Materials and Methods
After obtaining institutional review board approval from Kaiser Permanente, we performed a retrospective review of consecutive patients undergoing minimally invasive surgical staging for endometrial or cervical cancer between January 2008 and December 2011 at the West Los Angeles or Los Angeles Medical Center. Investigators abstracted data from each patient’s electronic medical record (EMR), specifically from notes from outpatient clinic, admission, anesthesia, and the postanesthesia care unit (PACU) as well as from the operative and pathology reports.
Investigators abstracted data regarding clinical and demographic information, perioperative data, and postoperative contacts, including emergency room visits, readmissions, and unscheduled office visits, e-mails, and phone calls within 6 weeks of surgery. Patients were excluded if they had known metastatic disease, were intended for overnight admission, or were converted to laparotomy at the time of the procedure.
Surgical procedure
All patients with endometrial or cervical cancer who underwent comprehensive surgical staging with either a robotic-assisted or traditional laparoscopic technique were included in the analysis. All endometrial cancer patients underwent total hysterectomy, bilateral salpingo-oophorectomy, bilateral pelvic lymphadenectomy, and bilateral paraaortic lymphadenectomy to the level of the inferior mesenteric artery.
For cervical cancers, the procedure included cystoscopy with bilateral ureteral stent placement, radical hysterectomy, bilateral pelvic lymphadenectomy, and ureteral stent removal; bilateral salpingo-oophorectomy was performed in a subset of patients. All cervical cancer patients were discharged to home with a Foley catheter in place. Robotic procedures were performed using the Model S Da Vinci surgical system (Intuitive Surgical, Sunnyvale, CA) at the West Los Angeles campus. Traditional laparoscopic procedures were performed at either of the Kaiser Permanente campuses. Two faculty surgeons were the primary surgeons for all of the procedures with gynecological oncology fellows assisting.
All patients received prophylactic cefazolin, or appropriate alternatives, prior to skin incision to prevent wound infections. All patients also received dexamethasone and ondansetron to prevent nausea and vomiting. Anesthetic management was not specifically mandated, but in general, patients were given a benzodiazepine prior to surgery, sevoflurane and fentanyl intraoperatively, and ketorolac and local anesthesia with 0.5% Marcaine at the conclusion of the procedure. An orogastric tube was placed after intubation. In addition, sequential compression devices were placed on the lower extremities of all patients during and after surgery to prevent deep venous thrombosis.
Patients who were discharged home before midnight on the same day as their surgery were categorized as having same-day discharge. The midnight demarcation was chosen based on the fact that Kaiser Permanente, as well as Medicare, tracks hospital days from midnight to midnight. During the study period, patients were discharged the same day if they met standard Kaiser Southern California PACU discharge criteria: normal vital signs, adequate oxygenation, able to ambulate independently, adequate pain control with oral medications, able to tolerate oral intake, and able to void postoperatively. The exception was for cervical cancer patients, all of whom were discharged with a Foley catheter in place.
Predictors of overnight stay
Multiple perioperative variables were assessed as possible predictors of overnight stay, including characteristics of the patient, tumor, and surgery. Operative time was defined as time from examination under anesthesia to the completion of skin closure; thus, the operative time for the robotic cases encompassed docking, console, and undocking time. Intraoperative complications included damage to organs, nerves, or vasculature. Anesthesia parameters included PACU pain scores; usage of antiemetics, nonsteroidal antiinflammatory drugs, and narcotics; status of catheter discontinuation; time to void; and time to oral intake. To facilitate a comparison, narcotic doses were converted to morphine intravenous equivalent doses using standardized tables.
Outcome measures
Postoperative complications and unscheduled contacts with health care providers were compared between patients undergoing same-day discharge and those requiring overnight admission. Postoperative complications included thromboembolic events, lymphedema, vaginal cuff complications, fistulas, femoral nerve neuropathy, bowel obstruction, non–wound-related infections (eg, pneumonia or sepsis), acute renal failure, requirement of blood transfusion, and intensive care unit admission.
The first postoperative appointment with the surgeon for gynecological oncology patients was scheduled 1 week after surgery. Data were collected regarding readmissions, emergency room visits without readmission, unscheduled office visits, and any verbal contacts (telephone or e-mail) within 6 weeks after surgery.
Statistics
Stata11 software (StataCorp LP, College Station, TX) was utilized to perform the statistical analysis. Descriptive statistics were used to analyze baseline characteristics. Categorical data were analyzed using a Fisher exact test, and nonparametric data were analyzed using the Mann–Whitney U test. A 2-tailed value of P < .01 was considered statistically significant.
To assess the combined effects of multiple variables in a manner that controls for multiple chances to find differences without assuming independence of dependent variables, a multivariate logistic regression model with backward elimination was used to identify predictive factors for same-day discharge. A second model was developed, also through backward elimination, to identify factors associated with unscheduled postoperative contacts. Unscheduled postoperative contacts was defined to include any hospital readmission, any emergency room visit without readmission, and any office visit other than those routinely scheduled; this included all patients with a postoperative complication. A value of P < .05 was required for inclusion in the final model. Given the original assessment of multiple variables, P < .01 was designated as statistically significant in the final multivariate models.
Results
Study population and patient characteristics
One hundred fifty-three consecutive patients with endometrial and cervical cancer were scheduled for comprehensive surgical staging via a minimally invasive method with a planned same-day discharge during the study period. Twelve patients (7.8%) were converted to laparotomy during the MIS procedure. All patients who were converted to laparotomy were excluded; thus, 141 patients met eligibility criteria and were included in the analysis. One hundred eighteen patients (83.7%) underwent same-day discharge and 23 patients (16.3%) required overnight admission. There were no significant statistical differences in demographic or clinical characteristics between the same-day and overnight groups ( Table 1 ).
| Characteristic | Same-day discharge (n = 118) | Overnight stay (n = 23) | P value |
|---|---|---|---|
| Age, y, median (range) | 60 (19–84) | 61 (36–78) | .97 |
| BMI, median (range) | 26 (16–48) | 27 (21–48) | .23 |
| Hypertension, n (%) | > .999 | ||
| Yes | 44 (37) | 9 (39) | |
| No | 74 (63) | 14 (61) | |
| Diabetes, n (%) | |||
| Yes | 17 (14) | 4 (17) | .75 |
| No | 101 (86) | 19 (83) | |
| Cardiovascular disease, n (%) | .69 | ||
| Yes | 10 (8) | 1 (4) | |
| No | 108 (92) | 22 (96) | |
| Lung disease, n (%) | .70 | ||
| Yes | 11 (9) | 3 (13) | |
| No | 107 (91) | 20 (87) | |
| Smoking, n (%) | .11 | ||
| Yes | 8 (7) | 4 (7) | |
| No | 110 (93) | 19 (83) | |
| Prior abdominal surgery, n (%) | .48 | ||
| Yes | 48 (41) | 7 (30) | |
| No | 70 (59) | 16 (70) | |
| Stage, n (%) | .77 | ||
| I | 96 (81) | 18 (78) | |
| II or greater | 22 (19) | 5 (22) | |
| Cancer type, n (%) | .77 | ||
| Endometrial | 96 (81) | 20 (87) | |
| Cervical | 22 (19) | 3 (13) |
Perioperative variables
Examining operative variables, same-day discharge was associated with robotic-assisted laparoscopy (vs traditional laparoscopy) and surgery initiation before 2:00 pm ( P < .01; Table 2 ). With regard to postoperative variables, same-day discharge was associated with lower pain scores in the PACU, lower narcotic requirements, ability to void if the Foley catheter was discontinued, and shorter median time to oral intake (all P < .01). Lower estimated blood loss ( P = .03) and shorter median time to void approached significance in predicting same-day discharge ( P = .01).
| Variable | Same-day discharge | Overnight stay | P value |
|---|---|---|---|
| Mode of surgery, n (%) | .001 | ||
| Laparoscopic | 20 (62.5) | 12 (37.5) | |
| Robotic | 98 (89.9) | 11 (10.1) | |
| Surgery start time, n (%) | .005 | ||
| Before 2:00 pm | 109 (87.2) | 16 (12.8) | |
| After 2:00 pm | 9 (56.2) | 7 (43.8) | |
| Operative time, cut to close, min, median (range) | 146 (91–253) | 170 (93–403) | .11 |
| Uterine weight, g, median (range) | 123.5 (39–885) | 104 (32–1332) | .38 |
| EBL, mL, median (range) | 50 (10–400) | 50 (10–950) | .03 |
| Pelvic nodes, median (range) | 13 (0–38) | 9 (0–25) | .04 |
| Paraaortic nodes, median (range) a | 8 (0–28) | 6 (0–15) | .14 |
| Intraoperative complication(s), n (%) | .51 | ||
| Yes | 3 (2.5) | 1 (4.4) | |
| No | 115 (97.5) | 22 (95.6) | |
| Narcotics (morphine equivalents), n (%) | .004 | ||
| <125 mg | 98 (89) | 12 (11) | |
| ≥125 mg | 20 (62.5) | 11 (37.5) | |
| Antiemetics, n (%) | .71 | ||
| <4 | 106 (84.1) | 20 (15.8) | |
| ≥4 | 12 (80) | 3 (20) | |
| PACU pain score, n (%) | .002 | ||
| Mild (0-2) | 51 (86.4) | 8 (13.6) | |
| Moderate (3-6) | 40 (95.2) | 2 (4.8) | |
| Severe (7-10) | 27 (67.5) | 13 (32.5) | |
| Void status, n (%) | < .0001 | ||
| Foley removed, able to void | 93 (87.7) | 13 (12.3) | |
| Foley removed, unable to void | 3 (30) | 7 (70) | |
| Foley left in place (cervical cancer) | 22 (88) | 3 (12) | |
| Time to void, min, median (range) | 222.5 (39–529) | 385 (48–676) | .01 |
| Time to PO intake, min, median (range) | 235.5 (13–523) | 316 (108–943) | .009 |
| PACU stay, min, median (range) | 225 (113–569) | 200 (109–685) | .34 |
a Total n = 116; endometrial cancer only; cervical cancer patients did not have paraaortic nodes removed.
Multivariate model to predict same-day discharge
A multivariate logistic regression model was then constructed via backward elimination to identify the factors that were predictive of same-day discharge ( Table 3 and Figure 1 ). Surgery start time at 2:00 pm or later, severe pain (pain score of 7 of ≥10), laparoscopic (vs robotic) approach, and more than 6 hours to oral intake were all significantly associated with an increased likelihood of an overnight stay ( P < .01, Figure 1 , A). Surgery start at 2:00 pm or later had the strongest impact, with an odds ratio (OR) of 36.8 ( P < .0001) for an overnight stay. An inability to void after catheter removal approached significance in predicting increased likelihood of an overnight stay ( P = .012). Once controlling for these variables in a multivariate analysis, there were no other factors significantly associated with an overnight stay.
| Variable | OR | SE | z | P > |z| | 95% CI lower bound | 95% CI upper bound |
|---|---|---|---|---|---|---|
| Lsc (vs robotic) | 9.05 | 6.25 | 3.19 | .001 | 2.34 | 35.10 |
| Surgery start time 2:00 pm or later | 36.80 | 33.5 | 3.96 | < .0001 | 6.19 | 219.3 |
| Severe pain in PACU | 6.81 | 4.74 | 2.76 | .006 | 1.74 | 26.60 |
| Time to oral intake longer than 6 h | 9.31 | 6.75 | 3.08 | .002 | 2.25 | 38.60 |
| Unable to void (if Foley removed) | 9.70 | 8.75 | 2.52 | .012 | 1.66 | 56.80 |
Postoperative outcomes and contacts
Postoperative complications
Within the same-day discharge group, 5.9% of patients developed a postoperative complication within 1 week, compared with 8.7% in the overnight group ( P = NS; Figure 2 , A). There was no difference in the overall frequency or type of complications 2 weeks postoperatively between the same-day and overnight admission groups (12.7% vs 8.7%, P = NS, Table 4 ).
| Variable | Same-day discharge, n, % (n = 118) | Overnight stay, n, % (n = 23) | P value |
|---|---|---|---|
| Pelvic abscess or vaginal cuff cellulitis | 4 (3.4) | 0 (0) | > .999 |
| Hematoma | 1 (0.8) | 0 (0) | > .999 |
| Venous thromboembolism | 1 (0.8) | 0 (0) | > .999 |
| Fistula | 1 (0.8) | 0 (0) | > .999 |
| Pneumonia | 1 (0.8) | 0 (0) | > .999 |
| Neuropathy/nerve palsy | 2 (1.7) | 1 (4.3) | .416 |
| Wound infection | 5 (4.2) | 1 (4.3) | > .999 |
| Lymphedema | 1 (0.8) | 0 (0) | > .999 |
| Death | 1 (0.8) | 0 (0) | > .999 |
| Any complication | 15 (12.7) | 2 (8.7) | .739 |
Readmissions
Seventeen patients in the entire cohort (12%) were readmitted to the hospital within 6 weeks; 1 patient was admitted within 7 postoperative days and 2 others were admitted within 14 postoperative days. There was no difference in readmission rates between the same-day discharge and overnight groups at 1, 2, or 6 weeks. Reasons for readmission included pelvic abscess, vaginal cuff dehiscence, hematoma, and venous thromboembolism ( Figure 2 , A and B). Causes for readmission in any time frame were also not significantly different between the 2 groups ( Table 5 ).
