Background

Patients admitted to the ICU are the sickest and most complex in the health care system often requiring expensive lifesaving technologies and interventions (e.g., invasive monitoring, intubation and mechanical ventilation, vasoactive medications, dialysis etc.)(Jacobs and Noesworthy, 1990). Further, care within an intensive care unit is provided by a collaborative team of health professionals, including nurses, respiratory therapists, pharmacists, occupational and physiotherapists, and social workers. Many efforts have been made to develop ICU metrics to reflect the performance of a healthcare system. One definition of a physician’s clinical performance (PCP) is “the quantitative assessment of physician performance based on the rates at which their patients experience certain outcomes of care and the rates at which physicians adhere to evidence-based processes of care during their actual practice of medicine”(Street, 2006). Traditionally, physicians participate in performance evaluations that are called 360 evaluations. The 360 model of feedback utilizes information from self-assessment, colleagues, non-physicians and patients. These assessments are useful, but fail to consider additional sources of data (e.g. patient specific outcomes including ICU or hospital length of stay, complications, and mortality) that may provide a more encompassing picture of physician performance.  We are interested in developing a model for assessing the performance of an individual physician. Performance is a confluence of multiple factors, particularly knowledge, skills, and behavioral competence. PCP assessment is beneficial to ensuring high-quality medical care, and may inform quality improvement initiatives amongst physicians.

Physician Participants:

In this dataset, each physician had 360 feedback evaluations completed anonymously by a random sample of allied health professionals working in the ICU, and peer physicians including medical directors, physicians from non-critical care disciplines. Questions are linked to 7 domains (Medical Expert, Communicator, Leader, Advocate, Professional, Scholar, and Collaborator). Each question was scored on a 5-points likert scale.  


Physician characteristics available included: (1) age, (2) background medical discipline before critical care training, (3) an administrative leader defined as a physician, (4) an education leader in the university, (5) university rank.

Patient Participants:

At hospital discharge, trained health record technicians abstract patient-specific data, including demographic, clinical, and administrative information. All multisystems ICU’s conform to the same standards for nursing, respiratory therapy and other allied health care professional training, have uniform departmental policies and protocols, standard equipment (eg, monitoring, drug infusion, and ventilators), and a standard drug formulary. However, each ICU might have different working environments, including variability in patient profiles (some which may not be measured/available e.g. socio-economic status, ethnic background, chronic disease burden). 

Patient Status and Outcome Scores:

APACHE II is an ICU-specific acuity of illness score calculated at the time of ICU admission based on the derangement of 12 physiological variables, age, and a chronic health score.(Naved et al., 2011) 
The SOFA (Sepsis-associated organ failure assessment) score was developed by an expert consensus panel to describe the severity of organ failure at and following ICU admission. In addition, SOFA was designed to complement the acuity of illness score.(Lambden et al., 2019).