Effects of Sepsis on Morbidity and Mortality in Critically Ill Patients 2 Years After Intensive Care Unit Discharge

Livia Biason, MD; Cassiano Teixeira, MD, PhD; Jaqueline Sangiogo Haas, RN, MS; Cláudia da Rocha Cabral, RN; Gilberto Friedman, MD, PhD


Am J Crit Care. 2019;28(6):424-432. 

In This Article


Study Design

This was a cohort study in which baseline data were collected routinely during patients' ICU stays and long-term data were collected by telephone interview 2 years after ICU discharge. The study was approved by the ethics committees of the 2 participating institutions. Written informed consent was obtained from the patients or next of kin in the first week after ICU discharge.


Eligible participants were patients who stayed for at least 24 hours in a mixed medical-surgical ICU in 1 of 2 hospitals in the city of Porto Alegre, southern Brazil, in 2014. One study site was an 87-bed general ICU in a 1042-bed university hospital that is part of a complex of 9 hospitals. From the other study hospital (471 beds and 48 ICU beds), data were obtained in 3 ICUs (general, cardiovascular, and neurocritical). Patients younger than 18 years and those who refused consent were excluded from the study. For patients who were readmitted to the ICU, only data from the first admission were included in the analysis.

On ICU admission, patients were classified as having sepsis or not having sepsis. The diagnoses of severe sepsis and septic shock were based on 2004 Surviving Sepsis Campaign definitions that were current at the time of study launch. Severe sepsis was defined as 2 or more markers of systemic inflammatory response syndrome plus at least 1 organ dysfunction. Septic shock was defined as persistent hypotension requiring vasopressor infusion.[15] Because we initiated the study before the publication of Sepsis-3 guidelines, we did not stipulate lactate testing or require a specific lactate level. Patient inclusion occurred immediately after ICU discharge, so to confirm sepsis we also evaluated the need for antibiotic therapy, diagnostic examinations performed during the ICU stay, and the clinical evolution of the patients. In our study, severe sepsis and septic shock were termed sepsis or patients with sepsis.

"Patients with sepsis appear to have higher mortality rates after ICU discharge than patients without."

Variables and Measurements

The study was conducted in 2 phases: (1) data extraction from patients' ICU records and (2) prospective data collection. The following data are routinely collected by the ICU medical team and were extracted for analysis: demographic variables; functional status before ICU admission (Karnofsky Performance Status and Lawton instrumental activities of daily living [IADL] scales); previous diseases; reason for ICU admission; severity scores (Acute Physiology and Chronic Health Evaluation II [APACHE II], Therapeutic Intervention Scoring System, and Sequential Organ Failure Assessment); lengths of ICU and hospital stays; need for life support (mechanical ventilation, vasoactive drugs, and dialysis therapy, among others); and outcomes during ICU stay.

Outcome Variables

Mortality, functional status, and self-perceived pain were evaluated 2 years after ICU discharge. The interviewers (1 physician and 6 nurses) contacted the participants or next of kin by telephone for consent renewal and further explanation of the study goals. Next of kin was considered the closest family member (parent, child, or spouse) living with the patient. Caregivers were not interviewed. Interviewers were previously trained in the administration of structured questionnaires (Karnofsky and IADL scales), subjective reporting of the presence and intensity of pain, need for hospital readmission, and patients' self-perceived quality of life to test the reliability of the scales used.

"Patients who stayed in the ICU > 1 day were followed up for 2 years after discharge to assess morbidity and mortality."

The Karnofsky Performance Status scale assesses the degree of functional impairment. It was initially designed to assess the physical performance of patients with cancer, but its use was extended to patients with other chronic disabling diseases.[16] On the basis of their scores, individuals were classified as follows: 100, normal, reporting no problems and having no evidence of disease; 90, capable of normal activity and having few symptoms of disease; 80, capable of normal activity with some difficulty and some symptoms of disease; 70, capable of self-care but not capable of normal activity or work; 60, occasionally requiring some assistance but able to take care of most personal needs; 50, requiring considerable assistance or frequent medical care; 40, disabled and requiring special care and assistance; 30, severely disabled with indicated hospital admission, although death is not imminent; 20, very ill, requiring hospital admission; and 10, moribund, with fatal processes progressing rapidly. The Lawton IADL scale[17] assesses more complex tasks (using the phone, shopping, using transportation, taking responsibility for one's own medication, handling finances, preparing food, housekeeping, and doing laundry). Both scales were completed by questioning patients or their relatives.


The fieldwork coordinator reassessed a random sample of about 5% of the interviews to ensure the accuracy of data capture. Data were entered by 2 independent operators and reviewed patient by patient in the questionnaires to avoid postcompletion errors. The same person evaluated the quality-of-life scales to avoid measurement bias in the interpretation of the scales. To avoid gender bias, we used the version of the Lawton IADL scale that assesses only 5 activities in men, with different result categories for men (scores of 5, 4, 2–3, 1, and 0 corresponding to independent, mildly dependent, moderately dependent, severely dependent, and totally dependent, respectively) and women (scores of 8, 6–7, 4–5, 2–3, and 0–1 corresponding to independent, mildly dependent, moderately dependent, severely dependent, and totally dependent, respectively).[17]

Statistical Analysis

Categorical variables were expressed as frequencies and percentages. Quantitative variables with normal distribution were expressed as means and SDs, and variables with skewed distribution were expressed as medians and interquartile ranges (25th-75th percentiles). Categorical variables were compared by using the χ2 test or Fisher exact test. Adjusted residual analysis was performed, and values greater than 1.96 or less than −1.96 indicated a higher or lower frequency than expected. We used the Student t test for independent samples to compare quantitative variables with normal distribution and the Mann-Whitney test to compare variables with skewed distribution. We used the log-rank test to compare Kaplan-Meier curves. For mortality risk, we calculated adjusted hazard ratios with 95% CIs by using Cox regression, adjusting for potential confounders. A covariance analysis was adjusted to evaluate the factors associated with functional scales. All statistical analyses were performed using statistics software (SPSS version 20.0, IBM). The level of significance was set at 5%.