Ambulatory-Based Bladder Outlet Procedures Offer Significant Cost Savings and Comparable 30-Day Outcomes Relative to Inpatient Procedures

Abstract

Introduction and Objectives: Budgetary constraints and novel minimally invasive surgical approaches have resulted in surgical care being increasingly provided at ambulatory centers rather than traditional inpatient settings. Despite increasing use of ambulatory-based procedure for bladder outlet obstruction (BOO) procedures, little is known about the effect of care setting on perioperative outcomes and costs. We sought to compare 30-day readmissions rates and costs of BOO procedure performed in the ambulatory vs inpatient setting.

Methods: Using Florida and New York all-payer data from the 2014 Healthcare Cost and Utilization Project State Databases, we identified patients who underwent transurethral resection, thermotherapy, or laser/photovaporization for BOO. Patient demographics, regional data, 30-day readmissions rates, and costs (from converted charges) associated with the index procedure and revisits were analyzed. Predictors of 30-day revisits were also identified by fitting a multivariate logistic regression model with facility-level clustering.

Results: Of the 15,094 patients identified, 1444 (9.6%) had a 30-day revisit at a median cost of $4263.43. The 30-day readmission rate for inpatient cases was significantly higher than that of surgeries performed in the ambulatory setting (12.0% vs 8.1%, p < 0.001). Payer status (private vs Medicare: odds ratio [OR] = 0.77, 95% confidence interval [CI] = 0.62–0.95; p = 0.02) and index care setting (ambulatory vs inpatient: OR = 0.48, 95% CI = 0.40–0.57; p < 0.001) predicted 30-day revisits.

Conclusions: We identified that index care setting and payer status are independent predictors of 30-day revisit after BOO procedure, with the inpatient setting and Medicare insurance associated with higher odds of revisit. Ambulatory procedures are significantly less costly than procedures performed in the inpatient setting, even after accounting for ambulatory procedures leading to an admission. There is an obvious cost benefit of offering BOO procedure in the ambulatory setting to the appropriate patient. In the context of value-based health care initiatives, our findings have important implications for policymakers seeking to reduce variation in nonclinical sources of perioperative costs and outcomes.

Introduction

With an aging population, the prevalence and economic burden of benign prostatic hyperplasia (BPH) are increasing.¹ Cost pressures and newer minimally invasive outlet procedures have led to a migration away from the traditional inpatient care setting toward ambulatory surgical centers for bladder outlet obstruction (BOO) procedure.² However, there is limited literature evaluating the impact that this transition in care setting has had on perioperative costs and outcomes after BOO procedure. Literature to date has largely consisted of prospective studies comparing the effectiveness of various surgical techniques (e.g., transurethral resection of the prostate [TURP] vs photo/laser therapy) or retrospective case series describing the feasibility of ambulatory procedure,³ with little attention given to associated costs or comparison of care settings.

Elucidating the factors influencing the cost of BOO procedure is a necessary first step in developing higher quality pathways of care for BPH patients, which will grow in importance as value-based purchasing practices are applied to an ever-growing number of health care services.⁴ However, it remains unclear which patient- and provider-level characteristics influence the index setting of operation (inpatient or outpatient), and whether perioperative outcomes and costs differ between inpatient and outpatient cohorts. Using all-payer claims data, we sought to identify factors associated with index surgical care setting and its impact on risk-adjusted perioperative outcomes and costs after elective surgical therapy for BPH. We hypothesize that ambulatory-based BOO procedure is associated with significant cost savings and comparable 30-day revisit rates relatively to procedures conducted in the inpatient setting.

Methods

Data source

The 2014 Healthcare Cost and Utilization Project (HCUP) State Ambulatory Surgery and Services Database (SASD), State Inpatient Database (SID), and State Emergency Department Database (SEDD) from Florida and New York were used to identify all patients. The SID, SEDD, and SASD are prospectively maintained administrative databases from the HCUP of the Agency for Healthcare Research and Quality (AHRQ) and contain encounter-level data on hospital-affiliated and/or owned facilities.⁵ A limited number of state datasets contain de-identified patient-linkage numbers that can track patients across care settings over the course of a year. The date of all index procedures and subsequent health care encounters are recorded, allowing for the identification of index vs revisit encounters.

Patient selection

The cohort selection process is given in Figure 1. All patients undergoing an index TURP (Current Procedural Terminology [CPT] codes 52601 and 52630; International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] code 60.29), photo/laser therapy (CPT code 52648; ICD-9-CM code 60.21), or thermotherapy (CPT codes 53850 and 53852; ICD-9-CM codes 60.96 and 60.97) were included, yielding 3182 revisits among 15,283 unique individuals. As BOO procedure performed in the inpatient setting represents a more costly alternative to ambulatory procedure, index surgical care setting was dichotomized into either inpatient (n = 5470) or ambulatory (n = 8902). For the purpose of our cost and 30-day revisit analyses, ambulatory procedures were further categorized into those requiring an admission (n = 3234) vs not requiring an admission (n = 5668). Index BOO procedures were defined as the first operation performed on an individual over a 365-day period. Of note, patients undergoing a repeat TURP within 90 days of their index procedure (CPT code 52620) were excluded from our analysis.

FIG. 1.

Consort diagram depicting cohort selection process.

Covariates

Baseline patient variables consisted of patient's age at time of operation, race (White, Black, Hispanic, other race), insurance type (Medicare, Medicaid, private, uninsured), type of residence (urban or rural), county-level median household income quartiles, comorbidity status using the modified Charlson Comorbidity Index (CCI) score, and education level (median county-level rate of less than high school education quintiles derived from linkage to Federal Information Processing Series codes). Facility-level characteristics included state in which the facility was located, length of index hospitalization, and total episode-related charges.⁶ Index surgical and 30-day acute care costs borne in the inpatient/emergency department (ED) setting were calculated by multiplying total charges associated with the hospitalization by the state and year-specific HCUP cost-to-charge ratio (CCR). Although CCRs are not available for outpatient procedures, CCRs for hospital-owned and freestanding ambulatory facilities can be extrapolated from the inpatient facility's CCR, as has been previously described in the orthopedic literature.⁷ Thirty-day acute care costs consisted of index surgical and all postoperative costs borne by hospital admission, ED visits, and/or repeat procedures. All encounters with missing covariates (n = 911) were dropped and considered to be missing at random as we controlled for all observable variables affecting the probability of missingness.⁸

Statistical analyses

Our co-primary outcomes were receipt of index procedure within the ambulatory rather than inpatient setting, index surgical and 30-day costs, and all-cause 30-day revisit rates. A revisit was defined as a postprocedural visit to the ED, admission to a hospital, and/or re-intervention in the ambulatory setting. Frequencies and proportions were reported for categorical variables, whereas median and/or adjusted mean and interquartile ranges were reported for continuous variables. Bivariate differences in categorical and continuous variables between index care setting and 30-day revisit groups were examined using the Rao–Scott chi-square and the Wilcoxon rank-sum tests, respectively.

We performed a multivariable logistic regression model adjusted for the abovementioned covariates and facility clustering to identify predictors of receiving index surgical care within the ambulatory vs inpatient care setting. Gamma generalized linear and logistic regression models adjusted for facility clustering were used to compare index surgical/30-day costs and odds of 30-day readmission between the ambulatory and inpatient procedure cohorts, respectively. In an attempt to adjust for confounding by indication, a subgroup analysis comparing odds of 30-day revisit between ambulatory procedures requiring admission and inpatient-based operation was performed.

All analyses were performed using StataMP 14 (StataCorp, College Station, TX) with a two-sided significance level set at p < 0.05. An institutional review board waiver was obtained before conducting this study.

Results

Univariate differences between individuals treated in the inpatient vs outpatient settings are given in Table 1. A total of 14,372 individuals underwent an outlet procedure during the study period, of which 61.9% (n = 8902) were performed in the ambulatory setting. Among ambulatory procedures, 36.3% (n = 3234) resulted in an admission ranging in length from 1 to 3 days. In multivariate logistic regression analysis, negative predictors of undergoing operation in the ambulatory care setting included a CCI score ≥2 (odds ratio [OR] = 0.45, confidence interval [CI] = 0.38–0.53, p < 0.001), Medicaid insurance (OR = 0.48, CI = 0.3–0.75, p < 0.001), and Black race (OR = 0.75, CI = 0.59–0.96, p = 0.02) (Table 2).

Table 1.

Patient and Facility Characteristics According to Index Surgical Care Setting

	Inpatient, n (%)	Ambulatory, n (%)	p
	n = 5470	n = 8902
Age, years, median (IQR)	74 (67–82)	72 (66–79)	<0.001
Length of stay, days, median (IQR)	2 (1–6)	0 (0–1)	<0.001
Gender, n (%)
Male	5470 (100.0)	8902 (100.0)
CCI, n (%)			<0.001
0	2279 (41.7)	5559 (62.4)
1	1056 (19.3)	1606 (18.0)
≥2	2135 (39.0)	1737 (19.5)
Payer, n (%)			<0.001
Medicare	4246 (77.6)	6878 (77.3)
Medicaid	279 (5.1)	119 (1.3)
Private	889 (16.3)	1835 (20.6)
Uninsured	56 (1.0)	70 (0.8)
Race, n (%)			<0.001
White	3614 (66.1)	7111 (79.9)
Black	582 (10.6)	523 (5.9)
Hispanic	800 (14.6)	1038 (11.7)
Other	474 (8.7)	230 (2.6)
Income, n (%)			<0.001
$1–$39,999	1634 (29.9)	2966 (33.3)
$40,000–$50,999	1490 (27.2)	3251 (36.5)
$51,000–$65,999	1000 (18.3)	1974 (22.2)
≥$66,000	1346 (24.6)	711 (8.0)
Education, n (%)			<0.001
<9.7%	1035 (18.9)	2155 (24.2)
9.7–11.6%	1082 (19.8)	1907 (21.4)
11.7–12.9%	1354 (24.8)	2598 (29.2)
≥13%	1999 (36.5)	2242 (25.2)
Urban/rural status, n (%)			<0.001
Large metro area, ≥1,000,000	4172 (76.3)	4425 (49.7)
Small metro area, <1,000,000	1029 (18.8)	4015 (45.1)
Micropolitan	163 (3.0)	314 (3.5)
Rural	106 (1.9)	148 (1.7)
State, n (%)			<0.001
New York	2676 (48.9)	388 (4.4)
Florida	2794 (51.1)	8514 (95.6)
Procedure, n (%)			<0.001
TURP	4524 (82.7)	5136 (57.7)
Laser	912 (16.7)	3699 (41.6)
Thermotherapy	34 (0.6)	67 (0.8)

CCI = Charlson Comorbidity Index; IQR = interquartile range; TURP = transurethral resection of the prostate.

Table 2.

Multivariate Logistic Regression Analysis Predicting Receipt of Index Procedure in the Ambulatory Care Setting

	OR	95% CI	p
Age, years	0.97	0.96–0.97	<0.001
CCI
0	Reference
1	0.85	0.72–1.00	0.06
≥2	0.45	0.38–0.53	<0.001
Payer
Medicare	Reference
Medicaid	0.48	0.30–0.75	0.001
Private	1.10	0.90–1.35	0.3
Uninsured	0.55	0.30–1.02	0.06
Race
White	Reference
Black	0.75	0.59–0.96	0.02
Hispanic	0.92	0.69–1.24	0.6
Other	0.75	0.53–1.07	0.1
Income
$1–$39,999	Reference
$40,000–$50,999	1.15	0.98–1.35	0.08
$51,000–$65,999	1.35	1.08–1.69	0.008
≥$66,000	0.90	0.64–1.26	0.5
Education
<9.7%	Reference
9.7–11.6%	0.85	0.54–1.32	0.5
11.7–12.9%	1.24	0.87–1.76	0.2
≥13%	0.51	0.34–0.79	0.002
Urban-rural status
Large metro area, ≥1 million	Reference
Small metro area, <1 million	2.78	1.83–4.23	<0.001
Micropolitan	4.55	1.54–13.46	0.006
Rural	3.16	1.50–6.66	0.003
Procedure
TURP	Reference
Laser	4.86	3.73–6.34	<0.001
Thermotherapy	1.38	0.45–4.28	0.6
State
New York	Reference
Florida	21.60	9.95–46.86	<0.001

CI = confidence interval; OR = odds ratio.

Index care setting and 30-day acute care costs

Adjusted mean index surgical costs were significantly lower among patients undergoing operation within the ambulatory compared with inpatient setting ($5777.64 vs $7926.71, p < 0.001). Adjusted mean 30-day acute care costs were also significantly lower among patients undergoing index procedure within the ambulatory compared with inpatient setting ($5212.27 vs $6212.12, p < 0.001). Differences in index ($6134.84 vs $7926.71, p < 0.001) and 30-day ($4066.35 vs $6212.12, p < 0.001) costs persisted even after accounting for ambulatory procedures requiring admission (Fig. 2).

FIG. 2.

Unadjusted cost figures by index care setting. Following risk adjustment, patients undergoing operation in the ambulatory setting still experienced significantly lower (A) index ($5777.64 vs $7926.71, p < 0.001) and (B) 30-day costs ($5212.27 vs $6212.12, p < 0.001) relative to inpatient-based care. Differences in index ($6134.84 vs $7926.71, p < 0.001) and 30-day ($4066.35 vs $6212.12, p < 0.001) costs persisted even after accounting for ambulatory procedures requiring admission (*p < 0.001).

Revisit analysis

The overall 30-day revisit rate after BOO procedure was 9.6% (n = 1376), with 8.0% of ambulatory procedures experiencing a 30-day revisit compared with 12.2% of inpatient-based procedures (p < 0.001). The most common diagnosis associated with a revisit was a urinary tract infection (UTI). UTIs collectively accounted for ∼15.0% of all primary/secondary diagnoses. In adjusted analysis, ambulatory-based procedure was found to be protective against experiencing a 30-day revisit (ambulatory vs inpatient: OR = 0.51, 95% CI = 0.41–0.62; p < 0.001). A sensitivity analysis comparing 30-day revisit rates between inpatient- and ambulatory-based procedure requiring admission revealed similar findings (OR 0.50, 95% CI 0.40–0.61; p < 0.001) (Supplementary Table S1).

Conversely, higher comorbidity burden (0 vs >2: OR = 1.63, 95% CI = 1.39–1.92; p < 0.001), lower income (highest vs lowest quartile: OR = 0.71, 95% CI = 0.57–0.88; p = 0.002), and undergoing operation in Florida rather than New York (OR = 2.47, 95% CI = 1.93–3.15; p < 0.001) predicted greater odds of experiencing a revisit (Table 3).

Table 3.

Multivariate Logistic Regression Analysis Identifying Predictors of 30-Day Revisit

	OR	95% CI	p
Age, years	1.01	1.01–1.02	0.001
Length of stay, days	1.02	1.01–1.03	0.002
CCI
0	Reference
1	1.33	1.12–1.59	0.002
≥2	1.63	1.39–1.92	<0.001
Payer
Medicare	Reference
Medicaid	0.89	0.56–1.42	0.6
Private	0.80	0.64–1	0.05
Uninsured	0.48	0.2–1.16	0.1
Race
White	Reference
Black	0.95	0.77–1.22	0.7
Hispanic	1.19	0.99–1.4	0.05
Other	0.66	0.48–1	0.03
Income
$1–$39,999	Reference
$40,000–$50,999	0.75	0.63–0.89	0.001
$51,000–$65,999	0.67	0.55–0.81	<0.001
$66,000+	0.71	0.57–0.88	0.002
Education
<9.7%	Reference
9.7–11.6%	1.18	0.96–1.45	0.1
11.7–12.9%	0.95	0.77–1.16	0.6
≥13%	0.96	0.78–1.19	0.7
Urban-rural status
Large metro area, ≥1 million	Reference
Small metro area, <1 million	1.06	0.90–1.26	0.5
Micropolitan	1.02	0.73–1.44	0.9
Rural	0.59	0.34–1.05	0.08
Index care setting
Inpatient	Reference
Ambulatory	0.51	0.41–0.62	<0.001
Procedure
TURP	Reference
Laser	1.15	0.98–1.34	0.1
Thermotherapy	1.45	0.69–3.07	0.3
State
New York	Reference
Florida	2.47	1.93–3.15	<0.001

Discussion

As the population ages, demand for the surgical management of BPH will continue to grow.⁹ In a health care system already strained by high costs¹⁰ with increasing pressure to deliver high-value care, untangling the cost of surgical care is an important step toward containing it.¹¹ There is a need for solutions maximizing the yield on the value equation, which seeks to maximize surgical outcomes per dollar spent. One such step is the provision of care in the appropriate care setting.¹² Using all-payer claims data from the states of New York and Florida, we identified several clinical (e.g., age and CCI score) and nonclinical (e.g., insurance status and race) predictors of the care setting in which individuals undergo operation for BOO. We demonstrate that ambulatory procedure is indeed significantly less costly than procedures performed in the inpatient setting, even after risk adjustment and accounting for ambulatory-based procedures requiring hospital admission. Furthermore, patients who had their index surgical procedure in the ambulatory setting were less likely to experience a 30-day revisit compared with patients receiving care in the inpatient setting. These findings hold significant policy implications in the setting of recent value-based purchasing reforms being undertaken in the United States.⁴

In our multivariable logistic regression analysis, Medicaid coverage was a predictor of receiving care in the inpatient setting relative to Medicare coverage. Provider-level biases may be at play here. Studies have demonstrated that physicians at physician-owned ambulatory centers were more likely to refer well-insured patients to these facilities and direct Medicaid patients to inpatient settings for administrative and financial reasons.^13,14 Such disadvantages of Medicaid include delayed payments and often lower reimbursement rates for surgical procedures, all which act as disincentives for care in more profitable ambulatory centers.¹⁵ With the Affordable Care Act leading to increased Medicaid enrollment, routing of Medicaid beneficiaries to high-resource inpatient hospital settings may have future consequences that remain to be seen.¹⁶ Another predictor of care setting was patients' geographic status. Small metro area, micropolitan, and rural patients were more likely to receive care in the ambulatory setting likely because of reduced access to inpatient care.¹⁷

When seeking to address the rising costs of care, treating patients in the optimal care setting is one way to reduce low-value care, especially in the context of cost-intensive surgical care.¹⁸ As has been shown by others, this analysis demonstrates that ambulatory procedures were less costly than procedures performed in the inpatient setting.^7,15,19 This was true for both the index episode of care and associated 30-day revisits. As these analyses were adjusted for comorbidity status and thus attempted to mitigate confounding by indication, it is unlikely that the etiology of this cost difference is because of routing of sicker patients to the inpatient setting. Rather, it is possible that patients receiving care in the inpatient setting are confronted with structural barriers that were not accounted for in our analyses. Such barriers may include no access to a usual source of care²⁰ and low health literacy,²¹ both associated with increased visits to the ED resulting in higher 30-day costs. Regarding lower index episode of care costs in the ambulatory setting, this can potentially be explained by lower surgical facility fees¹⁹ and fewer overnight admissions.²² Our analysis also highlights that a nonclinical factor, Medicaid coverage, was an independent predictor of inpatient care. It has been shown that Medicaid beneficiaries spend more on medical services.²³

Another important consideration for high-value care is the surgical approach selected. Although laser photovaporization of the prostate was more commonly carried out in the ambulatory setting, there were no differences in 30-day revisit rates for laser compared with traditional TURP. This is in line with other studies that have compared these modalities in both the inpatient and outpatient setting and found generally similar short-term outcomes.^24,25 Similar to Strope and colleagues, there were also no differences in revisits when comparing both therapies by care setting.²⁵

Medicaid was not correlated with increased 30-day revisits, whereas patients with private insurance, on the contrary, had increased odds of a 30-day revisit. Although it is tempting to attribute these results to inherent differences between the inpatient and ambulatory cohorts, our analysis was risk adjusted and took into account comorbidity status. Rather, these differences in 30-day revisits between payer status may be explained by factors that we did not control for such as the patient's health literacy,²⁶ care coordination,²⁷ and social relationships.²⁸ In a similar vein, a sensitivity analysis was performed comparing 30-day outcomes between inpatient-based procedures and patients treated in the ambulatory setting that required a hospital admission. As receiving care in the ambulatory setting was also protective against experiencing a 30-day revisit in the sensitivity analysis, the etiology of this finding is less likely attributable to disease burden but rather may reflect the aforementioned provider- and facility-level factors not accounted for in the analysis.

This study contributes to the growing body of evidence assessing the impact of care setting on the quality and cost of surgical care.¹⁵ In the context of value-based health care reforms being undertaken in the United States, our findings highlight the cost and outcome implications of directing patients to the optimal treatment setting. Indeed, we show that care in the ambulatory setting achieves cost savings without compromising 30-day outcomes. These findings are encouraging in the setting of the recent rule change proposal by CMS that will reimburse inpatient and ambulatory procedure similarly.²⁹ Consequently, it can be argued that developing policies that encourage high-value pathways of surgical care represents a promising means by which to slow, if not reduce, the economic burden of health care in the United States.

This study has some limitations that are common to all retrospective, observational studies using administrative datasets. In this particular study, the lack of granular data is the main limitation. Clinical factors that influence operative decision-making or outcomes were not available, such as prostate size, history of TURP, or secondary indications for operation beyond BPH/lower urinary tract symptoms (retention, infections, bleeding, bladder stones, etc.). Laser and TURP are also too general and do not capture the cost differences between different modalities. Identifying specific cost drivers, such as the use of anesthesia or time spent in the OR in different care settings, was also not possible. Finally, although the 30-day revisit metric is commonly used to evaluate outcomes, it is not the only measure of quality of care and is only an aspect of value of care assessment. Future studies should focus on identifying more granular clinical and nonclinical factors that may help to further risk-stratify patients and thereby direct patients to the highest value care setting possible.

Conclusion

We identified that index care setting and payer status are independent predictors of 30-day readmission after BOO procedures, with the inpatient setting and Medicare insurance associated with higher odds of readmission. Ambulatory procedures are significantly less costly than procedures performed in the inpatient setting, even after accounting for ambulatory procedures leading to an admission. There is an obvious cost benefit of offering BOO procedure in the ambulatory setting to the appropriate patient. In the context of value-based health care initiatives, our findings have important implications for policymakers seeking to reduce variation in nonclinical sources of perioperative costs and outcomes.

Abbreviations Used

BOO

bladder outlet obstruction

BPH

benign prostatic hyperplasia

CCI

Charlson Comorbidity Index

CCR

cost-to-charge ratio

CI

confidence interval

CPT

current procedural terminology

ED

emergency department

HCUP

Healthcare Cost and Utilization Project

ICD-9-CM

International Classification of Diseases, Ninth Revision, Clinical Modification

OR

odds ratio

SASD

State Ambulatory Surgery and Services Database

SEDD

State Emergency Department Database

SID

State Inpatient Database

TURP

transurethral resection of the prostate

UTI

urinary tract infection

Author Disclosure Statement

Q.-D.T. reports personal fees from Astellas, Bayer, Janssen, Insightec, and Intuitive Surgical. All other authors have no competing financial interests.

Funding Information

Q.-.D.T. is supported by the Brigham Research Institute Fund to Sustain Research Excellence, the Bruce A. Beal and Robert L. Beal Surgical Fellowship, the Genentech Bio-Oncology Career Development Award from the Conquer Cancer Foundation of the American Society of Clinical Oncology, a Health Services Research pilot test grant from the Henry M. Jackson Foundation for the Advancement of Military Medicine, the Clay Hamlin Young Investigator Award from the Prostate Cancer Foundation. D.F.F. is supported by a National Institutes of Health T32 training grant (2T32DK007527-33). D.-D.N. participated in this project as a recipient of an Endourological Society Summer Student Scholarship.

Supplementary Material

Supplementary Table S1