Comparison of surgical and conservative treatment outcomes for type a aortic dissection in elderly patients

We performed a retrospective cohort study using the Japanese Diagnosis Procedure Combination (DPC) database. The DPC system was introduced in Japan in 2003 as a comprehensive daily evaluation system for acute-phase hospital medical care. In 2016, DPC-participating hospitals accounted for approximately 80% of acute-phase beds in Japan. Clinical information and other data for patients were registered in the DPC database and classified by a 14-digit DPC code. The system holds the following clinical information: patient demographics, diagnoses, comorbidities and complications [registered using the International Classification of Disease, 10th revision (ICD10)], drugs and devices used, procedures performed, healthcare costs and status at admission and discharge.

Data were extracted from the DPC database between April 2014 and March 2016. First, we identified patients aged 80 years or older with acute aortic dissection (ICD10 code, I71.0) recorded as either the “main diagnosis,” “admission-precipitating diagnosis,” or “most resource-consuming diagnosis” and identified patients with dissection in the ascending aorta with “Stanford type A,” “DeBakey I,” or “DeBakey II” as the disease name. We excluded patients who had planned admission and unknown admission pathway. The final cohort was divided into the conservative and surgical treatment groups.

We matched propensity scores between the surgical and conservative treatment groups. Propensity scores were calculated using a logistic regression model with the following variables: age, sex, body mass index (divided into the four following groups: < 18.5, 18.5–25.0, > 25.0, and data missing), admission pathway (home, other hospital, or clinic and nursing-care facility), ambulance use, hypertension, diabetes mellitus, chronic obstructive pulmonary disease (COPD), chronic kidney disease, ischemic heart disease, chronic liver disease, cancer, dementia, cardiopulmonary arrest, cardiac tamponade, shock, coronary malperfusion, limb ischemia, aortic insufficiency, hospital volume [low (< 10 cases), middle (10–20 cases), and high (>20cases)] and consciousness level upon admission (Japan coma scale: 0, alert; 1–3, delirious; 10–30, somnolent and 100–300, coma). Outcome measures were all-cause in-hospital death, stroke, and acute kidney injury (AKI) requiring renal replacement therapy (RRT), tracheotomy, composite adverse events (all-cause in-hospital death, stroke, AKI, and tracheotomy), length of hospital stay, length of intensive care unit (ICU) stay, inpatient costs, home discharge rate, and activities of daily living (ADL) of surviving patients at discharge. ADL was evaluated using the Barthel index (BI), which assesses the ADL using a 100-point scale comprising the following 10 items: “Bowels,” “Bladder,” “Grooming,” “Toilet use,” “Feeding,” “Transfer,” “Mobility,” “Dressing,” “Stairs,” and “Bathing,” and we set the BI cut-off at 60 points, which indicates having functional dependency. [12, 13].

Summary statistics are presented as frequencies and percentages for categorical variables and median and interquartile ranges for continuous variables (with non-normal distribution in the Shapiro–Wilk test). Data were compared using the Mann–Whitney U test for continuous variables and the chi-squared test for categorical variables before matching; after matching, the Wilcoxon signed rank and McNemar tests were performed. We performed one-to-one pair matching using nearest neighbor matching without replacement within 0.2 standard deviations of the logit of the propensity score as caliper width [12]. Covariate balance between the two groups was checked depending on whether the absolute standardized difference was < 10% [13]. Adjusted odds ratios (AOR) for the primary outcome were estimated using multivariable logistic regression in all patients aged > 80 years to investigate the influence of the type of surgery. As a subgroup analysis, the status of surviving patients at discharge in both the groups was compared before after matching. The robustness of the result was confirmed using the following method as sensitivity analysis for the BMI missing value (749 cases): list-wise deletion method (excluded the BMI missing data), median imputation for BMI missing data and multiple imputation for BMI missing data. All analyses were performed using the R statistical software (version 3.5.1; R Foundation for Statistical Computing, Vienna, Austria), and the packages used for analyses were “Matching” for propensity score matching and “mice” for missing value complement. P < 0.05 was considered significant, and hypothesis tests were two sided. This study was performed with the approval of the Ethics Committee, Kyoto University Graduate School and Faculty of Medicine (Approval number: R0135).