Association between ICU admission during off-hours and in-hospital mortality: a multicenter registry in Japan

Study design and data collection

This was a multicenter observational study based on the national ICU registry, the Japanese Intensive care PAtient Database (JIPAD), which the Japanese Society of Intensive Care Medicine launched in 2014, and prospectively collected individual patient data and information on the participating facility in a predefined manner [23]. As of 2017, a total of 21 ICUs with 217 beds had participated in the registry.

The following patient information was registered in the JIPAD: age, sex, height, weight, date and time of admission, admission source (emergency department, operating room, ward, other care units, or transferred from other hospitals), type of admission (elective, emergency, and ICU medical procedure), diagnosis at ICU admission (cardiovascular disease, respiratory disease, gastrointestinal disease, metabolic disease, neurological disease, hematological disease, musculoskeletal disease, obstetric and gynecological disease, genitourinary disease, trauma, etc.), after cardiac resuscitation, Acute Physiology and Chronic Health Evaluation (APACHE) II score [24], and chronic organ insufficiencies (acquired immunodeficiency syndrome, heart failure, respiratory failure, liver cirrhosis, acute leukemia or multiple myeloma, lymphoma, metastatic cancer, immunosuppression, and maintenance dialysis). We also extracted the following facility-level information: number of intensivists dedicated to ICUs, number of full-time nurses designated to ICUs, number of ICU beds, number of hospital beds and type of hospital (university, public, or private).

Selection of participants

Out of all the patients in the JIPAD registry, we enrolled adult patients aged 18 years and older from April 2015 to March 2019. Referring to previous studies, we excluded those who were admitted to ICUs after elective surgery, or who were readmitted to ICUs, or who were admitted to ICUs for ICU clinical procedures and discharged alive within 24 h [4, 5, 9, 14, 18, 19]. Further, we excluded patients with missing data on ICU admission time, in-hospital mortality, or covariates for statistical analysis.

Off-hours vs. office-hours

We defined office-hours as from 09:00 to 17:00, weekdays, Monday to Friday, and regarded official public holidays and all other times as off-hours. Although there is no standard definition of office-hours, we referred to previous studies and finally adopted the practical definition of office-hours and off-hours as above, which has been commonly adopted in Japan as a practical approach [5, 12].

Outcome measures

We defined in-hospital mortality as the outcome of interest, and compared the outcome between two groups—an off-hours group and an office-hours group.

Statistical analysis

We described continuous variables with medians and interquartile ranges (IQRs), and categorical variables with counts and proportions. We examined continuous variables using Mann–Whitney U test, and categorical variables using the Chi-square test.

Primary analysis

In the primary analysis, we compared patient outcomes between ICU patients admitted during off-hours and office-hours, using a multilevel logistic regression model which allows for the random effect of each hospital (a random-intercept model).

In this model, we used a complete data set and adjusted both patient-level variables and facility-level variables as follows: age, sex, body mass index (< 18.5, 18.5 to 25, ≥ 25) [25], APACHE II score, the most common three diagnoses at ICU admission (cardiovascular disease, respiratory disease, and neurological disease), trauma, surgery, after cardiac resuscitation, admission source (emergency department, operating room, ward, other care units, or transferred from other hospitals), number of intensivists in relation to the number of ICU beds, number of full-time ICU nurses in relation to the number of ICU beds, number of hospital beds (categorized into tertiles) and type of hospital (university, public, or private). Further, we conducted a sensitivity analysis for missing covariates with multivariate imputation by chained equations. To impute the missing data, we used all measured variables, including outcomes, and generated 20 imputed data sets based on the assumption that the data were missing randomly.

Additional analysis

In order to evaluate the associations between the ICU admission time and in-hospital mortality in detail, we calculated predicted in-hospital mortality for each ICU admission time (every hour from 0:00 to 23:00). Initially, we divided patients into those who were admitted during weekdays and weekends (including official public holidays). We then estimated in-hospital mortality for both of these groups using a multilevel logistic regression model, which adjusted the same variables as the primary analysis and allowed for each hospital as a random effect.

Subgroup analysis

In order to evaluate the heterogeneity of the target population in the present study, we decided to perform an a priori subgroup analysis. In the subgroup analysis, we took into consideration patient factors (disease categories and requirement of surgery) and facility factors (the number of intensivists dedicated to ICUs, the number of full-time ICU nurses, and type of hospital) as potential sources of heterogeneity. Here, we adopted the most prevalent diseases (cardiovascular, respiratory and neurological disorder) as the subgroups of the disease categories [11].

We conducted all statistical analyses using STATA version 16.1 (StataCorp, College Station, TX), and all hypothesis tests were 2-tailed with a significance level of P < 0.05.

The ethics committee of Kobe City Medical Center General Hospital approved this study protocol (zn200717). The JIPAD project itself was approved by the ICU Functional Assessment Committee of the Japanese Society of Intensive Care Medicine. We conducted the present study with an opt-out policy from patients, their relatives, or proxies, and written informed consent was waived.

留言 (0)

沒有登入
gif