Data source

Data from the Perspective database were analyzed. Perspective is a nationwide, voluntary, fee-supported database developed to measure quality and resource utilization. Perspective captures data from >600 acute-care hospitals located across the United States. Each participating center reports data on a quarterly basis. The database includes comprehensive information on patient demographics, clinical characteristics, procedures, and all billed services. Submitted data undergo a rigorous process of quality assurance. The Perspective database is validated and has been used in a number of outcome studies. In 2006, the Perspective database recorded approximately 5.5 million hospital discharges, which represents roughly 15% of nationwide hospitalizations. Perspective includes a sample of acute care hospitals from throughout the United States. Like the national composition of hospitals in general, the facilities sampled by Perspective are predominately midsized, nonteaching facilities more frequently found in urban areas. There is a larger representation of hospitals from the southern United States. All data were deidentified and the study deemed exempt from human subjects approval by the Columbia University Institutional Review Board.

Patients and procedures

Women between 18-90 years of age who underwent laparoscopic hysterectomy from 2000 through the first quarter of 2010 were analyzed. Patients who underwent either a laparoscopically assisted vaginal hysterectomy ( International Classification of Diseases, Ninth Revision [ ICD-9 ] 68.51) or a total laparoscopic hysterectomy ( ICD-9 68.41) were included in the cohort. Patient characteristics that were analyzed included: age (<45 years vs 45-60 years vs >60 years), year of surgery, race (white vs black vs other), primary insurance (Medicare vs Medicaid vs commercial vs uninsured vs unknown), and marital status (married vs single vs unknown).

Hospital characteristics including location (metropolitan vs nonmetropolitan), teaching status (teaching vs nonteaching), size (<400 beds vs 400-600 beds vs >600 beds), and region (East vs Midwest vs South vs West) were recorded. Comorbidity was captured using the Deyo et al modification of the Charlson et al index. The specialty of the primary surgeon for each procedure was noted and classified as gynecologist, gynecologic oncologist, other, or unknown. Potential indications for surgery, including leiomyomata, endometriosis, abnormal bleeding, pelvic organ prolapse, benign ovarian neoplasms, and cancer were recorded based on ICD-9 coding for each patient. These entities were not mutually exclusive. Likewise, concomitant procedures performed at the time of laparoscopic hysterectomy, including oophorectomy, anterior colporrhaphy, posterior colporrhaphy, incontinence repair, and lymphadenectomy were noted.

To assess the effect of volume on discharge, annualized hospital and surgeon procedural volume were calculated. The number of procedures performed by each physician or hospital was calculated and the total divided by the number of years in which the given physician or hospital contributed at least 1 procedure. The volumes were then inspected visually and separate volume-based tertiles created. Physician volume tertiles were: low (<6 procedures), intermediate (6-14.625 procedures), and high (>14.625 procedures). Hospital volume tertiles were: low (<49.2 procedures), intermediate (49.2-108 procedures), and high (>108 procedures).

Outcomes

For each patient Perspective captures the total length of hospital stay. The length of stay is based on presence of the patient in a facility after a set point during a day. Patients who remain hospitalized after this point in time, which varies across hospitals, are considered to have an additional day of hospitalization. We classified patients as: same-day discharge (surgery and discharge within the same 24-hour period), length of stay 1 day (admission and discharge later than on the reported hospital day after surgery), and length of stay ≥2 days.

We measured major acute complications and transfusion during the admission for the laparoscopic hysterectomy. Complications included intraoperative complications (bladder, ureteral, intestinal, vascular, or other intraoperative injury), surgical-site complications (wound complications, abscess, hemorrhage, bowel obstruction, ileus), medical complications (venous thromboembolism, myocardial infarction, cardiopulmonary arrest, respiratory failure, renal failure, stroke, bacteremia/sepsis, shock, pneumonia), and transfusions as previously reported. Perioperative complications included any intraoperative, surgical-site, or medical complication. We also examined hospital-based reevaluation within 60 days of the index hospitalization for the procedure. Reevaluation rates are based only on evaluation within the same facility as the initial procedure. Each reevaluation was classified as either outpatient (emergency department or other urgent care facility) or inpatient (hospital readmission). ICD-9 diagnoses at the time of reevaluation were recorded and included: nausea/emesis, wound complications, abscess, bleeding, pain, failure to thrive, respiratory infection, and urinary tract infection. These diagnoses were not mutually exclusive.

Cost was examined for each patient. Perspective contains an itemized log of all items that are billed to a patient during hospitalization, as well as hospital service costs. Among the hospitals included in Perspective, approximately three quarters report direct costing data taken from internal accounting systems while the remainder report cost estimates based on Medicare cost-to-charge ratios. Cost data from Perspective have been validated and used in a number of reports examining resource utilization. All costs were adjusted for inflation using the Consumer Price Index and reported in 2010 US dollars. After conversion, cost data were visually inspected and those patients with spurious costs (<$500) were removed from the analyses. Cost estimates are reported for the index admission for each group and a second estimate is reported examining all costs during the 60-day perioperative period (index admission as well as reevaluations within 60 days). The reported economic data represent cost and not charges.

Statistical methods

Frequency distributions between categorical variables were compared using χ ² tests. The association between patient (age, year of surgery, race, comorbidity, marital status, insurance status), physician (volume), and hospital (location, region, teaching status, size, volume) characteristics and same-day discharge was examined using multivariate mixed effects log binomial regression models. To account for hospital-level clustering, these models included a random intercept for the hospital in which the procedure was performed. A separate model was developed that only included patients who did not experience a perioperative complication. Similar models were developed to examine factors predictive of reevaluation. Results are reported as risk ratios (RRs) and 95% confidence intervals (CIs).

Data on costs were estimated for each group based on hospital length of stay (0, 1, and ≥2 days). Since cost data were heavily right-skewed, we report costs as median with interquartile ranges (IQRs). Multivariable adjustment of costs across length of stay was compared based on quantile (median) regression methods. This method directly estimates the adjusted median costs and 95% CI were derived based on bootstrap resampling methods. All analyses were performed with software (SAS, version 9.2; SAS Institute Inc, Cary, NC). All statistical tests were 2-sided.