Cochrane review: post-operative procedures for improving fertility following pelvic reproductive surgery

Transcription

1 Human Reproduction Update 2000, Vol. 6 No. 3 pp European Society of Human Reproduction and Embryology Cochrane review: post-operative procedures for improving fertility following pelvic reproductive surgery N.P.Johnson 1,* and A.Watson 2 1 University Department of Obstetrics & Gynaecology, National Women s Hospital, Auckland, New Zealand and 2 Tameside General Hospital, Ashton-Under-Lyne, Lancashire OL6 9RW, UK Received on September 21, 1999; accepted on February 10, 2000 The objectives of the study was to determine the effectiveness of post-operative procedures following female pelvic reproductive surgery. A systematic review employing the principles of the Cochrane Menstrual Disorders and Subfertility Group was used. Five randomized controlled trials were included. Participants were women undergoing pelvic reproductive surgery; interventions were any post-operative procedure designed to improve fertility; outcomes were pregnancy, live birth, ectopic pregnancy and miscarriage rates and the rates of tubal patency and procedure-related complications. Summary statistics were expressed as odds ratios. The results showed that the odds of pregnancy, live birth, ectopic pregnancy and miscarriage were not significantly altered by post-operative hydrotubation nor second-look laparoscopy with adhesiolysis. Whether hydrotubation was early or late and whether hydrotubation fluid contained steroid or not had no significant impact on the odds of pregnancy, live birth, ectopic pregnancy or miscarriage. The odds of pregnancy and live birth were significantly increased and infective complications significantly decreased by hydrotubation with fluid containing antibiotic compared with hydrotubation with fluid containing no antibiotic, in late hydrotubation following tubal stent removal 6 weeks after tubal surgery. The odds of at least one patent Fallopian tube were significantly increased with late hydrotubation following tubal stent removal compared with early hydrotubation in women who had no tubal stenting, but this intervention had no significant impact on the odds of pregnancy, live birth, ectopic pregnancy or miscarriage. In conclusion, there is insufficient evidence to support the routine practice of hydrotubation or second-look laparoscopy following female pelvic reproductive surgery. The studies on which this conclusion is based were either poor quality or underpowered. Post-operative hydrotubation with fluid containing antibiotic following tubal surgery may offer benefit over hydrotubation fluid without antibiotic. A randomized controlled trial of post-operative hydrotubation with antibiotic-containing fluid versus no hydrotubation for improving fertility following tubal surgery is justified. Key words: adhesiolysis/hydrotubation/infertility/post-operative/second-look laparoscopy TABLE OF CONTENTS Introduction 259 Literature search 260 Description and methodological quality of studies 260 Data analysis 264 Evidence from randomized controlled trials: discussion 264 Conclusions 267 Acknowledgements 267 References 267 Introduction The importance of evidence-based practice in infertility was highlighted by the Royal College of Obstetricians and Gynaecologists (RCOG) in the publication of evidence-based guidelines for the initial investigation and management of the infertile couple and the management of infertility in secondary care (RCOG, 1998a,b). These guidelines did not address the issue of whether post-operative procedures following female pelvic reproductive surgery are of benefit. Pelvic reproductive surgery may be defined as surgery on the pelvis intended to improve fertility. It may include surgery on the uterus, ovaries, pelvic peritoneum (in the case of endometriosis) and, most commonly, the Fallopian tubes. A variety of procedures is collectively known as tubal surgery; salpingo-ovariolysis is division of adhesions involving Fallopian tube and ovary; salpingostomy is the refashioning of a distal tubal ostium for distal tubal occlusion and is designed to keep the Fallopian tube open; tubal reanastomosis is the rejoining of Fallopian tubes typically performed for reversal of sterilization; cornual anastomosis and utero tubal implantation are recognized surgical treatments for cornual occlusion. Whilst surgery on the Fallopian tubes has been * To whom correspondence should be addressed at: University Department of Obstetrics & Gynaecology, National Women s Hospital, Claude Road, Epsom, Auckland, New Zealand. Phone: ; Fax: ; n.johnson@auckland.ac.nz

2 260 N.P.Johnson and A.Watson performed less frequently since the advent of IVF (Novy, 1995), other types of pelvic reproductive surgery, including ovarian drilling as a treatment for clomiphene-resistant polycystic ovarian syndrome, are being performed more widely. The pregnancy and live birth rates following female pelvic reproductive surgery can be disappointing, in spite of apparent normal anatomical relationships at the time of surgery and restoration of Fallopian tube patency in tubal surgery. Pre-existing damage to the sensitive fimbrial end of the Fallopian tube and intrinsic inflammatory tubal damage limit the success of reproductive surgery (Winston and Margara, 1991; Strandell et al., 1995). Other causes of failure to conceive following pelvic surgery include postoperative adhesion formation or (re)occlusion of the Fallopian tubes. Failure to capture or to pass the released egg down the Fallopian tube may result. Hydrotubation is the transcervical flushing of fluid through the Fallopian tubes. It is widely used in the technique of hysterosalpingography (HSG) to provide diagnostic information about the Fallopian tubes in the investigation of infertility. Hydrotubation with oil-soluble contrast media is of proven primary benefit in infertile couples (Vandekerckhove et al., 1999) and may be due to the flushing out of occluding tubal plugs. Hydrotubation could theoretically flush out debris which may accumulate within the lumen of the Fallopian tube after tubal surgery, thus reducing the likelihood of tubal reocclusion. Increased pregnancy rates have been claimed in women undergoing early post-operative hydrotubation following tubal surgery (Grant, 1971). Laparoscopy is widely used for diagnostic purposes in infertility investigation, usually in association with hydrotubation to determine tubal patency. It has the advantage over hysterosalpingography that division of pelvic adhesions may be performed at the same time. Adhesiolysis, whether performed at laparotomy (Tulandi et al., 1990) or laparoscopically (Reich, 1987), is of primary benefit in tubal infertility. It has been claimed that secondlook laparoscopy with adhesiolysis following pelvic reproductive surgery may increase the intrauterine pregnancy rate (Surrey and Friedman, 1982) and decrease the ectopic pregnancy rate (Trimbos- Kemper et al., 1985), although others have failed to demonstrate increased pregnancy rates (Gomel et al., 1986; Luber et al., 1986). The objective of this review was to assess, from the best available evidence, the value of post-operative procedures performed following female pelvic reproductive surgery in the belief that they would improve fertility; hydrotubation and second-look laparoscopy (with adhesiolysis where appropriate) are examples of these procedures. Literature search Search strategy This review has employed the search strategy developed by the Cochrane Menstrual Disorders and Subfertility Group. Relevant trials were identified from the Trial Register of the Review Group. The electronic databases MEDLINE, EmBASE, PsychLIT, Current Contents, Biological Abstracts, Social Sciences Index and CINAHL were searched. The following terms were included in the electronic search strategy: pelvic surgery, tubal surgery, fertility surgery or reproductive surgery with, post-operative, post-operative procedures, hydrotubation, hysterosalpingogram, HSG, post-operative laparoscopy, second-look laparoscopy or adhesiolysis. Handsearching of relevant journals was performed and citation lists of included trials, eligible studies, conference abstracts and relevant review articles were also searched. For the included trials either the first or corresponding author was contacted for additional information on trial methodology or actual original trial data where a trial appeared to meet the eligibility criteria, when aspects of methodology were unclear. Identification of included trials Inclusion eligibility criteria were as follows: (i) randomized controlled trials; (ii) interventions, post-operative hydrotubation, different modalities of post-operative hydrotubation and secondlook laparoscopy with adhesiolysis (whether or not the trials included other intraoperative or post-operative interventions); and (iii) primary outcomes of pregnancy or live birth specified. Secondary outcomes, i.e. ectopic pregnancy rate (per woman), miscarriage rate (per intrauterine pregnancy), infection rate, other complication rate and tubal patency rate (at least one Fallopian tube remaining open) were also studied. The selection of trials for inclusion in the review was performed independently by the two reviewers after employing the search strategy described previously. The search strategy yielded 114 studies and of these, 28 papers required closer scrutiny based on the title and abstract. A total of 14 studies examined pregnancy outcome following the interventions but only five of these were found to be randomized controlled trials eligible for inclusion. Included studies were also assessed independently for predefined quality criteria and methodological details (Tables I and II). Data extraction All data was extracted independently by the two reviewers, entered into the Review Manager (RevMan 4.0.3) computer software (Update Software, Oxford, UK) and double-checked for accuracy. Additional information on trial data was sought from the corresponding authors where data were in a form unsuitable for metaanalysis. Statistical analysis was performed in accordance with the guidelines for statistical analysis developed by the Cochrane Menstrual Disorders and Subfertility Group (1999). For dichotomous data, odds ratios and 95% confidence intervals were calculated and, in the absence of heterogeneity (examined by inspecting the scatter in the data points on the graphs and the overlap in their confidence intervals and, more formally, by checking the results of χ 2 tests), results were pooled using a fixed effects model. Description and methodological quality of studies Five trials with a total of 588 women met the criteria for inclusion in the review. All trials were fully published in peer-reviewed journals. No trials which met the eligibility criteria were excluded. Three trials examined post-operative hydrotubation (Soihet, 1974; Comninos, 1977; Rock et al., 1984); two trials examined secondlook laparoscopy with adhesiolysis (Tulandi et al., 1989; Gurgan et al., 1992). Additional information has been sought for all five of these trials and has been forthcoming, to date, only in two cases (Tulandi et al., 1989;Gurgan et al., 1992). The authors are unaware of any ongoing trials relevant to the review. A detailed description of the three trials assessing post-operative hydrotubation is given in

3 Cochrane review: improving fertility 261 Table I. Trials assessing post-operative hydrotubation Comninos, 1977 Rock et al., 1984 Soihet, 1974 Trial characteristics Method of randomization Truly randomized (pack of cards) Allocation concealment Inadequate Unclear Inadequate Timing of randomization Multicentre/single centre Single centre Multicentre (total 5 centres) Single centre Utilization of blinding No No No No. of women analysed Withdrawals/losses to follow-up Unclear Unclear Unclear Intention to treat analysis Uncertain Uncertain Uncertain Power calculation Not done Not done Not done Location of the trial Marika Eliadi Maternity Hospital, Athens, Greece USA (Baltimore, Brooklyn, Durham), Netherlands (Amsterdam), Colombia (Bogota) San Marcos University, Lima, Peru Duration and timing of trial Jan 1978 to Dec 1981 Duration of follow-up At least 24 months 24 months 36 months Source of funding Baseline characteristics of the studied groups Age of women in the trial Range years (mean not stated) Mean 28 years (all under 36 years) Definition and duration of preexisting infertility Primary or secondary infertility; duration 2 4 years (mean not stated) Primary or secondary infertility; mean duration 12 years (range 1 19 years) Primary or secondary infertility; duration not stated Investigative work-up SA, PCT, BBT, cervical mucus examination, endometrial biopsy, HSG +/ laparoscopy SA,PCT,ovulationdetection (method not stated), HSG and laparoscopy in 90% Primary cause of infertility Bilateral distal tubal occlusion Distal tubal disease Tubal immobility or occlusion Previous administered treatments Therapeutic hydrotubation with hydrocortisone/ampicillin solution three times over each of four cycles Exclusion criteria Fibroids, endometriosis or ovarian cysts Nature of primary surgery Macrosurgical salpingostomy covered by antibiotics (not specified) Bilateral salpingoneostomy or fimbrioplasty (unilateral if only 1 residual tube) Tubal surgery involving cuff salpingostomy, tubal reanastomosis, tubal reimplantation, adhesiolysis or any combination of the above, in addition to ventrosuspension, by laparotomy. The procedure was covered by intraperitoneal instillation of chloramphenicol or penicillin and the same antibiotics systemically for 5 days

4 262 N.P.Johnson and A.Watson Table I. Continued Interventions Nature and timing of intervention Absence of other procedures in treatment or control group Early postoperative hydrotubation on days 3, 5 and 7 (using normal saline with 50 mg hydrocortisone but no antibiotics) versus intraoperative insertion of prosthesis (Cognat s device) followed by postoperative transabdominal peritubal irrigation via the device on days 3, 5and7(usingnormalsalinewith 50 mg hydrocortisone and antibiotics not specified) followed by late post-operative hydrotubation (using normal saline with 50 mg hydrocortisone but no antibiotics) following removal of the prosthesis on day 12 Post-operative hydrotubation on post-operative days 1, 2 and 3 and on the day of discharge home with 50 ml Ringer s lactate versus 50 ml Ringers lactate with 150 mg hydrocortisone versus no hydrotubation Early post-operative hydrotubation (solution not specified) twice weekly from day 5 to 6 weeks versus late post-operative hydrotubation twice weekly from week 6 12 where the surgery had included teflon stenting of the Fallopian tubes (removed electively after 6 weeks) versus late post-operative hydrotubation twice weekly from weeks 6 12 with a solution containing gentamicin where the surgery included tubal suspension in addition to teflon tube stenting of the Fallopian tubes (removed electively after 6 weeks) Other procedures present Absent Other procedures present Outcomes Method of expressing primary outcomes Ratesperwomaninagiventime period Ratesperwomaninagiventime period Ratesperwomaninagiventime period Primary outcomes Total pregnancy rate Total pregnancy rate Live birth rate Pregnancy or live birth rate Method of diagnosis of pregnancy Term delivery Secondary outcomes Ectopic pregnancy rate Miscarriage rate Tubal patency rate Ectopic pregnancy rate Miscarriage rate Infection and complication rates Ectopic pregnancy rate Tubal patency rate Infection rate Other outcomes Total number of pregnancies and live births in each group Bilateral tubal obstruction rate SA = semen analysis; PCT = post-coital test; BBT = basal body temperature charting; HSG = hysterosalpingography. Table I and that of the two trials assessing second-look laparoscopy with adhesiolysis in Table II. There were trial quality concerns, particularly with the older trials, although these were not sufficient to warrant exclusion of any eligible randomized controlled trials from the review. Heterogeneity in the design and quality of these five trials should be noted and the reasons for particular concern over trial quality are outlined below. Three of the five randomized controlled trials explicitly stated a genuine method of randomization, but this was unclear for two trials (Soihet, 1974; Comninos, 1977). Allocation concealment was unclear in one trial (Rock et al., 1984) and probably inadequate in two trials (Soihet, 1974; Comninos, 1977). Two trials had unusual numbers in their groups: three groups with 86, 60 and 60 participants (Rock et al., 1984) and three groups with 67, 100 and 91 participants respectively (Soihet, 1974) which raise concerns over the randomization process. The uneven numbers per group could represent deliberately unequal randomization, failure to use permuted blocks, abuse of the process, post-randomization exclusions or combinations of the above. None of the trials used blinding, indeed by the nature of the interventions (hydrotubation or second-look laparoscopy), blinding would be difficult. However, two trials (Soihet, 1974; Rock et al., 1984) could have built double-blinding into two arms of each of their studies. Scope for bias from not blinding includes the possibility of more thorough follow-up by investigators to find outcomes in couples not attending follow-up clinics. Independent singleblinding of investigators performing follow-up could have been adoptedinalltrials. Only two trials claimed a power calculation: in the discussion section of the published paper (Gurgan et al., 1992) and in correspondence (Tulandi et al., 1989) respectively. However, it is unclear to what extent such power calculations to determine sample size were adhered to, since these trials were both underpowered. Withdrawals after randomization were mentioned in only one trial (Gurgan et al., 1992); one participant declined second-look laparoscopy and was excluded from the analysis. In this case an intention-to-treat analysis was not done by the authors but was possible and was performed in this review. Since no withdrawals or losses to follow-up were mentioned in any of the other trials, it is

5 Cochrane review: improving fertility 263 Table II. Trials assessing second-look laparoscopy with adhesiolysis Trial characteristics Gurgan et al., 1992 Tulandi et al., 1989 Method of randomization Truly randomized (table of random numbers) Truly randomized (computer-generated random numbers) Allocation concealment Adequate Adequate Timing of randomization After primary surgery 1 year after primary surgery Multicentre/single centre Single centre Single centre Utilization of blinding No No No.ofparticipantsanalysed Withdrawals/losses to follow-up 1 withdrawal (declined intervention) Some lost to follow-up after randomization but taken into account in life table analysis Whether an intention to treat analysis was: (i) done by author (ii) not done by author but possible (iii) not possible (iv) uncertain Power calculation Not done by author but possible Described in discussion section possibly performed post-trial (trial underpowered) Uncertain Claimed in correspondence (trial underpowered) Location of the trial University of Hecettepi, Ankara, Turkey Fertility Centre, McGill University, Montreal, Canada Duration and timing of trial Recruitment took 6 months 2 years ( ) Duration of follow-up 6 months 36 months Source of funding Baseline characteristics of the studied groups Age of women in the trial Mean 25.2 years (range years) Mean 30 years Definition and duration of pre-existing infertility 33 primary, 7 secondary infertility; mean duration 4.4 years Duration at least 1 year Investigative work-up SA, HSG SA, endometrial biopsy, luteal phase progesterone, HSG Primary cause of infertility (and other causes) Polycystic ovarian syndrome resistant to clomiphene Tubal disease Previous administered treatments Clomiphene Exclusion criteria Endometriosis Nature of primary surgery Interventions Nature and timing of intervention Absence of other procedures in treatment or control group Laparoscopic Nd YAG laser photocoagulation of the ovaries Post-operative second-look laparoscopy and adhesiolysis 3 4 weeks following surgery versus no second-look laparoscopy Absent Tubal microsurgery with instillation of 32% Dextran (38 had salpingo ovariolysis and 36 bilateral terminal salpingostomy) Laparoscopic tubal patency testing +/ laparoscopic adhesiolysis where indicated versus no laparoscopy 12 months after initial tubal microsurgery Absent

6 264 N.P.Johnson and A.Watson Table II. Continued Outcomes Method of expressing primary outcomes Rates per woman in a given time period Continuous time survival analysis Primary outcomes Method of diagnosis of pregnancy (if stated) Secondary outcomes Other outcomes Total pregnancy rate Live birth plus ongoing pregnancy rate Serum β HCG concentrations and real time ultrasound Ectopic pregnancy rate Miscarriage rate Ovulation rate Multiple pregnancy rate Cumulative probability of pregnancy Cumulative probability of live birth Urine pregnancy test and ultrasound Cumulative probability of ectopic pregnancy Tubal patency rate SA = semen analysis; HCG = human chorionic gonadotrophin; HSG = hysterosalpingography. possible that only women completing the study were used in the results analysis, meaning that an intention-to-treat analysis was probably not done, nor would be possible for those trials. A major concern when assessing the effect of the intervention hydrotubation, is that its effect was clouded by the presence of other interventions in the control groups in two trials (Soihet, 1974; Comninos, 1977). The number of pregnancies was expressed in all trials and the pregnancy rate per woman could be extracted from the data in all trials this was done for all outcomes to maintain homogeneity of outcomes as far as possible for this review. Comparison of these rates per woman within a defined time period was, therefore, possible in all studies except one (Soihet, 1974) in which live birth rates were specified. The preferable method of reporting success rates in any trial would be discrete time survival analysis, followed by continuous time survival analysis, then rates per woman within a given time period. Most authors collect insufficient information to analyse (or meta-analyse) on the basis of survival techniques. It was considered likely that a comparison of rates within a defined time period would be realistic and was used in this review. Only one trial expressed the outcomes as continuous time survival analysis (Tulandi et al., 1989). Data analysis Table III expresses the results for primary and secondary outcomes for which data was available from the trials. There were differences in most outcomes between the five trials. For example the total pregnancy rates per woman in each trial were 27% (Comninos, 1977), 50% (Gurgan et al., 1992), 21% (Rock et al., 1984), 20% (Soihet, 1974), 42% (Tulandi et al., 1989), although these rates were expressed over different time periods (see Tables I and II). Trial results have been combined for only five outcomes, with not more than two studies in any analysis. There was no evidence of statistical heterogeneity. The intervention hydrotubation did not significantly alter the odds of pregnancy, live birth, ectopic pregnancy or miscarriage. Addition of hydrocortisone to the hydrotubation fluid did not significantly alter the odds of pregnancy, live birth, ectopic pregnancy or miscarriage. Addition of gentamicin to the hydrotubation fluid significantly increased the odds of pregnancy and live birth, decreased the odds of infection, but failed to significantly alter the odds of ectopic pregnancy or tubal patency. However, the women who received hydrotubation with gentamicin received the additional intervention of tubal suspension at initial surgery. In the metaanalysis, early hydrotubation versus tubal stent and late hydrotubation was associated with significantly increased odds of infective morbidity and significantly reduced odds of patency of at least one Fallopian tube at subsequent assessment, which suggested a possible benefit of the use of tubal stent or prosthesis with late hydrotubation following its removal, but failed to significantly alter the odds of pregnancy, livebirth, ectopic pregnancy or miscarriage. The odds of infection were reduced in the case of teflon stenting of the Fallopian tubes for 6 weeks (Soihet, 1974). The odds of tubal patency were increased when evidence from the use of the salpingostomy device of Cognat for 12 days (Comninos, 1977) in addition to that from tubal stenting (Soihet, 1974) was considered. Whether considered independently or combined in a metaanalysis, early second-look laparoscopy with adhesiolysis and late second-look laparoscopy with adhesiolysis did not significantly alter the odds of pregnancy, livebirth, ectopic pregnancy or miscarriage. Three trials reported that there were no post-treatment infections or complications in either treatment or control groups (Rock et al., 1984; Tulandi et al., 1989; Gurgan et al., 1992), whilst in one case, in which febrile morbidity was stated specifically as one of the outcomes of the trial, infections were reported in as many as 60% of women in the early hydrotubation group (Soihet, 1974). The wide heterogeneity in the occurrence of febrile morbidity in these trials suggests these data should be interpreted cautiously. Evidence from randomized controlled trials: discussion This systematic review of randomized controlled trials has failed to show a significant benefit or detriment of hydrotubation or secondlook laparoscopy with adhesiolysis following pelvic reproductive surgery. Hydrotubation with antibiotic versus hydrotubation without antibiotic may increase the odds of pregnancy and livebirth and reduce the odds of infection. The use of a tubal stent or prosthesis followed, after removal, by late hydrotubation may reduce the

7 Cochrane review: improving fertility 265 Table III. Summary data of primary and secondary outcomes for all interventions considered Hydrotubation versus no hydrotubation References Experiment n/n (%) Control n/n (%) Odds ratio 95% confidence interval Total pregnancy rate Rock et al. (1984) 26/120 (21.7) 17/86 (19.8) Live birth rate Rock et al. (1984) 9/120 (7.5) 10/86 (11.6) Ectopic pregnancy rate Rock et al. (1984) 9/120 (7.5) 4/86 (4.7) Miscarriage rate Rock et al. (1984) 8/17 (47.1) 3/13 (23.1) Infection rate Rock et al. (1984) 0/120 (0) 0/86 (0) Not calculable Complication rate Rock et al. (1984) 0/120 (0) 0/86 (0) Not calculable Hydrotubation with steroid versus without steroid Total pregnancy rate Rock et al. (1984) 12/60 (20) 14/60 (23.3) Live birth rate Rock et al. (1984) 3/60 (5) 6/60 (10) Ectopic pregnancy rate Rock et al. (1984) 6/60 (10) 3/60 (5) Miscarriage rate Rock et al. (1984) 3/6 (50) 5/11 (45.5) Infection rate Rock et al. (1984) 0/60 (0) 0/60 (0) Not calculable Complication rate Rock et al. (1984) 0/60 (0) 0/60 (0) Not calculable Hydrotubation with antibiotic versus without antibiotic Total pregnancy rate Soihet (1974) 26/91 (28.6) 15/100 (15) 2.23* Live birth rate Soihet (1974) 21/91 (23.1) 11/100 (11) 2.37* Ectopic pregnancy rate Soihet (1974) 5/91 (5.5) 4/100 (4) Infection rate Soihet (1974) 0/91 (0) 23/100 (23) 0.12* Tubal patency rate Soihet (1974) 56/91 (61.5) 65/100 (65) Early hydrotubation versus tubal stent and late non-antibiotic hydrotubation Total pregnancy rate Soihet (1974); Comninos (1977) 14/82 (17.1) 20/115 (17.4) Live birth rate Soihet (1974) 7/67 (10.4) 11/100 (11) Ectopic pregnancy rate Soihet (1974); Comninos (1977) 5/82 (6.1) 5/115 (4.4) Miscarriage rate Comninos (1977) 1/4 (25) 2/7 (28.6) Infection rate Soihet (1974) 40/67 (59.7) 23/100 (23) 4.72* Tubal patency rate Soihet (1974); Comninos (1977) 27/82 (32.9) 73/115 (63.5) 0.30* Second-look laparoscopy with adhesiolysis versus no second-look laparoscopy Total pregnancy rate Tulandi et al. (1989); Gurgan et al. (1992) 25/56 (44.6) 26/59 (44.8) Live birth rate Gurgan et al. (1992) 7/20 (35) 9/20 (45) Ectopic pregnancy rate Tulandi et al. (1989); Gurgan et al. (1992) 8/56 (14.3) 6/58 (10.3) Miscarriage rate Gurgan et al. (1992) 2/9 (22.2) 2/11 (18.2) Infection rate Tulandi et al. (1989); Gurgan et al. (1992) 0/56 (0) 0/58 (0) Not calculable Complication rate Tulandi et al. (1989); Gurgan et al. (1992) 0/56 (0) 0/58 (0) Not calculable Tubal patency rate Tulandi et al. (1989) 36/36 (100) 38/38 (100) Not calculable Experiment group refers to the first intervention, control group refers to the second intervention listed in the left hand column.

8 266 N.P.Johnson and A.Watson odds of infection and increase the odds of at least one Fallopian tube remaining patent but there is no evidence that it improves fertility. Whilst this review did not demonstrate a significant increase in the odds of infection or complications following the post-operative procedures, the interventions studied are not without invasive characteristics. Hydrotubation is uncomfortable, embarrassing and carries a cost. Laparoscopy causes discomfort, carries the hazards of general anaesthesia, the rare but recognized hazards of gastrointestinal, urinary tract or vascular injury and, even when performed on a day-case basis, is costly for the health service and the patient. The interventions should not be performed without evidence of benefit. Increased pregnancy rates have been claimed in women undergoing post-operative hydrotubation with a solution containing hydrocortisone compared with a control group (Grant, 1971) but this trial studied consecutive patients and used a historical control group. There are reports of a variety of types of hydrotubation fluid used post-operatively (Arronet et al., 1969; Jessen, 1971; Rock et al., 1978), but no other studies have claimed benefit of postoperative hydrotubation over a control group. Whilst the number of women in the three trials examining post-operative hydrotubation was 474, there are serious quality concerns over all three trials (Soihet, 1974; Comninos, 1977; Rock et al., 1984), particularly as none of the authors of these trials responded to requests for additional information. It is difficult to obtain reliable data from the older trials which tend to be the poorest quality trials and the least likely to be clarified in correspondence owing to the corresponding authors no longer being contactable (or, if contactable, unlikely to make meaningful comments about research performed, in some cases, >20 years previously). The apparent benefit of antibiotic in hydrotubation fluid was based on trial data from a trial of poor quality (Soihet, 1974). In particular, the group of women receiving hydrotubation fluid containing gentamicin had also undergone tubal suspension as part of the additional surgery (an additional intervention not performed in the control group). The trial author emphasized the importance of avoiding infection for the success of tubal surgery (Soihet, 1974). It may simply be that administration of antibiotic versus no antibiotic confers benefit following tubal stent removal 6 weeks after tubal surgery, the route of antibiotic administration (in this case, by hydrotubation) being coincidental. Increased odds of pregnancy and livebirth in late hydrotubation after tubal stent removal where hydrotubation fluid contains antibiotic versus no antibiotic must therefore be interpreted cautiously but is worthy of further evaluation. The use of a tubal stent or prosthesis at the time of tubal surgery was not an intervention under primary consideration in this review. It was considered only because it was a co-intervention with late hydrotubation when comparing early and late hydrotubation. The use of a tubal stent or prosthesis, whether associated with hydrotubation (Soihet, 1974; Comninos, 1977) or not (DeCherney and Kase, 1981), has not been shown to increase pregnancy rates this conclusion is consistent with a systematic review of pelvic surgical techniques (Watson et al., 1999) A future randomized controlled trial examining post-operative hydrotubation for improving fertility following tubal surgery would be justifiable if the hydrotubation fluid was to contain antibiotic and if the trial was adequately powered. Livebirth rates of 23 and 11% occurred with late post-operative hydrotubation with antibiotic and without antibiotic respectively (Soihet, 1974). In order to have 80% power to detect such an increase in the livebirth rate at the 95% confidence level, a trial would need to randomize 338 women undergoing post-operative hydrotubation with antibiotic compared with no hydrotubation. This would be the absolute minimum number to confer sufficient weight to the results and conclusions of such a trial. The total number of women in the two randomized controlled trials examining second-look laparoscopy (Tulandi et al., 1989; Gurgan et al., 1992) was only 56. With underpowered studies, conclusions about the ineffectiveness of these procedures must be cautious this is absence of evidence of benefit rather than evidence of absence of benefit. Much consideration was given to the pooling of data from these two trials in a meta-analysis. The primary surgery was very different (tubal surgery and laser photocoagulation of polycystic ovaries) and the participants differed in that the primary surgery had already failed to reverse infertility in those undergoing tubal surgery (not the case for the women undergoing ovarian photocoagulation), in addition to the difference in the timing of the second-look laparoscopy. The decision to pool the data was based on the fact that both trials used the single intervention of secondlook laparoscopy with adhesiolysis, the absence of evidence with respect to early and late second-look laparoscopic division of adhesions that one offers benefit over the other and the fact that results of the trials are very similar. The most appropriate time to perform post-operative pelvic adhesiolysis is uncertain and has varied from 8 days to 2 years after the primary surgery (Raj and Hulka, 1982; Surrey and Friedman, 1982; DeCherney and Mezer, 1984; Trimbos-Kemper et al., 1985; Gomel et al., 1986). It is very clear that, even with the most scrupulous of surgical techniques, adhesions do often form after pelvic reproductive surgery and that early second-look laparoscopy with adhesiolysis reduces the extent of such adhesions at a later date (Raj and Hulka, 1982; Trimbos-Kemper et al., 1985; Jansen, 1988). It has been suggested that the ideal time for early post-operative adhesiolysis would be between the time of serosal healing (8 days) and established adhesion fibrosis (21 days) (Jansen, 1988), whilst others have suggested these early fine fibrinous adhesions are a normal consequence of tissue repair and may eventually disappear in time due to local release of breakdown mediators in the remodelling process (Gillett et al., 1994). In the randomized controlled trials of this review, early second-look laparoscopy with adhesiolysis was undertaken after 3 4 weeks (Gurgan et al., 1992). Late second-look laparoscopy has additional diagnostic value in the rational planning of future fertility treatment, for example, accelerating referral for IVF for women deemed unlikely to conceive based on the laparoscopic appearance of the pelvis. Late second-look laparoscopy was performed 1 year after the primary surgery in the randomized controlled trials in this review (Tulandi et al., 1989). It is also interesting to speculate, but premature to conclude, that that the lysis of adhesions due to surgery gives a different outcome from the lysis of non-surgical adhesions, the latter being concluded as beneficial based on non-randomized studies (Reich, 1987; Tulandi et al., 1990). A future randomized controlled trial examining second-look laparoscopy with adhesiolysis following pelvic reproductive surgery would be justifiable if it were of adequate power and if able to randomize to early (8 21 days post-operatively), late or no

9 Cochrane review: improving fertility 267 second-look laparoscopy with adhesiolysis. The overall pregnancy rates (including second-look laparoscopy and no second-look laparoscopy) were 45% following laparoscopic laser ovarian photocoagulation (with 40% livebirth rate) (Gurgan et al., 1992), 22% following salpingostomy (Tulandi et al., 1989) and 61% following salpingo-ovariolysis (Tulandi et al., 1989). An increase in the total pregnancy rate of 25% (risk ratio 1.25) would be clinically important after any of these procedures, giving pregnancy rates of 56, 28 and 76% respectively. In order to have 80% power to detect such increases in the total pregnancy rates at the 95% confidence level, a trial would need to randomize 652 women following laparoscopic laser ovarian photocoagulation, 2002 following salpingostomy and 414 women following salpingo-ovariolysis to second-look laparoscopy with adhesiolysis or no second-look laparoscopy. Randomized controlled trials with the three groups suggested above (early versus late second-look laparoscopy with adhesiolysis versus no secondlook laparoscopy) could increase the power of the study. Conclusions There is insufficient evidence to support the routine practice of hydrotubation or second-look laparoscopy following female pelvic reproductive surgery. The studies on which this conclusion is based were either poor quality or underpowered. These interventions should be performed in the context of a good quality, adequately powered randomized controlled trial. Post-operative hydrotubation with fluid containing antibiotic following tubal surgery may offer benefit over hydrotubation fluid without antibiotic. A randomized controlled trial of post-operative hydrotubation with antibioticcontaining fluid versus no hydrotubation for improving fertility following tubal surgery is justified. Acknowledgements The authors acknowledge the helpful comments of those who have proofed this review and the authors of included trials who supplied additional information or data. We wish to thank Sarah Hetrick and Cindy Farquhar (Cochrane Menstrual Disorders and Subfertility Group Co-ordinator and Co-ordinating Editor respectively); Sue Furness for assistance in identifying trials; Anne Lethaby for advice on statistics; Patrick Vandekerckhove and Richard Lilford for their input in the initial draft of the protocol for the review. References Arronet, G.H., Eduljee, S.Y. and O Brien, J.R. (1969) A nine-year survey of Fallopian tube dysfunction in human infertility: diagnosis and therapy. Fertil. Steril., 20, Cochrane Menstrual Disorder and Subfertility Group (1999). Collaborative review group statistical guidelines. In The Cochrane Library, Issue 4, Update Software, Oxford, UK Comninos, A.C. (1977) Salpingostomy: results of two different methods of treatment. Fertil. Steril., 28, DeCherney, A.H. and Kase, N. (1981) A comparison of treatment for bilateral fimbrial occlusion. Fertil. Steril., 35, DeCherney, A.H. and Mezer, H.C. (1984) The nature of posttuboplasty pelvic adhesions as determined by early and late laparoscopy. Fertil. Steril., 41, Gillett, W.R., James, C., Jetha, N. et al. (1994) Removal of the ovarian surface epithelium from the rabbit ovary a cause of adhesions following a standard injury. Hum. Reprod., 9, Gomel,V.,Taylor,P.J.,Yuzpe,A.A.et al. (eds) (1986) Surgical endoscopy. In Laparoscopy and Hysteroscopy in Gynecologic Practice. Yearbook Medical Publishers, Chicago, USA, 158 pp. Grant, A. (1971) Infertility surgery of the oviduct. Fertil. Steril., 22, Gurgan, T., Urman, B., Aksu, T. et al. (1992) The effect of short-interval laparoscopic lysis of adhesions on pregnancy rates following Nd-YAG laser photocoagulation of polycystic ovaries. Obstet. Gynecol., 80, Jansen, R.P. (1988) Early laparoscopy after pelvic operations to prevent adhesions: safety and efficacy. Fertil. Steril., 49, Jessen, H. (1971) Forty-five operations for sterility. Acta Obstet. Gynecol. Scand., 50, Luber, K., Beeson, C.C., Kennedy, J.F. et al. (1986) Results of microsurgical treatment of tubal infertility and early second-look laparoscopy in the post-pelvic inflammatory disease patient: implications for in vitro fertilisation. Am. J. Obstet. Gynecol., 154, Novy, M.J. (1995) Tubal surgery or IVF making the best choice in the 1990s. Int. J. Menopausal Stud., 40, Raj, S.G. and Hulka, J.F. (1982) Second-look laparoscopy in infertility surgery: therapeutic and prognostic value. Fertil. Steril., 38, Reich, H. (1987) Laparoscopic treatment of extensive pelvic adhesions, including hydrosalpinges. J. Reprod. Med., 32, Rock, J.A., Katayama, K.P., Martin, E.J. et al. (1978) Factors influencing the success of salpingostomy techniques for distal fimbrial obstruction. Obstet. Gynecol., 52, Rock, J.A., Siegler, A.M., Meisel, M.B. et al. (1984) The efficacy of postoperative hydrotubation: a randomized prospective multicenter clinical trial. Fertil. Steril., 42, Royal College of Obstetricians and Gynaecologists (1998a) The Initial Investigation and Managament of the Infertile Couple. Evidence-based guidelines no. 2. RCOG Press, London, UK. Royal College of Obstetricians and Gynaecologists (1998b) The Management of Infertility in Secondary Care. Evidence-based guidelines no. 2. RCOG Press, London, UK. Soihet, S. (1974) Three comparative techniques on tubo-plasty. Int. J. Fertil., 19, Strandell, A., Bryman, I., Janson, P.O. et al. (1995) Background scoring systems in relation to pregnancy outcome after fertility surgery. Acta Obstet. Gynecol. Scand., 74, Surrey, M.W. and Friedman, S. (1982) Second-look laparoscopy after reconstructive pelvic surgery for infertility. J. Reprod. Med., 27, Trimbos-Kemper, T.C.M., Trimbos, J.B. and van Hall, E.V. (1985) Adhesion formation after tubal surgery: results of the eighth-day laparoscopy in 188 patients. Fertil. Steril., 43, Tulandi, T., Falcone, T. and Kafka, I. (1989) Second-look laparoscopy one year following reproductive surgery. Fertil. Steril., 52, Tulandi, T., Collins, J.A., Burrows, E. et al. (1990) Treatment-dependent and treatment-independent pregnancy among women with periadnexal adhesions. Am.J.Obstet.Gynecol.,162, Vandekerckhove, P., Watson, A., Lilford, R. et al. (1999) Therapeutic effect of oil-soluble and water-soluble media used for tubal patency testing (hysterosalpingography or laparoscopy) on pregnancy rates in infertility patients (Cochrane Review). In The Cochrane Library, Issue 4. Update Software, Oxford, UK. Watson, A., Vandekerckhove, P. and Lilford, R. (1999) Techniques for pelvic surgery in subfertility (Cochrane Review). In The Cochrane Library, Issue 4. Update Software, Oxford, UK. Winston, R.M. and Margara, R.A. (1991) Microsurgical salpingostomy is not an obsolete procedure. Br. J. Obstet. Gynaecol., 98, Received on September 21, 1999; accepted on February 10, 2000