The methods of the United States Collaborative Review of Sterilization have been described in detail elsewhere. 3,4 Briefly, women aged 18 to 44 years who had tubal sterilization in medical centers in nine cities were enrolled between 1978 and 1987 and followed up periodically; no follow-up took place after 1994. The method of sterilization was determined by the woman and her physician; all women were eligible for enrollment if they chose a method under study at their institution. All enrollees gave written informed consent, and the study protocol was approved by the institutional review board at each center.
Before sterilization, a trained nurse-interviewer obtained a detailed history for each participant. The nurse also recorded details of the sterilization procedure based on direct observation or a record review. Follow-up information was obtained by telephone 1 month after sterilization and annually thereafter. We intended to follow all women for 5 years; women enrolled early in the study were eligible for a final telephone interview 8–14 years after sterilization.
At each follow-up interview we asked whether, since her last interview, she had had a positive pregnancy test or been told by a physician that she was pregnant. If she reported a pregnancy, the nurse obtained detailed information from her about it and, if possible, pertinent medical records. We used the best available information to classify the woman's pregnancy status. If we were uncertain whether the pregnancy was a true sterilization failure or a luteal phase pregnancy (pregnancy conceived before sterilization but identified after sterilization), we classified her as having a luteal phase pregnancy, thus potentially underestimating the risk of true failure of sterilization. Follow-up was discontinued when a pregnancy or any of the following events occurred: repeat tubal sterilization, tubal anastomosis, hysterectomy, or death. We considered each woman to be at risk for pregnancy from sterilization until the date, if known, that one of those events occurred. If the date was unknown, we considered her to be at risk until the midpoint between the last interview before the event and the first interview after it.
We restricted our analysis to women who had bipolar electrocoagulation of both fallopian tubes. Because one objective was to determine whether the procedural effectiveness changed over the study period, we separately estimated failure probabilities of women enrolled in the early and later years of study enrollment. We previously reported that sterilized women were enrolled in 1978 through 1986. 3–6 More specifically, women who had sterilization by bipolar coagulation and all other techniques were enrolled in five separate periods: August 1978 to December 1979 (n for bipolar coagulation = 771), January 1980 to September 1981 (n = 629), December 1981 to December 1982 (n = 262), November 1985 to September 1986 (n = 280), and October 1986 to September 1987 (n = 325). Because no women were enrolled in 1983 and 1984, we compared women enrolled in 1978–1982 (n = 1662) with those enrolled in 1985–1987 (n = 605). Because women enrolled in 1985–1987 were eligible for only 5 years of follow-up, we restricted our analyses of women enrolled in the earlier period to 5 years of follow-up. We used a standard life-table technique to estimate the 5-year cumulative probability of pregnancy and Cox proportional hazards analysis with the SAS statistical package 7 to assess risk factors for failure of bipolar electrocoagulation. The only risk factor related to the extent of tubal coagulation that was collected for both periods of study enrollment was the number of sites on each fallopian tube to which current was applied. For analyses of number of sites coagulated, if the number differed between the right and left fallopian tubes, the lower number was used.
Of the total 10,863 women enrolled in the United States Collaborative Review of Sterilization who were eligible for analyses of pregnancy after tubal sterilization, 178 were excluded. Thirty-four of those women had luteal phase pregnancies and the other 144 were excluded for the following reasons: no follow-up (n = 116), refusal to be interviewed at 1-month follow-up or after prolonged loss to follow-up (n = 20), hysterectomy (n = 4), repeat tubal sterilization (n = 1), or death (n = 3) at 1-month follow-up. Of the remaining 10,685 women, 2267 had bipolar coagulation (Table 1). Most (65.2%) of them were 33 years old or younger, and equal proportions were white non-Hispanic (47.5%) and black non-Hispanic (50.1%). Nearly all the women had coagulation without division of the fallopian tube; almost all those enrolled in 1978–1982 had a procedure with a coagulation waveform as opposed to a cutting or blend waveform. Waveform data were not available for women enrolled in 1985–1987.
Thirty-seven (1.6%) of the 2267 women who had bipolar coagulation became pregnant. Thirty of these had pregnancies (18 ectopic and 12 intrauterine) identified within the 5-year follow-up period used for this analysis; 27 of these occurred among the 1662 women enrolled in 1978–1982 and three occurred among the 605 women enrolled in 1985–1987. The 5-year cumulative probability of pregnancy for women enrolled in 1978–1982 was 19.5 per 1000 procedures (95% CI 12.2, 26.9) (Table 2); the comparable probability for women enrolled in 1985–1987 of 6.3 per 1000 procedures (95% CI 0.0, 13.5) was significantly lower (one-tailed P = .01). Women enrolled in 1985–1987 who had fewer than three sites of coagulation had a probability of failure of 12.9 per 1000 procedures (95% CI 0.0, 38.0), which was similar to that for women enrolled in 1978–1982 who had fewer than three sites coagulated (17.1 per 1000) (P = .76). In contrast, women enrolled in 1985–1987 who had three or more sites coagulated had a very low probability of pregnancy (3.2 per 1000 procedures; 95% CI 0.0, 9.6), which was significantly lower than that for women enrolled in 1978–1982 who had three or more sites coagulated (27.1 per 1000 procedures; 95% CI 9.4, 44.7) (one-tailed P = .01).
We used a multivariate analysis to assess the following potential risk factors for pregnancy after sterilization: study period, age, race-ethnicity, study site, education, marital status, gravidity, history of pelvic inflammatory disease, history of previous abdominal or pelvic surgery, presence of adhesions noted at surgery, and the number of sites of tubal coagulation. Of the 11 factors assessed, only a history of pelvic inflammatory disease was significant in a multivariate context (relative risk = 2.99; 95% CI 1.15, 7.79) (Table 3).
Initially, all laparoscopic tubal sterilizations used unipolar electrocoagulation systems in which both jaws of the grasping forceps served as an active electrode and current flowed through the fallopian tube to a return electrode underneath the patient's buttocks or thigh. These systems were considered highly effective, but frequent reports of thermal bowel injuries and other electrical complications led to the development of bipolar systems for sterilization in the mid-1970s. 8–10 With these systems, one jaw of the grasping forceps serves as the active electrode and the other as the return electrode. Bipolar systems reduced the risk of electrical accidents during sterilization, but they also cause a more discrete injury to the fallopian tube. 11–14
In the mid-1980s, pregnancy rates as high as 4–10% were reported after sterilization with bipolar coagulation instruments 15,16 raising questions regarding the effectiveness of bipolar systems for sterilization. Subsequently, Soderstrom et al 13 and Tucker et al 14 provided evidence that bipolar systems have the potential to be highly effective if an appropriate amount of energy is delivered to the fallopian tube. For example, Soderstrom and colleagues coagulated the fallopian tubes of women who had hysterectomy. All 16 women who had less than 25 W of current applied with a coagulating waveform had incomplete desiccation of the tube on histologic examination; conversely, all 20 women who had more than 25 W of current applied with a cutting waveform had complete desiccation. The cutting waveform produces more power at a given setting than the coagulation waveform. 13,14 Soderstrom et al 13 have stated that bursts of high-peak voltage are produced with the coagulating waveform, which can desiccate the outer layers of the tube too quickly, thereby preventing deep penetration by the electrons delivered. When a cutting mode is used, however, the steady inflow of electrons with no change in peak voltage appears to reach the core of the tube before the surface impedance becomes too great.
Because the extent of desiccation of the fallopian tube is determined by the total amount of energy delivered to it, duration of current delivery and wattage are both important factors to consider. In general, the lower the power setting, the longer that current needs to be applied, 14 but the precise amount of time required for complete desiccation at a given power setting is difficult to determine. Soderstrom et al 13 demonstrated that the visual changes of “blanch, swell, and collapse” are unreliable indicators that enough energy has been delivered to cause complete desiccation. An inline ammeter confirmed, in several test cases, that current continued to flow after the tube appeared on gross inspection to be sufficiently coagulated. Although use of an inline ammeter is helpful in establishing the length of time that current should be applied, no commercially available monitor can establish precisely the amount of energy delivered to the tube. 14
The strengths and weaknesses of the United States Collaborative Review of Sterilization for estimating the effectiveness of tubal sterilization have been discussed in detail elsewhere. 3,4 Although many women (n = 2267) who had bipolar coagulation were studied, the numbers of women in key subgroup analyses were relatively small. In particular, there were only 91 women who had fewer than three coagulation sites in 1985–1987. Thus, there was little power to detect genuine differences; even large differences in the point estimates between women with and without three or more sites of coagulation in 1985–1987 were not statistically significant. Nonetheless, a focus on the larger group (n = 475) of women with three or more sites of coagulation in 1985–1987 indicates that the cumulative probability of pregnancy among these women at 5 years after sterilization is highly likely to be less than 1%.
We found that a history of pelvic inflammatory disease was associated with increased risk of pregnancy after bipolar coagulation. A history of pelvic inflammatory disease might not accurately reflect whether pelvic inflammatory disease actually occurred. Theoretically, pelvic infection could either reduce the risk of pregnancy after bipolar coagulation by decreasing fecundity or increase the risk by making it more difficult to identify, isolate, and adequately coagulate the fallopian tube. The present finding is consistent with that of our previous report 4 in which a history of pelvic inflammatory disease increased the risk of ectopic pregnancy in women in the entire study cohort, ie, when all methods of tubal occlusion were combined.
We postulated that sterilizations done earlier in the study (1978–1982), when bipolar systems were fairly new, might have been less effective than those done later. We found this to be the case in our cohort, but we have no ready explanation of why the difference was so marked between our 1978–1982 and 1985–1987 groups among those with three or more sites of coagulation. Perhaps because clinicians better understood bipolar systems, their efforts to assure adequate coagulation improved over time. One finding indicated that clinicians became more concerned about adequate coagulation, namely, the proportion of women reported to have had three or more sites coagulated increased from 26% in 1978–1982 to 84% in 1985–1987.
Whereas applying current to at least three sites is likely to be a critical determinant of effectiveness, applying insufficient energy to three or more sites will likely lead to inadequate coagulation and reduced effectiveness of the procedure. These findings indicate that bipolar coagulation systems are capable of delivering sufficient energy to desiccate the fallopian tube. The 5-year cumulative probability of pregnancy we found for women with three or more sites of bipolar coagulation in 1985–1987 (3.2 per 1000 procedures; 95% CI, 0.0, 9.6) was similar to the 5-year cumulative probability of pregnancy (2.3 per 1000; 95% CI, 0.0, 4.8) in the United States Collaborative Review of Sterilization for women sterilized with unipolar coagulating systems. 3
1. Peterson LS. Contraceptive use in the United States: 1982–90. Advanced data from vital and health statistics. No. 260. Hyattsville, Maryland: National Center for Health Statistics, 1995. (DHHS publication no. PHS 95-1250).
2. Peterson HB, Hulka JF, Phillips JM, Surrey MW. Laparoscopic sterilization: American Association of Gynecologic Laparoscopists' 1991 Membership Survey. J Reprod Med 1993;38:572–3.
3. Peterson HB, Xia Z, Hughes JM, Wilcox LS, Tylor LR, Trussell J, for the US Collaborative Review of Sterilization Working Group. The risk of pregnancy after tubal sterilization: Findings from the U.S. Collaborative Review of Sterilization. Am J Obstet Gynecol 1996;174:1161–70.
4. Peterson HB, Xia Z, Hughes JM, Wilcox LS, Tylor LR, Trussell J for the U.S. Collaborative Review of Sterilization Working Group. The risk of ectopic pregnancy after tubal sterilization. N Engl J Med 1997;336:762–7.
5. Hillis SD, Marchbanks PA, Tylor LR, Peterson HB for the U.S. Collaborative Review of Sterilization Working Group. Tubal sterilization and long-term risk of hysterectomy: Findings from the United States Collaborative Review of Sterilization. Obstet Gynecol 1997;89:609–14.
6. Hillis SD, Marchbanks PA, Tylor LR, Peterson HB for the U.S. Collaborative Review of Sterilization Working Group. Higher hysterectomy risk for sterilized than nonsterilized women: Findings from the United States Collaborative Review of Sterilization. Obstet Gynecol 1998;91:241–6.
7. SAS Institute, SAS/STAT Software: Changes and enhancements. Technical report P/229. Cary, North Carolina: SAS Institute, 1992.
8. Corson SL, Patrick H, Hamilton T, Bolognese RJ. Electrical considerations of laparoscopic sterilization. J Reprod Med 1973;11:159–64.
9. Rioux JE, Cloutier D. A new bipolar instrument for laparoscopic tubal sterilization. Am J Obstet Gynecol 1974;119:737–9.
10. Kleppinger RK. Ancillary uses of bipolar forceps. J Reprod Med 1977;18:254–6.
11. Soderstrom RM. Sterilization failures and their causes. Am J Obstet Gynecol 1985;152:395–403.
12. Soderstrom RM, Levy BS. Bipolar systems—do they perform? Obstet Gynecol 1987;69:425–6.
13. Soderstrom RM, Levy BS, Engel T. Reducing bipolar sterilization failures. Obstet Gynecol 1989;74:60–3.
14. Tucker RD, Benda JA, Mardan A, Engel T. The interaction of electrosurgical bipolar forceps and generators on an animal model of fallopian tube sterilization. Am J Obstet Gynecol 1991;165:443–9.
15. Gunston KD, van Coeverden de Groot HA, Goldberg GL, Bromhall MR. Macroscopic and microscopic changes in the fallopian tube after bipolar cauterization. S Afr Med J 1983;63:518–9.
16. Ayers JWT, Johnson RS, Ansbacher R, Menon M, LaFerla JJ, Roberts JA. Sterilization failures with bipolar tubal cautery. Fertil Steril 1984;42:526–30.