Image: Wikipedia, Receiver Operating Characteristic
Statistics are central to the human usability of data. We use statistics to describe and characterize findings and to make inferences about larger tracts of data from their smaller samples. Biostatistics focuses on those topics most often encountered in the medical and biological sciences. Such topics range from descriptive statistics, hypothesis testing, and t-tests to linear regression, correlation and non-parametric tests.
Building upon my surgical background and clinical decision support system project, I mocked a realistic study assessing the impact of clinical test sensitivity and specificity on clinician ordering practices in the diagnosis of acute pulmonary embolism (PE). Computerized physician order entry (CPOE) is commonplace and serves as a viable medium for decision support interventions.
CPOE is regularly used to guide medication dosing for patient with renal insufficiency. I suggest that similar principles can be used to address concerns in health economics and efficient delivery of care.
The display of the sensitivity and specificity of diagnostic tests, such as ventilation-perfusion lung scan, computed tomography angiography of the chest, chest radiograph, echocardiography, d-dimer, for the diagnosis of a PE in surgical patient in the CPOE can supplement patient knowledge at time of ordering. With regular maintenance of the test sensitivity and specificity knowledge base, this could become a powerful tool to improve upon the time, cost, and number of tests to diagnosis in PE and other disease processes. As the number of available medical tests and research around the power of each test grows, the use of intelligently placed statistics could lead the way towards more efficient health care delivery.
CPOE as a knowledge platform: Impact of diagnostic test sensitivity and specificity on the number of tests required in diagnosis of acute pulmonary embolism
(Note: This paper uses simulated data.)