A randomised comparison and economic evaluation of laparoscopic-assisted hysterectomy and abdominal hysterectomy
Correspondence: Dr M. A. Lumsden, Department of Obstetrics and Gynaecology, University of Glasgow, The Queen Mother's Hospital, Glasgow G3 8SJ, UK.
Objectives To determine the safety, cost effectiveness and effect on quality of life of laparoscopicassisted vaginal hysterectomy (LAVH) compared with total abdominal hysterectomy (TAH) in the management of benign gynaecological disease.
Design Randomised controlled trial and economic evaluation.
Setting Three hospitals in the West of Scotland.
Participants Two hundred women scheduled for an abdominal hysterectomy for benign gynaecological disease.
Main outcome measures Conversion rate of LAVH to TAH, complication rates, NHS resource use and costs, quality of life using EuroQol 5 D visual analogue scale, and achievement of milestones.
Results The overall incidence of operative complications was 14% in the TAH group and 8% in the LAVH group, with an 8% conversion rate. Length of operation was significantly greater in the women having LAVH at 81 ±30 min vs 47 ±16 min (P < 0.001). There was no difference in analgesic requirements between the groups although there was a significantly shorter hospital stay for those having LAVH. The rate of post-surgery recovery, satisfaction with operation and quality of life at four weeks post-operative were similar in the two groups of women. LAVH was significantly more expensive than TAH and remained more expensive for all but the most extreme scenario.
Conclusions This study demonstrates that despite the decreased length of hospital stay, LAVH is more expensive than TAH. In addition, recovery following operation and patient satisfaction were not affected by the route chosen. It is unlikely that LAVH represents an efficient use of NHS resources.
Hysterectomy is one of the most commonly performed operations in developed countries1. Until the 1990s, there were two techniques for hysterectomy: abdominal and vaginal. Observational studies suggest lower morbidity and quicker recovery2–4 in women having vaginal hysterectomy. However, most surgeons in the UK and elsewhere5–8 perform 75% to 80% of procedures by the abdominal route9,10 particularly when dealing with pelvic pathology or carrying out oophorectomy. Laparoscopic hysterectomy (LAVH) has been developed to allow laparoscopic techniques to be used to separate the uterus from the surrounding pelvic structures, the uterus then being removed through the vagina11, allowing rapid recovery and enabling oophorectomy to be achieved more easily than at vaginal hysterectomy.
Several small randomised controlled trials have compared laparoscopic and abdominal hysterectomy (TAH)12–17. These concluded that LAVH usually took longer but involved a shorter hospital stay and convalescence with an incidence of major complications of 3% to 5%18,19. Studies have found higher costs for LAVH than TAH, due to longer operation times and the use of disposable equipment12,13,20,21. We hypothesised that the laparoscopic approach would be associated with significantly greater patient satisfaction and more rapid recovery which would compensate for the greater operative cost.
Patients were recruited over a two year period from the gynaecological outpatient clinics of three hospitals in Glasgow. Patients were recruited if they were scheduled for an abdominal hysterectomy for benign gynaecological disease and if they were not suitable for vaginal hysterectomy, because of a uterine size in excess of 14 weeks or a requirement for oophorectomy. Since many of the women were to have an oophorectomy, those in whom hormone replacement therapy was not appropriate were excluded as this might have affected lifestyle assessment. Informed consent for the study was obtained by medical staff and randomisation was performed by the research nurse using a computer-generated schedule provided by the Department of Biostatistics, University of Glasgow, for each individual hospital. A date for operation was allocated to the patient within three months of randomisation.
Five consultant gynaecologists who had performed a minimum of 50 laparoscopic-assisted vaginal hysterectomies before starting the study were involved in all of the operations. Information was collected at the time of operation on anaesthesia and details of operation, including the presence of pathology, complications, duration, and extra surgery required. All items of disposable equipment used were recorded.
Immediate post-operative progress was recorded including the time the patient spent on the gynaecological ward and analgesic requirements. Assessment was made of bladder function, pyrexia, urinary tract infection with positive culture, chest infection requiring antibiotics, wound infection with a positive culture and superficial wound breakdown. Blood was taken to measure the haemoglobin concentration on the day before operation and 48 hours post-operatively. Patients were discharged after they had passed urine, were apyrexial and felt able to cope at home.
Patients were asked to keep a diary of the dates when simple ‘milestones’ to recovery were achieved. These included when the woman felt able to make a cup of tea, cook a meal, drive a car, resume sexual intercourse and return to work. The patients were then reviewed by the research nurse four weeks after surgery.
The restricted core of the Euroqol Health Questionnaire22 was used to measure women's evaluation of their health state before surgery and at one, six and twelve months after surgery. Evaluation of quality of life involves the use of a visual analogue thermometer when zero is labelled worst imaginable health state and hundred is labelled best imaginable health state.
Data were collected prospectively on all resources used during the initial inpatient stay including items of disposable equipment. This information was then confirmed by checking the case records for all hysterectomy-related resource use. This included preoperative stay, all blood tests, operation details, pain relief, post-operative stay, complications, additional surgery and re-admissions. A single set of unit costs was then applied to each unit of resource used to provide a NHS cost for each woman. A single set of costs was used to reflect true differences in resource use rather than different costing practices. All resource use was valued using 1997/1998 prices. Sensitivity analysis was used to test the robustness of results to changes in assumptions made.
Assuming complication rates similar to those reported previously23,24, 120 patients per arm would allow an 80% chance of detecting a 15% difference in complication rates at a 5% level using a two-sided test. Twenty-five patients per arm were required to detect a two day difference in hospital stay (one-tailed test at 5% significance). All data were entered on to the Statistical Package for the Social Sciences and analysis was on the basis of intention to treat. All data were assessed for normality and where appropriate nonparametric tests were used. Analysis was by intention to treat. The women were analysed (including all costs) in the groups to which they wee originally randomised.
Two hundred women were recruited to the study, 100 in each group. Recruitment in the first six months was rapid. However, the same time was required to recruit the last 10 women as to recruit the first 50. Seven women did not attend for operation and the case records were not available for a further three women, resulting in a sample size for analysis of 190 women (95 TAH, 95 LAVH). The response rate for the patient questionnaire was 87% and that for Euroqol was 78%, 64% and 47% at one, six and twelve months, respectively.
There was no significant difference in the general demographic characteristics of the patients or in the presence of pelvic pathology associated with hysterectomy (Table 1). The principal indications for operation were menstrual problems (TAH = 55% and LAVH = 59%) and pelvic pain (TAH = 17% and LAVH = 22%). Eight women did not have LAVH as randomised. Only three of the conversions between the two procedures were carried out for per-operative difficulty or complications, the remainder were converted either because of patient preference pre-operatively (n= 2) or because at diagnostic laparoscopy a very difficult LAVH was anticipated. Uterine arteries were divided laparoscopically in 76 women using a disposable stapling device (Endo GIA, Autosuture, Basingstoke, UK).
Table 1. The demographic characteristics, the incidence of previoussurgery and the presence of intra-abdominal pathology in the studygroups. Values are given as % or mean (SD). TAH = total abdominalhysterectomy; LAVH = laparoscopic-assisted vaginal hysterectomy.
|Age (years)||42.7 (6.4)||41.1 (5.6)|
|Body mass index (kg/m2)||26.6 (4.74)||26.3 (4.49)|
|Previous significant vaginal surgery*||3||6|
|Previous abdominal surgery†||29||21|
|Previous caesarean section||13||17|
The duration of operation was significantly less for TAH than LAVH, (Table 2). The overall complication rate, was 8% in the LAVH group and 14% in the TAH group (Table 3). Major complications, defined as those which were potentially life threatening, occurred in three women. These were two cases of haemorrhage requiring blood transfusion in the LAVH group (associated with pulmonary embolism in one woman) and gram-negative septicaemia unrelated to bowel injury. All three women made a full recovery. Urinary tract damage occurred on three occasions.
Table 2. Major items of resource use. Values are given as n, median (mean) or mean difference [95% CI]. ITU = TAH = total abdominalhysterectomy; LAVH = laparoscopic-assisted vaginal hysterectomy.
|Length of operation (min)||45 (47.3)||80 (81.9)||−34.6 [−41.1 to −27.8]|
|Total length of stay (days)||6.0 (5.7)||4.0 (4.0)||1.7 [1.2 to 2.2]|
|Admission to ITU||0||2|| |
|Women requiring additional surgery||2||3|| |
|Blood transfusions||2||1|| |
Table 3. Complications of surgery encountered in the study groups.Values are given as n. TAH = total abdominal hysterectomy;LAVH = laparoscopic-assisted vaginal hysterectomy;ITU = intensive care unit.
|Major complications|| || |
| Haemorrhage (requiring transfusion)||0||2|
| Urinary tract damage||1||2*|
| Pulmonary embolus||0||1|
| Bowel damage||0||0|
| Severe infection (ITU admission)||0||1|
|Minor complications|| || |
| Positive urine culture||6||4|
| Chest infection||4||0|
| Wound infection||4||1|
| Erythema wound||9||3†|
Two women experienced ureteric damage, one included in each group. However, the woman analysed in the LAVH group had opted to have TAH following randomisation but prior to being admitted for operation. The second case occurred during a routine TAH. A single case of bladder damage occurred in a woman having LAVH which required a laparotomy and surgical repair.
There were no instances of bowel damage. Minor problems also occurred, including pyrexia and wound infection and erythema, the latter being commoner after abdominal than laparoscopic hysterectomy (P= 0.014).
Complications unrelated to the hysterectomy also occurred during the follow up period. These included death from pancreatic cancer, pancreatitis with formation of a splenic abscess leading to splenic rupture and endometrial cancer requiring post-operative radiotherapy. The costs of these additional hospital treatments were excluded.
Of those women readmitted, five required further treatment (LAVH 3, TAH 2) due to a vault haematoma in three (TAH 2, LAVH 1) and reimplantation of ureter (LAVH 1, TAH 1). A leaking wound (TAH 1), chest pain (LAVH 1) and abdominal pain not requiring treatment accounted for the remainder of cases.
Duration of hospital stay was significantly less for women having LAVH (Table 2). However, the total amount of opiate analgesia used in the immediate postoperative period did not differ between the two groups (45.8 mg of morphine in the TAH group and 43.3 mg in the LAVH group). This Fig. included opiates given during the operation as well as those given either as bolus injections or by percutaneous pump. There was no difference in oral analgesic use between the treatment groups.
Table 2 shows major items of resource use and Table 4 shows NHS costs. TAH was associated with significantly lower total costs to the NHS than LAVH, resulting principally from the difference in operation costs. When the cost of disposable equipment was removed the difference, while still large, was nonsignificant. The results were sensitive to changes in underlying assumptions with LAVH associated with higher costs than TAH for all scenarios except the most extreme (Table 4).
Table 4. Costs (in £) to NHS. Values are given as n, median (mean) or mean difference [95% CI]. IP = inpatient.
|Operation cost||535.5 (565.5)||1252 (1280)||−715 [−795 to −635]|
|Inpatient stay||1227 (1178)||848 (1028)||151 [−191 to 493]|
|Readmission||0 (88)||0 (171)||−83 [−325 to 160]|
|Total cost||1667 (1832)||2112 (2479)||−647 [−1181 to −113]|
|Cost excluding disposables||1667 (1832)||1740 (2173)||−344 [−881 to 193]|
|Sensitivity analysis|| || || |
|Assumption|| ||Mean cost difference (TAH - LAVH)|| |
|Cost per minute in theatre halved|| ||−443 [−973 to 87]|| |
|Cost per minute in theatre doubled|| ||−858 [−1400 to −317]|| |
|Cost per IP day halved|| ||−796 [−1321 to −272]|| |
|Cost per IP day doubled|| ||−464 [−1015to 88]|| |
|Cost per minute in theatre halved and cost per IP day doubled|| ||−259 [−807 to 288]|| |
|Cost per minute in theatre halved, cost per IP day doubled and no disposable costs|| ||44 [−505 to 594]|| |
There were no differences in self-reported post-operative problems, with the exception of the category ‘other problems’, such as tiredness or constipation, where women who had LAVH reported higher levels (Table 5). Despite their earlier discharge home, patients having LAVH did not achieve post-operative milestones earlier than those having TAH (data not shown). Table 6 shows the changes in women's valuations of their present health state compared with before surgery. There were no significant differences in the change at one month, six months or a year after surgery.
Table 5. Patient reported outcomes obtained from the self-completed patient questionnaire. Values are given as n or mean(SD). GP = general practitioner.
|Oral analgesic use||68||75||0.14|
|No. of visits to GP||1.71 (0.92)||1.75 (0.94)||0.9|
|Antibiotics prescribed (all causes)||23||35||0.11|
|Urinary symptoms (dysuria)||12||18||0.38|
|Difficulty with micturition||22||21||0.59|
|Operation fulfilling expectation||63||59||0.28|
Table 6. Changes in valuation of health state using the Euroqol visual analogue scale, post-op = post operation; pre-op = pre-operation.
|One month post-op minus pre-op|| || || |
| TAH (n= 76)||6.8 (19.2)||8 (−50 to +60)||−16 (−7.2 to 6.9)|
| LAVH (n= 74)||7 (24.1)||10 (−50 to +50)|| |
|Six months post-op minus pre-op|| || || |
| TAH (n= 61)||14.9 (16.7)||15 (−20 to 60)||3.7 (−3.7 to 11)|
| LAVH (n= 62)||11.3 (23.9)||15 (−50 to 60)|| |
|One year post-op minus pre-op|| || || |
| TAH (n= 47)||15.9 (21)||15 (−40 to 60)||4.9 (−6.7 to 12.8)|
| LAVH (n= 43)||12.6 (25)||14 (−40 to 73)|| |
In this study we performed an economic analysis and quality of life assessment alongside a randomised trial of LAVH and TAH performed by gynaecologists experienced in laparoscopic surgery. LAVH was converted to TAH for per-operative difficulty on three occasions only. The complication rate compared well with those of other randomised studies14–18, but was significantly lower than anticipated. From data published prior to 19944,23,24 we estimated a possible difference in complication rates of 25% and based our sample size calculation on this.
However, we experienced difficulty in recruitment as the study proceeded, since women began to express a preference for either TAH or LAVH. Consequently, it was not possible to conduct a study with sufficient power to detect a difference in complication rates between LAVH and TAH.
LAVH took longer than TAH. When the ratio of the duration of LAVH and TAH was compared for each surgeon, the median value was 1.78 (range 1.5–2.6) and the mean duration of LAVH for each individual surgeon did not decrease during the study. This suggests that although it is possible to reduce the time of surgery, LAVH is likely to take longer than TAH even with increased experience.
Length of hospital stay was a mean of two days less in women having LAVH. Initial analgesic requirements did not differ between the two groups suggesting patient expectation could be of importance as well as the bias of individual surgeons as to the appropriate time for discharge. The rate of recovery after LAVH was similar to TAH and women were equally satisfied with the two types of operation at all stages of follow up.
LAVH was significantly more expensive than TAH due to the longer operating time and the use of disposable equipment. The finding that LAVH was associated with higher costs than TAH, except under the most extreme scenario, supports observations from other comparisons of laparoscopic versus open surgery25. Other studies have justified the use of LAVH suggesting that the shorter recovery time compensates for increased operation costs20,21,26. Our study does not support this since we did not observe such an effect on time taken to return to normal activity.
For a procedure to be cost effective requires that either costs are minimised for a given outcome, or outcomes maximised for a given cost. Although LAVH was only associated with significantly higher costs in the baseline case, costs for LAVH were higher than for TAH for all but the most extreme scenario. We also observed no difference in clinical outcomes, patient reported outcomes, including time taken to return to normal activity, and patient reported quality of life. Although the number of women completing the questionnaires decreased with time, we do not feel that this negates our results since no difference in rate of recovery was demonstrated at four weeks when over 80% of women replied. We can therefore conclude that LAVH is unlikely to be cost effective.
We feel our results are robust since our conversion rate and complication rate were low and we were performing operations which make up the bulk of the average gynaecological practice. We feel that the conclusions are of significance for policy makers and health service managers in determining allocation of resources for development of laparoscopic surgery. We had anticipated that the reduced hospital stay would offset the increased cost of operation, but this was not the case, although this may be important in units with limited numbers of inpatient beds. However, when considering the results as a whole, we do not feel that we can advocate the wholesale replacement of TAH by LAVH for routine hysterectomy.
The increased cost of laparoscopic hysterectomy over abdominal hysterectomy is justified by assuming that recovery will be quicker following the laparoscopic procedure since the traditional total abdominal hysterectomy involves an abdominal incision.
However, we have demonstrated that there was no difference in patient satisfaction or recovery time between the two types of operation. LAVH was significantly more expensive than TAH due to the longer operating time as well as the use of disposable equipment. Patients left hospital more quickly following LAVH but reached milestones for recovery after a similar length of time. Both routes were equally safe and the overall complication rate was low. Our results fail to suggest a need for all gynaecologists to develop the skills to perform LAVH.
We would like to thank Ms K. Jack and Mr N. O'Shaughnessy for their contribution to data entry and analysis. Sandra Quinn, Cost Accountant at Stobhill Hospital NHS Trust is thanked for the help with cost analysis. Finally, we thank the women who participated in the study.
This work was funded by a grant from the Chief Scientist's Office, Scottish Home and Health Department (Grant No. K/MRS/50/C2206).