Underutilization of Guideline-Concordant Smoking Cessation Treatments in Surgical Patients: Lessons From a Learning Health System

INTRODUCTION

Cigarette smoking is the leading cause of preventable death in the United States¹. Compared to nonsmokers, individuals who smoke at the time of surgery have a significantly higher risk of postoperative complications and mortality². Preoperative guideline-concordant smoking cessation interventions (ie behavioral support and/or pharmacotherapy^3,4) have been shown to increase short-term abstinence from smoking and consequently to decrease the risk of postoperative complications⁵. Despite this, it is unclear how frequently these therapies are prescribed to surgical patients, especially since prior studies have demonstrated variability in smoking prevalence and treatment patterns across various medical subspecialties⁶. Learning health systems, through which real-world data generation is translated into actionable knowledge-based practice, can address such gaps while improving patient outcomes and experiences⁷. The objective of this study is to describe the smoking prevalence and treatment rates across surgical specialties.

METHODS

We performed a cross-sectional study of adult patients seen in outpatient surgery clinics in 2019 (available in supplemental material, http://links.lww.com/AOSO/A106) and 2020 at Barnes Jewish Hospital (St. Louis, MO), one of the largest tertiary care academic hospitals in the United States. Our institution, in conjunction with our cancer center (Siteman Cancer Center, St. Louis, MO), instituted the Electronic Health Record-Enabled Evidence-based Smoking Cessation Treatment (ELEVATE) program as part of the National Cancer Institute (NCI) Cancer Moonshot Project and the Cancer Center Cessation Initiative (C3I)⁸. This program is a hospital-wide, electronic health record (EHR)-based tool for addressing smoking cessation with demonstrated efficacy⁸. As part of this program, we prospectively collect several longitudinal data elements on patient smoking behaviors and treatment compliance in a centralized database for quality improvement purposes. The study protocol was approved by the Washington University in St. Louis Human Research Protection Office and Institutional Review Board.

From this dataset, we extracted various data elements from our EHR (Epic, Verona, WI) including age, sex, race, number of comorbidities, and the surgical specialty by which the patient was seen. We also extracted several smoking-related variables, including whether a tobacco assessment occurred (ie, if the patient was asked about smoking) and the results of that assessment (ie, currently smoking); whether smoking deterrent medications (nicotine-replacement therapy, varenicline, and bupropion³) were prescribed and/or documented (within the same year); and whether behavioral support was given. We assessed the medication and behavioral support interventions together as our primary composite outcome (ie, “any treatment”). Detailed data collection methods have been described previously (additional information available in supplemental material, http://links.lww.com/AOSO/A106)⁸. We used multivariable logistic regression analyses to assess factors associated with our primary outcome. Analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC).

RESULTS

A total of 164,673 unique patients were seen in outpatient surgery clinics in 2020. The overall smoking assessment rate was 96.1% (n = 158,212, Table 1). Of those patients assessed, the overall smoking prevalence was 14.7% (n = 23,276). Smoking prevalence was highest in trauma (25.8%) and vascular surgery (25.1%) and lowest in transplant surgery (9.2%) and surgical oncology (9.9%).

TABLE 1.

Smoking Prevalence, Assessment Rates, and Treatment Rates Across Surgical Specialties, 2020

	Number of Patients	Assessment		Smoking		Medication		Behavioral Support		Any Treatment*
Surgical Specialties	N	n	%	n	%	N	%	n	%	N	%
Neurosurgery	11112	10614	95.5	2041	19.2	236	11.6	1146	56.1	1275	62.5
Cardiothoracic	6203	6004	96.8	1176	19.6	265	22.5	307	26.1	509	43.3
Surgical oncology	9302	9257	99.5	913	9.9	128	14.0	301	33.0	389	42.6
Vascular surgery	8551	8366	97.8	2098	25.1	349	16.6	612	29.2	851	40.6
Urology	15923	14891	93.5	2361	15.9	292	12.4	695	29.4	911	38.6
Plastic surgery	6907	6575	95.2	1255	19.1	151	12.0	332	26.5	449	35.8
Hepatobiliary surgery	2731	2579	94.4	379	14.7	59	15.6	85	22.4	133	35.1
Minimally invasive surgery	4837	4776	98.7	568	11.9	117	20.6	92	16.2	199	35.0
Transplant	5894	5472	92.8	503	9.2	94	18.7	85	16.9	166	33.0
Gynecologic oncology	6499	6280	96.6	802	12.8	117	14.6	152	19.0	257	32.0
Colon and rectal surgery	4783	4700	98.3	761	16.2	97	12.7	146	19.2	220	28.9
Otolaryngology	29059	27849	95.8	3622	13.0	430	11.9	688	19.0	1043	28.8
Trauma surgery	1812	1646	90.8	424	25.8	66	15.6	66	15.6	119	28.1
General surgery	12915	12837	99.4	2485	19.4	278	11.2	431	17.3	672	27.0
Orthopedic surgery	58007	56198	96.9	6716	12.0	813	12.1	891	13.3	1593	23.7
Overall	164673	158212	96.08	23276	14.71	2954	12.7	5014	21.5	7383	31.7

^*Any treatment is defined as patients receiving medication and/or behavioral support.

Among individuals who were smoking, cessation pharmacotherapy was provided to 2954 (12.7%) patients and behavioral support was provided to 5014 (21.5%) patients. The overall tobacco treatment rate (“any treatment”) was 31.7% (n = 7383). Any treatment was highest in neurosurgery (62.5%) and cardiothoracic surgery (43.3%) and lowest in orthopedic (23.7%) and general surgery (27.0%). Factors associated with receiving treatment in multivariable analyses were older age (70+ years old vs. 18–49, adjusted odds ratio [aOR] 2.04, 95% CI 1.87–2.23, Table 2), female sex (male vs. female, aOR 0.83, 95% CI 0.79–0.88), and higher comorbidity quartile (eg, Q4 vs. Q1, aOR 2.74, 95% CI 2.55–2.95).

TABLE 2.

Multivariable Analysis of Factors Associated with Receiving any Cessation Treatment, 2020

	Number of Patients	Assessment		Smoking		Medication		Behavioral Support		Any Treatment*		Univariable(OR, 95% CI)	Multivariable (OR, 95% CI)†
Age
18–49	47107	44924	95.37%	8025	17.86%	783	9.76%	1315	16.39%	1967	24.51%	[1 Reference]	[1 Reference]
50–59	32671	31387	96.07%	5868	18.39%	947	16.14%	1133	19.31%	1890	32.21%	1.46 (1.36–1.58)	1.38 (1.28–1.49)
60–69	42860	41306	96.37%	6030	14.31%	906	15.02%	1464	24.28%	2170	35.99%	1.73 (1.61–1.86)	1.62 (1.50–1.75)
70+	42035	40595	96.57%	3353	8.12%	318	9.48%	1101	32.84%	1356	40.44%	2.09 (1.92–2.28)	2.04 (1.87–2.23)
Sex
Female	93494	90245	96.52%	11801	12.85%	1677	14.21%	2602	22.05%	3965	33.60%	[1 Reference]	[1 Reference]
Male	71167	67958	95.49%	11472	16.65%	1277	11.13%	2410	21.01%	3417	29.79%	0.84 (0.79–0.89)	0.83 (0.79–0.88)
Race
White	135725	130448	96.11%	17588	13.25%	2056	11.69%	3740	21.26%	5460	31.04%	[1 Reference]	[1 Reference]
Black	24699	23863	96.62%	5321	22.09%	862	16.20%	1215	22.83%	1831	34.41%	1.17 (1.09–1.24)	1.04 (0.97–1.11)
Other	2984	2830	94.84%	213	7.42%	23	10.80%	37	17.37%	58	27.23%	0.83 (0.61–1.12)	0.84 (0.61–1.15)
Number of Comorbidities
Q1	67723	64201	94.80%	10520	16.19%	976	9.28%	1740	16.54%	2601	24.72%	[1 Reference]	[1 Reference]
Q2	23768	22256	93.64%	3136	13.90%	297	9.47%	622	19.83%	889	28.35%	1.20 (1.10–1.32)	1.22 (1.11–1.33)
Q3	37977	36710	96.66%	4925	13.21%	575	11.68%	1151	23.37%	1621	32.91%	1.49 (1.39–1.61)	1.48 (1.37–1.59)
Q4	35205	35045	99.55%	4695	13.06%	1106	23.56%	1500	31.95%	2272	48.39%	2.85 (2.66–3.07)	2.74 (2.55–2.95)

OR, odds ratio; CI, confidence interval

^*Any treatment is defined as patients who received medication, brief advice, or additional counseling referrals.

^†Modeling any treatment (yes vs. no) adjusting for age, sex, race, and comorbidity quartile.

Analyses using data from 2019 (n = 175934), before the COVID-19 pandemic, yielded similar conclusions (Supplemental Table 1, http://links.lww.com/AOSO/A106).

DISCUSSION

This study examined the smoking prevalence and treatment rates across various surgical specialties using a large EHR-based dataset set, including over 160,000 unique patients seen in outpatient surgery clinics. Despite the relatively high smoking prevalence among surgical patients (14.7%), guideline-concordant treatment rates were very low, with only 12.7% receiving pharmacotherapy and 31.7% receiving any treatment³. This is concerning given the well-established relationship between smoking and multiple adverse outcomes following surgery^2,9. Further, while our study assessed any treatment, some guidelines suggest that both behavioral support and pharmacotherapy should be the standard of care⁴; if so, then the guideline-concordant treatment rates in our study are much lower (<5%). These real-world data demonstrate that smoking treatments vary widely across different surgical specialties and patient groups⁶.

Some have advocated for restricting elective surgeries for patients who smoke, thereby delaying nonurgent procedures until mandatory smoking cessation is achieved (or exhausted)¹⁰. While the ethicality of rationing care in such scenarios is highly debated, perhaps a far more reasonable focus would be for routine implementation of guideline-concordant tobacco treatment programs for patients being considered for surgery. Our data suggest that, despite strong evidence supporting the use of such programs⁵, tobacco treatment among surgical patients remains severely underutilized¹¹. Implementing low-burden interventions that systematically address smoking cessation in surgical clinics may reduce this variation⁸.

This study has several strengths including the acquisition and analysis of large-scale, real-world, EHR-based data. Limitations to this study include the single-institutional study design and a lack of biochemical validation for self-reported smoking status.

In conclusion, guideline-concordant smoking cessation treatments are underutilized in surgery. Addressing disparities in smoking cessation treatments are critical given the disproportionate impact of smoking on surgical outcomes.