Results of Introducing a Rapid Recovery Program for Total Abdominal Hysterectomy

Abstract

Objective:

To review the impact of implementing a rapid recovery protocol (RRP) for patients undergoing abdominal hysterectomy.

Setting:

Metropolitan teaching hospital.

Population:

Women undergoing abdominal hysterectomy for non-malignant indications.

Methods:

We conducted a retrospective review of consecutive cases performed during study periods before and after the introduction of an elective rapid recovery program emphasizing regional anesthesia. To control for universal improvements in medical practice, charts from a comparable local hospital without an RRP were also reviewed.

Results:

400 charts were reviewed and 366 cases met inclusion criteria and had sufficient information. Patients were well matched for demographic and medical variables between the study periods and between the institutions. The median length of stay (LOS) fell dramatically from 3 (range 1–12) days prior to RRP introduction to 1 (range 1–17) day after RRP (p < 0.001). LOS among patients at the ‘control’ institution remained unchanged at 3 days during the same time frame, indicating that external pressures contributed minimally to the observed changes. There were no significant differences in estimated blood loss, duration of surgery, or complication rate between the groups in either time period.

Conclusions:

Introducing a rapid recovery program was associated with shorter hospitalization and did not appear to compromise surgical outcome.

Introduction

Over 600,000 hysterectomies are performed annually in the United States making this operation the second most commonly performed surgery among women [1]. Despite the increasing use of less invasive approaches (vaginal hysterectomy and/or laparoscopic), nearly 2/3 of hysterectomies are still performed via traditional laparotomy which is associated with generally slower recovery and longer lengths of postoperative hospitalization, highlighting the potential for both clinical improvement and cost savings [2].

Rapid recovery protocols (RRPs) seek to optimize postsurgical morbidity outcomes by returning a patient to normal physiology as quickly as possible following surgery [3]. Though numerous specific strategies have been employed with success, the core elements of these ‘fast-track’ approaches consistently include: extensive preoperative education; optimizing the use of intraoperative regional anesthesia; early ambulation and feeding, and optimizing pain control in the first 24 h following surgery [4].

Benefits of an RRP observed in prospective studies involving colorectal patients include shorter hospitalization, decreased overall and nonsurgical complication rates, and possibly a decrease in complication severity [5]. In a study of 100 patients undergoing colonic surgery Zargar-Shoshtari et al. [6] reported that patients enrolled in the RRP arm had a lower incidence of urinary tract infections, postoperative ileus, and cardiopulmonary complications when compared to ‘standard practice’ [7]. Interestingly, similar benefits to patients appear to accrue across multiple RRP strategies, suggesting that the institution on an RRP may be more important to outcomes than the specific components introduced.

Though multiple prospective studies have demonstrated the feasibility of RRPs in orthopedic [8], colorectal [9], and gynecologic [10] and laparoscopic procedures [11], few have compared RRP to standard therapy directly. Information on abdominal hysterectomy is particularly sparse, and we were unable to find any studies comparing fast-track and standard therapy directly (PubMed search, keywords: hysterectomy, fast track, rapid recovery). In the present study, we sought to determine the impact of initiating an abdominal hysterectomy RRP by comparing postoperative morbidity and length of hospital stay (LOS) in patients who underwent abdominal hysterectomy before and after the implementation of an RRP.

Materials and Methods

The RRP concept was introduced to the gynecologic service at our hospital in early 2000 by a single surgeon who established a multidisciplinary team that included anesthesiology as well as nursing support. The core tenets of our RRP consist of preoperative counseling to prepare patients for early discharge, use of long-acting spinal anesthesia consisting of bupivacaine hydrochloride (typically 2 ml of 0.75% solution) and morphine sulfate (Astramorphi, 0.06–0.1 mg) with or without general anesthesia during surgery, early ambulation (day of surgery), and unrestricted diet starting the evening of surgery. Though each RRP component is clearly important, the management of patient comfort with minimization of systemic narcotic appears to be central to achieving the other components. Discharge criteria were at the discretion of the treating surgeon.

Between 2000 and 2004 our RRP was increasingly employed by other surgeons, and is currently employed in over 80% of all abdominal hysterectomies. Given the broad adoption of this practice we sought to quantify the benefits of introducing this strategy.

After obtaining approval from the appropriate Institutional Review Boards we performed a retrospective chart review. In order to eliminate selection bias we prospectively determined to survey 100 consecutive patients undergoing abdominal hysterectomy for non-malignant indications in each of two time periods: period 1 (beginning on January 1, 1999), and period 2 (beginning January 1, 2004). Charts were identified by persons blinded to study intent using CPT and ICD-9 coding. Patients undergoing vaginal, laparoscopically assisted vaginal, total laparoscopic, or supra-cervical hysterectomy, or who underwent hysterectomy for cancer were excluded to reduce the effect of alternate surgical strategies on our targeted outcomes. We elected to review consecutive cases despite the non-uniform use of RRP tenets to avoid bias from ‘over-selecting’ favorable patients, and in an attempt to assess the global impact of adopting the RRP philosophy with the understanding that no single strategy would be appropriate for all patients. Given the retrospective nature of the design, and in the absence of a known variance in the length of stay or treatment effect size, sample size was selected to provide sufficient data points to detect gross trends and a priori power calculations were not attempted. In post hoc analysis using the observed variance from the period 1 group, our sample size (200 patients) would provide a power of 0.94 to detect a reduction in LOS of ≥50%. Adoption of this regimen was voluntary during the study period; and for the purposes of this analysis, use of regional anesthesia (which occurred in ≤5% of patients before the introduction of the RRP) was considered a surrogate for use of the RRP, as this was used rarely prior to the study period and was felt to be the central component of the rapid recovery process. Patients treated at our institution were designated the ‘experimental group’ regardless of whether or not they were treated using the RRP.

To confirm that the observed results from this analysis, especially with regard to LOS, were derived from the implementation of an RRP, and not the product of global changes in medical-surgical management during the study periods or of regional social and/or cost-containment pressures, a ‘control group’ with a comparable patient and physician makeup, but which was not exposed to an RRP, was identified at an alternate local hospital of similar size and procedural capacity, and subjected to the same analyses. Though surgeons may have operated at both institutions during the study period, there was no surgeon cross-over in the data analyzed; that is no surgeon had study patients at both sites. Further it was expected that any influence of ‘philosophical cross-over’ would blunt rather than amplify observed differences between the study groups.

Abstracted data included: demographics; intraoperative anesthesia (drug and delivery mechanism); incision type; length of surgery; estimated blood loss; uterine weight; intra- and postoperative complications; use of postoperative narcotic; time to ambulation; resumption of diet; LOS (measured in whole numbers with the day of surgery being ‘day 0’ and each subsequent day initiating at midnight); post-discharge complications (30 days), and readmission rate (up to 30 days). Fever in the absence of documented infection was not considered a postoperative complication. The Mann-Whitney rank sum method was used where appropriate to analyze continuous variables; and χ² tests were used to analyze proportions. A two-tailed analysis was performed in all cases and a p value of <0.05 was considered statistically significant. SigmaStat for Windows (version 3.11) was the software used to perform statistical analyses.

Results

After excluding patients with malignancy or insufficient available data, we abstracted data from 86 and 96 charts from the control and experimental groups, respectively, in period 1, and 90 and 94 charts, respectively, in period 2. The groups were well matched for surgical indications, uterine weight, and most demographic features between institutions and time periods (table 1). Similarly, there were no significant differences in the distribution of surgical procedures between the groups or time periods with approximately 60% undergoing hysterectomy with bilateral salpingo-oophorectomy in each group and period (p = 0.39).

Table 1.

Patient characteristics

	Period 1	Period 2	p value
Age, years
Control	42 (30–69)	45 (24–81)	0.16
Experimental	45 (26–82)	45.5 (23–79)	0.95
	p < 0.01	p = 0.85
Diabetes, %
Control	2	8	0.19
Experimental	4	9	0.35
	p = 0.78	p = 0.93
Hypertension, %
Control	7	19	0.04
Experimental	21	25	0.75
	p = 0.02	p = 0.43
>2 previous surgeries, %
Control	30	34	0.72
Experimental	22	26	0.74
	p = 0.29	p = 0.31
Uterine weight, g
Control	202 (52–956)	182 (19–1,339)	0.99
Experimental	205 (35–1,400)	183 (17–1,853)	0.57
	p = 0.18	p = 0.6

Continuous variables reported as medians (ranges).

The use of regional anesthesia during period 1 was similar at 5 and 8% in the experimental and control hospitals, respectively. Use increased significantly in the experimental group to 83% in period 2 reflecting an increased use of rapid-recovery modalities (p < 0.001). Use actually decreased from 8 to 1% over the same period in the control group though this difference was not statistically significant. Concomitant with these changes the experimental group experienced significant decreases in LOS (p < 0.0001), and median length of surgery (p < 0.01) that were not experienced in the control group (table 2).

Table 2.

Surgical outcomes

	Period 1	Period 2	p value
Surgical duration, min
Control	125 (70–310)	130 (55–260)	0.7
Experimental	150(95–485)	130 (80–270)	<0.01
	p < 0.01	p = 0.87
Estimated blood loss, ml
Control	300 (50–1,800)	250 (50–2,500)	0.85
Experimental	200 (10–3,500)	150 (50–2,100)	0.4
	p = 0.09	p < 0.001
Postoperative hemoglobin, g/dl
Control	10.5 (7.1–12.7)	10.7 (6.0–13.9)	0.07
Experimental	10.8 (7.5–14.0)	11.1 (6.9–14.1)	0.28
	p = 0.02	p = 0.19
Complication rate1, %
Control	22	18	0.6
Experimental	16	19	0.65
	p = 0.35	p = 0.96
LOS, days
Control	3 (2–8)	3 (2–13)	0.04
Experimental	3 (1–12)	1 (1–17)	<0.001
	p = 0.8	p < 0.001

Continuous variables reported as medians (ranges).

¹Includes both intraoperative and postoperative complications.

Major postoperative complications were rare and there were no statistically significant differences between the 2 groups in either period 1 or period 2 (table 3). In absolute terms, however, the complication rate increased in the experimental group from 11 to 17% between the 2 study periods while the rate in the control group remained more stable, rising from 9 to 11%. Complications in the immediate post-discharge period likewise appeared similar between the groups with readmission rates between 1 and 2% across both study periods.

Table 3.

Postoperative complications

	Period 1	Period 2	p value
ICU admission, %
Control	0	0	NS
Experimental	2	2	0.62
	p = 0.52	p = 0.50
Pulmonary (pneumonia or PE), %
Control	0	0	NS
Experimental	1	3	0.59
	p = 0.95	p = 0.26
Infection (blood or urine), %
Control	3	1	0.58
Experimental	0	0	NS
	p = 0.20	p = 0.98
Gastrointestinal (ileus or SBO), %
Control	0	7	0.04
Experimental	3	1	0.63
	p = 0.28	p = 0.11
Renal (ARF or stricture), %
Control	0	0	NS
Experimental	1	0	NS
	NS	NS
Readmission, %
Control	1	2	0.97
Experimental	1	2	0.99
	p = 0.53	p = 0.64
Total1, %
Control	9	11	0.88
Experimental	11	17	0.38
	p = 0.82	p = 0.35

NS = Not statistically significant.

¹Includes complications not presented above.

In subgroup analysis, 79 patients in period 2 of the experimental group underwent spinal anesthesia with or without the use of concomitant general anesthesia (our surrogate marker for RRP usage). The median LOS for this group was 1 (range 1–11) compared to 2.5 (range 1–17) days for the 15 patients who underwent general anesthesia without spinal anesthetic (p = 0.02). The median LOS for control group patients undergoing general anesthesia alone in period 2 (n = 89) was 3 days and was not significantly different from the experimental group (p = 0.60), again suggesting that improved LOS was likely attributable to the RRP and not unmeasured improvements at our hospital.

Discussion

Our review demonstrates that the introduction of a rapid recovery protocol was associated with improvements in length of hospitalization without apparent detriment to patient outcomes. These benefits accrued despite the absence of significant changes in any other surgical parameters studied, and without a formalized introduction of the program or complete penetration of the RRP among potentially eligible patients.

The philosophy of rapid recovery is founded on the notion that much of what slows post-surgical recovery is independent of the surgery itself, including the effects of systemic analgesia, immobility, and nutritional deprivation. The ideal, evidence-based, cost-effect strategy to abrogate these secondary effects has been sought since Basse et al. [12] first reported a successful RRP strategy in the treatment of patients with colorectal diseases. Perhaps, however, implementing the ‘ideal’ strategy should be a secondary goal; a meta-analysis of multiple prospective studies by Wind et al. [13], identified improvements in LOS and morbidity (without increases in readmission rates or mortality) across multiple RRP formats, suggesting that the initiation of an RRP is likely more important than the specific components of the program. Though some success in RRP development has been reported in the gynecologic literature, we were unable to identify any controlled studies comparing accelerated recovery programs directly to traditional recovery in the setting of abdominal hysterectomy [3].

Most RRPs, including the one used in this study, incorporate components of anesthesia, surgery, nursing, and physical therapy with the stated goal of reducing surgical morbidity though evidence-based optimization of one or more of the following: pre-operative preparation; anesthesia; analgesia; fluid management; use minimally invasive surgery; nutrition, and/or early ambulation [14]. Fundamental to these programs is a focus on decreasing the use of systemic narcotic analgesics, predominantly through the use of regional anesthesia, in an effort to limit the deleterious side effects of these medications, especially impairment of bowel motility, respiratory drive, and mobility. Later refinements included the addition of intraoperative regional anesthesia [15], combination therapy non-opioid analgesics and increasing used of short- and ultrashort-acting narcotic agents in regional anesthesia [11]. Virtually all reports stress the importance of a multi-specialty ‘team’ approach, involving the patient, surgeon, anesthesiologist, and recovery nursing staff.

Though retrospective, our study does demonstrate the feasibility of instituting an RRP, and that benefit to patients' LOS can be achieved without an increase in morbidity. Between the 2 study periods the use of regional anesthesia for routine hysterectomy in the experimental hospital increased from 5 to 82% of consecutive cases while the rate at the control institution was relatively static, implying that there was ‘buy-in’ by both surgeons and anesthesiologists concomitant with the demonstration of clinical benefit. There was no formal promotion of the RRP during or after its introduction, but the process was subject to the usual departmental quality assurance reviews; and in the absence of specific discharge criteria it can be inferred that the improvement in LOS resulted from an accelerated return to normal function. Interestingly, our experience was similar to that of Jottard et al. [16] who reported an increase in the use of regional anesthesia from 46 to 94% over 1 year following the introduction of a more formalized RRP for colonic surgery.

Another weakness of our study is the inability to detect to which components of the RRP a specific patient was exposed, thus limiting our ability to analyze the relative contributions of pre-operative counseling, early feeding, surgeon enthusiasm, regional anesthesia, etc. It was, however, our intent to demonstrate both the feasibility of introducing the RRP approach and to identify potential ‘global’ benefits. Given that the patient characteristics, surgical outcomes, and complication rates were similar in both the experimental and control groups before and after the introduction of the RRP, and the overwhelming adoption of regional anesthesia at the experimental hospital, but not the control hospital, there can be little doubt that the outcome differences observed accrued at least in part from the application of RRP tenets. Further evidence that the benefits observed result from the RRP can be gleaned from the analysis of the experimental hospital's period 2 subgroup which demonstrated that patients who did not have regional anesthesia, i.e. who should have benefitted from all non-RRP medical advances but none of those associated with the RRP, had outcomes very similar to those in all three of the other ‘non-exposed’ cohorts. As such, the authors anticipate that the results reflect what might be expected following introduction of a similar program.

Though we do not believe that all hysterectomy cases are appropriate for the use of regional anesthesia or RRP, our data suggest that there are significant populations of women undergoing hysterectomy who may benefit from some or all components of the program. Further, our data support the observations of Jottard et al. [16] that successful introduction of an RRP can lead to rapid, widespread adoption of the program with benefits of improved resource utilization and, perhaps improved clinical outcomes. Prospective study of this approach in a randomized controlled trial is warranted.