Source of data
We used the reimbursement claims data from the Taiwan’s National Health Insurance Program, which was implemented in March 1995 and covers more than 99% of the 23 million Taiwan residents [9,10,11,12]. The National Health Research Institutes of Taiwan established a National Health Insurance Research Database that records all beneficiaries’ medical services, including inpatient and outpatient demographics, primary and secondary diagnoses, procedures, prescriptions and medical expenditures for public research interest [9,10,11,12]. The validation of Taiwan’s National Health Insurance Research Database has been evaluated in previous studies [9, 10]. The validity of this database has been favorably evaluated, and research articles based on it have been accepted in prominent scientific journals worldwide [11, 12].
The data that support the findings of this study are available from the Ministry of Health and Welfare, Taiwan but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of the Ministry of Health and Welfare, Taiwan. We have made the formal application (included application documents, study proposals, and ethics approval of the institutional review board) of the current insurance data. The authors of the present study had no special access privileges in accessing the data which other interested researchers would not have [11, 12].
We conducted this study in accordance with the Helsinki Declaration. In the original insurance data, every patient had identification number. For protecting personal privacy, the Ministry of Health and Welfare decoded the identification number in the insurance research database. Therefore, the researchers used insurance research database could not understand the identification of patients and they also could not identify specific patients. The requirement of informed consent to participate was deemed unnecessary according to the regulations of the Ministry of Health and Welfare. The requirement of informed consent to participate was waived by the Institutional Review Board of Taipei Medical University that also evaluated and approved this study (TMU-JIRB-202203134; TMU-JIRB-201905042; TMU-JIRB-201902053; TMU-JIRB-201705063).
From the three million surgical patients who underwent nonhepatic elective surgeries between 2008 and 2013 in Taiwan, we identified 32,548 surgical patients with ALD who were aged 20 years and older. These elective surgeries were nonhepatic surgeries that required general, epidural, or spinal anesthesia and hospitalization for at least 1 day. To identify patients with ALD more clearly, the current study required at least one medical care visit with a physician’s diagnosis of ALD within the 24-month preoperative period of the index surgery. Using a matching propensity score procedure with age, sex, low income, hospital volume, types of surgery, types of anesthesia, number of inpatient care visits within the past 2 years, number of emergency care visits within the past 2 years, and coexisting medical conditions (including mental disorders, hypertension, diabetes, peptic ulcer disease, chronic obstructive pulmonary disease, ischemic heart disease, hyperlipidemia, chronic kidney disease, heart failure, and renal dialysis), we selected 32,548 surgical patients from the surgical patient populations who were without ALD preoperatively.
Measures and definitions
We identified income status by defining low-income patients as those qualifying for waived medical copayment because this status is verified by the Taiwan Bureau of National Health Insurance. Additionally, whether the surgery was performed in a teaching hospital and the types of surgery and anesthesia used were also recorded. In this study, we excluded sucgrical patients received hepatic surgeries, such as wedge biopsy of liver, partial hepatectomy, segemental hepatectomy, hepatorrhaphy, hepato-enterostomy, portocavo shunt, Warren's shunt, right lobectomy, left lobectomy, and liver transplantation. The details of procedure codes were showed in Additional file 1: Table S1.
We used the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) to define preoperative medical diseases and postoperative complications. Preoperative ALD was defined as the main exposure and included alcoholic fatty liver (ICD-9-CM 571.0), acute alcoholic hepatitis (ICD-9-CM 571.1), alcoholic cirrhosis of the liver (ICD-9-CM 571.2), and alcoholic liver damage (ICD-9-CM 571.3). Coexisting medical conditions, including mental disorders (ICD-9-CM 290-319), hypertension (ICD-9-CM401-405), diabetes (ICD-9-CM 250), peptic ulcer disease (ICD-9-CM 531, 532, 533), chronic obstructive pulmonary disease (ICD-9-CM 491, 492, 496), ischemic heart disease (ICD-9-CM 401-414), hyperlipidemia (ICD-9-CM 272.0, 272.1, 272.2), chronic kidney disease (ICD-9-CM 585, 586), heart failure (ICD-9-CM 428), and renal dialysis (administration codes D8, D9), were determined from medical claims for the 24-month preoperative period. Jaundice (ICD-9-CM 782.4), hepatic coma (ICD-9-CM 572.2), gastrointestinal hemorrhage (ICD-9-CM 578), ascites (ICD-9-CM 789.5) and alcohol dependence syndrome (ICD-9-CM 303) were also identified as clinical symptoms of patients with ALD.
Thirty-day in-hospital mortality after the index surgery and postoperative complications were considered as the study’s outcomes. These complications included septicemia (ICD-9-CM 038 and 998.5), pneumonia (ICD-9-CM 480-486), urinary tract infection (ICD-9-CM 599.0), acute renal failure (ICD-9-CM 584), stroke (ICD-9-CM 430-438), deep wound infection (ICD-9-CM 958.3), postoperative bleeding (ICD-9-CM 998.0, 998.1 and 998.2) and pulmonary embolism (ICD-9-CM 415). Admission to the intensive care unit, length of hospital stay and medical expenditure during the index nonhepatic surgery were also considered as secondary outcomes in this study.
We used a nonparsimonious multivariable logistic regression model to estimate propensity scores for preoperative ALD, irrespective of outcome. Clinical significance guided the initial choice of covariates in this model: age, sex, low income, hospital volume, types of surgery, types of anesthesia, number of inpatient care visits within the past 2 years, number of 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, and renal dialysis. We used a structured iterative approach to refine this model with the goal of achieving covariate balance within the matched pairs. Chi-square tests were used to measure covariate balance, and p < 1.0 was suggested to represent covariate imbalance. We matched patients with ALD to patients without ALD using a greedy-matching algorithm with a caliper width of 0.2 standard deviation of the log odds of the estimated propensity score. This method has been estimated to remove 98% of the bias from measured covariates. Adjusted odds ratios (ORs) with 95% CIs for 30-day postoperative complications and mortality between patients with and without ALD were analyzed with multivariate logistic regression. We controlled for age, sex, low income, hospital volume, types of surgery, types of anesthesia, number of inpatient care visits within the past 2 years, number of 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, and renal dialysis. We performed a stratified analysis and calculated an adjusted HR and 95% CI to examine the association between ALD and 30-day in-hospital mortality after surgery in the multivariate logistic regressions. SAS version 9.1 (SAS Institute Inc., Cary, NC, USA) statistical software was used; two-sided p < 0.05 indicated significant differences between surgical patients with and without ALD.