Possible Bias in Treatment Effectiveness Evaluating by Evidence-based Medicine Method in Serial Studies

<p>In many countries in the world economical and organization decisions in healthcare such as drugs approval, health technology assessment, financing (providing) of medical interventions are based on the methods of science being used in evidence-based medicine.</p><p>In this diploma paper the biases that can influence on efficacy assessment of medical interventions are analyzed. Several biases were included in the analysis: a publication bias, a bias in recording of adverse events of medical interventions, a problem of the quality o descriptions of treatments and a problem of accepting the false-positive results of clinical trials.</p><p>In the paper the following tasks were assigned: to analyze existing biases in application of methods of science for efficacy assessment of medical interventions; to develop an algorithm for calculation of probability of false-positive result in the serial trials; to analyze existing serial studies for calculation of probability that obtained positive result is false-positive; to form recommendations for consideration of possible biases in treatment effectiveness evaluating in serial trials.</p><p>The empiric part of the paper was based on the registers of clinical trials, publication of results of registered trials or information that results weren&rsquo;t published, information about new drugs approval.</p><p>In the paper 70 trials, registered in the registers of clinical trials and conducted for FDA approval of five drugs, were analyzed. The probabilities of false-positive results were calculated based on rates of type I error and type II error in trials, proportion of trials in series that showed positive result and having used Bayes formula. As consistent with the findings of the investigation the probability of false-positive result of a trial in the series was higher than p-value in each individual trial and took the value form 0,005% to 24%.</p>