With 1.1 million induced abortions in the United States each year,1 accurate estimates of abortion complications are paramount to assess and improve quality of care and determine how public policies can most effectively safeguard women's health. Although national abortion-related mortality data exist for the United States,2 no surveillance system captures abortion-related morbidity. Studies find varying complication rates3–7 depending on the procedure, weeks of gestation, length of follow-up, and protocols used to detect complications. Furthermore, complication rates are underestimated by low follow-up rates.5,7–9
Published complication rates are considered incomplete because they usually do not include those diagnosed at sites other than the original source of care.10 Limited research focuses on emergency department (ED) use when examining postabortion care. Because the abortion care delivery system is concentrated in urban centers maldistributed across states,1 women often travel to obtain abortion care. Thus, women are likely to seek postabortion care at an ED near their home.
Using state Medicaid data, we examine reasons for postabortion visits, contributing to the literature on ED use among patients with public insurance.11,12 Then, using a standardized methodology for identifying and classifying complications, we estimate the incidence of postabortion complications diagnosed, treated, or diagnosed and treated at all clinical sites to see whether the rate of complications in a closed system with complete follow-up differs from the rates found in other studies.
MATERIALS AND METHODS
We used patient-level billing data from Medi-Cal, California's State Medicaid program. Medi-Cal provides pregnant low-income women immediate, temporary Medicaid coverage. California is one of 17 states that covers abortion and subsequent care for women enrolled in Medicaid. In 2011 an estimated 512 facilities in California performed 181,730 abortions,1 approximately 51% of which were covered by Medi-Cal.13
The study was approved by the institutional review boards of the University of California, San Francisco and the California Health and Human Services Agency.
Medi-Cal is administered on a fee-for-service or managed care arrangement, split approximately in half across the two. Only the fee-for-service billing records contain complete information for care provided to a particular beneficiary; therefore, we received data only for those beneficiaries with fee-for-service coverage. We obtained billing records for every fee-for-service beneficiary who had an abortion in 2009 and 2010 and all billing records for all clinical encounters for each of those beneficiaries up to 6 weeks after the abortion. Each clinical encounter results in multiple (
=12) billing records; thus, we acquired 659,361 records.
Data included an encrypted beneficiary identification number, date of birth, race or ethnicity, city, state, zip code, date(s) of service, type of facility, diagnosis (International Classification of Diseases, 9th Revision [ICD-9] codes), procedure or treatment (Healthcare Common Procedure Coding System, Current Procedural Terminology [CPT] codes), facility type, and amount paid per treatment. For each abortion, we calculated the beneficiary's age and urban or rural residence (determined by zip code).
Abortions were identified using Healthcare Common Procedure Coding System codes (59840–59841, 59850–59852 and 59855–59857, X7724, Z0336); these codes also indicated the abortion type: 1) “medication abortions,” which includes use of mifepristone and misoprostol up to 9 weeks of gestation; 2) “first-trimester aspiration,” which includes both manual and electric aspiration abortions as well as dilation and curettage “in the first 12–14 weeks of gestation”14; and 3) “second-trimester or later procedures.” Included in this latter category are medical and surgical abortions performed using multiple abortion techniques such as dilation and evacuation, with or without osmotic dilators and misoprostol as well as full or partial inductions used “after 12–14 weeks of gestation.”14 The billing data that we had did not allow us to make a determination of weeks of gestation nor the specific technique used.
We searched for additional claims made to Medi-Cal up to 6 weeks after the abortion for each abortion identified. Postabortion ED visits (not including urgent care) within 6 weeks were identified using codes associated with ED use (Healthcare Common Procedure Coding System codes: 99281–99285, and Z7502). Clinically trained reviewers evaluated all available billing data (including all ICD-9 and Healthcare Common Procedure Coding System or CPT procedure codes) for each beneficiary who had an ED visit and assigned the visit to one of three categories: 1) not abortion-related; 2) not abortion-related but for a concurrent problem (diagnosed, treated, or diagnosed and treated at the time of the abortion); and 3) abortion-related. An ED visit was classified as abortion-related based on the constellation of ICD-9 and Healthcare Common Procedure Coding System or CPT procedure codes for that visit. For example, in many cases, it was a combination of an ICD-9 code for an abortion, postabortion complication, or abdominal pain that indicated that the visit was abortion related along with a Healthcare Common Procedure Coding System or CPT procedure code for a pregnancy test, pelvic examination, transvaginal ultrasonography, abdominal ultrasonography, or dose of misoprostol.
Each abortion-related visit was then classified as 1) woman received an abortion-related diagnosis, treatment, or both; or 2) woman presented with abortion-related symptoms such as abdominal pain or cramping but received no pathologic diagnosis or treatment. When ED visits took place for multiple reasons, only the reason most closely related to the abortion was recorded (Fig. 1).
Additionally, we identified all ambulance transfers (Healthcare Common Procedure Coding System codes: X0030, X0034, X0036, X0400, X0402, X0412) and all self-referred ED visits on the day of the abortion regardless of whether the visit resulted in an abortion-related diagnosis or treatment.
The process of identifying and classifying abortion complications involved several steps. We defined a complication as any postabortion adverse event that received an abortion-related diagnosis or treatment at any source of care, including EDs and the original abortion facility within 6 weeks of an abortion procedure. To identify complications, the clinically trained reviewers evaluated all: 1) abortion-related diagnoses and treatments identified through the ED visit analysis described previously (excluding visits having no pathologic diagnosis or treatment); 2) ICD-9 codes that indicate abortion-related complications (635.00–635.82) and Healthcare Common Procedure Coding System or CPT codes for laparoscopy, laparotomy, and hysterectomy surgeries (49000, 49320, 49329, 58150, 58578, 58960), blood transfusions (86970, P9016–P9021, P9048, 390), and antibiotics commonly used to treat abortion-related infections and sepsis at least 1 day after the abortion; and 3) subsequent medication abortions and aspirations within 6 weeks.
The reviewers examined each case identified through this process and applied a systematic classification scheme developed by several of this study's authors and used in a recent study of abortion safety.3 The classification system comprised a list of known abortion complications with standard definitions that included specific criteria (signs, symptoms, laboratory findings) to indicate the complication diagnosis. To validate the system, first, outside experts who work with the U.S. Agency for Health Research & Quality Evidence-Based Practice Centers and from abortion-related research or service delivery reviewed the classification system. Second, a Data and Clinical Safety Monitoring Committee reviewed incident data to further clarify complication definitions and criteria.
For this study, the clinician reviewers categorized each identified case into one of seven diagnoses: incomplete abortion, failed abortion, hemorrhage, infection, uterine perforation, anesthesia-related, and other or undetermined. The clinically trained reviewers examined all available billing data for the beneficiary, including laboratory tests ordered and medications, to validate each diagnosis. For example, to confirm a diagnosis of failed abortion, they checked for additional confirmatory evidence such as codes for aspiration or prenatal care. For diagnoses of hemorrhage, the reviewers looked for treatments such as aspiration, Methergine, or blood transfusion. One diagnosis category was assigned per abortion; when the billing records indicated more than one diagnosis, the highest level diagnosis was selected. Cases identified based on subsequent medication abortion, misoprostol dose, or aspiration within 6 weeks of the initial abortion without any ICD-9 code indicating a complication were categorized as “other or undetermined.” Cases with the ICD-9 code “635.8 Abortion with unspecified complication” were also categorized as other or undetermined. To produce the most conservative estimate, we included undetermined diagnoses in the overall complications estimate.
Each complication was then classified as receiving one of six treatment categories: no treatment, medication (including mifepristone and misoprostol, misoprostol alone, or other medications), uterine aspiration, both medication and aspiration, other treatment, or undetermined. Abortion-related complications were classified as major if they required hospital admission (vendor codes 50, 60), surgery, or blood transfusion with all others classified as minor.
The data analysis was done in several steps. Using Stata 13, first we described the sample characteristics: age, race, residence, abortion procedure type, facility type, and year of abortion. Second, we estimated ED visits on the day of the abortion and within 6 weeks and present reasons for visits. Third, we estimated the abortion-related complication rate by the sample characteristics and the relative risk of a complication adjusted for all other characteristics using a generalized linear mixed model that accounts for lack of independence between multiple abortions by the same woman and those performed by the same health care provider; P values were determined from z tests derived from the model. Women who had missing data for any characteristic were retained in the model. Fourth, we compared major and minor complications by abortion procedure type using Pearson χ2 tests. Finally, we described complication diagnoses by abortion procedure type and treatment. The abortion was the unit of analysis because 8.3% (n=4,165) of women in the data set had more than one abortion. Statistical significance was set at P<.05 for all comparisons; 95% confidence intervals (CIs) are reported.
Among the 659,361 records received, we identified 54,911 abortions among 50,273 fee-for-service Medi-Cal beneficiaries in 2009 and 2010. The largest proportions of women were ages 20–29 years, Hispanic, urban, had a first-trimester aspiration abortion, and were seen at an outpatient clinic (Table 1).
Among all 54,911 abortions, one in 1,036 (0.10%, n=53) were followed by an ED visit on the day of the abortion, including 1 of 5,491 (0.03%, n=15) transferred by ambulance for immediate care, although not all resulted in an abortion-related diagnosis or treatment.
Among all abortions (N=54,911), 1 of 16 (6.4%, n=3,531) was followed by an ED visit within 6 weeks of the abortion. Of these, 49.8% (n=1,758) were unrelated to the abortion, 9.4% (n=332) were conditions unrelated to but concurrent with the abortion, and 40.5% (n=1,431) were abortion-related. Among abortion-related visits, two thirds (66.6%, n=953) were cases in which a patient presented with abortion-related symptoms but did not receive a pathologic diagnosis or treatment. Thus, 1 of 115 (0.87%, n=478) abortions resulted in an ED visit receiving a diagnosis, treatment, or diagnosis and treatment. Among all abortion-related ED visits (n=1,431), 21.8% (n=770) followed a medication abortion, 64.2% (n=2,266) followed a first-trimester aspiration abortion, and 14.0% (n=495) followed a second-trimester or later procedure (Table 2).
Among all abortions (N=54,911), 1,156 (2.1%, 95% CI 1.99–2.23) resulted in an abortion-related complication diagnosed or treated at any source of care, including EDs and the original abortion facility. The unadjusted complication rate was 5.2% (n=588) for medication abortions, 1.3% (n=438) for first-trimester aspiration abortions, and 1.5% (n=130) for second-trimester or later procedures. Adjusted results indicate that women ages 30–39 years were 1.20 (95% CI 1.02–1.40) times as likely to have a complication compared with women ages 20–24 years, and Hispanic women were significantly less likely to have a complication compared with white women. Medication abortions were 5.96 (95% CI 5.11–6.94) times as likely to result in a complication as first-trimester aspiration abortions. Women receiving abortion care at hospitals or physician's offices or groups were significantly more likely to have a complication than women receiving care at outpatient clinics (Table 1).
The rate of major complications among all 54,911 abortions was 0.23% (95% CI 0.19–0.27) (n=126, 1/436), 0.31% (n=35) among women who had medication abortions, 0.16% (n=57) among women who had first-trimester aspiration abortions, and 0.41% (n=34) among women who had second-trimester or later procedures (Table 3). Among all women, 0.20% (n=108) were admitted to hospitals, 0.02% (n=13) had surgery, and 0.09% (n=50) received blood transfusions (data not shown). These three categories are not mutually exclusive; some women were admitted to a hospital and had surgery, received a blood transfusion, or had surgery and a blood transfusion.
The most common complications were other or undetermined diagnoses (1.20%, n=658), comprised mostly of undetermined diagnoses that lead to repeat abortion, and incomplete abortions (0.42%, n=231) (Table 4). The majority of incomplete abortions (85.7%, n=198) and failed abortions (73.3%, n=22) were treated with uterine aspiration. The majority of hemorrhage cases were treated by other treatments (including blood transfusion) (39.5%, n=34) or medication (23.3%, n=20). All anesthesia-related cases were treated with medication as were the majority of infections (72.5%, n=100).
We observed a 2.1% abortion-related complication rate after nearly 55,000 abortions diagnosed or treated at all sources of care. The majority were minor. Rates of transfers to an ED, hospital admissions, surgeries, and blood transfusions were low. The complication rate is much lower than that found during childbirth15 and comparable to that found in the literature3,7 even when ED visits are included and there is no loss to follow-up.
We observed the highest rate of complications among women obtaining medication abortions (5.2%, n=588), the vast majority of which were minor and expected. This rate may be overestimated with aspirations performed presumptively or to alleviate bleeding or cramping symptoms.16 Nevertheless, this rate is consistent with intervention rates found in other studies.17,18
The complication rate for second-trimester or later procedures is lower than other studies.4,19 This may be because the second-trimester or later category includes a large number of abortions performed earlier in the second trimester when complication rates are closer to those in the first trimester. Only 1.4% of all abortions nationally occur after 20 weeks of gestation20; our data should not be used to make determinations about the complication rate among the small number of procedures performed later in pregnancy.
We found a high rate (6.4%) of ED visits after abortion, half of which were not abortion-related. This finding is consistent with previous research and could reflect Medi-Cal beneficiaries' use of the ED as the health care provider of first resort.11 Additionally, the Medi-Cal fee-for-service population has been noted as having greater health risks and more costly use compared with the Medi-Cal managed care population.21
Two thirds of abortion-related ED visits did not result in a diagnosis or treatment, representing visits primarily for symptoms, not complications. Strategies to reduce ED visits include increasing the number and types of Medi-Cal primary care providers, particularly in underserved areas, who can provide abortion care, postabortion care, or both, and improving health care provider–patient communication on nonurgent postprocedure side effects.
This study examines postabortion ED visits and complications up to 6 weeks and across multiple facilities without loss to follow-up, addressing a common methodologic limitation of other studies. In other studies, follow-up periods ranged from 2 to 4 weeks with most considering follow-up a return visit to the original abortion facility. When reported, loss to follow-up varies widely: 9% at 1 week,7 2–34% at 2 weeks,5,8,9 and between 8 and 65% in studies where duration is not specified.22,23 Additionally, this study has the sample size necessary to robustly estimate rare events.
Using billing codes to identify complications has methodologic limitations. Administrative data sets often contain erroneous codes24; however, our clinical reviewers examined all related billing records for patients with complications for inconsistencies and errors. Also, by relying on Medi-Cal codes, we could not assess whether any of the complications lead to deaths or detect complications not documented by billing codes. It is possible that complications seen or treated at the original abortion facility did not result in any Medi-Cal reimbursements, thereby underestimating the complication rate.
We were unable to determine the exact week of gestation of each abortion, which is known to be a strong predictor of complication risk.6 Additionally, repeat abortions performed within 6 weeks were assumed to be for the same pregnancy, but they may have been for new pregnancies. This, too, may lead to overestimating the complication rate, although the proportion of new pregnancies is likely too small to have much effect.
Regarding generalizability to the national abortion population, demographically, our sample had higher rates of Hispanics, lower rates of blacks,25 and presumably higher rates of low-income women. Medi-Cal beneficiaries may have more health problems than the general population21 and given that the sample had insurance coverage, they may differ from women whose follow-up care is self-pay. These differences would mean that the reported complication rate is overestimated.
These new data can inform policy debates regarding abortion regulation in the United States. State legislatures have passed regulations such as ambulatory surgical center requirements (23 states), transfer agreement laws (eight states), and hospital admitting privileges requirements (13 states)26 with the stated intent to increase safety. Given that in practice their ultimate effect often is the closure of abortion facilities,27 there is a need to consider the public health effect of these policies, weighing any theoretical incremental reduction in patient risk that may occur against any increases in risk that may occur with reduced access to abortion care.
1. Jones RK, Jerman J. Abortion incidence and service availability in the United States, 2011. Perspect Sex Reprod Health 2014;46:3–14.
2. Bartlett LA, Berg CJ, Shulman HB, Zane SB, Green CA, Whitehead S, et al.. Risk factors for legal induced abortion-related mortality in the United States. Obstet Gynecol 2004;103:729–37.
3. Weitz TA, Taylor D, Desai S, Upadhyay UD, Waldman J, Battistelli MF, et al.. Safety of aspiration abortion performed by nurse practitioners, certified nurse midwives, and physician assistants under a California legal waiver. Am J Public Health 2013;103:454–61.
4. Grossman D, Blanchard K, Blumenthal P. Complications after second trimester surgical and medical abortion. Reprod Health Matters 2008;16(suppl):173–82.
5. Paul ME, Mitchell CM, Rogers AJ, Fox MC, Lackie EG. Early surgical abortion: Efficacy and safety. Am J Obstet Gynecol 2002;187:407–11.
6. Cates W Jr, Schulz KF, Grimes DA, Tyler CW Jr. The effect of delay and method choice on the risk of abortion morbidity. Fam Plann Perspect 1977;9:266–8, 273.
7. Bennett IM, Baylson M, Kalkstein K, Gillespie G, Bellamy SL, Fleischman J. Early abortion in family medicine: clinical outcomes. Ann Fam Med 2009;7:527–33.
8. Meirik O, My Huong NT, Piaggio G, Bergel E, von Hertzen H; WHO Research Group on Postovulatory Methods of Fertility Regulation. Complications of first-trimester abortion by vacuum aspiration after cervical preparation with and without misoprostol: a multicentre randomised trial. Lancet 2012;379:1817–24.
9. Goldman MB, Occhiuto JS, Peterson LE, Zapka JG, Palmer RH. Physician assistants as providers of surgically induced abortion services. Am J Public Health 2004;94:1352–7.
10. Cree D, Jelsema R. The comparative safety of legal induced abortion and childbirth in the United States. Obstet Gynecol 2012;119:1271.
11. McConville S, Lee H. Emergency department care in California: who uses it and why? San Francisco (CA): Public Policy Institute of California; 2008.
12. Hoot NR, Aronsky D. Systematic review of emergency department crowding: causes, effects, and solutions. Ann Emerg Med 2008;52:126–36.
13. California Department of Health Care Services. Medi-Cal funded induced abortions, 2010. Sacramento (CA): California Department of Health Care Services; 2011.
14. California Department of Health Care Services. Medi-cal provider manual. Sacramento (CA): Department of Health Care Services; 2008.
15. Raymond EG, Grimes DA. The comparative safety of legal induced abortion and childbirth in the United States. Obstet Gynecol 2012;119:215–9.
16. Cleland K, Creinin M, Nucatola D, Nshom M, Trussell J. Significant adverse events and outcomes after medical abortion. Obstet Gynecol 2013;121:166–71.
17. Medical management of first-trimester abortion. Practice Bulletin No. 143. American College of Obstetricians and Gynecologists. Obstet Gynecol 2014;123:676–92.
18. Winikoff B, Dzuba IG, Creinin MD, Crowden WA, Goldberg AB, Gonzales J, et al.. Two distinct oral routes of misoprostol in mifepristone medical abortion: a randomized controlled trial. Obstet Gynecol 2008;112:1303–10.
19. Peterson WF, Berry FN, Grace MR, Gulbranson CL. Second-trimester abortion by dilatation and evacuation: an analysis of 11,747 cases. Obstet Gynecol 1983;62:185–90.
20. Pazol K, Creanga AA, Burley KD, Hayes B, Jamieson DJ; Centers for Disease Control and Prevention (CDC). Abortion surveillance—United States, 2010. MMWR Surveill Summ 2013;62:1–44.
21. Riner RM. Challenging the cost effectiveness of Medi-Cal managed care. West J Emerg Med 2009;10:124-e8.
22. Hakim-Elahi E, Tovell HM, Burnhill MS. Complications of first-trimester abortion: a report of 170,000 cases. Obstet Gynecol 1990;76:129–35.
23. Westfall JM, Sophocles A, Burggraf H, Ellis S. Manual vacuum aspiration for first-trimester abortion. Arch Fam Med 1998;7:559–62.
24. Fjerstad M, Westhoff C, Lifford KL. Immediate complications after medical compared with surgical termination of pregnancy. Obstet Gynecol 2010;115:660.
25. Jones R, Finer LB, Singh S. Characteristics of US abortion patients, 2008. New York (NY): Guttmacher Institute; 2010.
26. Guttmacher Institute. State policies in brief: targeted regulation of abortion providers. New York City (NY): Guttmacher Institute; 2014.
27. Grossman D, Baum S, Fuentes L, White K, Hopkins K, Stevenson A, et al.. Change in abortion services after implementation of a restrictive law in Texas. Contraception 2014;90:496–501.