Objective
We sought to develop and evaluate a risk-adjusted perioperative morbidity model for vaginal hysterectomy.
Study Design
Medical records of women who underwent vaginal hysterectomy during 2004 and 2005 were retrospectively reviewed. Morbidity included hospital readmission, reoperation, and unplanned medical intervention or intensive care unit admission; urinary tract infections were excluded. Multivariate logistic regression identified factors associated with perioperative morbidity (adjusted for urinary tract infection). The resulting model was validated using a random 2006 sample.
Results
Of 712 patients, 139 (19.5%) had morbidity associated with congestive heart failure or prior myocardial infarction, perioperative hemoglobin decrease >3.1 g/dL, preoperative hemoglobin <12.0 g/dL, and prior thrombosis (c-index = 0.68). Predicted morbidity was similar to observed rates in the validation sample.
Conclusion
History of congestive heart failure or myocardial infarction, prior thrombosis, perioperative hemoglobin decrease >3.1 g/dL, or preoperative hemoglobin <12.0 g/dL were associated with increased perioperative complications. Quality improvement efforts should modify these variables to optimize outcomes.
As in other surgical specialties, quality improvement efforts in gynecologic surgery are gaining momentum. Recently, a large national prospective assessment of morbidity after benign hysterectomy was performed in Finland. The overall complication rate was 19%, and this was further stratified among different approaches. Intraoperative complication rates for abdominal, vaginal, and laparoscopic procedures were 2.3%, 3.0%, and 3.2%, respectively. The corresponding postoperative inhospital complication rates were 12.3%, 16.1%, and 8.4%, respectively. Although that study described morbidity from vaginal hysterectomy (VH), there was no description of the impact of concomitant urogynecologic procedures. Complications from urogynecologic surgery range from 20% for minor complications to 0.2% for major complications, but they are often compared with those from abdominal hysterectomy. As such, the baseline morbidity associated with VH and the effect of additional reconstructive surgery have been underinvestigated in the literature.
The relatively elective nature of benign gynecologic surgery has led to a delay in the application of morbidity models. In particular, most women undergoing VH are generally healthy, and surgery frequently is performed because of quality-of-life considerations. Consequently, the surgical risk should be low. In this context, quality improvement processes are imperative for identifying characteristics or comorbid conditions that predispose patients undergoing VH to increased perioperative complications. Morbidity can be decreased if specific modifiable variables are identified that can categorize patients at risk. We hypothesized that specific variables could be identified and that, by creating a risk-adjusted model of morbidity, we could predict patients at risk for adverse outcomes in VH.
Materials and Methods
This study was a planned secondary analysis on the data of women who underwent VH for benign indications, which was a retrospective cohort study approved by our institutional review board. We retrieved medical records of all women who underwent VH for benign indications from January 2004 through December 2005. Patients were included if they underwent VH with or without salpingectomy, oophorectomy, pubovaginal sling placement, or reconstructive pelvic surgery (anterior, posterior, or combined anteroposterior colporrhaphy). Any patient undergoing additional nongynecologic procedures, including laparoscopy, was excluded. Medical records were reviewed for demographics, baseline medical status, perioperative findings, surgical procedures, and complications within 9 weeks after the index surgery. Morbidity outcomes included hospital readmission, reoperation, unplanned intensive care unit admission, or any medical problem requiring intervention (eg, antibiotics for infection, blood transfusion for symptomatic anemia, and diuretics for fluid overload). Because urinary tract infection (UTI) was considered a common and relatively minor complication, UTI and antibiotics given for UTI were not considered among the morbidity outcomes.
Composite medical and surgical diagnoses, including a history of congestive heart failure, myocardial infarction, or a thrombotic event, were identified through the initial preoperative surgical consultation and anesthesia medical examination. Standard preoperative laboratory tests included a complete blood cell count, creatinine, and serum glucose. A complete blood cell count was obtained the day after surgery, allowing for comparison of preoperative and postoperative hemoglobin values to determine the absolute decrease in hemoglobin.
Multivariate logistic regression models were fit, using stepwise and backward selection, to identify factors significantly associated with non-UTI perioperative morbidity after forcing type of surgery into the model (VH with or without salpingectomy/oophorectomy vs more complex surgery). A P value criterion of < .05 was set for variables to be included in the model. Associations were summarized by calculating odds ratio (OR) and corresponding 95% confidence interval (CI) using the parameter estimates from the model. The predictive ability of the final model was summarized using the c-index (“c” for concordance) proposed by Harrell et al. A c-index ranges from 0-1; 1.0 indicates that the variables in the model perfectly separate women with and without perioperative non-UTI morbidity, and 0.5 indicates that the variables contain discriminant information equal to that obtained by chance alone.
The resulting model was validated with data from a computer-generated random sample of 100 women who underwent VH for benign indications at our institution in 2006 and had complete records through the postoperative evaluation. In this way, we calculated expected and observed complication rates for validation. All calculated P values were 2-sided, and P < .05 was considered statistically significant. Analyses were performed using a software package (SAS; SAS Institute Inc., Cary, NC).
Results
Of the 736 women who met the inclusion criteria, 712 had complete records through the postoperative evaluation and were included in the final analysis. Table 1 summarizes patient characteristics. The model was developed using 712 patients, of whom 139 (19.5%) had non-UTI morbidity within 9 weeks after VH. Table 2 summarizes the most important variables predicting morbidity. The following variables were found to be significantly associated with increased morbidity: congestive heart failure, prior myocardial infarction, or both (OR, 13.0; 95% CI, 2.7–63.5); difference in postoperative hemoglobin value of >3.1 g/dL from preoperative levels (OR, 2.4; 95% CI, 1.6–3.8); a preoperative hemoglobin value <12.0 g/dL (OR, 3.3; 95% CI, 1.9–5.5); and prior thrombosis (OR, 2.7; 95% CI, 1.0–6.8).
| Characteristic | Model development cohort (n = 712) a | Model validation cohort (n = 100) a |
|---|---|---|
| Mean (SD) age at surgery, y | 52.0 (13.3) | 54.6 (13.5) |
| White | 608 (85.4) | 95 (95.0) |
| Primary insurance type | ||
| Private | 589 (82.7) | 74 (74.0) |
| Public | 110 (15.4) | 24 (24.0) |
| None | 8 (1.1) | 2 (2.0) |
| Not documented | 5 (0.7) | 0 |
| Mean (SD) BMI, kg/m 2 | 28.2 (6.0) | 28.2 (4.7) |
| Prior incontinence surgery | 9 (1.3) | 2 (2.0) |
| Prior prolapse surgery | 7 (1.0) | 1 (1.0) |
| Stage 3 or 4 prolapse on initial examination b | 287/462 (62.1) | 40/55 (72.7) |
| Indication for surgery | ||
| Prolapse | 344 (48.3) | 50 (50.0) |
| Bleeding | 351 (49.3) | 52 (52.0) |
| Incontinence | 157 (22.1) | 22 (22.0) |
| Other (eg, pain) | 190 (26.7) | 26 (26.0) |
| Medical comorbidities | ||
| Hypertension | 199 (27.9) | 30 (30.0) |
| Congestive heart failure | 4 (0.6) | 0 |
| Prior myocardial infarction | 7 (1.0) | 1 (1.0) |
| Asthma | 60 (8.4) | 11 (11.0) |
| Chronic obstructive pulmonary disease | 6 (0.8) | 1 (1.0) |
| Emphysema | 3 (0.4) | 0 |
| Gastrointestinal reflux disease | 71 (10.0) | 13 (13.0) |
| Gastric bypass surgery | 7 (1.0) | 2 (2.0) |
| Diabetes mellitus | 36 (5.1) | 2 (2.0) |
| Thyroid disorder | 106 (14.9) | 12 (12.0) |
| Cerebrovascular accident | 9 (1.3) | 2 (2.0) |
| Dementia | 2 (0.3) | 0 |
| Preoperative creatinine value ≥1.2 mg/dL | 47/692 (6.8) | 5/99 (5.1) |
| Hemoglobin <12.0 g/dL | 95/694 (13.7) | 12 (12.0) |
| Prior thrombosis | 23 (3.2) | 3 (3.0) |
| Tobacco use | 87 (12.2) | 7 (7.0) |
| Corticosteroid use within past year | 18 (2.5) | 2 (2.0) |
| ASA classification 1 or 2 | 638/706 (90.4) | 92/98 (93.9) |
| Surgical procedures performed | ||
| Oophorectomy (bilateral or unilateral) | 504 (70.8) | 72 (72.0) |
| Salpingectomy (bilateral or unilateral) | 522 (73.3) | 77 (77.0) |
| Reconstructive pelvic surgery | 336 (47.2) | 49 (49.0) |
| Anterior colporrhaphy | 24 (3.4) | 11 (11.0) |
| Posterior colpoperineorrhaphy | 16 (2.3) | 3 (3.0) |
| Combined anteroposterior colporrhaphy | 296 (41.6) | 35 (35.0) |
| Midurethral sling placement | 118 (16.6) | 20 (20.0) |
| Suprapubic catheter | 271 (38.1) | 43 (43.0) |
| Noncomplex surgery (VH with or without salpingectomy or oophorectomy) | 348 (48.9) | 48 (48.0) |
| Anesthesia type | ||
| General | 528 (74.2) | 80 (80.0) |
| Spinal | 181 (25.4) | 20 (20.0) |
| Epidural | 3 (0.4) | 0 |
| Mean (SD) estimated blood loss, mL | 274.7 (154.2) | 289.8 (126.2) |
| Transfusion | 29 (4.1) | 2 (2.0) |
| Mean (SD) change in hemoglobin (preoperative–postoperative), g/dL | 2.7 (1.2) | 2.6 (0.9) |
| Change in hemoglobin (preoperative–postoperative) >3.1 g/dL | 203/693 (29.3) | 22/96 (22.3) |
| Mean (SD) uterine weight, g | 141.4 (109.7) | 126.0 (104.5) |
| Uterine weight >250 g | 81 (11.4) | 8 (8.0) |
| Mean (SD) length of hospitalization, d | 2.4 (1.1) | 2.2 (1.2) |
| Any complication | 224 (31.5) | 26 (26.0) |
| Any non-UTI complication | 139 (19.5) | 15 (15.0) |
| Complications | ||
| Unplanned ICU admission | 7 (1.0) | 1 (1.0) |
| Hospital readmission | 24 (3.4) | 3 (3.0) |
| Reoperation | 17 (2.4) | 2 (2.0) |
| Medical problem requiring intervention c | 97 (13.6) | 14 (14.0) |
| Type of complication | ||
| Cardiopulmonary | 19 (2.7) | 1 (1.0) |
| Thromboembolic | 15 (2.1) | 0 |
| Infectious (not including UTI) | 18 (2.5) | 4 (4.0) |
| Genitourinary | 37 (5.2) | 7 (7.0) |
| Bleeding | 34 (4.8) | 3 (3.0) |
| Neurologic | 14 (2.0) | 2 (2.0) |
| Anesthesia related | 0 | 0 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
