Adverse outcomes after non-hepatic surgeries in patients with alcoholic liver diseases: a propensity-score matched study

Alcohol is commonly consumed in surgical patients during the perioperative period [1, 2]. The prevalence of alcohol consumption during the year before surgery is approximately 5–16%, with the level of alcohol consumption recognized as severe misuse in 2–4% of these patients [1], even though there are many unrecorded users [2]. Previous studies suggested that alcohol consumption during the perioperative period may increase complications and medical expenditures after surgery [3, 4].

Previous reports indicated that acute alcohol use was more harmful than chronic alcohol misuse among patients who underwent surgery [5]. In contrast, reducing the consumption of alcohol during the perioperative period improves postoperative outcomes [6]. A systematic review and meta-analysis including 55 studies suggested that preoperative alcohol consumption was associated with risks of postoperative morbidities, general infections, wound complications, pulmonary complications, prolonged hospital stays, and intensive care unit admission [7]. However, another meta-analysis suggested that alcohol drinking was not a risk factor for surgical site infection and anastomotic leakage [8].

These inconclusive findings imply the need for further research. In addition, the effects of alcoholic liver disease (ALD) on perioperative outcomes are not completely understood [1, 3, 7, 8]. Therefore, we used claims data from Taiwan’s National Health Insurance to explore the outcomes after nonhepatic surgeries in patients with and without alcoholic liver disease (ALD).

Table 1 shows the matched characteristics of surgical patients with and without ALD. There was no difference in age, sex, low income, hospital volume, types of surgery, types of anesthesia, inpatient care visits within the past 2 years, emergency care visits within the past 2 years, mental disorders, hypertension, diabetes, peptic ulcer disease, chronic obstructive pulmonary disease, ischemic heart disease, hyperlipidemia, chronic kidney disease, heart failure, or renal dialysis between the ALD patients and non-ALD patients. The liver-related characteristics of surgical patients with and without preoperative ALD was showed in Additional file 1: Table S2.

After adjusting for all covariates listed in Table 1 using multivariate logistic regression analyses (Table 2), we found that patients with ALD had higher risks of postoperative acute renal failure (OR 2.74, 95% CI 2.28–3.28), postoperative bleeding (OR 1.64, 95% CI 1.34–2.01), stroke (OR 1.51, 95% CI 1.34–1.70), septicemia (OR 1.47, 95% CI 1.36–1.58), pneumonia (OR 1.43, 95% CI 1.29–1.58), and 30-day in-hospital mortality (OR 2.64, 95% CI 2.24–3.11) compared with patients without ALD. The medical expenditures (3548 ± 6578 vs. 2939 ± 5187 US dollars; p < 0.0001) and hospital length of stay (10.9 ± 15.1 vs. 10.0 ± 15.8 days; p < 0.0001) of the index surgical admission were comparatively greater for the ALD patients than the non-ALD patients.

In Table 3, the association between ALD and 30-day in-hospital mortality after nonhepatic surgeries was more significant in females (OR 4.07, 95% CI 1.63–10.1) than in males (OR 2.60, 95% CI 2.20–3.07). The 30-day in-hospital mortality after nonhepatic surgeries was associated with ALD in patients aged 30–39 years (OR 3.26, 95% CI 1.88–5.66), 40–49 years (OR 3.04, 95% CI 2.34–3.96), 50–59 years (OR 2.76, 95% CI 2.05–3.70), and 60–69 years (OR 2.00, 95% CI 1.26–3.17). The adjusted ORs for the association between ALD and 30-day in-hospital mortality after nonhepatic surgeries for patients with 0, 1, 2, and ≥ 3 medical conditions were 2.79 (95% CI 2.24–3.49), 2.70 (95% CI 2.06–3.55), 1.49 (95% CI 0.83–2.70), and 2.65 (95% CI 0.49–14.3), respectively.

Further analysis (Table 4) of the correlation between 30-day in-hospital mortality after nonhepatic surgeries and different severities of ALD showed that acute alcoholic hepatitis (OR 1.96, 95% CI 1.47–2.62), alcoholic cirrhosis of the liver (OR 3.94, 95% CI 3.30–4.70), alcoholic liver damage (OR 2.03, 95% CI 1.64–2.52), and suffering more than two alcoholic liver diseases (OR 2.64, 95% CI 2.01–3.48) were associated with 30-day in-hospital mortality. For patients with ALD, alcohol dependence syndrome (OR 3.07, 95% CI 2.39–3.94), jaundice (OR 4.82, 95% CI 3.68–6.32), ascites (OR 4.57, 95% CI 3.64–5.74), gastrointestinal hemorrhage (OR 3.84, 95% CI 3.09–4.79), hepatic coma (OR 4.41, 95% CI 3.44–5.67), and ≥ 2 types of alcoholic liver diseases (OR 4.58, 95% CI 3.67–5.71) were significant factors contributing to 30-day in-hospital mortality after nonhepatic surgeries.

Additional file 1: Table S3 showed the risks of postoperative mortality for surgical patients with the severity of alcoholic liver disease. Additional file 1: Table S4 showed the risks of postoperative adverse events for surgical patients with the severity of alcoholic liver disease.

This is the first nationwide population-based study to discuss the postoperative risks in patients with ALD receiving nonhepatic surgery and present the association between ALD and 30-day in-hospital mortality. Patients with ALD displayed higher risks of postoperative acute renal failure, postoperative bleeding, postoperative stroke, postoperative septicemia, postoperative pneumonia, 30-day in-hospital mortality, length of stay, and medical expenditures compared with patients without ALD who underwent nonhepatic surgeries.

There were several strengths in this study, such as large sample size, propensity-score matching, validated database, and integrative assessment for postoperative outcomes. However, the study limitations suggest that caution should be taken when interpreting our study findings. First, physical examinations, laboratory data, and the patients’ sociodemographic and lifestyle characteristics could not be extracted from the reimbursement data in Taiwan’s National Health Insurance Research Database. Thus, we could not evaluate the influence of these factors on the perioperative outcome in patients with ALD. Second, we have no information regarding the severity of ALD, such as the Child–Pugh score or the Model score for End-stage Liver Disease score [13, 14]. Third, to identify ALD cases correctly, we required at least 2 medical care visits with a physician’s primary diagnosis of ALD for inclusion. We could not exclude the possibility that some ALD cases without medical care may have been included in the control group. Fourth, there were no details on the alcohol drinking status of participants because of the unavailable information in the claim database. We admitted this is also one of study limitations. Fifth, although the procedure and diagnosis codes of ALD has not been validated in the previous studies, we considered that the codes used in this study is reliable because the insurance claims and payment was strictly examined and reviewed by the Ministry of Health and Welfare, Taiwan. In addition, we could not analyze the duration of ALD for more the two years because of the unavailable data of insurance database. We also could not evaluate the influence of emergency surgeries on the postoperative outcomes in patients with ALD because we have no related information of emergency surgeries in this study. Finally, residual confounding bias may be possible, although we used propensity-score matching and multivariate regression adjustments.

To our knowledge, advanced age, sex, low income, hospital volume, medical conditions (such as mental disorders, hypertension, diabetes, peptic ulcer disease, chronic obstructive pulmonary disease, ischemic heart disease, hyperlipidemia, chronic kidney disease, heart failure and renal failure), type of surgery, and type of anesthesia were associated with postoperative outcomes [15‐18]. After adjusting for these confounding factors, our findings suggest that ALD increased the risks of postoperative complications and 30-day in-hospital mortality following nonhepatic surgery.

Some possible reasons may help to explain the association between ALD and postoperative complications. First, acute renal failure is a common presentation in patients hospitalized for advanced liver cirrhosis with acute decompensation. Hepatorenal syndrome is one of the comorbidities of end-stage liver disease due to severe portal hypertension [19]. Moreover, during the perioperative period, prerenal hypoperfusion is worsened by the hypotension induced by anesthesia or surgical blood loss. Therefore, postoperative acute renal failure is more common in patients with alcoholic liver disease. Second, coagulation and thrombocytopenia are common problems in ALD patients [20]. Spontaneous deep bleeding into muscles and the retroperitoneum due to disordered and unstable coagulation has been reported [21]. Obviously, more postoperative bleeding problems are expected in patients with ALD than in patients without ALD. Third, some studies have reported that liver cirrhosis may cause endothelial dysfunction and chronic inflammation [22, 23]; as a result, both were suggested to be associated with increased ischemic and hemorrhagic stroke risks [11]. Fourth, chronic inflammation usually occurs in patients with ALD due to hepatic steatosis, oxidative stress, and acetaldehyde-mediated toxicity [24]. The long-term inflammation problem induced by cytokines and chemokines may impair the autoimmune system and may be associated with increased risks of septicemia and pneumonia [25]. Fifth, preoperative alcohol consumption is related to intraoperative and postoperative delirium, which will affect the sensitivity of symptom recognition and quality of care [26]. ALD patients usually have poor self-care abilities, family support, health knowledge, attitudes, and practices of disease prevention and treatments [26, 27]. Malnutrition is common in patients with chronic liver disease, which can also explain the poorer prognosis in ALD patients [28]. In addition, the comorbidities of patients with ALD, such as liver cirrhosis, encephalopathy, and ascites, may contribute to postoperative complications [12, 29]. Acute chronic liver failure presenting in advanced stages, precipitated by some special events, including bacterial infection, acute alcoholic, and drug-induced or viral hepatitis, is associated with increased morbidities and mortality [30]. These factors explain how ALD patients have worse postoperative complications, longer hospital length of stay, more medical expenditure, and increased mortality.

Our study suggested that ALD increases 30-day in-hospital mortality at all ages, especially in women. Sex-related differences in inflammatory and immune biomarker concentrations in ALD patients were reported [31]. Previous studies revealed that female drinkers are at greater risk of ALD than men due to their increased likelihood of developing acute liver failure from excessive alcohol use and more rapid progression of ALD [32]. Patients with ALD have an approximately 2.6-fold increased risk of 30-day in-hospital mortality, regardless of the number of medical conditions. This result implied that ALD affects patient mortality more than other diseases. In the subgroup analysis, ALD patients with acute alcoholic hepatitis, alcoholic liver damage, alcoholic liver cirrhosis and more than two alcoholic liver diseases had a 2–4-fold increase in 30-day in-hospital mortality. Among those different types of alcoholic liver disease, alcoholic liver cirrhosis most affected mortality in patients who underwent nonhepatic surgery. There are systemic comorbidities related to cirrhosis, such as hepatic encephalopathy, hepatorenal syndrome, and cirrhotic cardiomyopathy [19, 30, 33]. These comorbidities and their associated systemic organ damage may greatly affect surgical patient mortality following nonhepatic surgery. Furthermore, the ORs for 30-day in-hospital mortality were higher in ALD patients with alcohol dependence syndrome, gastrointestinal hemorrhage, hepatic coma, ascites, jaundice, and more than 2 of the above indicators. Alcohol dependence syndrome is related to chronic alcohol consumption, which has been previously determined to increase postoperative mortality [7, 34]. Liver function is usually assessed by the Child-Turcott-Pugh (CTP) classification and the Model for End-Stage Liver Disease (MELD) scores [13]. The Child-Turcott-Pugh classification considers five variables (serum bilirubin, albumin, prothrombin time, ascites, and hepatic encephalopathy), with higher classifications indicating worse prognosis in patients with cirrhosis [14]. Based on the above, our findings suggested that more severe ALD (with cirrhosis-related complications) increased 30-day in-hospital mortality in patients who underwent nonhepatic surgery.

The influence of ALD on perioperative outcomes is incompletely understood, and previous studies have focused on the effects of alcohol consumption instead of ALD [3, 7]. Although some studies mentioned that patients with liver cirrhosis have increased risks of postoperative complications and mortality [12, 35, 36], ALD does not consist of cirrhosis alone. ALD is the most prevalent type of chronic liver disease worldwide, and it can progress to alcoholic fatty liver and alcoholic steatohepatitis. Chronic alcoholic steatohepatitis eventually leads to fibrosis, cirrhosis, or hepatocellular carcinoma [24]. Our study described a more comprehensive design and explanation of the relationship between ALD and outcomes after nonhepatic surgery.

In conclusion, we found that patients with ALD had more complications and higher mortality after nonhepatic surgery compared with non-ALD controls, particularly those with severe liver symptoms. Further studies are needed to develop effective preventions for postoperative adverse events among patients with ALD.