Statistical computation of numerical indices for grading the severity of asthma
1966; Elsevier BV; Volume: 37; Issue: 1 Linguagem: Inglês
10.1016/0021-8707(66)90105-5
ISSN1878-2213
AutoresConstantine J. Falliers, Hyman Chai, Donald T. Searls, Charles O. Robertson,
Tópico(s)Scientific Measurement and Uncertainty Evaluation
ResumoAn accurate and reproducible system of measurements is essential for the longitudinal assessment of changes in asthma and for the evaluation of the role of various provocative or therapeutic agents upon it. To achieve this goal, a statistical index of improvement “I,” and a numerical factor “C,” expressing the patient's condition have been developed, using an IBM 709 electronic computer system. The clinical history and physical examination, taken weekly and graded on a 0–40 scale, the peak flow expiratory rates measured with Wright's Peak Flowmeter, and the type and dosage of medication needed by the patient were the main variables utilized. A discriminant function analysis revealed that for a linear model the lowest estimated probability of misclassification with the fewest variates was a combination of history scores and peak flow rates. A nonlinear model was investigated for this combination and an even lower estimated probability of misclassification was attained. Scaling the nonlinear models for a range of 0–100 for “C,” and +100 to −100 for “I” produced two simple tables which can be used for a prompt calculation of “I” and “C.” The observed frequency of various values during a two-year project in clinical pharmacology is reported.
Referência(s)