Goldberg, Alisa B. MD, MPH*†; Dean, Gillian MD‡; Kang, Mi-Suk MPH‡; Youssof, Sarah‡; Darney, Philip D. MD, MSc‡
Approximately 1.31 million abortions were performed in the United States in the year 2000.1 Of these procedures, 88% occurred in the first trimester, and 97% used suction curettage.2 Although the majority were done with an electric suction device, there is increasing interest among abortion providers in using the manual vacuum aspirator for early first-trimester elective abortion procedures. (Manual vacuum aspiration technique draws interest. Contracept Technol Update 1999;20:8–9.)
The manual vacuum aspirator is a hand-operated, 60-mL syringe in which vacuum is produced by sealing one end of the syringe and retracting a plunger at the other. The syringe is connected to either a rigid or flexible cannula for uterine aspiration. In contrast to the commonly used electric suction device, the manual vacuum aspirator is more portable, inexpensive, and quiet, and it does not require electricity. These benefits might be most important in settings where access to an electric suction device is limited or difficult. The manual vacuum aspirator also might cause less disruption of evacuated tissue, making identification of the products of conception easier for very early gestations.3
The manual vacuum aspirator has been in use for more than 30 years for the management of incomplete abortions, for endometrial sampling in nonpregnant women, and for elective abortion. Despite extensive use and several large case series that report low failure and complication rates with the device when used for elective abortion,4–6 until recently only the original clinical trial that compared the modified IPAS syringe (IPAS, Chapel Hill, NC) for manual vacuum aspiration with electric suction curettage had compared these methods for abortion.4 In this trial, manual aspiration was associated with a 3% uterine reaspiration rate and electric suction curettage with a 1% reaspiration rate.4
Three randomized controlled trials recently compared these devices. One trial measured patient pain levels and procedure time in 56 women undergoing manual and 52 women undergoing electric suction curettage.7 Another study measured patient and physician acceptability in 41 women undergoing manual and 42 women undergoing electric suction curettage,8 and a third study measured efficacy in 99 women undergoing manual and 98 women undergoing electric suction curettage.9 Neither the efficacy study, which was the largest, nor the other studies found any difference in complete abortion rates or complication rates between the methods. However, because complications of first-trimester abortions are uncommon, none of these trials was sufficiently large to adequately compare complication rates. In this study, we compared immediate complication rates in women undergoing first-trimester abortion at up to 10 weeks’ gestation by manual and electric suction curettage.
MATERIALS AND METHODS
We conducted a retrospective cohort study to compare complication rates for elective abortion procedures performed by manual vacuum aspiration or electric suction. Using current procedural terminology code 59840, we identified all women undergoing dilation and suction curettage at San Francisco General Hospital over a 3.5-year period, between September 1998 and March 2002. We then used a procedural log book in the family-planning unit to confirm this list and to identify women at 10 weeks’ gestation or less whose procedure was performed in the unit, rather than in the main operating room, emergency department, or elsewhere in the hospital. Records were considered unavailable for review if they were not located after requests on three separate occasions. All clinical data were abstracted from medical records onto standardized data-collection forms by one abstractor (SY). This study was approved by the University of California, San Francisco, Committee on Human Research.
Patients were excluded if they were more than 10 weeks’ gestation, if the method of abortion was not recorded, if they had a spontaneous abortion (either threatened, missed, or incomplete), a possible ectopic pregnancy, a suspected molar pregnancy, a failed medical abortion, or if both abortion methods were used. We excluded women whose procedure was completed with both methods, because in most cases we could not determine which method was attempted first. Our primary outcome was the frequency of uterine reaspiration within 6 weeks from the original procedure, as recorded in the medical record. No attempt was made to locate records from locations outside the hospital or to determine whether women had received care elsewhere after their abortion. Secondary outcomes included all hospitalizations, other immediate complications occurring before discharge after the abortion procedure, estimated blood loss, and procedure time. Blood loss was estimated by the operating physician, not directly measured, and in routine cases was usually estimated at between 25 and 50 mL. Procedure time was routinely measured, with a stopwatch, as the time from initiation of cervical dilation until the final instrument was removed from the uterine cavity.
Demographic variables, past medical and reproductive history, and operative details of the abortion procedure were all collected from the medical record. Gestational age was determined either by last menstrual period, confirmed by physical examination or ultrasound. Whenever ultrasound was performed, the gestational age determined by ultrasound was recorded. Most procedures were scheduled for patients at least 6.0 weeks’ gestation; however, a small number of procedures were performed for earlier gestations.
At San Francisco General Hospital, first-trimester abortions are routinely performed with manual dilation with Pratt dilators followed by suction curettage. The manual vacuum aspirator and the electric suction device are readily available in all procedure rooms, and selection of the device is left to physician discretion. All manual vacuum aspiration procedures are performed with the 60-mL syringe with a double-locking valve mechanism, manufactured by IPAS, and electric procedures are performed with the Synevac Vacuum Curettage System 10 (Richmond, CA). Nearly all procedures are performed with a paracervical block with chloroprocaine and moderate sedation with fentanyl and midazolam dosed according to physician discretion. Patients also routinely receive 0.4 mg of intravenous atropine and 2.5–5 units of vasopressin in the paracervical block. Aside from this dose of vasopressin, patients do not routinely receive other uterotonic agents, nor are cervical ripening agents routinely used before abortions at up to 10 weeks’ gestation. All patients receive antibiotic prophylaxis with doxycycline or another agent if they are allergic to tetracyclines. All products of conception are sent to the pathology department for evaluation.
First-trimester procedures are performed either by an attending gynecologist or a third-year obstetrics and gynecology resident under the supervision of an attending physician. A small number of procedures are performed by other residents completing an elective rotation in family planning.
Because ours is a tertiary-care, hospital-based, family-planning unit, no women are referred elsewhere on the basis of medical problems, and only the most severely ill women have their abortion procedures in the main hospital operating room.
Routine postabortion follow-up is not conducted within the family-planning unit but usually occurs either in the general gynecology clinic or at one of many community health centers affiliated with the hospital. A small number of patients follow up with private or health maintenance organization physicians. At the time of this study, outpatient medical records from community health centers were not part of the hospital medical record. In this public health system, nonsurgical management of abortion complications, including antibiotic treatment for endometritis, occurs at multiple locations; however, patients with complications that require surgical management, including uterine reaspiration, or hospitalization are managed at San Francisco General Hospital.
In the original clinical trial that compared the modified IPAS syringe for manual vacuum aspiration with electric suction curettage for abortion, manual aspiration was associated with a 3% uterine reaspiration rate and electric suction curettage with a 1% reaspiration rate.4 Using a two-tailed hypothesis with an μa error of .05 and a power of 80%, we would require a sample size of 1730 women to detect this three-fold difference. We estimated that reviewing approximately 2000 records would provide us with this number of eligible patients.
Data were collected from the medical records, double-entered into an American Standard Code for Information Interchange (ASCII) file, and imported into Stata 7.0 (College Station, TX) for analysis. We used the χ2 and Fisher exact tests to compare binary outcomes and the two-tailed Student t test and two-sample median test to compare continuous outcomes. Logistic regression models were then created to control for possible confounding factors. After initial analysis and review, the records of patients coded as having uterine reaspiration, uterine perforation, hospital admission, hemorrhage requiring transfusion, cervical laceration, and “other” were reviewed again for additional information.
A total of 2298 first-trimester suction curettage procedures were performed in the family-planning unit at San Francisco General Hospital between September 1998 and March 2002. Of the women having these procedures, 86 did not have records available to review, and 2040 met inclusion criteria after chart review (10 weeks or less pregnant and having a primary suction curettage in the family-planning unit). We then excluded 175 women because the method used to perform the abortion procedure was not documented; 119 more were excluded because of predefined criteria, primarily spontaneous abortions and suspected ectopic pregnancies; and 20 more were excluded because both methods were used. We excluded women in whom both methods were used, because in 15 of 20 cases we were unable to determine which method was used first unsuccessfully. In the five cases in which the initial method could be determined, four switches were due to inadequate suction; three switched from manual to electric and one from electric to manual. The remaining case in which both methods were used resulted in uterine perforation and will be discussed later.
We included 1726 women in the analysis: 1002 who underwent manual vacuum aspiration and 724 who underwent electric suction aspiration. This sample size provided us with 80% statistical power to detect a 2% difference in the frequency of uterine reaspiration between the methods, given an α error of .05.
Of all the abortion procedures, 40% (n = 687) were performed by 15 attending gynecologists, 49% (n = 849) were done by 31 gynecology residents, and 11% (n = 188) were completed by 11 other residents.
Women who underwent abortion with the manual vacuum aspirator were demographically similar to women for whom the electric suction device was used, except slightly fewer women in the manual group were married (Table 1). Neither past medical nor reproductive history were associated with the type of method used. Overall, 73% of women reported that they had no medical problems, and the remainder had medical conditions of widely varying severity (Table 1).
Gestational age was not normally distributed: Women in the manual group had an earlier median gestational age (53.0 days versus 59.0 days, P < .001, Table 1). There were 37 procedures performed for women less than 6 weeks’ gestation: Two patients were 4.5 weeks’ gestation and 35 were between 5 and 6 weeks’ gestation. Of these 37 very early cases, 35 were completed with manual vacuum aspiration, and there were no reaspirations. Gestational age was determined by ultrasound in 53% of cases and by last menstrual period confirmed by examination in 47%. Ultrasound dating was more frequently used before manual procedures (57.6% manual versus 47.5% electric, P < .001). Gestational age was also associated with uterine reaspiration but not with other complications. Patients undergoing reaspiration had a median gestational age of 59 days, compared with 55 days among women not requiring reaspiration (P = .03).
Although both the manual vacuum aspirator and the electric suction device can be used with either a flexible or rigid cannula, physicians more frequently used flexible cannulas with the manual device (87% flexible, 12% rigid, 1% both) and rigid cannulas with the electric device (74% rigid, 20% flexible, 6% both, P < .001). Physician level of training was also associated with choice of suction device. Attending physicians chose the manual aspirator for 52% of their patients, whereas gynecology residents chose it for 59%, and other residents chose it for 76% (P < .001).
There were no differences in the amounts of intravenous medications received for pain control and sedation during the procedure (Table 2). Procedure times were also similar in both groups (Table 2). Although blood loss was statistically lower with manual vacuum aspiration, the difference between an estimated blood loss of 35 and 42 mL is not clinically important, and both procedures were associated with very low blood loss (Table 2). Heavy bleeding was uncommon, and there were a similar number of cases in each group in which estimated blood loss was reported as greater than 150 mL (seven women in each group, 0.7% manual versus 1.0% electric, P = .54). No transfusions were required.
Overall, we found no difference in the rate of uterine reaspiration with manual or electric aspiration. There were 22 reaspirations in the manual group and 12 in the electric group (2.2% versus 1.7%, P = .43, Table 3). Of these 34 cases, 21 reaspirations were for postabortal syndrome, four for retained products of conception, three for inability to accurately identify the pregnancy after initial aspiration, and six for unclear final diagnoses (Table 3). Significantly more reaspirations for inability to accurately identify the pregnancy occurred in the electric group.
Similarly, we found no difference in the frequency of other immediate complications. Combining major complications, including uterine reaspiration, uterine perforation, hospital admission, hemorrhage requiring transfusion, and cervical laceration, there were 25 procedures with complications in the manual group and 15 in the electric group (2.5% versus 2.1%, P = .56). Individual complications included in the analysis are listed in Table 4. There were a total of seven cases of uterine perforation in the data set; however, two were excluded from the manual versus electric vacuum analysis because either the original method was not recorded or both methods were used. Of the seven patients with uterine perforation, two underwent additional surgery, and five were managed conservatively. Six were diagnosed when an instrument was felt to go beyond the uterine fundus, and one was diagnosed with possible omental fat in the suction cannula. The patient with suspected bowel injury was admitted to the hospital and underwent exploratory laparotomy; uterine perforation was confirmed, but no bowel injury was found. Another stable patient who underwent manual and then electric aspiration, neither successfully, was identified with a sharp curette to have retained tissue and a uterine perforation. She underwent diagnostic laparoscopy; no intraperitoneal injury was identified, and the perforation site was not bleeding. The abortion was then completed, no further intervention was required, and she was discharged home from the recovery room. The remaining five stable patients were managed conservatively, in most cases with several hours of observation in the family-planning unit, serial hematocrits, and instructions for close follow-up. If a procedure required completion after perforation, it was done under ultrasound guidance. There were four women, two in each group, with complications coded as “other.” Two of these were drug reactions, one was a syncopal episode, and one was a presyncopal episode in a patient with suspected Marfan syndrome.
We did not find a significant association between uterine reaspiration and the level of training of the physician performing the procedure. Attending gynecologists had a uterine reaspiration rate of 1.6%, gynecology residents 2.0%, and other residents 3.2% (P = .38). Neither did we find a significant difference in the overall rate of major complications in procedures performed by attending physicians and residents (attending physicians 1.6%, gynecology residents 2.6%, and other residents 3.2%, P = .39).
Because we found gestational age to be associated with both procedural method and reaspiration, it is a confounding variable. No other predictors of reaspiration or possible confounders were identified. We found level of physician training to be associated with method choice but not with an increased risk of reaspiration or other complications. In contrast, several other studies have shown an association between complication rates and procedures performed by resident physicians.10–12 Because of these earlier findings, we created a logistic regression model to control for gestational duration and level of physician training (attending physician, gynecology resident, other resident). We found no difference in the frequency of uterine reaspiration with manual or electric vacuum aspiration (uterine reaspiration with electric suction, odds ratio [OR] = 0.71, 95% confidence interval [CI] .32, 1.6). Controlling for the same two variables, we found no difference in overall immediate complication rates between the methods (complications with electric suction, OR = 0.81, 95% CI 0.40, 1.7).
Within our sample of 1726 procedures, 172 were repeat elective abortions in women with at least one other abortion in the data set. Because these procedures are not completely independent observations, we evaluated this population separately before deciding to include them in the main data set. Of these repeat procedures, 62.2% were by manual aspiration, and there were four uterine reaspirations (2.3%), similar to the reaspiration rate for first occurrences (2.0%, P = .72). Additionally, 49.8% of all women in our sample report a history of a prior induced abortion, and there was no difference in the distribution of these women between groups (Table 1). For these reasons, we included repeat procedures in the data set for all our analyses.
Our data demonstrate that the rates of uterine reaspiration and of other immediate major complications are similar when early first-trimester abortions are performed with either manual or electric suction curettage. Three smaller, randomized, controlled trials were also unable to identify a difference in complication rates between the devices.7–9 These randomized trials demonstrated similar efficacy,9 patient acceptability, and pain perception7,8 when procedures were performed by manual or electric suction curettage. The trial by Dean et al8 also found both procedures to be equally acceptable to physicians. Together, these data suggest that manual vacuum aspiration and electric suction aspiration are comparable procedures for early first-trimester abortions.
Although we found that manual aspiration was associated with a statistically significant lower estimated blood loss, the difference we identified (7 mL) is probably not clinically important. There were no differences in the number of women with a subjective estimated blood loss greater than 150 mL, and no transfusions were required.
We found an inverse association between level of training and choice of the manual vacuum aspirator. This might reflect previous experience and comfort levels with new technology. Most attending physicians had more extensive previous experience with the electric suction device and thus might have felt more comfortable with it. In contrast, most of the gynecology residents, before their rotation in the family-planning unit, had more previous experience with the manual vacuum aspirator for evacuating incomplete abortions in the emergency department. Given the limited duration of suction generated by the manual vacuum aspirator, there also might be a perception of increased safety of the manual device among residents.
Some investigators have suggested that identification of the gestational sac and chorionic villi is easier after manual vacuum aspiration compared with electric. Our finding of fewer uterine reaspirations in the manual group for inability to accurately identify the pregnancy supports this. This effect might be particularly pronounced in very early (less than 6 weeks) gestations.3,13 Unfortunately, we did not have sufficient numbers of very early gestations to adequately address this question. In our sample, nearly all of the procedures performed at less than 6 weeks’ gestation were by manual vacuum aspiration, demonstrating a physician preference for this method in these early cases.
Although we did not detect a difference in complication rates between the manual and electric vacuum aspiration procedures, our data have several limitations. First, because we did not account for reaspirations that might have occurred at other locations, our reaspiration rate might be falsely low. There is no reason to suspect that if we missed reaspiration cases they were unevenly distributed between the manual and electric groups; however, if the absolute numbers are high, this could affect our results.
Our data on overall complication rates are limited by our retrospective study design and by the fact that we chose to evaluate only uterine reaspiration and other immediate complications about which we were more likely to have complete information. We know that any patient who had a complication occurring in the hospital on the day of her abortion procedure would have documentation of that event in her medical record. Additionally, given the structure of the public health care system in San Francisco, we suspected that most, if not all, women who required a repeat suction curettage shortly after their abortion procedure would return to our institution for uterine reevacuation. In contrast, we were unlikely to have complete information on delayed postabortion complications, including postabortal endometritis, because women receive care for these complications at many locations outside the hospital setting.
Given the large number of abortions occurring in the United States and a relative paucity of abortion providers, many women have limited access to services.1 Although both electric and manual vacuum aspiration can easily be performed in the outpatient setting, many physicians do not have an electric suction machine in their offices. In these cases, physicians either refer patients to abortion clinics or use a hospital operating room, which is much more expensive.14 Whereas purchasing an electric vacuum machine requires a large initial investment, the manual aspiration equipment is inexpensive and might be easily stored in a small office. These factors, together with the simplicity of use, and now proof that manual vacuum aspiration has a safety and efficacy profile similar to that of electric vacuum aspiration, could increase the number of physicians who offer abortion to their patients in an office setting. Early first-trimester medical abortion with mifepristone and misoprostol is also highly effective and easily administered in an office setting. Manual vacuum aspiration might be particularly well suited to serve as the surgical backup method for office medical abortions that fail.15