Advance care planning billing codes in patients undergoing TAVR is infrequent and associated with adverse TAVR outcomes

Introduction

Transcatheter aortic valve replacement (TAVR) is an emerging technology often used in a geriatric population with multiple comorbid conditions. Alignment of patient health care goals and preferences with actual care provided is particularly important for patients considering TAVR because of the risks of postoperative complications in this older cohort (1). Advance care planning (ACP) – discussions between patients or their surrogates and physicians, other health care providers, and family to determine patient goals and values to guide decision-making – is an important tool to facilitate this alignment.

ACP has been associated with increased patient satisfaction, reduced intensity of care at the end of life, and other patient-centered benefits (2, 3). The Centers for Medicare and Medicaid Services began reimbursing clinicians and allied health professionals for conducting ACP on January 1, 2016. We sought to understand the frequency, timing, and clinical factors associated with billed ACP use in fee-for-service (FFS) Medicare beneficiaries undergoing TAVR.

Methods

Data sources and study population

This study was approved by the Mass General Brigham institutional review board. We examined a national sample of Medicare beneficiaries at least 66 years old with continuous fee-for-service coverage from January 1, 2016 to death or December 31, 2017 undergoing TAVR. Inpatient, outpatient, and carrier files were obtained for analysis. TAVR was defined using International Classification of Disease, Tenth Revision (ICD-10) codes. Patients who underwent TAVR from July 1, 2016 to June 30, 2017 were included to allow for a 6-month lookback and follow-up period for all patients.

Variables

ACP was identified using Current Procedural Terminology (CPT) codes 99497 and add-on code 99498. Only the first ACP code within each patient’s individualized observation window was used for analysis. Timing of ACP was dichotomized as before surgical admission for TAVR versus during or after surgical admission for TAVR. A list of major surgical complications was generated from existing lists with expert review (4)

Statistical Analysis

We performed two primary analyses: 1) to identify factors associated with the presence of ACP in the entire study population and 2) to identify factors associated with the timing of ACP among patients who had at least one ACP. We used univariate and multiple logistic regression models for both. Covariates included in the multiple logistic regression models are listed in Supplemental Table 1.

Results

We identified 2,124,989 unique Medicare beneficiaries who underwent inpatient surgery from July 1, 2016 to June 30, 2017. Of these, 27,154 (1.3%) patients underwent TAVR and met the other criteria outlined above (Table 1). The mean age of the cohort was 82.3 years, 94.0% of patients were White. Most patients lived in either the American South (35.0%) or Northeast (24.8%).

Table 1.

Characteristics of FFS Medicare patients undergoing TAVR by ACP status July 1 2016 - June 30 2017

	Total	%	No ACP	%	Yes ACP	%
Total N	27,154		26,562	97.82%	592	2.18%
Sex
Male	14,638	53.91%	14,354	98.06%	284	1.94%
Female	12,516	46.09%	12,208	97.54%	308	2.46%
Age, Mean (SD)	82.3(6.8)		82.3(6.8)		83.4(6.8)
Length of Stay, Mean (SD)	6.2(5.4)		6.2(5.3)		7.7(6.3)
Race
White	25,520	93.98%	24980	97.88%	540	2.12%
Black	826	3.04%	801	96.97%	25	3.03%
Other	808	2.98%	781	96.66%	27	3.34%
Missing
Region
Northeast	6,732	24.79%	6,566	97.53%	166	2.47%
South	9,513	35.03%	9,292	97.68%	221	2.32%
Midwest	5,980	22.02%	5,914	98.90%	66	1.10%
West	*	*	*	*	*	*
OTHER	*	*	*	*	*	*
Comorbidities
Chronic pulmonary disease	7,638	28.13%	7,467	97.76%	171	2.24%
Congestive heart failure	19,903	73.30%	19,422	97.58%	481	2.42%
Depression	2,023	7.45%	1,971	97.43%	52	2.57%
Diabetes (uncomplicated)	4,680	17.24%	4,592	98.12%	88	1.88%
Diabetes (complicated)	5,223	19.23%	5,089	97.43%	134	2.57%
HIV/AIDS	*	*	*	*	*	*
Metastatic cancer	182	0.67%	*	*	*	*
End stage renal disease on chronic dialysis	1,004	3.7%	962	95.82%	42	4.18%
Solid tumor without metastasis	563	2.07%	549	97.51%	14	2.49%
Dementia	7,172	26.41%	6,951	96.92%	221	3.08%
None	154	0.57%	*	*	*	*
Elixhauser Comorbidity Index
0	154	0.57%	*	*	*	*
1	1105	4.07%	*	*	*	*
2	3,301	12.16%	3,239	98.12%	62	1.88%
3+	22,594	83.21%	22,093	97.78%	501	2.22%
Missing
Complications
Respiratory	4,634	17.07%	4,473	96.53%	161	3.47%
Cardiac	1,875	6.91%	1,810	96.53%	65	3.47%
Infectious	4,249	15.65%	4,076	95.93%	173	4.07%
Surgical Site	88	0.32%	*	*	*	*
Nervous System	1,350	4.97%	1,284	95.11%	66	4.89%
Bleeding/ Embolism/ Thrombosis	1,885	6.94%	1,831	97.14%	54	6.94%
GI	877	3.23%	836	95.32%	41	4.68%
Other	6,271	23.09%	6,069	96.78%	202	3.22%
Mortality up to 6 mos after discharge
Alive	24,854	91.53%	24,357	98.00%	497	2.00%
Dead	2,300	8.47%	2,205	95.87%	95	4.13%
Missing
Number of ACPs during study period
0	26,562	97.82	26,562	100.00%	-	0.00%
1	523	1.93%	-	0.00%	523	100.00%
2	56	0.21%	-	0.00%	56	100.00%
3+	13	0.05%	-	0.00%	13	100.00%
Missing
Timing of First Ever ACP
Before Study Period	118	0.43%	86	72.88%	32	27.12%
During Study Period	560	2.06%	-	0.00%	560	100%
After Study Period	237	0.87%	237	100.00%	-	0.00%
No ACP as of Dec 31 2017	26,239	96.63%	26,239	100.00%	-	0.00%
Missing

^*Cell values less than 11 or those allowing these to be calculated have been suppressed

Of the overall cohort, 83.2% had an Elixhauser comorbidity score of three or higher (Table 1). Congestive heart failure (73.3%), chronic obstructive pulmonary disease (28.1%), and dementia (26.4%) were the three most common comorbidities. Mortality was 8.5% within 6 months of surgical discharge.

In total, 592 (2.2%) received billed ACP; 523 (88.3%) had a single billed ACP while 69 (11.6%) had two or more within the study period (Table 1). Among beneficiaries with a billed ACP, 223 (37.7%) received it in the 6 months prior to their index admission for TAVR, while 369 (62.3%) received it during index admission or within 6 months after discharge from index admission. Characteristics of patients with billed ACP versus patients without billed ACP during the study period are similar (Supplemental Table 1).

Multivariable adjusted analysis demonstrated that receiving billed ACP was associated with complication or death within 6 months of discharge (OR 2.06, 95% CI 1.74–2.44) (Table 2).

Table 2.

Adjusted analysis of factors associated with receipt of billed ACP

	Adjusted OR (95% CI)	p-value
Sex
Male	REF
Female	1.22 (1.03–1.43)	0.02
Age, Years	1.02 (1.08–1.03)	0.0009
Race
White	REF
Black	1.44 (0.96–2.18)	0.08
Other	1.48 (0.99–2.02)	0.05
Region
Northeast	REF
South	0.97 (0.79–1.185)	0.73
Midwest	0.46 (0.34–0.61)	<0.0001
West	1.18 (0.78–1.49)	0.16
Elixhauser Comorbidity Index
Less than 3	REF
3+	0.98 (0.78–1.24)	0.89
Complication and/or death up to 6 months after discharge
No	REF
Yes	2.06 (1.74–2.44)	<0.0001

We repeated our multivariable analysis among patients who received ACP to determine factors associated with timing of ACP after surgical admission (Supplemental Table 2). We found that only complications or death within the 6 months after surgery was associated with ACP occurrence after index admission (OR 1.87, 95% CI 1.31–2.69).

Discussion

Within the last decade, TAVR has become the most common mode of aortic valve replacement for the management of symptomatic aortic stenosis (5). TAVR is common among elderly patients, who have increased risk of adverse outcomes following TAVR compared to younger patients (6). Patients with symptomatic aortic stenosis should be engaged in shared decision making to consider all viable modes of treatment, which is critical in helping patients understand the natural history of disease and the risks and benefits inherent to each treatment approach (7).

In 2016, CMS demonstrated a commitment to ACP by creating codes for reimbursement. Among patients undergoing TAVR, an elderly cohort with a life-limiting condition, ACP may help align patient goals with actual care delivered, thus improving quality of care. Nevertheless we demonstrated low use of ACP billing codes. There are several possible explanations for the low uptake of ACP billing codes: first, ACP is performed but not billed for a variety of reasons including ethical concerns over billing, concerns over charges related to billing, and lack of knowledge about billing (8). Second, healthcare teams may not have the resources to perform ACP within the constructs of clinical care. Third, health system incentives in a fee-for-service model may not align with ACP conversations, especially if de-escalation of care is employed after these conversations.

We also uniquely demonstrated that billed ACP occurrence tended to occur after TAVR, particularly after complications and before death. While ACP at any point may be helpful, pre-surgical ACP is most likely to be beneficial because it allows the patient to consider whether or not TAVR is consistent with goals of care and may reduce unwanted procedures. It is reasonable for billed ACP to be performed again after complications, since patient and family healthcare decisions may change depending on the clinical scenario.

Our findings regarding the low frequency and reactive use of billed ACP are identical to those that our group discovered in patients with dementia undergoing inpatient surgery(9). Taken together, these suggest broader patterns of billed ACP use surrounding surgery that will benefit from additional investigation. Areas of focus may include multidisciplinary approaches to ACP (surgeons and primary care physicians), optimal timing, and the potential for informatics to increase ACP use.

Our study has several important limitations. First, as we already noted, ACP may occur without billing. Analyses of ACP using claims data are based on using billed ACP as a proxy for ACP itself. The authors acknowledge that billed ACP is an imperfect proxy: our own unpublished qualitative work demonstrates that unbilled ACP often occurs in clinical settings. Even given that the majority of ACP is not captured, we feel that our broad conclusions still stand, viz., ACP occurs infrequently. Second, because our study period began immediately after the adoption of ACP billing codes, the increase in ACP billing over time may be related to provider uptake rather than purely reactive to adverse outcomes. Regardless, the association between adverse TAVR outcomes and increased ACP billing code use is notable. Third, we included death as a complication. ACP can affect timing of death while timing of death can affect ACP occurrence, a phenomenon that we cannot clarify. Last, our study was restricted to Medicare beneficiaries above 66 years with fee-for-service coverage. Generalization to non-Medicare fee-for-service beneficiaries and younger patients should be done with caution.

In conclusion, among FFS Medicare beneficiaries, patients undergoing TAVR are older and have multiple medical comorbidities. Billed ACP in this population is nonetheless low, occurs predominantly after TAVR, and is associated with complications, including death. Further investigation is needed to understand current patterns of billed ACP use.

Supplementary Material

supinfo

Supplemental Table 1. Characteristics of FFS Medicare patients undergoing TAVR by timing of ACP July 1 2016 - June 30 2017.

Supplemental Table 2. Adjusted analysis of factors associated with receipt of billed ACP after surgical admission.

Abbreviations List

ACP

Advance care planning

TAVR

Transcatheter aortic valve replacement

ICD-10

International Classification of Disease, Tenth Revision