Abstract
Introduction:
Few studies evaluate the impact of unhealthy alcohol and drug use on the risk and severity of postoperative outcomes following upper gastrointestinal and pancreatic oncologic resections.
Methods:
The National Inpatient Sample was queried to identify patients undergoing total gastrectomy, esophagectomy, total pancreatectomy, and pancreaticoduodenectomy between 2012 and 2015. Unhealthy alcohol and drug use was assessed by ICD-9 and NIS coder designation. Multivariable regression (MVR) was used to identify associations between alcohol and drug use and postoperative complication, length of stay (LOS), hospital cost and mortality.
Results:
59,490 patients met inclusion criteria; 2060 (3.5%) had unhealthy alcohol use; 1265 (2.1%) had unhealthy drug use. Postoperative complication rates were higher in patients with alcohol and drug use than in abstainers (67.5% vs 62.8% vs 57.2%; p < 0.01). On MVR, alcohol use was independently associated with increased risk of a non-withdrawal complication (OR 1.33 [1.05, 1.68]) while alcohol and drug use were independently associated with increased LOS (1.54 [0.12, 2.96]) and 2.22 [0.90, 3.55] days) and cost ($5,471 [$60, $10,881] and $4,022 [$402, $7,643]), but not mortality.
Conclusion:
Unhealthy substance use is associated with increased rates of postoperative complications, prolonged LOS, and costs in patients undergoing major upper gastrointestinal and pancreatic oncologic resections. Screening and abstinence interventions should be incorporated into the preoperative care pathways for these patients.
Introduction
Unhealthy alcohol and drug use affects a substantial proportion of the American population. In the 2018 National Survey on Drug Use and Health, 26.5% of Americans self-reported binge or heavy alcohol use within the month prior to the survey, while 12.0% of Americans reported non-prescribed or illicit drug use.1 This and other recent population based studies suggest that abuse of these substances is underdiagnosed and that the clinical effects of substance use are present in a large and growing proportion of patients.2–5
It is well known that abuse of various substances has a deleterious effect on health, increases risk of specific diseases, and is associated with increased morbidity in surgical patients. The term “unhealthy use” is defined by the United States Preventive Services Task Force (USPSTF) as referring to a spectrum of behavior ranging from regular excess consumption to actually meeting diagnostic criteria for alcohol and drug use disorders.6, 7 A history of alcohol use is associated with increased rates of postoperative complications including delirium, stroke, pneumonia, and surgical site infections, as well as increased rates of mortality following cardiac, orthopedic, and general surgery procedures such as laparoscopic cholecystectomy and hernia repair.8–10 Similarly, chronic drug users experience increased rates of respiratory, cardiovascular, and pain management complications.11 The importance of recognizing unhealthy alcohol and substance use preoperatively and providing abstinence intervention in surgical patients has been well documented.8
Major oncologic operations such as esophagectomy, gastrectomy, and pancreatectomy, are generally characterized by relatively high rates of postoperative morbidity with those rates ranging from 30 to 70% depending on the reported case series.12–14 The morbidity inherent in these procedures would be expected to be worsened by comorbid substance use. Few studies, however, examine the impact of alcohol and drug use on clinical outcomes in patients undergoing upper gastrointestinal and pancreatic oncologic operations. The purpose of the current study is to evaluate the effect of unhealthy alcohol and drug use on outcomes after major upper gastrointestinal and pancreatic oncologic resections in the United States using the Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS) between 2012 and 2015.
Methods
Study Design and Patient Population
NIS is the largest national all-payer (Medicare, Medicaid, Private, and Uninsured) database of discharge data and allows for assessments of national estimates of inpatient utilization. Maintained by the Agency for Healthcare Research and Quality,15 the NIS is a weighted sample drawn from the HCUP State Inpatient Databases by a complex, single-cluster survey design. The NIS was queried to identify patients age 18 years or older undergoing either total gastrectomy (International Classification of Diseases, Ninth Revision, Clinical Modification (ICD9-CM) procedure codes: 43.91 and 43.99), esophagectomy (ICD9-CM procedure code: 42.42), total pancreatectomy (ICD9-CM procedure code: 52.6), or pancreaticoduodenectomy (ICD9-CM procedure codes: 52.51 and 52.7) between 2012 and 2015. Patients were classified as unhealthy alcohol users or drug users by means of NIS comorbidity codes or ICD9-CM codes associated with use of alcohol or drugs (Supplemental Table 1). The various drugs included in analysis indicated by ICD-9CM code designations were amphetamines, cannabis, cocaine, opioids, hallucinogens, and diagnoses of drug abuse and dependence with unspecified substances. A small number of patients were identified as having codes for both alcohol and drug use. Given the small size of this cohort, any analysis involving them would be underpowered and these records were excluded from analysis. Similarly, records were excluded if cost, length of stay (LOS), or mortality data were missing.
Study Variables
Survey-adjusted prevalence of comorbid conditions were reported and evaluated individually. Elixhauser comorbidity scores were calculated from the variables described in other publications using the NIS dataset.16 Hospital volume quartiles were calculated using the total number of operations of interest for all upper gastrointestinal and pancreatic procedures at individual hospitals between 2012 and 2015. Postoperative complications were identified using ICD9-CM diagnosis codes (Supplemental Table 1) and evaluated as prevalence rates for each complication and for a composite outcome of any non-withdrawal complication. Complications included: myocardial infarction; hypotension and shock, pneumonia and respiratory failure, gastroparesis, total parenteral nutrition use, acute respiratory failure, sepsis, central line associated blood stream infection, urinary tract infection, surgical site infection, deep vein thrombosis or pulmonary embolism (DVT/PE), persistent postoperative fistula, postoperative gastrointestinal complication, and withdrawal from either alcohol or drugs. LOS quartiles were calculated for each specific operation. Total hospital charges for the index admission were converted to total hospital costs with use of the Agency for Healthcare Research and Quality provided HCUP Cost-to-Charge Ratio files for the year of the procedure. Items with less than 11 patients per variable were labeled as “<11” in accordance with the HCUP data use agreement.
Statistical Analysis
Chi-square and Student’s t-tests were used for univariate survey-adjusted comparisons between non-users, alcohol users, and drug users. Multiple multivariable survey-adjusted logistic regressions were used to identify the association between unhealthy alcohol and drug use and payer, resection type, malignancy status, income quartile, hospital location, hospital size by bed number, hospital procedure volume for all operations in aggregate, Elixhauser comorbidity score, unhealthy alcohol and drug use, and development of any non-substance related complication. Analyses were conducted using RStudio version 1.2.1335 and the package “survey” version 4.0.17–19
Results
Demographic and Disease Characteristics
A total of 61,235 patients underwent one of the four index operations included in our study between 2012 and 2015. Excluding patients missing mortality, LOS, and cost information resulted in a cohort of 59,740 (97.6%) patients. 250 (0.4%) patients had documented history of both unhealthy alcohol and drug use and were excluded from further analysis. The final analytic cohort consisted of 59,490 (97.2%) patients. Of these, 56,165 (94.4%) patients had no reported substance use, 2060 (3.5%) patients had unhealthy alcohol use, and 1265 (2.1%) patients had unhealthy drug use. The majority of patients diagnosed with unhealthy alcohol use were males aged 50–74 years and more likely to have chronic lung disease, diabetes, peripheral vascular disease, and either obesity or significant weight loss. Patients with unhealthy alcohol use were more likely to have Medicaid insurance and an income quartile below the median for their zip code. Similarly, patients with unhealthy drug use were more likely to be male and had higher prevalence of congestive heart failure, peripheral vascular disease, renal failure, and obesity relative to those without unhealthy drug use. The most common diagnoses in patients who had unhealthy use of alcohol or drugs included chronic pancreatitis (n = 825, 24.8%), cystic disease of the pancreas (n = 260, 7.8%), and benign neoplasms of the pancreas (n = 120, 3.6%). For all three groups, the majority of substance use groups had higher proportions of urgent/emergent surgery compared to those who did not use these substances (Table 1).
Table 1:
Characteristics of patients undergoing upper gastrointestinal and pancreatic oncologic resections with no substance use, unhealthy alcohol use, and unhealthy drug use.
| No substance use (%) | Alcohol use (%) | p-value (Alcohol use vs None) | Drug use (%) | p-value (Drug use vs None) | p-value (all) | |
|---|---|---|---|---|---|---|
| Characteristic, n (%) | n = 56165 | n = 2060 | n = 1265 | |||
| Age (%) | <0.01 | <0.01 | <0.01 | |||
| <50 | 7235 (12.9) | 385 (18.7) | 260 (20.6) | |||
| 50–74 | 38670 (68.9) | 1525 (74.0) | 800 (63.2) | |||
| 75+ | 10260 (18.3) | 150 (7.3) | 206 (16.3) | |||
| Gender (%) | <0.01 | <0.01 | <0.01 | |||
| Male | 32020 (57.0) | 1650 (80.1) | 800 (63.2) | |||
| Female | 24145 (43.0) | 410 (19.9) | 465 (36.8) | |||
| Race (%) | <0.01 | <0.01 | <0.01 | |||
| White | 39405 (70.2) | 1340 (65.0) | 835 (66.0) | |||
| Black | 5045 (9.0) | 270 (13.1) | 200 (15.8) | |||
| Hispanic | 4405 (7.8) | 215 (10.4) | 85 (6.7) | |||
| Asian or Pacific Islander | 1830 (3.3) | 30 (1.5) | <11 | |||
| Native American | 145 (0.3) | <11 | <11 | |||
| Other | 1860 (3.3) | 40 (1.9) | 50 (4.0) | |||
| Comorbidities (%) | ||||||
| Congestive Heart Failure | 2035 (3.6) | 85 (4.1) | 0.23 | 65 (5.1) | <0.01 | 0.39 |
| Chronic Lung Disease | 8440 (15.0) | 460 (22.3) | <0.01 | 195 (15.4) | 0.7 | <0.01 |
| Diabetes | 14010 (24.9) | 380 (18.4) | <0.01 | 285 (22.5) | 0.05 | 0.01 |
| Peripheral Vascular Disease | 2250 (4.0) | 140 (6.8) | <0.01 | 70 (5.5) | <0.01 | 0.01 |
| Renal Failure | 2905 (5.2) | 100 (4.9) | 0.52 | 90 (7.1) | <0.01 | 0.39 |
| Obesity | 6605 (11.8) | 115 (5.6) | <0.01 | 120 (9.5) | 0.01 | <0.01 |
| Weight Loss | 13135 (23.4) | 605 (29.4) | <0.01 | 290 (22.9) | 0.7 | 0.02 |
| Year of Procedure (%) | 0.04 | <0.01 | 0.31 | |||
| 2012 | 14580 (26.0) | 545 (26.5) | 290 (22.9) | |||
| 2013 | 15560 (27.7) | 575 (27.9) | 305 (24.1) | |||
| 2014 | 14840 (26.4) | 580 (28.2) | 415 (32.8) | |||
| 2015 | 11185 (19.9) | 360 (17.5) | 255 (20.2) | |||
| Payer (%) | <0.01 | <0.01 | <0.01 | |||
| Private Insurance | 21450 (38.2) | 715 (34.7) | 440 (34.8) | |||
| Medicare | 27640 (49.2) | 740 (35.9) | 570 (45.1) | |||
| Medicaid | 4160 (7.4) | 410 (19.9) | 190 (15.0) | |||
| Self-Pay | 1230 (2.2) | 100 (4.9) | 40 (3.2) | |||
| Other | 1545 (2.8) | 85 (4.1) | 25 (2.0) | |||
| Income Quartile by Zip Code (%) | <0.01 | <0.01 | <0.01 | |||
| 0–25 | 12605 (22.4) | 570 (27.7) | 325 (25.7) | |||
| 26–50 (median) | 13230 (23.6) | 605 (29.4) | 325 (25.7) | |||
| 51–75 | 14260 (25.4) | 465 (22.6) | 300 (23.7) | |||
| 76–100 | 14765 (26.3) | 360 (17.5) | 285 (22.5) | |||
| Diagnosis (%) | <0.01 | <0.01 | <0.01 | |||
| Benign Disease | 15450 (27.5) | 860 (41.7) | 510 (40.3) | |||
| Biliary Malignancy | 4360 (7.8) | 130 (6.3) | 90 (7.1) | |||
| Esophageal Malignancy | 4810 (8.6) | 300 (14.6) | 45 (3.6) | |||
| Gastric Malignancy | 10035 (17.9) | 235 (11.4) | 210 (16.6) | |||
| Pancreatic Malignancy | 21510 (38.3) | 535 (26.0) | 410 (32.4) | |||
| Resection Type (%) | <0.01 | <0.01 | 0.04 | |||
| Total Gastrectomy | 13485 (24.0) | 430 (20.9) | 285 (22.5) | |||
| Total Esophagectomy | 3940 (7.0) | 205 (10.0) | 50 (4.0) | |||
| Total Pancreatectomy | 3545 (6.3) | 145 (7.0) | 110 (8.7) | |||
| Pancreaticoduodenectomy | 35195 (62.7) | 1280 (62.1) | 820 (64.8) | |||
| Volume Quartile | <0.01 | <0.01 | 0.04 | |||
| First | 3265 (5.8) | 135 (6.6) | 65 (5.1) | |||
| Second | 4400 (7.8) | 160 (7.8) | 165 (13.0) | |||
| Third | 9470 (16.9) | 415 (20.1) | 185 (14.6) | |||
| Fourth | 39030 (69.5) | 1350 (65.5) | 850 (67.2) | |||
| Location/Teaching Status (%) | <0.01 | 0.04 | 0.42 | |||
| Rural | 675 (1.2) | 40 (1.9) | 25 (2.0) | |||
| Urban nonteaching | 5705 (10.2) | 180 (8.7) | 120 (9.5) | |||
| Urban teaching | 49785 (88.6) | 1840 (89.3) | 1120 (88.5) | |||
| AHA Hospital Bed Size (%) | 0.08 | <0.01 | 0.35 | |||
| Small | 4050 (7.2) | 175 (8.5) | 130 (10.3) | |||
| Medium | 10050 (17.9) | 355 (17.2) | 215 (17.0) | |||
| Large | 42065 (74.9) | 1530 (74.3) | 920 (72.7) | |||
| Hospital Location (%) | <0.01 | <0.01 | 0.11 | |||
| New England | 2895 (5.2) | 120 (5.8) | 85 (6.7) | |||
| Middle Atlantic | 9350 (16.6) | 300 (14.6) | 145 (11.5) | |||
| East North Central | 8640 (15.4) | 360 (17.5) | 220 (17.4) | |||
| West North Central | 4095 (7.3) | 155 (7.5) | 75 (5.9) | |||
| South Atlantic | 11765 (20.9) | 465 (22.6) | 355 (28.1) | |||
| East South Central | 3595 (6.4) | 125 (6.1) | 55 (4.3) | |||
| West South Central | 5995 (10.7) | 155 (7.5) | 120 (9.5) | |||
| Mountain | 3055 (5.4) | 120 (5.8) | 90 (7.1) | |||
| Pacific | 6775 (12.1) | 260 (12.6) | 120 (9.5) | |||
| Hospital Region (%) | 0.02 | <0.01 | 0.62 | |||
| Northeast | 12245 (21.8) | 420 (20.4) | 230 (18.2) | |||
| Midwest | 12735 (22.7) | 515 (25.0) | 295 (23.3) | |||
| South | 21355 (38.0) | 745 (36.2) | 530 (41.9) | |||
| West | 9830 (17.5) | 380 (18.4) | 210 (16.6) | |||
In compliance with the HCUP DUA, cells with fewer than 10 observations are marked “<11”.
Univariate Comparison of Postoperative Outcomes
Regarding short-term postoperative outcomes, both unhealthy alcohol and drug use groups experienced higher proportions of complications than those that had no reported substance use. There were 1390 (67.5%) patients with unhealthy alcohol use and 795 (62.8%) patients with drug use experiencing a complication as compared with 32,110 non-users (57.2%) (p < 0.01 for all comparisons). Patients with unhealthy alcohol use had higher rates of all complications, with the exception of DVT/PE and gastrointestinal complications, when compared to non-users. Higher rates of hypotension and shock; gastroparesis, nausea and vomiting; total parenteral nutrition use; DVT/PE; and gastrointestinal complications were observed in the group who used drugs when compared to non-users. The group that used alcohol experienced withdrawal type complications at higher proportions than patients with unhealthy drug use. Both groups had a longer LOS, higher hospital charges, and higher costs than non-users. The unhealthy alcohol use group also required more procedures during their index hospitalization than non-users. There was no difference in postoperative mortality rates for all three groups (Table 2).
Table 2:
Outcomes after upper gastrointestinal and pancreatic oncologic resections in nonsubstance, unhealthy alcohol users, and unhealthy drug users.
| No substance use (%) | Alcohol use (%) | p-value (Alcohol use vs None) | Drug use (%) | p-value (Drug use vs None) | p-value (all) | |
|---|---|---|---|---|---|---|
| Routine Case Status (%) | 47350 (84.3) | 1610 (78.2) | <0.01 | 1010 (79.8) | <0.01 | <0.01 |
| Mortality | 2140 (3.8) | 85 (4.1) | 0.46 | 30 (2.4) | <0.01 | 0.46 |
| Total Charges (median (IQR)) | $120,085 (80,704–198,384) | $141,693 (87,070–241,560) | <0.01 | $140,794 <0.01 | (91,856–211,883) | <0.01 |
| Total Cost (median (IQR) | $34,294 (24,94151,295) | $37,832 (26,782–62,836) | <0.01 | $39,630 (28611–58397) | <0.01 | <0.01 |
| LOS (median (IQR)) | 10 (7–16) | 11 (8–21) | <0.01 | 12.5 (8–20) | <0.01 | <0.01 |
| LOS Quartile (%) | <0.01 | <0.01 | <0.01 | |||
| First | 16865 (30.0) | 545 (26.5) | 260 (20.6) | |||
| Second | 13245 (23.6) | 395 (19.2) | 255 (20.2) | |||
| Third | 12465 (22.2) | 465 (22.6) | 325 (25.7) | |||
| Fourth | 13590 (24.2) | 655 (31.8) | 425 (33.6) | |||
| Number of Procedures (median (IQR)) | 5 (3–7) | 6 (3–9) | <0.01 | 5 (3–8) | 1.00 | <0.01 |
| Disposition (%) | <0.01 | <0.01 | <0.01 | |||
| Routine | 25085 (44.7) | 955 (46.4) | 570 (45.1) | |||
| Transfer to Short-term | ||||||
| Hospital | 420 (0.7) | 35 (1.7) | <11 | |||
| Transfer Other - SNF/ICF/Other | 8680 (15.5) | 350 (17.0) | 260 (20.6) | |||
| Home Health Care | 19800 (35.3) | 635 (30.8) | 395 (31.2) | |||
| AMA | 35 (0.1) | <11 | <11 | |||
| Died | 2140 (3.8) | 85 (4.1) | 30 (2.4) | |||
| Surgical Outcomes (%) | ||||||
| Any Complication | 32110 (57.2) | 1390 (67.5) | <0.01 | 795 (62.8) | <0.01 | <0.01 |
| Myocardial Infarction | 540 (1.0) | <11 | <0.01 | 15 (1.2) | 0.42 | 0.31 |
| Hypotension/Shock | 7415 (13.2) | 405 (19.7) | <0.01 | 175 (13.8) | 0.51 | <0.01 |
| Pneumonia/Acute Respiratory Failure | 9740 (17.3) | 565 (27.4) | <0.01 | 205 (16.2) | 0.29 | <0.01 |
| Gastroparesis/Nausea & Vomiting | 3140 (5.6) | 85 (4.1) | <0.01 | 110 (8.7) | <0.01 | 0.05 |
| Total Parenteral Nutrition | 7030 (12.5) | 340 (16.5) | <0.01 | 190 (15.0) | <0.01 | 0.04 |
| Acute Renal Failure | 5855 (10.4) | 285 (13.8) | <0.01 | 140 (11.1) | 0.45 | 0.08 |
| Any Infection | 13245 (23.6) | 585 (28.4) | <0.01 | 315 (24.9) | 0.27 | 0.07 |
| Urinary Tract Infection | 3945 (7.0) | 90 (4.4) | <0.01 | 105 (8.3) | 0.08 | 0.07 |
| Surgical Site Infection | 4225 (7.5) | 200 (9.7) | <0.01 | 100 (7.9) | 0.61 | 0.26 |
| Blood Transfusions | 12155 (21.6) | 505 (24.5) | <0.01 | 250 (19.8) | 0.11 | 0.28 |
| DVT/PE | 1240 (2.2) | 45 (2.2) | 0.94 | 40 (3.2) | 0.02 | 0.59 |
| Postoperative Fistula | 690 (1.2) | 40 (2.0) | <0.01 | 15 (1.2) | 0.99 | 0.43 |
| Gastrointestinal | 7600 (13.5) | 275 (13.3) | 0.83 | 200 (15.8) | 0.02 | 0.58 |
| Complication | ||||||
| Withdrawal | 0 (0.0) | 390 (18.9) | <0.01 | 60 (4.7) | <0.01 | <0.01 |
In compliance with the HCUP DUA, cells with fewer than 10 observations are marked “<11”.
IQR = interquartile range, LOS = length of stay, SNF = skilled nursing facility, ICF = intermediate care facility, AMA = against medical advice, DVT = deep vein thrombosis, PE = pulmonary embolism
Multivariable Analysis of Factors Associated with Postoperative Complication
Given the high proportion of non-substance use related complications on univariate analysis in patients with unhealthy substance use and the well-studied link between complications, LOS, cost, and mortality, we sought to better understand the relationship between unhealthy alcohol and drug use and development of complications. We performed multivariable survey adjusted analysis with complications as the dependent outcome. This model demonstrated that unhealthy alcohol use was associated with an increase in the occurrence of any non-withdrawal related complication (OR 1.33; 95% CI [1.05, 1.68], p = 0.02) independent of age, comorbid condition, disease pathology, surgical procedure, facility type and facility volume, while unhealthy drug use was not statistically associated with risk of a complication (OR 1.33; 95% CI [0.98, 1.81], p = 0.07) (Table 3).
Table 3:
Multivariable analysis of factors associated with development of a non-substance use related postoperative complication after upper gastrointestinal and pancreatic oncologic resections.
| Age | Odds Ratio [95% CI] | p-value |
|---|---|---|
| <50 (Ref) | 1 | - |
| 50–74 | 1.10 [0.96, 1.26] | 0.17 |
| 75+ | 1.37 [1.15, 1.64] | <0.01 |
| Sex | ||
| Male (Ref) | 1 | - |
| Female | 1.05 [0.96, 1.14] | 0.30 |
| Race | ||
| White | 1 | - |
| Black | 1.31 [1.14, 1.50] | <0.01 |
| Hispanic | 1.40 [1.18, 1.66] | <0.01 |
| Asian or Pacific Islander | 1.20 [0.94, 1.52] | 0.15 |
| Native American | 1.02 [0.44, 2.36] | 0.95 |
| Other | 1.16 [0.92, 1.45] | 0.21 |
| Primary Payer | ||
| Private Insurance (Ref) | 1 | - |
| Medicare | 1.22 [1.11, 1.35] | <0.01 |
| Medicaid | 1.22 [1.04, 1.43] | 0.02 |
| Self-Pay | 1.19 [0.90, 1.58] | 0.22 |
| Other | 0.96 [0.74, 1.24] | 0.76 |
| Resection Type | ||
| Whipple | 1 | - |
| Total Gastrectomy | 1.76 [1.44, 2.15] | <0.01 |
| Total Esophagectomy | 1.69 [1.31, 2.16] | <0.01 |
| Total Pancreatectomy | 0.99 [0.83, 1.17] | 0.90 |
| Malignancy Status | ||
| Benign Disease (Ref) | 1 | - |
| Biliary Malignancy | 0.90 [0.76, 1.07] | 0.22 |
| Esophageal Malignancy | 0.52 [0.42, 0.66] | <0.01 |
| Gastric Malignancy | 0.42 [0.34, 0.52] | <0.01 |
| Pancreatic Malignancy | 0.74 [0.66, 0.82] | <0.01 |
| Hospital Location | ||
| New England (Ref) | 1 | - |
| Middle Atlantic | 0.90 [0.72, 1.13] | 0.38 |
| East North Central | 0.96 [0.76, 1.20] | 0.71 |
| West North Central | 0.83 [0.64, 1.09] | 0.19 |
| South Atlantic | 1.04 [0.84, 1.29] | 0.73 |
| East South Central | 0.98 [0.75, 1.28] | 0.89 |
| West South Central | 1.06 [0.82, 1.37] | 0.65 |
| Mountain | 1.05 [0.81, 1.37] | 0.72 |
| Pacific | 1.11 [0.89, 1.40] | 0.36 |
| Hospital Location/Teaching Status | ||
| Rural (ref) | 1 | - |
| Urban/Non-teaching | 1.16 [0.74, 1.82] | 0.52 |
| Urban/T eaching | 0.90 [0.58, 1.40] | 0.65 |
| Hospital Bedsize | ||
| Small (ref) | 1 | - |
| Medium | 0.93 [0.76, 1.14] | 0.48 |
| Large | 0.99 [0.83, 1.19] | 0.95 |
| Hospital Volume Quartile | ||
| 1 | 1 | - |
| 2 | 0.95 [0.75, 1.19] | 0.63 |
| 3 | 0.88 [0.72, 1.08] | 0.22 |
| 4 | 0.66 [0.54, 0.80] | <0.01 |
| Elixhauser Mortality Score | 1.06 [1.05, 1.06] | <0.01 |
| Income Quartile by Zip Code | ||
| 1 | 1 | - |
| 2 | 0.93 [0.83, 1.04] | 0.21 |
| 3 | 0.93 [0.83, 1.04] | 0.21 |
| 4 | 0.93 [0.82, 1.05] | 0.22 |
| Substance Use | ||
| Alcohol Use | 1.33 [1.05, 1.68] | 0.02 |
| Drug Use | 1.33 [0.98, 1.81] | 0.07 |
Multivariable Analysis of Factors Associated with LOS, Cost and Mortality
To assess the association between unhealthy substance use and LOS, cost, and mortality relative to the effects of other determinates of these outcomes, we performed multivariable survey-adjusted analysis for the following variables: age, sex, race, primary payer, index procedure, pathology, income quartile by zip code, Elixhauser comorbidity score, hospital location, bed number, teaching status, procedure volume quartile, development of any non-substance related complication, and unhealthy substance use. In this analysis, unhealthy alcohol use was associated with a 1.54 day (95% CI [0.12, 2.96], p = 0.03) increase in LOS and a $5,471 (95% CI [$60, $10,881], p = 0.047) increase in cost independent of age, comorbidities, facility type or volume and surgical procedure. Similarly, unhealthy drug use was also independently associated with a longer LOS (2.22 days; 95% CI [0.90, 3.55], p < 0.01) and cost ($4,022; 95% CI [$402, $7,643], p = 0.03). There was no association between unhealthy alcohol or drug use and risk-adjusted mortality on multivariable analysis. Other factors independently associated with risk of prolonged LOS, cost, and mortality included (Table 4): total gastrectomy relative to pancreatectomy and esophagectomy, increased Elixhauser comorbidity score, and development of any non-withdrawal complication (any complication not related to alcohol or drug withdrawal).
Table 4:
Multivariable analysis of factors associated with postoperative outcomes after upper gastrointestinal and pancreatic oncologic resections.
| LOS Difference [95% CI] | p-value | Cost Difference [95% CI] | p-value | Mortality Odds Ratio [95% CI] | p-value | |
|---|---|---|---|---|---|---|
| Age | ||||||
| <50 (Ref) | 0 | - | 0 | - | 1 | - |
| 50–74 | 0.58 [−0.10, 1.25] | 0.10 | −$390 [−$2,763, $1,983] | 0.75 | 1.02 [0.69, 1.51] | 0.92 |
| 75+ | 0.22 [−0.67, 1.11] | 0.62 | −$2,822 [−$6,043, $399] | 0.09 | 1.62 [1.05, 2.51] | 0.03 |
| Sex | ||||||
| Male (Ref) | 0 | - | 0 | - | 1 | - |
| Female | −0.34 [−0.77, 0.10] | 0.13 | −$3,087 [−$4,684, −$1,489] | <0.01 | 0.64 [0.51, 0.80] | <0.01 |
| Race | ||||||
| White | 0 | - | 0 | - | 1 | - |
| Black | 0.61 [−0.25, 1.47] | 0.17 | $2,895 [−$119, $5,910] | 0.06 | 1.01 [0.70, 1.47] | 0.94 |
| Hispanic | 0.05 [−0.85, 0.95] | 0.91 | −$475 [−$4,673, $3,723] | 0.82 | 0.95 [0.65, 1.38] | 0.77 |
| Asian or Pacific | ||||||
| Islander | −0.6 [−1.60, 0.41] | 0.24 | −$2,151 [−$6,563, $2,261] | 0.34 | 0.90 [0.48, 1.71] | 0.75 |
| Native American | 2.37 [−4.89, 9.63] | 0.52 | $3,879 [−$15,282, $23,041] | 0.69 | 1.17 [0.12, 11.3] | 0.90 |
| Other | 0.27 [−1.02, 1.56] | 0.68 | $2,470 [−$2,366, $7,305] | 0.32 | 1.18 [0.67, 2.10] | 0.57 |
| Primary Payer | ||||||
| Private Insurance (Ref) | 0 | - | 0 | - | 1 | - |
| Medicare | 0.62 [0.13, 1.12] | 0.01 | $1,704 [−$53, $3,461] | 0.06 | 1.50 [1.13, 1.98] | <0.01 |
| Medicaid | 2.53 [1.44, 3.62] | <0.01 | $5,531 [$1,684, $9,378] | <0.01 | 1.45 [0.94, 2.25] | 0.09 |
| Self-Pay | 0.86 [−0.49, 2.21] | 0.21 | −$22 [−$4,443, $4,399] | 0.99 | 2.05 [1.13, 3.72] | 0.02 |
| Other | 0.83 [−0.50, 2.16] | 0.22 | $1,768 [−$3,601, $7,137] | 0.52 | 1.59 [0.80, 3.14] | 0.19 |
| Income Quartile | ||||||
| 1 | 0 | - | 0 | - | 1 | - |
| 2 | −0.67 [−1.33, −0.02] | 0.04 | −$1,822 [−$4,113, $469] | 0.12 | 0.96 [0.72, 1.28] | 0.80 |
| 3 | −0.57 [−1.25, 0.11] | 0.10 | −$662 [−$3,051, $1,727] | 0.59 | 0.95 [0.70, 1.27] | 0.72 |
| 4 | −0.54 [−1.22, 0.14] | 0.12 | $239 [−$2,273, $2,751] | 0.85 | 0.78 [0.57, 1.08] | 0.13 |
| Resection Type | ||||||
| Whipple | 0 | - | 0 | - | 1 | - |
| Total Gastrectomy | 2.67 [1.03, 4.32] | <0.01 | $10,697 [$4,996, $16,399] | <0.01 | 1.93 [1.27, 2.92] | <0.01 |
| Total Esophagectomy | 3.75 [2.01, 5.50] | <0.01 | $17,630 [$11,169, $24,090] | <0.01 | 1.28 [0.72, 2.28] | 0.41 |
| Total Pancreatectomy | −1.28 [−2.17, −0.38] | 0.01 | $4,154 [$590, $7,718] | 0.02 | 2.75 [1.85, 4.10] | <0.01 |
| Malignancy Status | ||||||
| Benign Disease (Ref) | 0 | - | 0 | - | 1 | - |
| Biliary Malignancy | −1.22 [−2.02, −0.41] | <0.01 | −$5,801 [−$8,719, −$2,882] | <0.01 | 0.75 [0.49, 1.16] | 0.19 |
| Esophageal Malignancy | −3.28 [−4.98, −1.57] | <0.01 | −$8,363 [−$14,504, −$2,221] | 0.01 | 0.58 [0.36, 0.94] | 0.03 |
| Gastric Malignancy | −4.05 [−5.67, −2.43] | <0.01 | −$14,572 [−$20,429, −$8,715] | <0.01 | 0.41 [0.27, 0.63] | <0.01 |
| Pancreatic Malignancy | −2.38 [−3.00, −1.76] | <0.01 | −$8,614 [−$10,772, −$6,457] | <0.01 | 0.64 [0.48, 0.87] | <0.01 |
| Hospital Location | ||||||
| New England (Ref) | 0 | - | 0 | - | 1 | - |
| Middle Atlantic | 1.08 [−0.11, 2.27] | 0.07 | −$4,157 [−$8,744, $429] | 0.08 | 1.31 [0.65, 2.66] | 0.45 |
| East North Central | −0.15 [−1.29, 1.00] | 0.80 | −$4,013 [−$8,452, $426] | 0.08 | 1.06 [0.52, 2.13] | 0.88 |
| West North Central | 0.20 [−1.20, 1.60] | 0.78 | −$12,485 [−$17,329, −$7,640] | <0.01 | 1.73 [0.74, 4.02] | 0.20 |
| South Atlantic | 1.02 [−0.12, 2.16] | 0.08 | −$5,175 [−$9,456, −$893] | 0.02 | 1.27 [0.64, 2.52] | 0.50 |
| East South Central | 0.69 [−0.80, 2.19] | 0.36 | −$11,678 [−$16,975, −$6,380] | <0.01 | 1.33 [0.56, 3.14] | 0.52 |
| West South Central | 0.90 [−0.30, 2.10] | 0.14 | −$5,982 [−$10,476, −$1,487] | 0.01 | 1.88 [0.92, 3.84] | 0.08 |
| Mountain | 0.81 [−0.65, 2.28] | 0.28 | −$1,543 [−$7,080, $3,994] | 0.59 | 1.70 [0.79, 3.64] | 0.18 |
| Pacific | 1.20 [−0.01, 2.40] | 0.05 | $20,207 [$14,392, $26,023] | <0.01 | 1.03 [0.50, 2.10] | 0.94 |
| Hospital Location/ Teaching Status | ||||||
| Rural (ref) | 0 | - | 0 | - | 1 | - |
| Urban/Non-teaching | 1.76 [−0.15, 3.67] | 0.07 | $1,815 [−$8,953, $12,582] | 0.74 | 1.16 [0.42, 3.21] | 0.77 |
| Urban/T eaching | 2.30 [0.44, 4.17] | 0.02 | $6,725 [−$4,145, $17,595] | 0.23 | 1.25 [0.45, 3.48] | 0.67 |
| Hospital Bedsize | ||||||
| Small (ref) | 0 | - | 0 | - | 1 | - |
| Medium | 0.47 [−0.58, 1.52] | 0.38 | −$2,438 [−$6,221, $1,344] | 0.21 | 0.82 [0.55, 1.24] | 0.35 |
| Large | 0.88 [−0.07, 1.82] | 0.07 | $1,232 [−$2,194, $4,658] | 0.48 | 0.79 [0.55, 1.13] | 0.19 |
| Hospital Volume Quartile | ||||||
| 1 | 0 | - | 0 | - | 1 | - |
| 2 | −0.50 [−1.83, 0.83] | 0.46 | −$1,499 [−$6,396, $3,398] | 0.55 | 1.47 [1.00, 2.17] | 0.05 |
| 3 | −1.00 [−2.20, 0.20] | 0.10 | −$5,727 [−$10,054, −$1,400] | 0.01 | 0.80 [0.54, 1.17] | 0.25 |
| 4 | −2.51 [−3.70, −1.33] | <0.01 | −$4,116 [−$8,527, $294] | 0.07 | 0.43 [0.29, 0.63] | <0.01 |
| Elixhauser Mortality | ||||||
| Score | 0.23 [0.21, 0.26] | <0.01 | $890 [$793, $988] | <0.01 | 1.05 [1.04, 1.06] | <0.01 |
| Complication | 7.29 [6.91, 7.67] | <0.01 | $22,557 [$21,211, $23,902] | <0.01 | 7.37 [4.56, 11.89] | <0.01 |
| Substance Use | ||||||
| Alcohol Use | 1.54 [0.12, 2.96] | 0.03 | $5,471 [$60, $10,881] | 0.05 | 0.85 [0.49, 1.47] | 0.55 |
| Drug Use | 2.22 [0.90, 3.55] | <0.01 | $4,022 [$402, $7,643] | 0.03 | 0.69 [0.30, 1.60] | 0.39 |
Discussion
In this study, we examine the effect of unhealthy alcohol and drug use on outcomes after total gastrectomy, esophagectomy, total pancreatectomy, and pancreaticoduodenectomy. Our major finding is that, in a national population-based analysis, unhealthy alcohol and drug use, both potentially modifiable risk factors, are associated with significantly increased LOS and increased cost of care in major upper gastrointestinal and pancreatic oncologic resections. On univariate analysis, both unhealthy alcohol and drug use were associated with an increased incidence of postoperative complication rates. This increase in complications was not only seen as an increase in the rate of withdrawal-like symptoms and overt delirium tremens but also as an increase in the number of complications traditionally associated with upper gastrointestinal and pancreatic oncologic resections: myocardial infarction, pneumonia, urinary tract infection, sepsis, and DVT. On adjusted multivariable analysis, unhealthy alcohol use was independently associated with an increased likelihood of development of a non-withdrawal complication, while both unhealthy alcohol and drug use were independently associated with increased LOS and increased hospital-related costs. Neither type of substance use was associated with an increased risk of mortality.
Our results suggest that patients with unhealthy substance use have an increased risk of a complication but also require prolonged hospitalizations for their recoveries independent of any postoperative complications they suffer. The exact reasons for this cannot be determined given the limitations of our dataset but they are likely multifactorial. Some of the effect on complication rate is likely due to compromised immune function and wound healing that happen with chronic alcohol and drug use. Some of the effect on LOS and cost is also likely related to such physiologic manifestations but also likely due in part to social circumstances whereby substance users are less likely to have appropriate support mechanisms at home and more likely to present disposition problems to health systems.
At base, our findings point to a need to incorporate substance screening and intervention into preoperative care pathways. We believe that, if anything, the effect observed here and the potential for intervention to be of benefit is understated by our study. There were no identifiable withdrawal events in the non-user cohort but, almost certainly, the incidence of alcohol and drug use reported in our cohort is significantly lower than would be expected based on current reports on national substance use.1 Given this low capture, we would expect that a number of patients with active substance use are included in the non-user cohort and that the observed increase in complication rate and the effect on LOS among substance users would be substantively more marked than we report here if these individuals were rightly categorized. Proper identification of these patients with improved preoperative screening would at very least allow for realistic conversations regarding treatment outcomes and the effects of abstinence on prevention of potential complications.
Several prior studies have identified a similar relationship between alcohol and drug use and postoperative outcomes.9, 20 One well powered analysis of all elective general surgical procedures reported to the American College of Surgeons’ National Surgical Quality Improvement Program (NSQIP) from 2005–2007 reported increased rates of pneumonia and shock as well as greater length of hospitalization when patients reported consumption of two drinks per day during the 2 weeks prior to operation.9 While alcohol use in the quantity identified as significant consumption in this study meet USPSTF guidelines for “risky use,” this study relied on patient self-reporting alcohol use likely leading to an underestimate of the true prevalence of alcohol use in this cohort. Our alcohol cohort, derived from the comorbidity variables for alcohol use provided by HCUP as well as ICD9-CM codes associated with chronic alcohol use, likely expands the population by being more inclusive of patients that may not necessarily be actively drinking to meet “risky use” criteria leading up to their operations but otherwise demonstrate sequelae of long-term, regular alcohol use.
Relatively few studies evaluate the impact of alcohol and drug use on outcome following major upper gastrointestinal and pancreatic oncologic operations specifically. One by Spies et al utilized the CAGE Questionnaire to identify patients with an alcohol use disorder admitted to the intensive care unit following esophagectomy for malignancy in a single center in Germany. The case-rate of alcoholism in their study was 60%. The subgroup of patients with alcohol use experienced increased rates of respiratory complications and sepsis and had a longer LOS than those without alcohol use.20 The prevalence of alcohol use identified in our population is significantly lower than in the Spies study but our findings with regard to prevalence rates of respiratory and septic complications and LOS are similar. The literature on the effect of drug use and postoperative outcomes is less clear. Far fewer studies have been performed on the effect of drug use and postoperative outcomes. There have been several case series regarding outcomes of patients with drug use after bariatric surgery suggest an increased rates of excess weight loss.21 Two separate case series performed on patients undergoing cardiac surgery for infectious endocarditis have demonstrated poorer outcomes among patients with a history of intravenous drug use, including increased mortality and reoperation rates.22, 23 There are no large population-level analyses evaluating the impact of drug use on surgical patients more broadly. In our study, we had a relatively small number of patients identified as unhealthy drug users, but this subgroup did demonstrate a prolonged LOS and increased costs of care. Although the collective results are more limited than those available regarding alcohol use, these studies do point to the fact that drug use likely contributes to risk of compromised clinical outcomes.
There have been two attempts made to study the effect of goal directed preoperative abstinence, harm reduction, or medication-assisted treatment on surgical outcomes. The interventional methods used in these studies vary widely and the results of these trials are mixed. One randomized controlled trial of 42 patients consuming 5 drinks daily and undergoing elective colorectal surgery demonstrated a significantly lower postoperative complication rate in patients randomized to a disulfiram abstinence program (31% vs 74%, p = 0.02).24 In a separate trial in an elective surgical population and utilizing behavioral modification (informational packet with educational material) rather than pharmaceutical intervention, no significant change in complication rate was detected.25 These results suggest that a more intensive targeted preoperative intervention on patients undergoing specific procedures with greater baseline morbidities has potential to improve outcomes and drive cost savings.26
There are several limitations to this work. It is by nature a retrospective review of a large institutional dataset may have omitted variables, selection bias, and sampling error. With regards to reporting of postoperative complications, we attempted to present both general and procedure-specific complications as possible within the confines of the NIS dataset. One notable omission relevant to our analysis is the rate of postoperative pancreatic fistula, which is not explicitly coded for in ICD-9CM. While we are able to present the rate of persistent postoperative fistulas, which we hope would capture pancreatic fistula to an extent, we are unable to subdivide and report specific organ fistula rates. Additionally, as alluded to above, it should be noted that our rate of unhealthy alcohol and drug use was lower than observed in several national surveys. This is likely due to a underdiagnosis and underreporting of substance use in this NIS population. The reason for the underdiagnosis is unknown. We would anticipate, that this would manifest as a significant muting of the magnitude of the observed increase in risk associated with substance use. Our expectation would be that there are many individuals in the subgroup of non-users that do have significant dependency and that the complication rate and LOS profiles of the non-user subgroup is driven in the direction of the unhealthy substance use subgroup. Additionally, we are unable to account for degree of unhealthy alcohol or drug use in any of our modeling. This limits our ability to determine what level of unhealthy use is associated with greatest risk of an untoward outcome. Similarly, we cannot explicitly grade the severity of the complication suffered by any individual patient. We believe that the LOS and cost data would indicate that substance use is associated with higher severity of complication, but, as stated above, the effect on LOS may reflect social determinates of outcome (a lack of support at home requiring more prolonged hospitalization). With respect to categorizing patients as either unhealthy alcohol or drug users, while we excluded those with documented simultaneous unhealthy alcohol and drug use for this analysis, there is very likely some degree of combined use within each individual group, whether clinically detectable or not. Lastly, in studying patients with unhealthy drug use we elected to group any type of drug use into a single patient group to develop a cohort suitable for meaningful analysis. This approach likely ignores the varied phenotypes present in patients who use specific individual drugs and outcomes.
All these limitations demonstrate the importance of and need for prospective study with accurate assessments of substance use in surgical patients, to better delineate the effects of different degrees of unhealthy use and identify specific targeted interventions to improve outcomes and minimize costs of care. Given the high degree of concern for stigmatization and negative consequences, study designs will need to be mindful of how substance use information is collected, to improve study participation and the quality of obtained results.
Conclusion
In patients undergoing major upper gastrointestinal and pancreatic oncologic resections, unhealthy alcohol and drug use are associated with increased LOS and hospital costs but not an increased risk of mortality. Increased focus on screening should be incorporated into the preoperative setting to improve postoperative outcomes for these patients.
Supplementary Material
Acknowledgments
Funding/Financial Support
Dr. S. Kulshrestha is supported by NIH T32 NIAAA 5T32AA013527-17. Dr. C. Bunn is supported by NIH T32 NIGMS 5T32GM008750-20.
Footnotes
COI/Disclosures
No competing interests are declared.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Administration SAaMHS. 2018 National Survey on Drug Use and Health: Methodological summary and definitions. In: Quality CfBHSa, editor. Rockville, MD: Substance Abuse and Mental Health Services Administration; 2019. [Google Scholar]
- 2.de Wit M, Jones DG, Sessler CN, Zilberberg MD, Weaver MF. Alcohol-Use Disorders in the Critically Ill Patient. Chest 2010. p. 994–1003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Rydon P, Redman S, Sanson-Fisher RW, Reid AL. Detection of alcohol-related problems in general practice. Journal of Studies on Alcohol. 1992;53:197–202. [DOI] [PubMed] [Google Scholar]
- 4.Rumpf HJ, Bohlmann J, Hill A, Hapke U, John U. Physicians’ low detection rates of alcohol dependence or abuse: a matter of methodological shortcomings? General Hospital Psychiatry. 2001;23:133–7. [DOI] [PubMed] [Google Scholar]
- 5.Center for Substance Abuse T. SAMHSA/CSAT Treatment Improvement Protocols. Substance Abuse Among Older Adults. Rockville (MD): Substance Abuse and Mental Health Services Administration (US); 1998. [Google Scholar]
- 6.Jin J. Screening for Unhealthy Drug Use in Adults and Adolescents. JAMA. 2020;323:2350-. [DOI] [PubMed] [Google Scholar]
- 7.Force UPST. Screening and Behavioral Counseling Interventions to Reduce Unhealthy Alcohol Use in Adolescents and Adults: US Preventive Services Task Force Recommendation Statement. JAMA. 2018;320:1899–909. [DOI] [PubMed] [Google Scholar]
- 8.Egholm JWM, Pedersen B, Møller AM, Adami J, Juhl CB, Tønnesen H. Perioperative alcohol cessation intervention for postoperative complications. Cochrane Database of Systematic Reviews. 2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Nath B, Li Y, Carroll JE, Szabo G, Tseng JF, Shah SA. Alcohol exposure as a risk factor for adverse outcomes in elective surgery. J Gastrointest Surg. 2010;14:1732–41. [DOI] [PubMed] [Google Scholar]
- 10.de Wit M, Goldberg S, Hussein E, Neifeld JP. Health care-associated infections in surgical patients undergoing elective surgery: are alcohol use disorders a risk factor? J Am Coll Surg. 2012;215:229–36. [DOI] [PubMed] [Google Scholar]
- 11.Moran S, Isa J, Steinemann S. Perioperative management in the patient with substance abuse. Surg Clin North Am. 2015;95:417–28. [DOI] [PubMed] [Google Scholar]
- 12.Bartlett EK, Roses RE, Kelz RR, Drebin JA, Fraker DL, Karakousis GC. Morbidity and mortality after total gastrectomy for gastric malignancy using the American College of Surgeons National Surgical Quality Improvement Program database. Surgery. 2014;156:298–304. [DOI] [PubMed] [Google Scholar]
- 13.Greenblatt DY, Kelly KJ, Rajamanickam V, Wan Y, Hanson T, Rettammel R, et al. Preoperative factors predict perioperative morbidity and mortality after pancreaticoduodenectomy. Ann Surg Oncol. 2011;18:2126–35. [DOI] [PubMed] [Google Scholar]
- 14.Dimick JB, Pronovost PJ, Cowan JA, Lipsett PA. Complications and costs after high-risk surgery: where should we focus quality improvement initiatives? J Am Coll Surg. 2003;196:671–8. [DOI] [PubMed] [Google Scholar]
- 15.(HCUP) HCaUP. HCUP National Inpatient Sample (NIS). Rockville, MD: Agency for Helathcare Research and Quality; 2012–2015. [Google Scholar]
- 16.Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36:8–27. [DOI] [PubMed] [Google Scholar]
- 17.Team R. RStudio: Integrated Development for R..1.2.1335 ed. Boston MA: RStudio, Inc.; 2015. [Google Scholar]
- 18.Lumley T. survey: Analysis of complex survey samples. R package version 4.0 ed2020. [Google Scholar]
- 19.Lumley T. Analysis of Complex Survey Samples. 2004. 2004;9:19. [Google Scholar]
- 20.Spies CD, Nordmann A, Brummer G, Marks C, Conrad C, Berger G, et al. Intensive care unit stay is prolonged in chronic alcoholic men following tumor resection of the upper digestive tract. Anaesthesia Scandinavica. 1996;40:649–56. [DOI] [PubMed] [Google Scholar]
- 21.Heinberg LJ, Ashton K. History of substance abuse relates to improved postbariatric body mass index outcomes. Surg Obes Relat Dis. 2010;6:417–21. [DOI] [PubMed] [Google Scholar]
- 22.Shrestha NK, Jue J, Hussain ST, Jerry JM, Pettersson GB, Menon V, et al. Injection Drug Use and Outcomes After Surgical Intervention for Infective Endocarditis. The Annals of Thoracic Surgery. 2015;100:875–82. [DOI] [PubMed] [Google Scholar]
- 23.Rabkin DG, Mokadam NA, Miller DW, Goetz RR, Verrier ED, Aldea GS. Long-Term Outcome for the Surgical Treatment of Infective Endocarditis With a Focus on Intravenous Drug Users. The Annals of Thoracic Surgery. 2012;93:51–7. [DOI] [PubMed] [Google Scholar]
- 24.Tonnesen H, Rosenberg J, Nielsen HJ, Rasmussen V, Hauge C, Pedersen IK, et al. Effect of preoperative abstinence on poor postoperative outcome in alcohol misusers: randomised controlled trial. BMJ. 1999;318:1311–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Shourie S, Conigrave KM, Proude EM, Ward JE, Wutzke SE, Haber PS. The effectiveness of a tailored intervention for excessive alcohol consumption prior to elective surgery. Alcohol Alcohol. 2006;41:643–9. [DOI] [PubMed] [Google Scholar]
- 26.Tønnesen H, Nielsen PR, Lauritzen JB, Møller AM. Smoking and alcohol intervention before surgery: evidence for best practice. British Journal of Anaesthesia. 2009;102:297–306. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.