Random survival forests for competing risks.

We introduce a new approach to competing risks using random forests. Our method is fully non-parametric and can be used for selecting event-specific variables and for estimating the cumulative incidence function. We show that the method is highly effective for both prediction and variable selection in high-dimensional problems and in settings such as HIV/AIDS that involve many competing risks.

A scenario-based, randomized trial of patient values and functional prognosis on intensivist intent to discuss withdrawing life support.

OBJECTIVES: To evaluate the effect of 1) patient values as expressed by family members and 2) a requirement to document patients' functional prognosis on intensivists' intention to discuss withdrawal of life support in a hypothetical family meeting. DESIGN: A three-armed, randomized trial. SETTING: One hundred seventy-nine U.S. hospitals with training programs in critical care accredited by the Accreditation Council for Graduate Medical Education. SUBJECTS: Six hundred thirty intensivists recruited via e-mail invitation from a database of 1,850 eligible academic intensivists. INTERVENTIONS: Each intensivist was randomized to review 10, online, clinical scenarios with a range of illness severities involving a hypothetical patient (Mrs. X). In control-group scenarios, the patient did not want continued life support without a reasonable chance of independent living. In the first experimental arm, the patient wanted life support regardless of functional outcome. In the second experimental arm, patient values were identical to the control group, but intensivists were required to record the patient's estimated 3-month functional prognosis. MEASUREMENTS AND MAIN RESULTS: Response to the question: "Would you bring up the possibility of withdrawing life support with Mrs. X's family?" answered using a five-point Likert scale. There was no effect of patient values on whether intensivists intended to discuss withdrawal of life support (p = 0.81), but intensivists randomized to record functional prognosis were 49% more likely (95% CI, 20-85%) to discuss withdrawal. CONCLUSIONS: In this national, scenario-based, randomized trial, patient values had no effect on intensivists' decisions to discuss withdrawal of life support with family. However, requiring intensivists to record patients' estimated 3-month functional outcome substantially increased their intention to discuss withdrawal.

Timing of limitations in life support in acute lung injury patients: a multisite study*.

OBJECTIVE: Substantial variability exists in the timing of limitations in life support for critically ill patients. Our objective was to investigate how the timing of limitations in life support varies with changes in organ failure status and time since acute lung injury onset. DESIGN, SETTING, AND PATIENTS: This evaluation was performed as part of a prospective cohort study evaluating 490 consecutive acute lung injury patients recruited from 11 ICUs at three teaching hospitals in Baltimore, MD. INTERVENTIONS: None. MEASUREMENTS: The primary exposure was proportion of days without improvement in Sequential Organ Failure Assessment score, evaluated as a daily time-varying exposure. The outcome of interest was a documented limitation in life support defined as any of the following: 1) no cardiopulmonary resuscitation, 2) do not reintubate, 3) no vasopressors, 4) no hemodialysis, 5) do not escalate care, or 6) other limitations (e.g., "comfort care only"). MAIN RESULTS: For medical ICU patients without improvement in daily Sequential Organ Failure Assessment score, the rate of limitation in life support tripled in the first 3 days after acute lung injury onset, increased again after day 5, and peaked at day 19. Compared with medical ICU patients, surgical ICU patients had a rate of limitations that was significantly lower during the first 5 days after acute lung injury onset. In all patients, more days without improvement in Sequential Organ Failure Assessment scores was associated with limitations in life support, independent of the absolute magnitude of the Sequential Organ Failure Assessment score. CONCLUSIONS: Persistent organ failure is associated with an increase in the rate of limitations in life support independent of the absolute magnitude of Sequential Organ Failure Assessment score, and this association strengthens during the first weeks of treatment. During the first 5 days after acute lung injury onset, limitations were significantly more common in medical ICUs than surgical ICUs.

Age and decisions to limit life support for patients with acute lung injury: a prospective cohort study.

INTRODUCTION: The proportion of elderly Americans admitted to the intensive care unit (ICU) in the last month of life is rising. Hence, challenging decisions regarding the appropriate use of life support are increasingly common. The objective of this study was to estimate the association between patient age and the rate of new limitations in the use of life support, independent of daily organ dysfunction status, following acute lung injury (ALI) onset. METHODS: This was a prospective cohort study of 490 consecutive patients without any limitations in life support at the onset of ALI. Patients were recruited from 11 ICUs at three teaching hospitals in Baltimore, Maryland, USA, and monitored for the incidence of six pre-defined limitations in life support, with adjustment for baseline comorbidity and functional status, duration of hospitalization before ALI onset, ICU severity of illness, and daily ICU organ dysfunction score. RESULTS: The median patient age was 52 (range: 18 to 96), with 192 (39%) having a new limitation in life support in the ICU. Of patients with a new limitation, 113 (59%) had life support withdrawn and died, 53 (28%) died without resuscitation, and 26 (14%) survived to ICU discharge. Each ten-year increase in patient age was independently associated with a 24% increase in the rate of limitations in life support (Relative Hazard 1.24; 95% CI 1.11 to 1.40) after adjusting for daily ICU organ dysfunction score and all other covariates. CONCLUSIONS: Older critically ill patients are more likely to have new limitations in life support independent of their baseline status, ICU-related severity of illness, and daily organ dysfunction status. Future studies are required to determine whether this association is a result of differences in patient preferences by age, or differences in the treatment options discussed with the families of older versus younger patients.

Sex, Race, and HIV Risk Disparities in Discontinuity of HIV Care After Antiretroviral Therapy Initiation in the United States and Canada.

Disruption of continuous retention in care (discontinuity) is associated with HIV disease progression. We examined sex, race, and HIV risk disparities in discontinuity after antiretroviral therapy (ART) initiation among patients in North America. Adults (≥18 years of age) initiating ART from 2000 to 2010 were included. Discontinuity was defined as first disruption of continuous retention (≥2 visits separated by >90 days in the calendar year). Relative hazard ratio (HR) and times from ART initiation until discontinuity by race, sex, and HIV risk were assessed by modeling of the cumulative incidence function (CIF) in the presence of the competing risk of death. Models were adjusted for cohort site, baseline age, and CD4+ cell count within 1 year before ART initiation; nadir CD4+ cell count after ART, but before a study event, was assessed as a mediator. Among 17,171 adults initiating ART, median follow-up time was 3.97 years, and 49% were observed to have ≥1 discontinuity of care. In adjusted regression models, the hazard of discontinuity for patients was lower for females versus males [HR: 0.84; 95% confidence interval (CI): 0.79-0.89] and higher for blacks versus nonblacks (HR: 1.17; 95% CI: 1.12-1.23) and persons with injection drug use (IDU) versus non-IDU risk (HR: 1.33; 95% CI: 1.25-1.41). Sex, racial, and HIV risk differences in clinical retention exist, even accounting for access to care and known competing risks for discontinuity. These results point to vulnerable populations at greatest risk for discontinuity in need of improved outreach to prevent disruptions of HIV care.