysterectomy when conducted in spinal anaesthesia. Women with high stress coping capacity had a sick leave that was shorter than 3 weeks, and postoperative symptoms were perceived to be less intensive and troublesome. It might be possible and desirable for the individual as well as the healthcare system to enhance recovery after surgery by intervention programmes aiming to improve or manage stress coping for, in particular, individuals with low stress coping capacity. This merits further investigation.

Disclosure of interests

The authors report no conflicts of interest.

Contribution to authorship

NBW, LN and PK planned and conducted the study. The data were processed and analysed by PK, NBW, LN, MF and KW. PK and KW are the primary authors of the article. All authors contributed to writing, revising and approving the final version of the article.

Details of ethics approval

The Regional Ethical Board at Linköping University (registration no. M159-06, approved 15 November 2006; amendment 1, registration no. T83-07, approved 3 October 2007; amendment 2, registration no. T19-08, approved 4 March 2008) and the Swedish Medical Products Agency (registration no. EudraCT no. 2006-002520-41) approved the study.

Funding

The study was supported financially by grants from the Medical Research Council of south-east Sweden, Linköping University and the County Council of Östergötland.

Commentary on ‘Propensity score analysis: a novel methodology to address selection bias in observational studies’

Observational studies are often fraught with a variety of biases that can limit valid inferences (for an excellent and comprehensive review, see Lash et al. Applying quantitative bias analysis to epidemiologic data. New York: Springer-Verlag; 2011). An evaluation of the causal effect of an intervention (or an exposure) in the setting of observational data can introduce selection bias. This bias can arise from the ‘non-random’ assignment of subjects to the interventions – an issue that is overcome in well-designed randomised controlled trials. Efforts to minimise selection bias can be achieved through a novel and relatively under-utilised method called ‘propensity score analysis’ (Rosenbaum and Rubin, Biometrika 1983;70:41–55; Rosenbaum, Observational studies. New York: Springer-Verlag; 1995). The propensity score (PS) is the weight (or probability) that is assigned to each subject to receive either the intervention (exposure) or the placebo (non-exposure), conditional on a set of observed covariates. All else being equal, the PS method assumes that the approach ensures a good balance in the assignment of intervention for measured covariates and confounding factors. Arguably, what is implicit with this approach is that the method is also likely to provide a good balance of unmeasured covariates between the two groups.

The propensity score is derived from fitting a logistic regression model (or any regression model within the framework of the class of generalised linear models) for the treatment or intervention based on observed covariates. From this model, the predicted probability of receipt of the intervention or, analogously, the ‘propensity’ to receive the intervention is estimated. The score is then used as a predictor in the second stage model for the outcome (with further adjustment for confounding factors, if necessary.) Four methods have been proposed for a PS analysis: covariate adjustment using the PS; stratification based on PS; PS matching; and PS weighting (Austin Statistics in Medicine 2007;27:2037–2049.) Of the four methods, the PS-matching approach has been a preferred choice, as this approach ensures the largest reduction in selection bias (Austin, Statistics in Medicine 2007;26:734–753.) An extension to the PS methods to accommodate adequate balance of several 100s of covariates between intervention or exposure groups have also been introduced (Schneeweiss et al., Epidemiology 2009;20:512–522.)

In this issue of BJOG, Kjölhde and colleagues report data on an evaluation of the effect of stress coping capacity in combination with mode of anaesthesia on postoperative recovery in fast-track abdominal hysterectomy. In this secondary analysis of data from a randomised controlled trial, the authors concluded that in patients for whom spinal anaesthesia was applied, high stress coping seemed to be a quality that helps the patients to manage the burden of surgery. They applied a PS analysis to match subjects with low and high stress-coping capacities on a variety of covariates. These analyses are useful to minimise the impact of selection bias between the two groups. An added benefit of this analytic approach is that the two groups are also likely to be ‘balanced’ with respect to the measured (and probably unmeasured) covariates at baseline.

Disclosure of interests

Cannde Ananth is the Editor-in-Chief of Perinatal Epidemiology, an international journal published by Wiley-Blackwell.

Cande V Ananth

Professor of Reproductive Sciences, Department of Obstetrics and Gynecology College of Physicians and Surgeons Columbia University Medical Center, New York, NY, USA

Email cva2111@columbia.edu

Ancillary