Bridging the Gap: Implementation and Uptake of AUDIT-C Alcohol Screening for Inpatient Surgery

Abstract

Objective:

To evaluate the feasibility of Alcohol Use Disorders Identification Test-Consumption (AUDIT-C) screening among surgical inpatients following hospital-wide implementation of a quality improvement initiative.

Background:

Risky alcohol use increases postoperative complications, yet alcohol screening for surgery remains inconsistently applied. As a brief, validated screening tool, the AUDIT-C may improve risk assessment and guide timely intervention, but its adoption in surgical settings is understudied.

Methods:

We retrospectively analyzed electronic health record data from inpatient surgical admissions between April 2021 and September 2023. The primary outcome was completion of a valid AUDIT-C score during admission. Scores were categorized as no use (0), low-risk (1–4), moderate-risk (5–8), and high-risk (9–12). Multivariable logistic regression assessed independent associations with AUDIT-C screening completion.

Results:

Among 30,714 encounters, 9540 (31.1%) had a completed AUDIT-C score. Screening completion varied by surgical service (15.2%–52.4%) and was higher in emergent (51.7%) and urgent (48.4%) cases than elective (29.1%) cases. The 3 most frequently performed procedures screened 53.8% (cesarean), 42.8% (free flap), and 11.6% (aortic valve) of patients. In multivariable models, emergent case status [marginal effect (ME): 0.14], urgent case status (ME: 0.09), and Medicare insurance (ME: 0.03) were independently associated with higher odds of screening completion. Among those screened, 6.5% were classified as moderate- or high-risk.

Conclusions:

Perioperative alcohol screening using AUDIT-C is feasible but inconsistently implemented. System-wide integration could support identification of at-risk patients and improve perioperative outcomes.

INTRODUCTION

Alcohol use is a significant contributor to adverse health outcomes worldwide, with higher levels of intake strongly associated with increased mortality.¹ Nevertheless, in the United States, rates of risky alcohol use and alcohol use disorder have steadily increased from 1990 to 2019, and increases surged by more than 20% during the COVID-19 pandemic.^1–3 Within surgical populations, high preoperative alcohol consumption, typically defined as ≥2 drinks/d or Alcohol Use Disorders Identification Test-Consumption (AUDIT-C) ≥5, has been linked to a range of negative outcomes, including increased infection rates, prolonged hospital stays, higher rates of intensive care unit (ICU) admissions, and elevated postoperative mortality.^4–7 These risks underscore the need for systematic screening to identify and mitigate alcohol-related complications in surgical patients.

Despite the well-known negative impact of alcohol intake on surgical outcomes, alcohol screening in perioperative settings is poorly characterized, with no universal guideline dictating when, how, or for whom screening should be performed. Some health systems incorporate screenings into preoperative visits, while others rely on informal clinician inquiries or omit screening entirely. Further, while early outpatient screening offers the opportunity to optimize care, it may not capture recent changes in alcohol consumption that influence perioperative risk, and it fails to account for patients undergoing emergent surgeries.^8,9 Same-day or inpatient screening plays a more essential role for patients undergoing emergent surgeries, where preoperative screening is not feasible, and the ability to identify hazardous use is still critical for anticipating withdrawal and managing complications.¹⁰ Additionally, previous studies on alcohol screening for surgery have primarily relied on data collected via mailed questionnaires, often administered up to a year before surgery, limiting their applicability to real-time clinical decision-making.^11,12 Research on structured alcohol screening in surgical populations has largely been conducted within the Veterans Health Administration (VHA) system.^13–16 Outside the VHA, the feasibility of screening remains largely unexamined in perioperative care. The lack of standardized screening approaches and generalizability across previous studies highlights gaps in understanding the feasibility and application of alcohol screening in surgical settings.

To address this gap in understanding alcohol screening feasibility and real-world uptake, we analyzed the recent real-world implementation of a hospital-wide alcohol screening initiative, which provides a unique opportunity to evaluate just-in-time alcohol screening associated with hospital admission for surgery. Unlike prior approaches, this initiative integrates AUDIT-C screening into routine admission workflows within the electronic health record (EHR), enabling an assessment of completion rates across diverse patient and procedural characteristics. This analysis aims to assess the adoption, feasibility, and reach of screening and compare patient characteristics by screening status at the time of hospital admission for surgery.

METHODS

This retrospective study utilized EHR data recorded during routine care, which was extracted from the Data Office of Clinical and Translational Research at the University of Michigan. This study was reviewed by the University of Michigan institutional review board and deemed exempt from the need to obtain patient-informed consent, as it evaluated a system-wide intervention implemented as part of routine clinical care. The study protocol and analytic approach were preregistered.

Study Population

The study population included all patients aged ≥18 years admitted for inpatient care on the same day as their surgical procedure at the University of Michigan Hospital between April 1, 2021 and September 30, 2023, when the screening initiative concluded. Patients were included if the admission date matched the surgery date. Exclusion criteria included admissions lasting <24 hours, given such encounters were unlikely to represent true inpatient surgical admissions. No other exclusions were made.

AUDIT-C Score

The primary outcome was the presence of a complete and valid AUDIT-C score recorded in the EHR during the surgical admission. The AUDIT-C is a concise 3-question screening tool derived from the longer World Health Organization AUDIT and has been shown to be effective in primary care settings.^17,18 AUDIT-C is a valid and reliable screening tool for risky alcohol use recommended for outpatient clinical settings.^18–20

As part of a hospital-wide alcohol screening initiative, AUDIT-C was integrated into the EHR and embedded within the admission navigator used by nursing staff. The initiative was implemented through institutional policy and disseminated via system-wide communications, including staff emails, daily huddles, and education campaigns.

A complete score required responses to all 3 AUDIT-C questions. A partial (completion of 1 or 2 questions) or declined response (ie, provider failed to administer screening, patient declined all questions) to any AUDIT-C question was considered incomplete and categorized as a nonrespondent. AUDIT-C scores were grouped into risk categories for use as follows: no alcohol use (0), low-risk (1–4), moderate-risk (5–8), and high-risk (9–12).¹⁴

Covariates

Additional chart-extracted measures included patient demographic and surgical characteristics. Demographic characteristics included: age in years, sex (male, female, and other/unknown), race (American Indian/Alaska Native, Asian/Pacific Islander, Black, Middle Eastern/North African, Unknown/Other, and White), ethnicity (Hispanic, non-Hispanic, and unknown), insurance type (Medicare, Medicaid, private, and unknown), and relationship status (partnered, nonpartnered, and unknown/other). Surgical characteristics included: hospital site (cardiovascular center, children’s hospital, university hospital, and women’s hospital), procedure type, American Society of Anesthesiologists (ASA) classification score (I, II, III, IV, V, and other/unknown), Elixhauser comorbidity index, and surgical urgency (emergent, urgent, elective, and multiple types).

Statistical Analysis

The primary analysis evaluated the feasibility of implementing the AUDIT-C screening tool by assessing the proportion of patients with completed scores and comparing the characteristics of patients with versus without a recorded AUDIT-C score. Descriptive statistics were used to evaluate the proportion of the sample with complete AUDIT-C scores and the subsequent distribution of scores. Continuous variables were summarized as medians and interquartile ranges (IQRs) and assessed for normality. Ordinal groupings were reported descriptively for interpretability but not used in comparative testing. Given the large sample size, between-group comparisons were conducted using standardized mean differences (SMDs). A threshold of 0.10 was used to indicate a meaningful imbalance. For comparisons restricted to AUDIT-C responders, χ² tests and Wilcoxon rank-sum tests were applied accordingly.

Subgroup analyses were also performed. Characteristics of AUDIT-C responders were stratified by surgical service, procedure type, hospital site, and case urgency (elective, urgent, and emergent procedures). Additionally, we calculated adjusted odds multivariable logistic regression with AUDIT-C completion (yes/no) as the dependent variable. Predictor variables included age, sex, race, insurance type, comorbidity index, ASA class, case urgency, hospital site, and surgical service. Marginal effects (MEs) were reported.

RESULTS

Among 30,714 unique admissions, patients had a median age of 61 years (IQR: 28) with the largest age group being 65–74 years [7713 (25.1%)] (Fig. 1). The sample included admissions for 15,970 (52.0%) female and 14,738 (48.0%) male patients. Most admissions were for patients identified as White [25,385 (82.6%)], followed by Black [2622 (8.5%)], and smaller proportions of other racial groups (Table 1). Most admissions were for Medicare or Medicaid beneficiaries [17,152 (55.8%)], and 11,532 (37.5%) were partnered at the time of surgery. Detailed demographic characteristics, stratified by AUDIT-C screening status, are presented in Table 1.

FIGURE 1.

Study cohort creation. *Among 21,174 nonrespondent encounters, 182 (0.86%) of them had a partial score with only 1 question completed, and 308 (1.45%) of them had a partial score with only 2 questions completed.

TABLE 1.

Completion of a Valid AUDIT-C Score Recorded in the EHR During Surgical Admission by Patient Characteristics

Patient Characteristic	Overall		Alcohol Screening With AUDIT-C						SMDs*
			Yes			No
	(n = 30,714)		[n = 9540 (31.1%)]			[n = 21,174 (68.9%)]
Continuous Variable	Median	IQR	Median	IQR		Median	IQR
Age, mean, in years	57.0	17.7	54.1	18.4		58.2	17.3		0.234
Elixhauser comorbidity index	5.0	5.0	5.0	7.0		6.0	6.0		0.209
Categorical Variable	Frequency	%	Frequency	%	Screening Rate	Frequency	%	Nonresponse Rate
Age category									N/A
18–24	1470	4.8	524	5.5	35.6	946	4.5	64.4
25–34	3234	10.5	1432	15.0	44.3	1802	8.5	55.7
35–44	3421	11.1	1305	13.7	38.1	2116	10.0	61.9
45–54	3745	12.2	1120	11.7	29.9	2625	12.4	70.1
55–64	6232	20.3	1764	18.5	28.3	4468	21.1	71.7
65–74	7713	25.1	2068	21.7	26.8	5645	26.7	73.2
75+	4899	16.0	1327	13.9	27.1	3572	16.9	72.9
Sex								100.0	0.147
Male	14,738	48.0	4098	43.0	27.8	10,640	50.3	72.2
Female	15,970	52.0	5440	57.0	34.1	10,530	49.7	65.9
Unknown/other	6	0.0	2	0.0	33.3	4	0.0	66.7
Race								100.0	0.084
White/Caucasian	25,385	82.6	7680	80.5	30.3	17,705	83.6	69.7
Black/African American	2622	8.5	914	9.6	34.9	1708	8.1	65.1
Asian/Pacific Islander	731	2.4	269	2.8	36.8	462	2.2	63.2
American Indian/Alaska Native	132	0.4	38	0.4	28.8	94	0.4	71.2
Middle Eastern/North African	70	0.2	26	0.3	37.1	44	0.2	62.9
Unknown/other	1774	5.8	613	6.4	34.6	1161	5.5	65.4
Ethnicity								100.0	0.021
Hispanic	921	3.0	305	3.2	33.1	616	2.9	66.9
Non-Hispanic	28,822	93.8	8920	93.5	30.9	19,902	94.0	69.1
Unknown	971	3.2	315	3.3	32.4	656	3.1	67.6
Insurance type								100.0	0.183
Medicare	13,430	43.7	3750	39.3	27.9	9680	45.7	72.1
Medicaid	3722	12.1	1352	14.2	36.3	2370	11.2	63.7
Private	12,898	42.0	4098	43.0	31.8	8800	41.6	68.2
Unknown	664	2.2	340	3.6	51.2	324	1.5	48.8
Relationship status								100.0	0.069
Partnered	11,532	37.5	3376	35.4	29.3	8156	38.5	70.7
Nonpartnered	7220	23.5	2262	23.7	31.3	4958	23.4	68.7
Unknown/other	11,962	38.9	3902	40.9	32.6	8060	38.1	67.4
Elixhauser comorbidity index score								100.0	N/A
0	1408	4.6	617	6.5	43.8	791	3.7	56.2
1–5	14,766	48.1	5063	53.1	34.3	9703	45.8	65.7
6–13	12,762	41.6	3389	35.5	26.6	9373	44.3	73.4
14+	1778	5.8	471	4.9	26.5	1307	6.2	73.5

^*Standardized mean difference comparing the mean or proportion of the characteristics among the group with an AUDIT-C score versus the group without and AUDIT-C score, using 0.10 to indicate meaningful imbalance.

EHR indicates electronic health record.

AUDIT-C Screening Completion (Primary Outcome)

A valid, complete AUDIT-C score was recorded for 9540 (31.1%) admissions, while 21,174 (68.9%) admissions did not complete AUDIT-C screening (Table 1). Among nonrespondent encounters, 182 (0.86%) of them had a partial score with only 1 question completed and 308 (1.45%) of them had a partial score with only 2 questions completed; the rest (20,684, 97.7%) either declined responses or were not screened by their provider. Reasons for nonresponse were not collected.

Between-group comparisons revealed that admissions with screening had a higher proportion of female patients (57.0% vs 49.7%; SMD = 0.15). On average, screened patients were younger than those not screened (54.1 vs 58.2 years; SMD = 0.23). Younger patients, particularly those aged 25–34 and 35–44 years, had the highest screening rates (44.3%), whereas screening was lowest among those aged 65–74 years (26.8%) and 75 years and older (27.1%) (Table 1).

Insurance status was also associated with differences in screening completion. Patients with unknown insurance had the highest screening rate (51.2%, SMD = 0.428), followed by those with Medicaid (36.3%, SMD = 0.127), private insurance (29.6%, SMD = 0.026), and Medicare (26.0%, SMD = 0.121) (Table S1, https://links.lww.com/AOSO/A569). Racial and ethnic differences were observed: 34.9% of Black patients, 30.3% of White patients, and 28.8% of American Indian/Alaska Native patients completed screening (Table S1, https://links.lww.com/AOSO/A569). Additionally, AUDIT-C scores were recorded in 43.8% of admissions for patients with Elixhauser comorbidity index scores of 0, compared to 34.3% for scores 1–5, 26.6% for scores 6–13, and 26.5% for scores ≥14 (SMD = 0.21).

Screening Completion by Surgical Characteristic

Screening completion varied by service, ranging from 15.0% for surgical admissions in internal medicine to 52.4% for Obstetrics and Gynecology (SMD = 0.59) (Table 2). Other high-screening specialties included ophthalmology (49.3%), acute care surgery (43.5%), and dental/oral surgery (37.5%). Cardiac (15.2%), internal medicine (15.0%), thoracic (17.7%), and vascular surgery (15.4%) had the lowest screening rates. Statistical differences in screening rates within the responder population are reported in Table S2, https://links.lww.com/AOSO/A569. Site-specific screening mirrored this variability, with the cardiovascular center reporting the lowest screening completion (15.1%, SMD = 0.55) and the women’s hospital the highest (53.8%, SMD = 0.52). Comprehensive surgical characteristics are detailed in Table 2.

TABLE 2.

Completion of a Valid AUDIT-C Score Recorded in the EHR During Surgical Admission by Surgical Characteristics

Surgical Characteristics	Overall		Has an AUDIT-C Score Recorded			Has no AUDIT-C Score Recorded			SMD*
	(n = 30,714)		[n = 9540 (31.1%)]			[n = 21,174 (68.9%)]
	Frequency	%	Frequency	%	Screening Rate	Frequency	%	Nonresponse Rate
Location of procedure									0.592
University Hospital	16,514	53.8	5968	62.6	36.1	10,546	49.8	63.9
Cardiovascular center	9861	32.1	1492	15.6	15.1	8369	39.5	84.9
Women’s Hospital	2532	8.2	1361	14.3	53.8	1171	5.5	46.2
Children’s Hospital	1793	5.8	714	7.5	39.8	1079	5.1	60.2
Multiple locations	14	0.0	5	0.1	35.7	9	0.0	64.3
Surgical service									0.593
Internal medicine†	4679	15.2	700	7.3	15.0	3979	18.8	85.0
Obstetrics and Gynecology	3382	11.0	1772	18.6	52.4	1610	7.6	47.6
Orthopedic surgery	3164	10.3	1080	11.3	34.1	2084	9.8	65.9
Neurosurgery	3106	10.1	1176	12.3	37.9	1930	9.1	62.1
Cardiac surgery	2401	7.8	366	3.8	15.2	2035	9.6	84.8
General surgery‡	2164	7.0	788	8.3	36.4	1376	6.5	63.6
Radiology	1700	5.5	584	6.1	34.4	1116	5.3	65.6
Thoracic surgery	1558	5.1	276	2.9	17.7	1282	6.1	82.3
Otolaryngology	1537	5.0	569	6.0	37.0	968	4.6	63.0
Urology surgery	1508	4.9	552	5.8	36.6	956	4.5	63.4
Colorectal surgery	1232	4.0	353	3.7	28.7	879	4.2	71.3
Vascular surgery	1124	3.7	173	1.8	15.4	951	4.5	84.6
Plastic surgery	857	2.8	306	3.2	35.7	551	2.6	64.3
Dental surgery, oral surgery	739	2.4	277	2.9	37.5	462	2.2	62.5
Minimally invasive surgery	567	1.8	163	1.7	28.7	404	1.9	71.3
Acute care surgery	531	1.7	231	2.4	43.5	300	1.4	56.5
Pediatric surgery	164	0.5	48	0.5	29.3	116	0.5	70.7
Pediatric internal medicine§	154	0.5	63	0.7	40.9	91	0.4	59.1
Ophthalmology	75	0.2	37	0.4	49.3	38	0.2	50.7
Podiatric surgery	6	0.0	1	0.0	16.7	5	0.0	83.3
Anesthesia	2	0.0	0	0.0	0.0	2	0.0	100.0
Multiple services	64	0.2	25	0.3	34.4	39	0.2	65.6

^*Standardized mean difference comparing the mean or proportion of the characteristics among the group with an AUDIT-C score versus the group without and AUDIT-C score, using 0.10 to indicate meaningful imbalance.

^†Internal medicine includes cardiology, gastrointestinal, nephrology, and pulmonary.

^‡General surgery includes endocrine, hepatopancreatobiliary, oncology, and transplant surgery.

^§Pediatric internal medicine includes pediatric cardiology, pediatric GI medicine, pediatric hematology/oncology, pediatric nephrology, and pediatric pulmonary medicine.

EHR indicates electronic health record.

Screening Completion by Case Status, Anesthesia Type, and Procedure Frequency

Among patients with a recorded AUDIT-C score, 14.7% were emergent or urgent cases, compared to 6.6% among those without a recorded score (SMD = 0.27) (Table 3). Similarly, a higher proportion of patients with recorded AUDIT-C scores underwent regional anesthesia (13.6%) compared to those without recorded scores (7.5%) (SMD = 0.26). General anesthesia was the most common technique in both groups, but was slightly less frequent among those with a recorded AUDIT-C score (71.3% vs 75.4%).

TABLE 3.

Completion of a Valid AUDIT-C Score Recorded in the EHR During Surgical Admission by Case Status and Anesthesia Status

Surgical Characteristics	Overall		Has an AUDIT-C Score Recorded			Has no AUDIT-C Score Recorded			SMD*
	(n = 30,714)		[n = 9540 (31.1%)]			[n = 21,174 (68.9%)]
	Frequency	%	Frequency	%	Screening Rate	Frequency	%	Nonresponse Rate
Case status									0.265
Elective	27,791	90.5	8097	84.9	29.1	19,694	93.0	70.9
Urgent	1449	4.7	702	7.4	48.4	747	3.5	51.6
Emergent	1351	4.4	699	7.3	51.7	652	3.1	48.3
Multiple statuses	123	0.4	42	0.4	34.1	81	0.4	65.9
Principal anesthesia technique									0.260
General	22,777	74.2	6803	71.3	29.9	15,974	75.4	70.1
Regional	2887	9.4	1293	13.6	44.8	1594	7.5	55.2
MAC	1887	6.1	348	3.6	18.4	1539	7.3	81.6
Moderate sedation	875	2.8	250	2.6	28.6	625	3.0	71.4
Multiple techniques	28	0.1	11	0.1	39.3	17	0.1	60.7
None	778	2.5	290	3.0	37.3	488	2.3	62.7
Unknown/other	1482	4.8	545	5.7	36.8	937	4.4	63.2
ASA class									0.179
I	337	1.1	131	1.4	38.9	206	1.0	61.1
II	5654	18.4	2113	22.1	37.4	3541	16.7	62.6
III	13,369	43.5	3857	40.4	28.9	9512	44.9	71.1
IV	2804	9.1	691	7.2	24.6	2113	10.0	75.4
V	92	0.3	43	0.5	46.7	49	0.2	53.3
Unknown/other	8458	27.5	2705	28.4	32.0	5753	27.2	68.0

^*Standardized mean difference comparing the mean or proportion of the characteristics among the group with an AUDIT-C score versus the group without and AUDIT-C score, using 0.10 to indicate a meaningful imbalance.

EHR indicates electronic health record.

Correspondingly, screening rates varied when stratified by case status and anesthesia technique. Approximately half of emergent (51.7%) and urgent (48.4%) cases recorded AUDIT-C scores. On the other hand, for elective cases, 8097 patients out of 27,791 total cases recorded AUDIT-C scores (29.1%) (Table 3). Screening also differed by anesthesia type, with regional anesthesia associated with the highest completion rate (44.8%) and monitored anesthesia care (MAC) with the lowest (18.4%). Screening completion declined with increasing ASA class, from 38.9% (ASA I) to 24.6% (ASA IV). Statistical significance of within-group screening rate is reported in Table S3, https://links.lww.com/AOSO/A569.

Among the top 10 most common surgical procedures, completion rates varied from 53.8% for cesarean sections to 11.6% for aortic valve replacements (SMD = 0.64), which were the most and second-most common procedures, respectively. The top 10 most common frequencies and completion rates are listed in Table 4.

TABLE 4.

Completion of a Valid AUDIT-C Score Recorded in the EHR During Surgical Admission by Top 10 Most Common Procedures

Primary Procedure (Listed by Overall N)	Overall N	Encounter With Completed AUDIT-C		SMD*
		Frequency	%
C-section	2088	1124	53.83	0.645
Aortic valve repair or replacement	673	78	11.59	−0.558
Free flap – tumor resection	594	254	42.76	−0.380
Total knee arthroplasty	583	181	31.05	0.301
EP ablation atrial fibrillation	565	21	3.72	0.939
Total hip arthroplasty	532	164	30.83	−0.294
Cath transcatheter aortic valve replacement	502	115	22.91	−0.164
Cath left heart catheterization w/coronary angiography	443	234	52.82	0.597
Mitral valve repair or replacement	443	55	12.42	0.108
Exploratory laparotomy	402	197	49.00	0.103

^*Standard mean differences comparing the mean or proportion of the characteristics among the group with an AUDIT-C score versus the group without and AUDIT-C score, using 0.10 to indicate meaningful imbalance.

EHR indicates electronic health record.

Multivariable Analysis of Factors Associated With AUDIT-C Completion

Medicare insurance (ME: 0.03), Elixhauser index (ME: 0.00), emergent case status (ME: 0.14), various surgical services, and urgent case status (ME: 0.09) were each independently associated with significantly higher odds of screening completion (P < 0.001). Age, sex, race, ethnicity, and ASA class did not show statistical significance after adjusting for covariates. All adjusted odds ratios (ORs) are outlined in Table 5.

TABLE 5.

Multivariable Marginal Effects of Factors Associated With Completion of a Valid AUDIT-C Score Recorded in the EHR During Surgical Admission

Characteristic	Marginal Effects for AUDIT-C Completion			P
	Marginal Effects	95% CI
Age, mean, in years	0.00	1.00	1.00	0.340
Sex (reference: female)
Male	0.00	0.95	1.07	0.738
Unknown/other	0.02	0.19	6.03	0.931
Race (reference: White/Caucasian)
Black/African American	0.00	0.92	1.11	0.790
Asian/Pacific Islander	0.01	0.88	1.22	0.628
Middle Eastern/North African	0.03	0.68	1.90	0.619
American Indian/Alaska Native	−0.03	0.58	1.27	0.456
Unknown/other	0.00	0.89	1.13	0.937
Ethnicity (reference: non-Hispanic)
Hispanic	−0.03	0.75	1.02	0.085
Unknown	0.01	0.88	1.20	0.721
Insurance type (reference: private)
Medicare	0.03	1.07	1.23	<0.001
Medicaid	0.02	1.01	1.19	0.036
Unknown	0.05	1.11	1.55	0.001
Relationship status (reference: partnered)
Nonpartnered	0.00	0.95	1.09	0.616
Unknown/other	0.03	1.11	1.25	<0.001
Elixhauser comorbidity index score	0.00	0.98	0.99	<0.001
Location of procedure (reference: University Hospital)
Cardiovascular center	−0.18	0.32	0.46	<0.001
Mott Children’s Hospital	−0.01	0.84	1.13	0.686
Von Voitlander Women’s Hospital	0.04	0.92	1.49	0.200
Multiple locations	−0.05	0.23	2.66	0.703
Surgical service (reference: obstetrics and gynecology)
Acute care surgery	−0.12	0.45	0.74	<0.001
Cardiac surgery	−0.19	0.30	0.52	<0.001
Colorectal surgery	−0.18	0.35	0.53	<0.001
Dental surgery, oral surgery	−0.12	0.46	0.71	<0.001
General surgery*	−0.11	0.49	0.73	<0.001
Internal medicine†	−0.24	0.23	0.39	<0.001
Minimally invasive surgery	−0.17	0.35	0.58	<0.001
Neurosurgery	−0.11	0.50	0.73	<0.001
Ophthalmology	−0.08	0.43	1.15	0.160
Orthopedic surgery	−0.13	0.45	0.66	<0.001
Otolaryngology	−0.12	0.47	0.71	<0.001
Pediatric internal medicine†, ‡	−0.07	0.52	1.07	0.111
Pediatric surgery	−0.18	0.30	0.62	<0.001
Plastic surgery	−0.12	0.46	0.73	<0.001
Podiatric surgery	−0.34	0.01	1.15	0.066
Radiology	−0.13	0.44	0.68	<0.001
Thoracic surgery	−0.16	0.38	0.63	<0.001
Urology surgery	−0.12	0.47	0.71	<0.001
Vascular surgery	−0.17	0.34	0.62	<0.001
Multiple services	−0.11	0.31	1.15	0.123
Principal anesthesia technique (reference: general)
Regional	0.00	0.90	1.16	0.692
MAC	0.02	0.98	1.31	0.096
Moderate sedation	0.04	0.98	1.46	0.071
Multiple techniques	0.32	0.59	3.43	0.436
None	−0.05	3.72	5.51	<0.001
Unknown/other	0.07	0.64	0.90	0.002
ASA class (reference: I)
II	0.00	0.78	1.25	0.899
III	0.00	0.79	1.28	0.943
IV	0.07	1.07	1.81	0.013
V	0.14	1.16	3.18	0.011
Unknown/other	0.01	0.83	1.34	0.677
Case status (reference: scheduled)
Emergent	0.14	1.75	2.24	<0.001
Urgent	0.09	1.39	1.76	<0.001
Multiple statuses	0.01	0.67	1.58	0.887

Standard mean differences comparing the mean or proportion of the characteristics among the group with an AUDIT-C score versus the group without and AUDIT-C score, using 0.10 to indicate a meaningful imbalance.

^*General surgery includes endocrine, hepatopancreatobiliary, oncology, and transplant surgery.

^†Internal medicine includes cardiology, gastrointestinal, nephrology, and pulmonary.

^‡Pediatric internal medicine includes pediatric cardiology, pediatric GI medicine, pediatric hematology/oncology, pediatric nephrology, and pediatric pulmonary medicine.

EHR indicates electronic health record.

Distribution of AUDIT-C Scores

Among patients with a valid AUDIT-C score (n = 9540), 3660 (38.4%) reported some level of alcohol use, with the largest group [3041 (31.9%)] classified as low-risk. A smaller proportion of patients fell into the moderate-risk [488 (5.1%)] and high-risk [131 (1.4%)] categories. The median AUDIT-C score among all screened patients was 0, with a mean score of 1.12 (SD: 1.97). The full distribution of AUDIT-C scores is shown in Table 6 and Figure 2.

TABLE 6.

Distribution of Valid, Complete AUDIT-C Score Recorded in the EHR During Surgical Admission Among Study Cohort (N = 9540)

AUDIT-C Score	N	% of AUDIT-C Responders
0	5880	61.6%
1	1164	12.2%
2	816	8.6%
3	586	6.1%
4	475	5.0%
Total 1–4	3041	31.9%
5	219	2.3%
6	123	1.3%
7	77	0.8%
8	69	0.7%
Total 5–8	488	5.1%
9	43	0.5%
10	42	0.4%
11	24	0.3%
12	22	0.2%
Total 9–12	131	1.4%

EHR indicates electronic health record.

FIGURE 2.

Distribution of AUDIT-C scores. *This figure represents the distribution of AUDIT-C scores among surgical admissions with a valid AUDIT-C recorded during hospitalization (N = 9540). Scores were categorized into 4 risk strata based on standard cutoffs: No risk (AUDIT-C: 0), low-risk (AUDIT-C: 1–4), moderate-risk (AUDIT-C: 5–8), and high-risk (AUDIT-C: 9–12).

DISCUSSION

In this analysis of real-world data on the implementation of perioperative AUDIT-C screening among surgical patients admitted to a large tertiary care system, approximately 1 in 3 surgical inpatients (31.1%) were successfully screened. Multivariable analysis identified several factors that were independently associated with AUDIT-C screening completion, underscoring the multifactorial nature of screening uptake. Emergent case status (OR: 1.98), urgent case status (OR: 1.56), Medicare insurance (OR: 1.15), and Elixhauser comorbidity index (OR: 0.99) had significant ORs, which aligns with our prior discussion. Together, these findings reinforce that successful integration of alcohol screening in perioperative care is shaped by a complex interplay of demographic, procedural, and organizational drivers, rather than any single factor alone.

These findings align with broader implementation efforts, suggesting that standardized alcohol screening using the AUDIT-C is both feasible and effective when integrated into routine workflows. For example, a multihospital study in London embedded AUDIT-C into admission workflows with automated referrals and educational materials, resulting in high uptake and sustainability.²¹ Similarly, Fernandez et al. reported that up to 95% of eligible patients record an AUDIT-C score when EHR systems incorporated screening with support from nursing workflows.^22,23 Despite the adaptability of AUDIT-C, our results reveal persistent disparities in screening completion, highlighting areas where further attention is needed. In our study, patient characteristics significantly influenced AUDIT-C screening completion. Although unadjusted analyses suggested that older adults (≥65 years) had lower screening rates than younger patients, age was not independently associated with screening completion in the multivariable model. This finding indicates that observed age-related differences may be partly attributable to other factors, such as surgical service, case urgency, or comorbidity burden. Nonetheless, consistent screening in older populations remains important given their elevated risk for alcohol-related complications.²⁴ National data show that 12.0% of adults ≥65 reported binge drinking and 3.3% reported heavy alcohol use in the past month, underscoring the ongoing need for risk assessment in this group.^25–29 Stigma and discomfort discussing alcohol use may also be heightened in older patients, potentially contributing to incomplete or deferred screening.^19,27,30 Similarly, while raw comparisons showed decreasing screening rates with higher ASA class (from 38.9% in ASA I to 24.6% in ASA IV), ASA class was not an independent predictor of screening in multivariable analyses, suggesting that organizational and workflow factors may have a greater influence on screening completion than physiological status alone. While this could reflect clinical prioritization of medically complex patients and less time for ancillary screening, it may also signal missed opportunities to identify risky alcohol use among patients at elevated baseline risk for postoperative complications. These findings underscore the need for consistent, system-level screening workflows that are not deprioritized in the context of either low-acuity or high-acuity surgical cases. In contrast, Medicare and Medicaid beneficiaries had significant MEs of being screened, consistent with national estimates showing a higher prevalence of alcohol use disorder (5.6%) and greater alcohol-related healthcare utilization, including hospitalizations, Emergency Department visits, and barriers to follow-up care.^31–35 Patients with an “unknown” insurance designation had the highest screening rate of 51.2% (ME: 0.05). When insurance status is unknown, care teams may prioritize comprehensive risk assessment, such as alcohol screening, to ensure holistic evaluation and mitigate potential health risks in the absence of a complete social or medical history.

From an organizational standpoint, screening varied widely across surgical services, reflecting a multifaceted set of clinical, structural, and systemic influences. 490 (2.3%) patients had partial responses, which may reflect patient discomfort, time constraints, or inconsistent administration, offering a potential proxy for patient- or workflow-level barriers. In particular, the absence of a “hard stop” in the EHR for AUDIT-C screening, unlike required fields for other assessments, was identified by hospital leadership as a key implementation barrier. Without a hard stop, staff could bypass the AUDIT-C questions without documenting a reason. Leadership reported that screening was frequently missed in cases where patients were unable to respond at the time of admission, such as those recovering from anesthesia, with altered mental status, or admitted to the ICU, and that no automated reminders prompted staff to return to complete the screening once patients stabilized. This gap was further substantiated by variability in how admission documentation was completed. For example, patients unable to respond at the time of admission due to anesthesia recovery, altered mental status, or ICU transfer were often missed. Even among the most common procedures, screening rates varied from nearly half Cesarean-sections (53.8%), to 1 in 9 aortic valve replacement patients (11.6%). These differences likely reflect variation in unit-level workflows, patient acuity, and perceived relevance of alcohol screening to postoperative outcomes. For example, behavioral health screening may be more culturally embedded in services like OBGYIN, which often use multidisciplinary care models, compared to more procedurally focused specialties. Obstetrics and Gynecology units may prioritize comprehensive behavioral risk assessment given potential fetal or neonatal implications, while cardiac units may prioritize physiological stabilization, especially in urgent or complex surgical cases. Procedures such as hysterectomies and free flap reconstructions also showed moderate screening rates, possibly reflecting service-level commitment to holistic preoperative assessment. Fostering a culture of routine preventive screening may have positive spillover effects.²² Establishing standardized workflows across services could help reinforce a culture that prioritizes comprehensive risk assessment and early intervention, ultimately enhancing perioperative care and patient safety.

Feasibility in this study reflects successful integration rather than universal completion; over 9500 AUDIT-C scores were documented without external incentive, embedded into perioperative admission workflows across more than 30 surgical services. This successful integration of AUDIT-C screening in perioperative settings has important clinical implications. Although most surgical patients reported no or low-risk alcohol use, a meaningful subset of 6.5% admissions had patients who were screened as moderate- or high-risk, highlighting a critical opportunity for targeted intervention. While preoperative screening may offer the advantage of informing intraoperative management and facilitating optimization before surgery, postoperative or same-day inpatient screening still provides considerable clinical value. In unscheduled or emergent cases where advance screening is not feasible, inpatient screening can support the early recognition and management of alcohol withdrawal, guide postoperative monitoring, and serve as a touchpoint for longer-term engagement in alcohol use treatment. These interventions may mitigate alcohol-related complications including impaired wound healing, infection, prolonged hospitalization, or acute alcohol withdrawal, which can significantly affect surgical recovery.^36–38 These implications are particularly relevant in emergent and urgent surgical settings. However, mandatory risk assessments in the EHR, such as those for fall risk and infection control, can create opportunity costs, where prioritizing 1 screening may divert resources from another. Thus, time-sensitive environments often rely on biochemical tests like blood or urine alcohol levels to detect acute intoxication but may overlook chronic high-risk alcohol use, which poses persistent perioperative risks. Despite logistical challenges in these scenarios, incorporating AUDIT-C could support clinicians in anticipating complications and tailoring perioperative management.

This study has several limitations. First, the analysis considers the conduct of a single tertiary care academic health system, which may limit generalizability to other settings with different patient populations, resources, and clinical workflows. The analysis also focused on inpatient surgical admissions, where screening practices may differ from outpatient and ambulatory procedures. As surgical care continues to shift toward outpatient settings, further evaluation of screening feasibility in these environments will be required. It is also important to note that the AUDIT-C cutoff of ≥5, used in this study to define moderate- or high-risk drinking, was originally validated among male surgical patients within the VHA.¹³ While this threshold has been shown to correlate with adverse postoperative outcomes in that population, it may not capture risk as accurately among women, who often experience alcohol-related harms at lower levels of consumption. Optimal AUDIT-C risk stratification for women in surgery is unknown. Future research should evaluate whether sex-specific AUDIT-C thresholds are warranted in surgical populations to ensure accurate risk stratification and similar intervention across patient groups.^39,40 Additionally, this study examined screening uptake but did not assess associations between AUDIT-C scores and postoperative outcomes or whether screening led to clinical intervention. Finally, although we identified differences in screening rates across patient demographics and surgical services, we did not explore the underlying drivers; future research should investigate structural, provider-level, and patient-related factors that influence screening.

CONCLUSION

This study demonstrates that the AUDIT-C is a feasible tool for alcohol screening in perioperative clinical settings. While screening completion varied substantially across surgical services and patient subgroups, adjusted analyses indicate that differences were driven by a complex set of system- and workflow-level factors, with some higher-risk groups (Medicaid and Medicare, emergent and urgent) achieving higher completion rates. These patterns highlight that variability in screening is not solely attributable to patient demographics but also reflects organizational priorities, service-specific culture, and the degree of EHR integration. Sustaining and expanding implementation will require standardized, system-wide processes that ensure all surgical patients, across services, insurance types, and patient populations, receive consistent screening. Without such standardization, alcohol-related risks may go unrecognized, leading to missed opportunities to improve surgical outcomes. Future efforts should expand implementation, address screening disparities, and link results to targeted interventions to mitigate alcohol-related surgical risks and improve outcomes.

ACKNOWLEDGMENTS

This project was supported by the Michigan Department of Health and Human Services under award E20242492-002. The authors would like to thank the Data Office for Clinical and Translational Research (DOCTR) at the University of Michigan for their assistance with data acquisition and extraction. We also acknowledge the contributions of the care teams at the University of Michigan who performed screening for patients admitted to the hospital. No medical writer or editorial assistant was used in the preparation of this manuscript.

Supplementary Material

as9-7-e647-s001.pdf ^{(594.3KB, pdf)}