Can prolonged sick leave after gynecologic surgery be predicted? An observational study in The Netherlands.

Because not all 52 patients completed all the questionnaires, 46 patients were available for analysis. Of the 46 patients, 23 (50%) returned to work completely after 8 weeks, 14 (30%) resumed their work partly, and 9 (20%) did not resume work at all. In the analysis, the patients with complete return to work were compared with those who did not return or returned partly. A subset of six questions was used because in the logistic regression analysis, only these questions, if controlled for type of surgery, had an independent value for predicting return to work after 8 weeks (Table  2 ). Table 2 Capacity of the distinguished questions in the Recovery Index 10 (RI-10), controlled for type of surgery, to predict sick leave for 46 patients after gynecologic surgery a Questions RI-10 Odds ratio b p Value Slight exertion makes me feel tired 2.6 0.006 During the day I need to rest regularly 3.0 0.001 Even without activity, I am bothered by abdominal pain 1.8 0.025 Any light work (e.g., making coffee) exhausts me 2.2 0.008 I feel completely recovered after surgery 0.14 0.001 I can finish my daily activities at home without effort 0.2 0.003 Since the operation, I have problems sleeping 1.1 0.5 Surgery and recovery have not been as uneventful as I expected 1.3 0.2 I had a lot of pain after surgery 1.4 0.13 The complaints for which I had surgery are completely resolved 0.8 0.5 a Patients indicated their agreement on a Likert scale from 1 (completely agree) to 5 (completely disagree). The total score was divided by 10 and ranges from 1 to 5 b Odds ratio for return to work, controlled for type of surgery by logistic regression analysis Capacity of the distinguished questions in the Recovery Index 10 (RI-10), controlled for type of surgery, to predict sick leave for 46 patients after gynecologic surgery a a Patients indicated their agreement on a Likert scale from 1 (completely agree) to 5 (completely disagree). The total score was divided by 10 and ranges from 1 to 5 b Odds ratio for return to work, controlled for type of surgery by logistic regression analysis Recovery as expressed in the RI-6 after 1, 2, 3, 4, and 6 weeks improved with time after surgery and differed between patients who completely returned to work after 8 weeks and those who did not (Fig.  1 ). Figure  2 shows the predictive value of the RI-6, expressed as an ROC curve assessed at various moments after surgery. It appeared that the measurement 2 weeks after surgery showed the best discriminative capacity to predict sick leave after 8 weeks, with an AUC of 0.88 (confidence interval [CI], 0.74-1.03). The most predictive RI-6 score (at week 2) then was chosen as the subjective recovery instrument. The Cronbach alpha of the RI-6 was 0.94, and the inter-item correlation ranged from 0.71 to 0.92. Fig. 1 The mean RI–6 score of 46 patients after gynecologic surgery according to postoperative weeks (1, 2, 3, 4, and 6) and return to work after 8 weeks Fig. 2 Receiver operating characteristic curve of the Recovery Index 6 (RI-6) scores 1, 2, 3, 4, and 6 weeks after surgery predicting complete return to work after 8 weeks for 46 patients after gynecologic surgery The mean RI–6 score of 46 patients after gynecologic surgery according to postoperative weeks (1, 2, 3, 4, and 6) and return to work after 8 weeks Receiver operating characteristic curve of the Recovery Index 6 (RI-6) scores 1, 2, 3, 4, and 6 weeks after surgery predicting complete return to work after 8 weeks for 46 patients after gynecologic surgery Table  3 shows that the subjective postoperative recovery, as expressed by the RI-6, is more closely related to the type of surgery ( p  = 0.001) than sick leave is ( p  = 0.14). With regard to prediction of return to work, we found no significant association of return to work with level of education or type of work (sedentary, light, heavy), as shown in Table  4 . The way of employment (self-employment vs employee) and the type of surgery showed p values less than 0.2. We did not include the way of employment in the multivariate analyses because of the small number of self-employed patients. Table 3 Cross-tabulation of the extent to which the type of surgery is associated with the Recovery Index 6 (RI-6) and with the sick leave data for 46 patients scheduled for elective gynecologic surgery Minor surgery ( n ) Intermediate surgery ( n ) Major surgery ( n ) p (χ 2) RI-6 • 3 5 0 0 Sick leave • Yes a 4 13 6 0.14 • No 10 10 3 a No return or a partial return to work Table 4 Sick leave or a partial return to work according to preoperative variables for 46 patients undergoing benign gynecologic surgery Variable Sick leave a ( n ) Return to work ( n ) p Value Employment b • Self-employed 0 5 0.05 • Employed 22 18 Level of education b • Low 5 6 0.96 • Middle 6 6 • High 11 11 Type of work b • Light 11 13 0.47 • Moderate 5 7 • Heavy 6 3 Type of surgery • Minor 4 10 0.14 • Intermediate 13 10 • Major 6 3 Recovery Index 6 • 3 0 5 a No return or a partial return to work b One patient did not answer this question Cross-tabulation of the extent to which the type of surgery is associated with the Recovery Index 6 (RI-6) and with the sick leave data for 46 patients scheduled for elective gynecologic surgery a No return or a partial return to work Sick leave or a partial return to work according to preoperative variables for 46 patients undergoing benign gynecologic surgery a No return or a partial return to work b One patient did not answer this question Using the ROC curve, we found an RI-6 score of 1.7 to be the optimal cutoff point, with its sensitivity of 68% and its specificity 74% regarding return to work after 8 weeks. A low score was associated with 25% of the patients completely returned to work (predictive value, 75%) compared with 72% of the patients in relation to a high score. Logistic regression was performed using the type of surgery and the RI-6 score after 2 weeks as independent variables. The RI-6 was dichotomized due to the small number of subjects, with 1.7 used as a cutoff level. Only the RI-6 was an independent predictor of return to work, although the number of subjects was small (Table  5 ). Table 5 Logistic regression using return to work after 8 weeks as the dependent variable and using type of surgery and RI-6 score as independent variables for 46 patients after gynecologic surgery a Predicting factor Odds rate b CI p Value Type of surgery c 3.4 0.4–32 0.28 RI-6 index 6.5 1.9–21 0.002 CI confidence interval a Variable or variables entered in step 1: type of surgery and RI-6 index b The odds rate represents the ratio change in the odds of not being completely returned to work for a one-unit change in the predicting factor c Minor, intermediate, and major surgery Logistic regression using return to work after 8 weeks as the dependent variable and using type of surgery and RI-6 score as independent variables for 46 patients after gynecologic surgery a CI confidence interval a Variable or variables entered in step 1: type of surgery and RI-6 index b The odds rate represents the ratio change in the odds of not being completely returned to work for a one-unit change in the predicting factor c Minor, intermediate, and major surgery

Sick leave 8 weeks after gynecologic surgery is unexpectedly high (median. 50%) in a working population of Dutch patients with benign disease. No return to work 8 weeks after elective gynecologic surgery can be considered generally as a prolonged sick leave. The RI-6 shows good internal consistency and construct validity because the score correlates well with the time until return to work after surgery. Because the subjective recovery (RI-6) is more closely related than sick leave to the type of surgery (Table  3 ), it seems to be a better tool for comparing laparoscopic and open surgery, confirming that sick leave may not be based on recovery alone. Reasons for the high proportion of patients with a prolonged sick leave after surgery cannot be given because they were not addressed in the questionnaire. However, many women reported tiredness and a need for rest during the day (questions 1 and 2 of the RI-10). The lack of consistent recommendations by professionals about return to work may not promote early recovery. There is a substantial correlation between the RI-6 after 2 weeks and sick leave at 8 weeks. In the multivariate analysis, the RI-6 at postoperative week 2 was able to predict prolonged sick leave with considerable accuracy. Patients with an RI-6 score below 1.7, reflecting an obstructed recovery, have a 25% probability of being at work after 8 weeks, and patients with an RI-6 score higher than 1.7 have a 72% probability of being completely at work after 8 weeks. However, the number of subjects was too small for a logistic regression with all potentially predictive variables. For example, in the univariate analysis, patients who were self-employed (employers) apparently were at less risk for prolonged sick leave. This study had some methodologic shortcomings. Questions from the original RI-10 [ 4 ] were selected for the RI-6 based on their value for predicting return to work. Therefore, the findings need to be externally validated by their confirmation in another population. This study currently is underway. The sick leave was reported by the patients and not retrieved from the employer. This may affect the reliability of the data. Finally, presentation of the return-to-work data is more appropriate by life table analysis, but we had no data on the number of sick leave days. Several reasons exist for prolonged sick leave after surgery. One reason is varying information on convalescence by medical professionals. In a survey to collect current opinions among surgeons and general practitioners about the time a patient should be absent from work, a wide range of times off work were recommended [ 6 ]. Also in randomized trials on laparoscopic hysterectomy, the return to work varied from 10 [ 7 ] to 39 days [ 8 ], at least partly reflecting varying recommendations with regard to sick leave. It has been shown that standardizing postoperative recommendations can result in shortening of the sick leave [ 9 , 10 ]. In studies comparing minimally invasive and conventional surgery, the complication rate, admission time, and return to daily activities and work usually are given as outcomes, with the assumption that both approaches are equally effective. Rarely is the quality of life reflecting the subjective recovery used. In a systematic review of randomized trials comparing laparoscopic hysterectomy with total abdominal hysterectomy, only in 7 of 30 trials was quality of life used as an outcome, and only in 1 trial was it used as the primary outcome [ 11 ]. In four studies, validated questionnaires were used, with a difference between the two study arms found in only two studies. This may be explained by the fact that generic quality-of-life questionnaires such as the Short Form 36 or the EuroQol 5 were used. It is well known that generic questionnaires are not very sensitive in specific situations. Therefore, we recommend the use of a validated postoperative recovery-specific questionnaire that is optimally able to measure recovery after different surgical approaches. We have shown that the RI-6 is able to predict sick leave after surgery. This enables the detection of patients with slow recovery and prolonged return to work already at 2 weeks after surgery. In conclusion, we found the subjective recovery scored by the patient in a questionnaire of six questions to be a better outcome measure than sick leave for evaluating a surgical approach (e.g., laparoscopy vs laparotomy). If administered 2 weeks after surgery, it may predict prolonged sick leave. In this way, patients at risk for prolonged sick leave can be subjected to an ergonomic intervention, thus doing justice to the advantages of the laparoscopic surgery. Larger-scale studies are needed to confirm the predictive value and validity of the RI-6.

After approval of the institutional review board and written informed consent, 52 patients scheduled for elective gynecologic surgery in 2006 at the VU University Medical Center in Amsterdam, The Netherlands, who had a paid job, were entered into the study. Patients with oncologic disease, acute disease that required surgery, or ambulatory surgery were excluded. Before surgery, the patients filled out questionnaires that provided data on their employment and education level. The Recovery Index-6 (RI-6) is a subset of questions from the Recovery Index-10 (RI-10). The RI-10 is a quality-of-life questionnaire for measuring subjective postoperative recovery. Patients can indicate their level of agreement with each item on a scale from 1 (completely agree) to 5 (completely disagree). The total score of the RI-10, controlled for positive and negative formulations, is divided by 10 and ranges from 1 to 5. A score of 1 means complete subjective recovery. The RI-10 has been validated recently [ 4 ]. The RI-10 was self-administered by the patients 1, 2, 3, 4, and 6 weeks after surgery. The selection of six questions from the RI 10 was based on their individual association with sick leave. Additionally, after 8 weeks, the patients were approached by telephone to give information on their return to work. Their surgery was classified as minor, intermediate, and major, as stated in Table  1 . Table 1 Classification of gynecologic operations according to the gravity of surgery • Minor surgery Number o Diagnostic laparoscopy 3 o Laparoscopic adnexal surgery 3 o Laparoscopic adhesiolysis 3 • Intermediate surgery o Total laparoscopic hysterectomy 5 o Laparotomy for adnexal surgery 9 o Vaginal hysterectomy 9 • Major surgery o Total abdominal hysterectomy 9 o Excision endometriosis stage 3–4 5 Classification of gynecologic operations according to the gravity of surgery The data were analyzed using SPSS 15.0 (SPSS inc., Chicago, IL, USA). Cronbach’s alpha was calculated as a measure of the internal consistency of the RI-6 based on the average inter-item correlation. The chi-square statistic was used to assess the relation of the RI-6 and sick leave to the type of surgery. Receiver operator characteristics (ROC) curves were constructed to display the value of the RI-6 measured at different intervals after surgery for predicting return to work after 8 weeks. The ROC curves were constructed using the scores of the RI-6 at 1, 2, 3, 4, and 6 weeks. By calculating the area under the curve (AUC), the most accurate ROC curve was identified. Subsequently, the most accurate cutoff score of the curve was established, and test characteristics (sensitivity and specificity) were calculated. Univariate analyses were performed to assess the relation between various variables and sick leave after 8 weeks. Variables showing a p value less than 0.2 were included in the multivariate logistic regression analysis. Statistical tests were performed using a two-sided approach, and a p value less than 0.05 was considered significant.