Assessment of outcome is critical in the success of outpatient surgery and anesthesia. Although major morbidity and mortality after ambulatory surgery are rare , less serious postoperative events may occur but go unreported. The impact of these events is difficult to measure as most continuous quality improvement programs focus on patient satisfaction, hospital admissions, and cancellations , but have limited data regarding return hospital visits once patients have been discharged. Postoperative phone calls, conducted usually within 72 hours after discharge, identify minor side effects and may fail to capture surgery-related difficulties . Additionally, issues that come to the surgeon's attention postdischarge are rarely shared with the ambulatory surgery unit (ASU). The objective of this study was to examine return hospital visits as an outcome measure and to identify predictor variables in a multispecialty ASU which serves a heterogeneous population of ages, races, and insurance classes in a large metropolitan area. The investigators focused specifically on patients returning to the hospital or emergency room for complications related to the prior ambulatory surgery in order to formulate interventions to reduce these returns.
An ASU information system is integrated into the database of the hospital from which a retrospective review was conducted to identify patients returning to the same hospital within 30 days after ASU admission. Data were collected from hospital medical records for 12 consecutive months. Demographic variables recorded included the following: patient age, gender, surgical service, procedure, anesthesia type, ASA physical status class, interval between discharge and return visit, length of stay of readmission, insurance status, and provider profiles. A return hospital visit was defined as either any readmission to the hospital as an inpatient (IP), to the ASU for another surgical procedure, or to the emergency room (ER) with subsequent discharge home. Location of return (ER, ASU, or IP) was identified by the hospital medical records coder, while reasons for the return visits were determined independently by two investigators using current procedural terminology, diagnostic codes, and medical chart review. Return visits were coded as either related to or unrelated to the initial ambulatory surgery and then categorized by their location, as follows: ER visits related to prior ambulatory surgery (e.g., bleeding, fever, pain, urinary retention); ER visits unrelated to the prior ambulatory surgery; surgical complication (e.g., bleeding, wound dehiscence), medical complication (e.g., perioperative cardiac or pulmonary events), or anesthesia complication (e.g., airway complications, nausea, vomiting) directly attributed to the ambulatory surgery requiring rehospitalization as an IP or to the ASU; IP or ASU admission related to the prior ambulatory surgery but not secondary to complications (e.g., mastectomy after breast biopsy, additional renal stone related procedures); and IP or ASU admission unrelated to the prior ambulatory surgery.
A matched case control design was used to identify factors associated with an increased likelihood of return. Because complications requiring rehospitalization and ASU-related ER visits were considered to be the more clinically relevant variables, subsequent analyses are reported for those categories only, and termed complications. For this sample, two case controls were obtained for each patient and were matched for time of surgery (within 1 wk). The case controls were obtained from medical records of patients who underwent ambulatory surgery during the same time period, but who did not return to the hospital within 30 days. Patient demographics recorded for the control group were the same as for the study group. The logistic regression performed in the matched case control data included both demographics as well as surgical service and type of anesthesia. Miscellaneous surgical services (e.g., ophthalmology, plastic and dental surgery) were combined into one category. A separate univariate analysis was conducted which considered the likelihood of return given the 13 most common surgical procedures for all ASU patients in the facility.
Statistical analyses were performed using SAS (SAS Institute, Cary, NC) and StatXact. SAS Proc PHREG (SAS Institute) was used to determine predictors of return in the matched case control data, using multiple and univariate logistic regressions. Odds ratios and their probabilities for the 13 surgical procedures were calculated using the program StatXact (Cytel Software Co., Cambridge, MA). A P value of <0.05 was considered significant for all comparisons.
Over the 12-mo period, from January through December 1994, 6243 cases were completed in our hospitalintegrated ASU, with 187 return hospital visits reported within 30 days (3.0%). Of these, 54% of patients returned to the ER, 46% returned as IP or ASU admissions (Table 1). Of these, 29% returned to the ER for treatment related to the ASU procedure and occurred with greater frequency than the other categories (P = 0.04); 25% returned to the ER for reasons unrelated to surgery. Surgical complications, resulting in IP or ASU readmission, accounted for 15% of the total return hospital visits, IP admissions not considered complications but related to previous surgery accounted for 21%, while IP admissions unrelated to previous ambulatory surgery occurred in 10% of patients. No medical or anesthesia complications related to surgery were reported. Two medical admissions did occur: one patient developed congestive heart failure with atrial fibrillation 17 days after a cataract extraction; the other patient, also after a cataract extraction, developed unstable angina 13 days postoperatively. Both were discharged home. Because of the time interval, both incidents were thought to be unrelated to the previous ambulatory procedures.
Of the four major ASU surgical specialties (general surgery, otolaryngology, gynecology, and urology (GU), only GU was found to have a significantly higher return rate of 5.8%, P < 0.001. The majority of return hospital visits were related to the prior ambulatory surgery for all specialties (Table 2). Miscellaneous surgical operations (plastic surgery, orthopedics, dental surgery, and podiatry) reported a low rate of 1.7%. The average return rate for "complications" was 1.3%. Similar to that found with the overall return rate, GU had a significantly higher rate for complications of 2.7%, P < 0.0013 (Table 2).
The result of the multivariate logistic regression analysis of the case-control population identified GU as an independent predictor of returns for complications; odds ratio 27.87, confidence interval (CI) 3.78-74.86, P = 0.0002. Miscellaneous surgical services were less likely to be readmitted, P = 0.0016. Age, ASA class, and gender were not significant predictors of return. Additionally, insurance class, day of the week of the reencounter, or specific providers were not significant predictor variables. In the multivariate analysis, monitored anesthesia care (MAC) was found to predict hospital returns as nearly five times more frequent than all other anesthesia type, P < 0.01 (CI 1.43-15.73) (Table 3). Subsequent exploration of the logistic regression model showed that MAC was associated with higher rates of return in the case of GU and lower rates in the case of miscellaneous surgical services. The number of patients in these breakdowns was rather small and so the results for MAC may not be reliable.
Although age was not a predictor of return, it did emerge as a predictor of the location of return. From the group that returned, patients under the age of 40 yr were 2.8 times more likely to be seen and treated in the ER (CI 1.86-4.37, P < 0.001) regardless of the reason for ER visit, while patients over the age of 65 yr were 2.6 more likely to be rehospitalized as an IP or for additional ASU procedure (CI 1.41-4.86, P = 0.0023).
To further evaluate return rates by specialty, the likelihood of return after 13 common surgical procedures was examined for all 6243 ambulatory surgery cases. Odds ratios for each procedure is given in Table 4. Two procedures were identified as more likely to result in complications: patients undergoing varicocelectomy and hydrocelectomy were 8.3 times more likely to return (CI 2.90-23.75, P = 0.0042), of which two thirds were evaluated for infection and fever. Patients undergoing dilation and curettage (D&C) were three times more likely to return (CI 1.78-5.55, P = 0.0002), and 56% were for bleeding. These included mostly termination of pregnancies and a few diagnostic procedures. Eight of nine (89%) D&C return hospital visits for bleeding, returned to the ER. Only one was significant enough to require rehospitalization and reoperation.
Reasons for return hospital visits are reported in Table 5. Of the 82 surgical complications reported, 10 (12%) required reoperation and 18 (22%) returned within 24 h of ASU discharge, all of whom had met discharge criteria. Extended observation or a 23-h admission was not even considered except in two cases (one patient developed urinary retention, the other patient had a suspected bladder perforation). However, both patients refused hospital admission. Seventy-two percent of return hospital visits occurred within the first 7 days, and no mortalities were reported within 30 days. Because of the small numbers involved with multiple procedures and reasons analyzed, it was not possible to determine statistically whether certain complications tended to be associated with specific procedures. However, certain complications occurred more frequently with one procedure than another. This was previously noted above for D&C, hydrocelectomy, and varicocelectomy procedures. Of all the returns, 34 or 41.5% of patients returned for bleeding, 76.5% of whom were treated and discharged through the ER. All the other patients who were rehospitalized secondary to bleeding were admitted through the ER. Eighty percent of return visits for bleeding after hernia repair, and 75% of return visits for bleeding after breast biopsies were to the ER. Other patients who returned to the ER after a variety of other procedures did so for various reasons. Fever and infection occurred after 10 different procedures. Surgical pain required returns to the ER after six different procedures. Swelling occurred in six cases, four of whom had breast-related procedures. Urinary retention occurred in five patients (four after urologic surgery, one after hernia repair), four of whom returned within 24 h for bladder catheterization.
The incidence of returning to the hospital after ASU discharge from a hospital-integrated ASU is reported to be 3%: 1.3% for complications requiring rehospitalization or ER treatment and 1.7% not considered as complications. Like major morbidities and mortalities and unanticipated hospital admissions, return hospital visits after ambulatory surgery appear to be fewer, but they nonetheless serve as an important indicator of outcome. The data reported here include complications with varying morbidities and interventions required. However, no recent reported data from the United States are available for benchmarking among facilities. ASUs should include in their continuous quality improvement programs the ability to evaluate and compare return hospital visits after ASU. A prospective monitor should include patient follow-up at seven to ten days postoperatively to identify ER visits or hospital admission. The impact of return hospital visits on patient outcome is not fully appreciated, especially with insurance providers, as pressure mounts to further expand the ambulatory surgery procedure list.
The most striking finding was of GU emerging as a significant predictor for all return hospital visits. Data suggest that changing trends in GU practice have occurred, the greatest change seen in kidney procedures and less significantly in those procedures involving the ureter, urethra, testes, and scrotum . Rates of return for common urological procedures from our facility are provided. Noted is the significantly higher return rate specifically in patients undergoing hydrocelectomy and varicocelectomy procedures, who returned within 14 days postoperatively with the diagnosis of infection. While we do not suggest that patients remain hospitalized for that extensive period of time, proper treatment and follow-up need to be incorporated into the discharge planning. Reevaluation of the appropriateness of these cases on an outpatient basis is strongly warranted as is the need to determine the cause of this phenomenon. We speculate that GU surgery is inherently vulnerable because of the potential for three of the most disturbing events requiring readmission: hematuria, infection, and urinary retention. Birch  discusses the implications of urological day surgery on European and American practices and identifies an extensive list of suitable urological procedures for outpatient surgery. However, no input regarding outcome of these procedures was provided. Of clinical importance to anesthesiologists is that hospital returns are independent of age, ASA class, and general anesthesia. No patients returned because of medical or anesthesia events directly related to perioperative care. Continued involvement by the anesthesiologist during the perioperative period could be a major factor for maintaining this high safety profile. From the reasons analyzed for hospital return (e.g., bleeding, pain, urinary retention, wound dehiscence), we infer that MAC emerges as a significant predictor because of its correlation with type of surgical procedure, rather than as an independent causative factor.
Bleeding was the most common reason for hospital return (41.5%), and occurred with greatest frequency after D&Cs. These procedures were three times more likely to result in return hospital visits, with the majority of these returns for bleeding which were evaluated in the ER for treatment and discharge. This is a clinically important finding that warrants further evaluation.
What was evident in the study is that ambulatory surgery is placing some degree of burden on emergency rooms. The gynecologic population is relatively young and healthy; however, our data did not suggest that these return hospital visits were more common among indigent patients, or on the weekend. Although not as frequent as D&Cs, returns to the ER for bleeding occurred in 11 other common ASU procedures. Standard postoperative instructions encourage patients to return to the ER should a significant problem arise. However, we further question the use of the ER resources for issues that could be addressed in a different medical care setting. Improper use of ER is a problem that continues to plague the medical system [6-8]. Interventions involving intense pre- and postoperative education, and advising patients when to call their physicians may be helpful in reducing the number of ER visits related to bleeding. Guidelines for what constitutes serious bleeding and what does not (for example, spotting or oozing), may significantly reduce these nonemergent return hospital visits. Provisions for postoperative care need to be provided by the primary care physician (as is done outside the United States) or through provisions of ambulatory care facilities or "urgent-care" clinics, rather than using the already overtaxed facilities of the ER.
Return hospital visits and complications were reported by Henderson et al.  in 1989 and by Natof  in 1980, but were descriptive in nature without any analysis to identify predictor variables. More recent studies examining patient outcome and follow-up using postoperative phone calls [11,12], questionnaires , and hospital medical records [14,15] have not specifically examined return hospital visits within 30 days after ambulatory surgery, nor attempted to identify any risk factors. In the study by Osborne and Rudkin , follow-up occurred in the first three postoperative days with 3.1% of patients returning to hospital emergency departments; however no information is provided for reason, or whether it was related to surgery. In addition, while the overall reported incidence is similar, their data only spanned three postoperative days. Ghosh and Sallam  reported that 2.5% of patients contacted the hospital within 24 hours of ASU discharge for minor problems. It is unclear whether any of these patients required a return to the hospital. Warner et al.  evaluated only patients sustaining major morbidity or mortality within 30 days, not all of whom were readmitted. A readmission rate comparable to this study was reported from the United Kingdom after either ambulatory or IP inguinal hernia repair and hemorrhoidectomy (Jarrett PEM, Department of Surgery, Kingston NHS Trust, England, personal communication, March 1995). Unlike United States data, this study permits comparisons to the same procedures on inpatients; however, analysis regarding risk factors have not been identified. We believe, therefore, that ours is the only recent study that focuses on analyzing return hospital visits with an attempt to identify risk factors to further enhance patient education and safety.
One limitation of this study was that it reported patients returning only to our health care facility. Therefore, there may be an underreporting of the actual return rate. However, based on our nursing follow-up phone calls to approximately 90% of the ASU patients, we found that patients rarely report seeking care at another facility.
In conclusion, the finding of a relatively low rate of return visits to the hospital within 30 days after ASU discharge is encouraging; however, improvement is still possible. Further benchmarking is needed among facilities, as current data are lacking. Identifying specific interventions which could improve patient outcome postoperatively, in particular for GU, is needed. The exceptionally high incidence of return hospital visits to the ER for bleeding suggests that a reduction could be gained by better informing patients about bleeding. Pre- and postoperative education that advises patients of alternative routes for seeking medical care under nonemergency conditions are warranted.
The authors acknowledge John Hartung, PhD, and Barbara McEwan, RN, for their editorial assistance.