Efficiency of Sorting Site of Care for Frail Patients Undergoing Mastectomy

TL;DR

Frail patients undergoing mastectomy may benefit economically from outpatient care despite slightly higher transfer risks.

Contribution

This study quantifies the economic benefits of outpatient mastectomy for frail patients and evaluates transfer risks.

Findings

Frailty is associated with increased odds of inpatient care (OR, 5.856; p < 0.001).

Prefrail or frail patients in ASCs have a 0.4% transfer rate, with cost savings of $8404 per patient.

Transfer rates would need to increase over 100-fold to negate cost savings from ambulatory procedures.

Abstract

Patients undergo mastectomy in both ambulatory surgery centers (ASCs) and inpatient settings. Guidelines for site selection are poorly defined. Older adults, particularly those with frailty, are at increased risk of adverse outcomes postoperatively. Transfer to an acute hospital is a unique adverse event suggesting potentially inappropriate ASC care.

The authors used logistic regression modeling to describe the association of frailty with site of care and transfer, and modeled expected costs associated with ambulatory mastectomy for robust and prefrail or frail patients.

In ASCs, 85.3% of all patients and 51.3% of prefrail or frail patients underwent mastectomy. Frailty or prefrailty was associated with increased odds of inpatient care (odds ratio [OR], 5.856; p < 0.001). Odds of transfer were higher among prefrail and frail patients (OR, 2.640; p < 0.05), but rates remained low (< 0.4%). Rates of transfer needed to negate cost-savings from ambulatory procedures are more than 100 times the current rate (38%; standard error, 4.7%). If all prefrail and frail patients received care at ASCs, expected cost savings would be $8404 per patient.

Despite slightly higher rates of transfer, clinicians should consider treating frail and prefrail older adults in ASCs given possible economic benefits.

Full text

INTRODUCTION

Approximately 300,000 patients are diagnosed with breast cancer annually in the United States, roughly one third of whom undergo mastectomy.^1,2^ Patients undergo mastectomies in a number of sites, including freestanding ambulatory surgery centers (ASCs) and hospital outpatient departments (HOPDs). ASCs are considered safe for appropriate patients^3–5^ and care in ASCs is associated with significant cost savings relative to care at hospitals,^6^ contributing to ongoing shifts in site of care selection. Financially constrained healthcare systems, patient preference, and overcrowded hospital wards have all contributed to an increasing proportion of mastectomies being performed on an ambulatory basis.^6^ Despite this shift towards ASC care, there is no accepted metric for determining a patient’s approriateness for ambulatory surgery.

Frailty, a clinical syndrome marked by decreased physiologic reserve and increased vulnerability to stressors, is present in an estimated 43% of women with breast cancer.^7–9^ Compared to robust patients, frail patients are more likely to suffer complications and use a greater amount of healthcare resources after surgery.^10–13^ As a result, frailty assessments are becoming increasingly integrated into surgical practice. The interplay of frailty and appropriateness for different sites of care, however, remains under-explored.

In previous studies, clinicians appear to be triaging patients with greater multimorbidity to settings with greater post-operative supervision,^14^ however, the impact of frailty, a distinct entity which is influenced by multimorbidity but not determined by it, on site of care has not yet been explicitly described. Furthermore, though ASCs are widely considered safe, frailty has not been investigated as a risk factor for undertriage to ASC care, including transfer to acute care hospitals following surgery, a key quality metric. Finally, while prior literature has investigated costs associated with ambulatory and inpatient care,^6^ analyses which take into account differential risk profiles for frail patients or that establish recommended care patterns, have yet to be conducted.

Our investigation had three aims 1) to characterize the current impact of frailty on site of care selection, 2) to describe the potential effects of undertriage to ASC care based on frailty status, and 3) to clarify economically optimal care patterns for patients undergoing mastectomy. Our results may ultimately be used to guide practice recommendations and inform further policy changes regarding coverage of inpatient stays associated with mastectomy.

Methods

Data Source

i.

We utilized the Agency for Healthcare Research Healthcare Cost and Utilization Program’s Nationwide Inpatient Sample (NIS) and Nationwide Ambulatory Surgery Sample (NASS) databases from 2019 to 2022. The project was deemed non-human subjects research by the Massachusetts General Brigham Institutional Review Board and was exempt from further review.

Study Population

ii.

Both databases (NIS and NASS) were queried for all female patients aged 18+ undergoing mastectomy for a diagnosis of breast cancer. Though frailty is more common in those over the age of 65, we included all adult women to capture any patients who may meet criteria for frailty despite being below the age of 65. We excluded patients with metastatic cancer or those receiving concurrent reconstruction, as these patients may have alternative factors playing a role in the clinician’s decision to treat them as inpatients. We identified ambulatory operations as recorded in the NASS and inpatient surgeries as those recorded in the NIS. Procedure codes used were utilized in prior studies to identify similar samples of patients.^15,16^

Variables

iii.

We collected demographic variables, including patient age, race, and urbanicity. We also considered clinical variables, including hospital teaching status. We defined frailty using the Hospital Frailty Risk Score^17^ which stratifies patients on the basis of ICD-10 coded comorbidities into low risk patients with a score <5 (robust), intermediate risk with a score 5–15 (prefrail), or high risk with a score >15 (frail). We grouped prefrail and frail patients given limited numbers of patients in these groups and prior studies suggesting minimal difference in outcomes between medium and high risk of frailty groups.^18,19^

Outcomes

iv.

The primary outcome was location of surgery determined by data file source with inpatient surgery identified in the NIS and ASC surgeries identified via the NASS. The NASS captures ambulatory surgeries performed in hospital owned facilities indepent of location and may include observation stays. The secondary outcome, transfer to an acute care hospital, was assessed only among patients receiving care in an ambulatory setting. Transfer to an acute hospital following ambulatory procedures was selected as it is a rare but essential quality indictor suggesting unsuitability of ambulatory care.^20–22^ Costs of care by site were calculated via weighted means from sample data per protocol surggested by the AHR.

Analysis

v.

Impact of Frailty on Site of Care Selection and Transfer Following Ambulatory Surgery

a.

Table 1 and Table 3 summarize categorical variables using frequencies with percentages and continuous variables with medians and interquartile ranges (IQRs). We conducted descriptive analyses using Chi-square tests to compare categorical characteristics and Wilcoxon rank-sum tests for continuous variables by site of care and post-surgery transfer to an acute care hospital, respectively. Subsequently, we performed adjusted multivariable logistic regression to assess the associations of frailty with site of care, accounting for demographic (age, race, urbanicity) and clinical variables (location and teaching status). Additional subset analysis were conducted separately within the robust and the prefrail and frail population to identify factors influencing site of care selection within each group. We furthermore examined factors associated with transfer to an acute care facility following care in an ASC using adjusted logistic regression model, accounting for the same demographic and clinical covariates factors. All analyses accounted for the complex survey design, including stratification and clustering. We applied discharge weights to produce national estimates. These analyses were performed in SAS software, v.9.4 (SAS Institute, Cary, NC).

Econometric Analysis

b.

For our econometric analysis, we utilized a decision tree to characterize clinician site of care selection and to determine expected costs associated with each site of care for women with breast cancer requiring a mastectomy. Expected costs were calculated based on weighted average charges by location, documented proportions of patients receiving care in both locations by frailty status and predicted probabilities of transfer from the previously described logistic regression models utilizing the following formula: $[eqn]$ . Charges, rather than patient facing costs were used to refect the economic efficiency of the system, rather than burden to the patient. Cost of transfer was estimated to be $45,390 (SE$5,518.93) based on prior literature reporting the cost differential between ambulatory surgery requiring a readmission after the procedure and ambulatory surgery which did not require a readmission following the procedure.^6^ The model relies on the simplifying assumption that care is provided either in inpatient settings or ambulatory surgical settings, aligned with how we analyzed and interpreted our data. Since care at ASC costs less than inpatient care, cost savings were estimated based on assuming constant rates of transfer while transitioning all mastectomy care to ASCs i.e. setting proportion^ASC^ to 1. Maximum allowable transfer risk was determined to be the rate of transfer such that expected cost savings for total ASC care were equivalent to expected costs associated with transfer to inpatient care. These analyses were performed in R version 4.4.1.

Results

Overview

i.

The sample consisted of 222,943 patients, a minority of whom (14.7%) underwent their mastectomy in inpatient settings (Table 1). The median age was 65.1 (IQR 54.8–73.3). Most patients (66.6%) were White and lived in metropolitan settings (82.2%). Most treating facilities were urban teaching centers (74.1%). Most patients (97.7%) were robust.

Impact of Frailty on Site of Care Selection and Transfer Following Ambulatory Surgery

ii.

Patients receiving inpatient care were younger (median age 64.9, IQR 54.4–73.8) compared to patients cared for in ASCs (median age 65.1, IQR 54.9 – 73.2, p = 0.05). The proportion of patients receiving inpatient care was highest among Black patients (19.4%) and lowest among White patients (12.9%, p < 0.001). Patients in metropolitan areas were more likely to receive care in inpatient settings (15.4%) compared to those in micropolitan settings (11.4%, p < 0.001). There were no significant differences in the proportion of patients receiving care in either setting based on location or teaching status of the hospital. Patients who were frail or prefrail were more likely to receive care in inpatient settings compared to robust patients (48.5% vs 13.9%, p < 0.001) and median frailty scores were significantly higher in the inpatient population (0.7, IQR 0.0–2.1, vs 0, IQR 0.0 – 1.4, p < 0.001)

On multivariable analysis, prefrailty and frailty were associated with 5.86 increased odds (95% CI 5.21 – 6.59, p < 0.001) of inpatient surgery relative to robust patients when adjusting for clinical and demographic variables (Table 2). Patients who were Black (OR 1.54, 95% CI 1.40 – 1.70), Hispanic (OR 1.56, 95% CI 1.38 – 1.76), or Other (OR 1.32, 95% CI 1.17 – 1.48) were all more likely to receive inpatient care than White patients (p < 0.001). Patients receiving care in non-metropolitan areas were less likely to receive inpatient care (Micropolitan OR 0.65, 95% CI 0.57 – 0.74; not Metropolitan or Micropolitan OR 0.68, 95% CI 0.59 – 0.78, p < 0.001). Patients receiving care at hospitals in rural areas were more likely (OR 1.33, 95% CI 1.15 – 1.53, p < 0.001) to receive inpatient care. Among robust patients, factors associated with increased odds of inpatient care included non-White Race, urbanicity, and rural hospital location (p < 0.001). Among pre-frail and frail patients, only Black patients were more likely to receive inpatient care (OR 1.44, 95% CI 1.12 – 1.86, p < 0.001).

Of patients treated in an ASC (190,148), 209 were transferred to an acute hospital following their procedure (0.1%) (Table 3). Age, race, urbanicity, location and teaching status of the hospital, and frailty were all associated with transfer following surgical treatment (Table 3). On multivariable analysis, being prefrail or frail (OR 2.64, 95% CI 1.14 – 6.13, p<0.05) and receiving care at a rural hospital (OR 5.41, 95% CI 2.73–10.78, p <0.001) were associated with increased odds of transfer, adjusting for age, race, and urbanicity (Table 4).

Expected Costs, Cost Savings, and Maximum Allowable Transfer Risk

iii.

Weighted average costs associated with mastectomy were $64,799.74 (SE$808.40) for inpatient care and $47,472.71 (SE$102.32) for ASC care. Estimated probability of transfer was 0.107% (SE: 0.001%) among robust patients and <0.4% (SE: 0.01%) among prefrail and frail patients. Total expected costs per patient associated with undergoing mastectomy were $49,929.54 (SE$142.89) among robust patients and $56,036.23 (SE$396.05) among prefrail and frail patients. If all prefrail and frail patients underwent ambulatory surgery with transfer rates at currently observed levels, total expected costs would be $47,632.62 (SE$104.00) yielding an expected cost savings of $8,404 per patient. The maximum allowable transfer risk such that there would be expected cost equivalency between ASC and inpatient care would be 38% (SE 4.7%), over 100 times greater than currently observed transfer rates seen among frail and prefrail patients.

Discussion

We found that among patients undergoing mastectomy, frailty is associated with inpatient surgery as opposed to ASC surgery. Overall, though transfer to an inpatient hospital following mastectomy in an ASC was a rare event in either population, transfer is more common among prefrail and frail patients. Average costs associated with inpatient care were higher than those associated with ASC care. Expected costs for robust patients were lower than those for prefrail and frail patients, attributable to the described differences in site of care selection. We describe opportunities for cost savings in excess of $8,000 per patient if all prefrail and frail patients were to undergo mastectomy in ASCs rather than inpatient settings. Finally, we demonstrate that in order for expected costs between the two sites to be equivalent, the transfer rate would need to be roughly 100x greater than is currently observed among prefrail and frail patients.

In our study, we found that prefrail and frail patients are more likely to undergo mastectomy in inpatient settings. This is consistent with prior studies that have described clinicians selecting lower risk patients for treatment in ASCs.^23–25^ Our study is, however, not able to disentangle the numerous factors that fuel site of care selection including clinician judgement and patient preference. For example, site of care selection often involves a collaborative process between anesthesiologists and surgeons and may be further influenced by institutional protocols that dictate who may or may not receive care in ambulatory surgery centers. Clinicians preferentially selecting prefrail and frail patients to undergo inpatient surgery may represent a laudable attempt to provide greater support to patients, however, inpatient stays in frail older adults are also linked to delirium, which can independently confer significant long term consequences.^26^ Prior work has already highlighted potential cost savings and psychological benefits of ambulatory surgery for breast cancer in all patients.^27^ Our results suggest that prefrail and frail patients are potentially being over-triaged to undergo inpatient care.

We also observed an association between race and inpatient surgery, with non-White patients more likely to undergo inpatient surgery. This impact was stronger among robust patients, where Black, Hispanic, and patients identified as “Other” were all more likely to undergo inpatient surgery. Among pre-frail and frail patients, however, a statistically significant impact was seen only among Black patients. These findings suggest that frailty may be a greater driver for site of care than race in certain groups.

Overall rates of transfer to an acute care hospital following ASC mastectomy were exceptionally low. While patients with prefrailty or frailty had statistically significantly higher odds of transfer, rates of transfer remained exceptionally low (<0.5%). These low rates of transfer to acute care at time of ASC discharge are similar to previously reported figures.^20–22^ Low rates of transfer, even among prefrail and frail patients, are reassuring and support the expansion of ASC care within this category of older adults.

Average costs of care were higher for inpatient compared to ambulatory mastectomy, aligned with the cost savings associated with ambulatory breast cancer care seen in other studies.^6,27^ Our study adds to the literature by describing differences in expected costs, taking into account transfer to acute care following ASC surgery to create a more nuanced understanding of total costs associated with ASC care. Given low rates of transfer, the cost savings associated with ASC care remained significant. Our analyses further demonstrate that, taking current rates of transfer as given, health systems have significant opportunities to economically optimize site of care, including for prefrail and frail older adults, by expanding access to ASC care.

Our study only captures the economic costs associated with care, creating a limited picture of relative advantages or disadvantages of each site of care. Prior work has highlighted the importance of capturing patient perspectives on ASC care, a domain than remains understudied.^28^ Previous research has also highlighted potentially uncaptured benefits, including psychological wellbeing,^27^ or uncaptured costs, like caregiver strain,^29^ associated with ASC care. Future work with a focus on capturing the patient and care partner perspective in the perioperative period, as well as describing features of more successful facilities, may strengthen the argument for tranisitioning more care to ASC settings. Additionally, this analysis does not include facility level detail about the services available at each ASC, which can exhibit variability in terms of proximity to inpatient hospitals, availability of overnight stays, or access to additional services, like radiology, which may influence suitability of ASC care for a given patient.

Our model relies on a number of simplications, including considering care only as inpatient or ASC, which does not capture the role of HOPDs or observation only inpatient stays, and use of a limited number of factors driving site of care selection. Our study is also limited by the shortcomings of a claims based frailty index, which captures comorbidity rather than clinical frailty. Our data were limited; we were not able to assess reasons for transfer nor were we able to assess outcomes related to the post-transfer hospitalization, including length of stay. In utilizing readmission costs as a proxy for transfer costs, we may have either over or underestimated how much a transfer alone would have cost. Transfer costs may vary with distance and outcome of transfer so may not universally apply. Additionally, our analysis did not include an investigation of the impact of primary payor and insurance coverage, which may influence where patients are eligible to receive care, nor were we able to perform state based assessments within either national database. Finally, we assume stable transfer risks with expanding pools of patients, however, if the highest risk patients are currently undergoing inpatient surgery the transfer risk may rise. In sensitivity analyses, however, this rate would need to be over 100x the currently observed rate to violate economic equivalence between sites of care. Shifts in care patterns during the study period, between 2019 and 2022, may have favored ambulatory care, potentially limiting our estimate of potential savings if surgeons were diverting more patients to ASCs who during non-pandemic conditions may be treated inpatient.

Our findings suggest that, at least from an economic perspective, clinicians should consider recommending ambulatory surgery to more prefrail and frail patients. As ASC mastectomy care becomes more common and is established as safe for an increasing portion of the population, states may need to revisit policies mandating inpatient coverage following breast surgery,^30^ weighing the need to ensure appropriate treatment remains available for patients best cared for in the hospital and the availability of ASCs with appropriate capacity including pathology or nuclear medicine for axillary breast surgery. Furthermore, as care for increasingly vulnerable patients shifts to the outpatient setting, developing supports in the community will be imperative.