Racial/Ethnic Differences in the Risk of Surgical Complications and Posthysterectomy Hospitalization among Women Undergoing Hysterectomy for Benign Conditions.

A total of 214,935 women ≥18 years old in the Florida and New York SID and SASD were coded for hysterectomy ≤1 day of admission from January 2011-September 2014. The analytic cohort included 183,697 women who underwent hysterectomy for benign indications. This was done after excluding 1,681 women with a delivery complication or history of delivery outcome and 29,557 women with a diagnosis of gynecological, colon/abdominal, bladder/kidney, or metastatic cancer, ( Supplemental Figure 1 ). Patient and hospital characteristics are presented in Table 1 . The majority of the women who underwent hysterectomy for benign conditions were White (58%), followed by Black (19%), Hispanic (16%), and Asian/PI (2%) women. Fifty-eight percent of the population were from Florida and 42% from NY. The majority of women underwent hysterectomy during an inpatient hospitalization (Black, 90%; Asian/PI, 86%; Hispanic, 84%; White, 83%). More Asian/PI (85%), Hispanic (85%), and Black (79%) women resided in a large metropolitan area than White women (52%), and more Asian/PI (75%) and Black women (65%) underwent hysterectomy at a teaching hospital than Hispanic (56%) and White women (55%). More Black (51%), Asian/PI (45%), and Hispanic women (34%) underwent abdominal hysterectomy than White women (24%). Compared to White women, Black, Hispanic, and Asian/PI race/ethnicity were associated with significantly increased risk of several comorbidities. These included chronic blood loss/deficiency anemias (Black, OR, 3.27, 95% CI, 3.12–3.43; Asian/PI, OR, 1.95, 95% CI, 1.79–2.13; Hispanic, OR, 1.78, 95% CI, 1.67–1.89), hypertension (Black, OR, 1.83, 95% CI, 1.75–1.91), obesity (Black, OR, 1.74, 95% CI, 1.65–1.84), diabetes (Black, OR, 1.69, 95% CI, 1.61–1.78; Asian/PI, OR, 1.61, 95% CI, 1.41–1.83; Hispanic, OR, 1.50, 95% CI, 1.41–1.60), and adhesions (Black, OR, 1.42, 95% CI, 1.36–1.48; Hispanic, OR, 1.27, 95% CI, 1.19–1.34; Asian/PI, OR, 1.22, 95% CI, 1.10–1.35). The incidence of 30- and 90-day surgical complications is shown in Figure 1 . For 30-day complications, intra-operative injury to abdominal/pelvic organs was most common (2.3%), followed by digestive system complications (1.5%), hemorrhage/hematoma (1.4%), surgical site infection (1.2%), and urologic complications (1.1%). The incidence of 90-day complications was similar. The incidence of 30-day post-hysterectomy hospitalization was 3.7% and 90-day hospitalization was 4.9% ( Figure 1 ). Post-hysterectomy hospitalization was highest following abdominal (30-day, 4.5%; 90-day 5.7%) and lowest following vaginal hysterectomy (30-day, 3.3%; 90-day, 4.5%).The laparoscopic 30-day hospitalization rate was 3.4% (90-day, 4.5%). Among women with a post-hysterectomy hospitalization, the most common indications for 30-day post-hysterectomy hospitalization were surgical site infection (24%), digestive system complications (8%), and hemorrhage/hematoma (7%). Reasons for post-hysterectomy hospitalization were similar at 90 days ( Table 2 ). The incidence of 30-day surgical complications and post-hysterectomy hospitalization by surgical approach and race/ethnicity are presented in Figure 2 , with 90-day outcomes in Supplemental Figure 2 . Compared to other groups, Asian/PI women had the highest incidence of intra-operative injury to abdominal/pelvic organs identified within 30 days following abdominal (4.0%), vaginal (3.7%), and laparoscopic procedures (3.1%). Compared to other groups, Black women had the highest incidence of 30-day post-hysterectomy hospitalization (abdominal, 5.2%; laparoscopic 4.2%; vaginal, 4.0%), digestive system complications (abdominal, 4.9%; laparoscopic, 1.2%; vaginal, 0.9%), and surgical site infection (abdominal, 2.4%; laparoscopic, 1.2%, vaginal, 1.1%). The incidence of all complications and post-hysterectomy hospitalization at 90 days was similar. In multivariable analyses ( Table 3 , see 90-day and complete results in Supplemental Table 2 ) Black race was independently associated with increased risk of 30-day digestive system complications (multivariable-adjusted odds ratio, aOR, 1.98, 95% CI, 1.78–2.21), surgical site infection (aOR, 1.34, 95% CI, 1.18–1.53), post-hysterectomy hospitalization (Black, aOR, 1.31, 95% CI, 1.22–1.40), and urologic complications (aOR 1.16, 95% CI, 1.01–1.34) compared to White. Black race was not associated with 30-day risk of hemorrhage/hematoma. Asian/PI race was independently associated with increased risk of 30-day urologic complications (aOR, 1.48, 95% CI, 1.08–2.03), intra-operative injury to abdominal/pelvic organs (aOR, 1.46, 95% CI, 1.23–1.75), and hemorrhage/hematoma (aOR, 1.33, 95% CI, 1.06–1.67). Hispanic ethnicity was independently associated with significantly increased risk of 30-day post-hysterectomy hospitalization (aOR, 1.11, 95% CI, 1.02–1.20). In the same multivariable analyses, multiple comorbidities/pre-existing conditions were found to be associated with significantly increased risk of 30-day surgical complications and post-hysterectomy hospitalization ( Supplemental Table 2 ). Fluid and electrolyte disorders were associated with significantly increased risk of all outcomes, ranging from 3.49-fold for digestive system complications to 1.53-fold for post-hysterectomy hospitalization. Adhesions were also associated with significantly increased risk of all outcomes. They ranged from 2.18-fold for intra-operative injury to abdominal/pelvic organs to 1.22-fold for post-hysterectomy hospitalization. Chronic blood loss/deficiency anemias were associated with significantly increased risk of 1.27-fold for hemorrhage/hematoma, 1.22-fold for digestive system complications, and 1.09-fold for post-hysterectomy hospitalization. Coagulopathy was associated with significantly increased risk of 3.96-fold for hemorrhage/hematoma, 1.65-fold for intra-operative injury to abdominal/pelvic organs, 1.62-fold for digestive system complications, and 1.27-fold for post-hysterectomy hospitalization. We found that chronic pulmonary disease was associated with significantly increased risk of 1.37-fold for digestive system complications, 1.33-fold for surgical site infection, and 1.22-fold for post-hysterectomy hospitalization. Obesity was associated with significantly increased risk of 1.49-fold for surgical site infection and 1.11-fold for post-hysterectomy hospitalization. Diabetes (1.16-fold) and hypertension (1.07-fold) were also associated with significantly increased risk of post-hysterectomy hospitalization. Surgical approach and concomitant procedures were associated with increased risk of 30-day surgical complications and post-hysterectomy hospitalization ( Supplemental Table 2 ), consistent with previous reports [ 2 ]. All results were similar at 90 days. The results of the sensitivity analyses focused on the most important surgical route-complication combinations from a clinical perspective are presented in Supplemental Table 3 .

We performed a retrospective cohort study using hospital discharge data from the Agency for Healthcare Research and Quality Healthcare Cost and Utilization Project (HCUP) State Inpatient (SID) and State Ambulatory Surgery Databases (SASD). The SID (inpatient stays) and SASD (ambulatory/observation stays) are state-specific files that contain all inpatient and ambulatory surgery billing records from acute-care, non-federal community hospitals [ 14 ]. These databases contain encrypted person identifiers to track patients over time within a state. Because the SID and SASD allow follow up of all individuals who seek care at any acute-care, non-federal hospital within a state, information for a patient who received follow-up care at another in-state hospital other than where their surgery was performed is captured in the data. Individuals could not be identified in the dataset, therefore, the Human Research Protection Office at Washington University considered this study exempt from institutional review. Women aged ≥18 years were included who underwent hysterectomy for benign gynecologic conditions from 2011 through the third quarter of 2014 from the Florida and New York SID and SASD. Data from Florida and New York were used because they had race/ethnicity information, SID and SASD data, and encrypted person identifiers to follow patients over time, and are geographically large and racially and ethnically diverse. Only hysterectomies performed ≤1 day of admission were included, to avoid procedures related to an adverse event during the hospitalization [ 12 ]. Only encounters in which the hysterectomy procedure was performed in the person’s home state were included to avoid missing outcomes for women who returned to their home state for follow-up care. We excluded hysterectomies performed for a complication of delivery, as previously described [ 12 ]. In addition, to maintain our focus on hysterectomies for benign conditions, we excluded encounters coded for uterine, cervical, ovarian, colon/abdominal, bladder/kidney, or metastatic cancer, as previously described [ 12 ]. However, a woman might have been identified with a non-abdominal malignancy, or metastatic cancer for other reasons (e.g., breast cancer). Hysterectomy procedures were classified as either abdominal, vaginal, or laparoscopic (including robotic), using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) procedure codes in the SID or Current Procedural Terminology (CPT) codes in the SASD ( Supplemental Table 1 ) [ 12 ]. The vast majority of women with a diagnosis of conversion to open surgery had a procedure code for abdominal surgery (n=3,893, 93.9%). We defined a procedure as inpatient if the length of stay (LOS) was ≥1 day, and outpatient if the LOS <1 day. If more than one procedure date for hysterectomy was recorded in a record, we used the first procedure date. Our primary exposure was race/ethnicity and categorized as White, Black, Hispanic, Asian/Pacific Islander (PI), Native American (NA)/Other, and missing. It was collected through patient self-report or hospital staff observation. When race and ethnicity were reported as separate data elements, ethnicity took precedence over race [ 15 ]. Five different surgical complications/categories within 30 and 90 days after hysterectomy were identified from the SID and SASD using ICD-9-CM diagnosis and procedure codes ( Supplemental Table 1 ). We chose to focus on racial/ethnic differences in potentially avoidable surgical complications with incidence of 1% or greater that have significant implications for postoperative morbidity, quality of life, and health care expenditures. They included digestive system complications (including paralytic ileus), hemorrhage/hematoma, surgical site infection, intra-operative injury to abdominal/pelvic organs (including accidental laceration, repair of ureter, suture of laceration of bladder, closure of cystostomy), and urologic complications (including ureteric stricture, ureteric obstruction, hydroureter, ureteral fistula, unspecified disorder of kidney and ureter, urethral stricture, urethral fistula, and urinary obstruction). All complications were identified during the hysterectomy admission and in readmissions (SID) or revisit encounters (SASD) through 30 and 90 days after surgery, with the exception of hemorrhage/hematoma which was only identified through 30 days. Subsequent surgical procedures within 90 days after hysterectomy were identified using ICD-9-CM and CPT-4 procedure codes as defined by the Centers for Disease Control and Prevention National Healthcare Safety Network [ 16 ].They were used to censor observations for complications and hospitalizations attributable to subsequent surgeries. Post-hysterectomy inpatient hospitalizations were identified in the SID from index hysterectomy hospital discharge through 30 and 90 days. Hospitalizations overlapping the hysterectomy admissions or hospitalizations ≤1 day of the hysterectomy discharge in which the discharge status was transfer to an acute care hospital were collapsed into a single admission (hysterectomy hospitalization). This was done because these post-operative admissions would be considered transfers from the acute care surgical admission to another hospital. To maintain our focus on non-elective hospitalizations, we excluded the most common elective post-hysterectomy hospitalizations from analysis (encounter for chemotherapy and rehabilitation procedures). Among women with a post-hysterectomy hospitalization, we identified the primary reason for post-hysterectomy hospitalization using the primary ICD-9-CM diagnosis code. Age, race/ethnicity, year of surgery, insurance status, residence status, median household income for zip code, as defined by HCUP, and ICD-9-CM diagnoses codes for benign indications for hysterectomy were identified using information from the index surgical encounter [ 12 ]. We included concomitant gynecological procedures performed at time of hysterectomy, defined by ICD-9-CM and CPT-4 procedure codes, as potential operative risk factors ( Supplemental Table 1 ). Women could have >1 indication for surgery and/or concurrent procedure [ 12 ]. Comorbidities (defined using the Elixhauser classification [ 17 ] and comorbidity index [ 18 ]), previous abdominopelvic surgery, adhesions, and tobacco use were identified during the hysterectomy encounter and hospitalization(s) in the year prior to hysterectomy [ 12 ]. Hospital identifiers in the SID and SASD were linked to American Hospital Association Annual Survey data (American Hospital Association, Chicago, IL) to identify hospital location (rural, urban), teaching status (association with a medical school), and hospital bed-size (small ≤300, medium 300–600, large ≥600). Total annual hospital hysterectomy procedure volume was based on all hysterectomies performed for any indication, and was calculated using the unique hospital identifier in the SID and SASD databases. Hysterectomy volume was categorized as low 1–50, medium 51–200, high ≥201 [ 12 ]. Hospitals were also stratified into quintiles based on the proportion of Black hospitalizations among all hospitalizations at the individual hospitals, ranging from quintile 1 (0–2% Black) to quintile 5 (>15% Black) [ 12 ]. We created categories for missing information for race/ethnicity (0.63%) and hospital-level variables (0.52%). We considered a priori all variables for inclusion in our analyses based on clinical significance, including those known to be independent risk factors; following that, we considered their statistical significance. Bivariate logistic regression was conducted to assess differences in patient, hospital, and surgical characteristics between White and other racial/ethnic groups. Multivariable analyses were performed for the association between race/ethnicity and each of the outcomes using SAS PROC GENMOD with binomial distribution and logit link function. We used the repeated statement for the hospital identifier to account for clustering of procedures within hospitals. All statistically significant variables in bivariate analysis were included in multivariable models. To create parsimonious models and avoid overfitting, we only retained variables that remained significant after controlling for other factors in the final models. Although we tested the interaction between race/ethnicity and surgical approach, it was not statistically significant and therefore not included in the final models. Further, we conducted five sensitivity analyses restricting to a single hysterectomy route and the most important surgical route-complication combinations from a clinical perspective. All statistical analyses were performed with SAS v.9.4 (SAS Institute, Inc., Cary, NC). All tests were two-sided, with statistical significance set at α= 0.05.

This state-based analysis of women who underwent inpatient or outpatient hysterectomy for benign conditions found Black women to have higher risk of 30- and 90-day digestive system complications, urologic complications, surgical site infection, and post-hysterectomy hospitalization than White women. Asian/PI women experienced higher risk of 30- and 90-day urologic complications, intra-operative injury to abdominal/pelvic organs, and hemorrhage/hematoma. Hispanic women were at higher risk of post-hysterectomy hospitalization than White women. Our finding of higher risk of complications among non-White women after hysterectomy are consistent with prior studies [ 9 – 11 , 19 – 21 ]. Our finding of post-hysterectomy hospitalization in 3.7% of women within 30 days of hysterectomy is also consistent with previous studies [ 4 , 8 ]. Black women had the highest 30- and 90-day post-hysterectomy hospitalization incidence. Black race and Hispanic ethnicity were independently associated with post-hysterectomy hospitalization, corroborated by previous studies [ 4 , 9 ]. Likewise, surgical site infection was the most commonly coded complication of post-hysterectomy hospitalization [ 4 , 8 ]. We examined surgical complications individually rather than in composite, and thus are able to provide more information than previous studies about the association between race/ethnicity and individual surgical complications and other risk factors for these adverse outcomes [ 9 , 10 ]. Several comorbidities/pre-existing conditions contributed to increased risk of surgical complications. These included obesity, diabetes, blood loss/deficiency anemias, and adhesions. Black women disproportionately have pre-existing conditions associated with increased risk for adverse outcomes. Consistent with previous studies [ 22 – 24 ], Black women in our study had a higher prevalence of obesity and diabetes. Both conditions were associated with significantly increased risk of surgical site infection and post-hysterectomy hospitalization. Black women have larger uterine weight on average which may make them more likely to have higher blood loss during surgery [ 25 , 26 ], and have a higher prevalence of blood loss/deficiency anemias [ 27 ]. Black women may be at higher risk of adhesions due to a higher prevalence of pelvic inflammatory disease [ 28 , 29 ], and Black and Asian/PI women are at higher risk of developing keloids, which can lead to denser intra-abdominal adhesions [ 30 ]. In this study, adhesions were associated with increased risk of all outcomes, including ≥2-fold increased risk of intra-operative abdominal/pelvic organ injury and urologic complications. Serious complications can result from intra-operative injury to abdominal/pelvic organs and urinary tract injuries (e.g., infection, fistula formation, renal failure [ 31 , 32 ]), including higher risk for urologic complications and post-hysterectomy hospitalization, as we demonstrated. This is important for Asian/PI women, who were at increased risk of intra-operative injury to abdominal/pelvic organs and urologic complications. Multiple socioenvironmental factors contribute to racial/ethnic disparities in pre-existing conditions, including access to prevention and health care services, other policy-related factors, health behaviors (e.g., smoking, nutrition, physical activity), and differential environmental exposures [ 33 ]. Medical optimization of pre-existing conditions could potentially reduce surgical complications and post-hysterectomy hospitalization. For example, blood loss/deficiency anemias can be managed pre-operatively with iron therapy, pharmacologic agents, or transfusion [ 34 , 35 ], and diabetes can be managed pre- and peri-operatively with glucose control [ 36 ]. To ensure that patients are medically optimized for surgery, surgeons should involve a patient’s primary care physician and specialist (e.g., endocrinologist for diabetes) as needed and potentially delay elective and non-urgent/emergent surgeries until patients are medically optimized. Intervention strategies targeted toward identifying and better managing disparities in pre-existing conditions have the potential to reduce differences in adverse outcomes, and, in the end, lower health care expenditures [ 7 ]. In addition, compared to abdominal surgery, MIGS is associated with fewer surgical complications [ 2 , 3 , 37 , 38 ] and post-hysterectomy hospitalization [ 4 , 5 ] and should be performed when feasible, especially in Black women, a population experiencing more surgical complications after open procedures. Our use of all-payer state databases, including inpatient and outpatient hysterectomies, allowed us to evaluate race/ethnicity as an independent predictor of surgical complications and post-hysterectomy hospitalization. By examining surgical complications individually, we were able to provide more information about race/ethnicity and other factors associated with 30- and 90-day adverse outcomes after hysterectomy than previous studies [ 9 , 10 , 39 , 40 ]. We controlled for several risk factors for surgical complications common among Black women (e.g., abdominal hysterectomy [ 9 , 10 , 12 ], obesity [ 22 ], diabetes [ 23 , 24 ], hypertension [ 41 ], chronic renal failure [ 41 ]), and included previously unaccounted for hospital factors. One limitation of our study includes lack of sensitivity of coding for complications and underlying conditions due to relying on data derived for billing purposes. In addition, we only had one year of prior hospital data to identify underlying risk factors, thus residual confounding may account for some of the elevated risk of surgical complications and post-hysterectomy hospitalization associated with Black, Asian/PI, and Hispanic race/ethnicity. For example, obesity is undercoded in administrative data [ 42 ], and is both more prevalent among Black women [ 22 ] and associated with surgical site infection [ 43 ]. Furthermore, we could not control for some clinical (uterine size/weight [ 44 ]) and surgeon characteristics (e.g., training [ 45 ], volume [ 46 – 48 ]) known to be associated with surgical complications. We found that Black and Asian/PI women had higher risk of some 30- and 90-day surgical complications after hysterectomy for benign conditions, and that Black and Hispanic women had higher risk of post-hysterectomy hospitalization. Identifying and developing better strategies to manage disparities in pre-existing conditions/comorbidities could potentially reduce racial/ethnic differences in adverse outcomes after hysterectomy and arguably lower rather than raise healthcare costs. Future studies should examine racial/ethnic differences in outcomes using more recent data and from more states, to determine whether differences in outcomes by race/ethnicity are narrowing or widening over time.

Hysterectomy is the most commonly performed non-obstetrical surgical procedure in women, with almost 400,000 procedures performed for benign gynecologic conditions annually [ 1 ]. Compared to minimally invasive gynecologic surgery (MIGS; vaginal, laparoscopic, and robot-assisted), abdominal hysterectomy is associated with poorer outcomes, including longer length of hospital stay [ 2 ], surgical complications [ 2 , 3 ], and hospital readmissions [ 4 , 5 ]. Because adverse outcomes may be avoidable and are major drivers of health care costs [ 6 , 7 ], they are important measures of healthcare quality [ 5 , 8 ]. Studies, including our prior work, have reported that non-White women, and Black women in particular, are less likely to undergo MIGS [ 5 , 9 – 12 ] and have worse post-operative outcomes, even after controlling for surgical approach [ 5 , 9 – 11 ]. It is possible, however, that reported differences in outcomes can be explained, at least partially, by pre-existing conditions not controlled for in prior studies and hospital factors not accounted for. In addition, analyses based on composite medical and surgical complication variables fail to provide information about significant risk factors by complication type [ 9 , 10 , 13 ]. For these reasons, it remains unclear whether race/ethnicity independently predicts poorer outcomes. The objectives of this study are to 1) determine whether surgical complication and post-hysterectomy hospitalization rates within 30 and 90 days of hysterectomy for benign conditions differ by race/ethnicity and 2) whether differences in outcomes remain after controlling for patient, hospital, and surgical characteristics, including procedure type/route of surgery.