RESUMO
In clinical research it is impossible and inefficient to study all patients with a specific pathology, so it is necessary to study a sample of them. The estimation of the sample size before starting a study guarantees the stability of the results and allows us to foresee the feasibility of the study depending on the availability of patients and cost. The basic structure of sample size estimation is based on the premise that seeks to demonstrate, among other cases, that the observed difference between two or more maneuvers in the subsequent state is real. Initially, it requires knowing the value of the expected difference (δ) and its data variation (standard deviation). These data are usually obtained from previous studies. Then, other components must be considered: a (alpha), percentage of error in the assertion that the difference between means is real, usually 5 %; and ß, error rate accepting the claim that the no-difference between the means is real, usually ranging from 15 to 20 %. Finally, these values are substituted into the formula or in an electronic program for estimating sample size. While summary and dispersion measures vary with the type of variable according to the outcome, the basic structure is the same.
Assuntos
Projetos de Pesquisa/estatística & dados numéricos , Tamanho da Amostra , HumanosRESUMO
OBJECTIVE: Patients with chronic heart failure (HF) develop important changes in body composition. Nevertheless, the usual methods of body composition assessment can be misleading in patients with HF because tissue hydration is altered. Bioelectrical impedance vector analysis (BIVA) works without making any assumption about constant soft tissue hydration. In this study, patients with HF and systolic dysfunction (HFS) and preserved systolic function (HFPSF) underwent a body composition evaluation by the BIVA method; the comparison was done between New York Heart Association (NYHA) functional classes I-II and III-IV. METHODS: We studied 243 patients with HF, 140 (101 in NYHA I-II and 39 in III-IV) with HFS and 103 (67 in NYHA I-II and 36 in II-IV) with HFPSF. Whole-body bioelectrical impedance was measured using BodyStat QuadScan 4000, which is tetrapolar and multiple-frequency equipment. RESULTS: In both HF categories, reactance and phase angle were significantly lower, the impedance ratio of 200 kHz to that at 5 kHz was higher, and had significantly shorter and downsloping impedance vector in the NYHA III-IV group compared with the NYHA I-II group by gender. CONCLUSION: Bioelectrical impedance analysis allows an easier evaluation of body composition and this might be particularly useful to stratify the severity of HF.