Avaliação dos ensaios clínicos randomizados em Ginecologia e Obstetrícia com publicação de protocolo de pesquisa
Carregando...
Data
2020-02-28
Autores
Nonato, Karen Bergoce
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
INTRODUÇÃO: A publicação dos resultados de ensaios clínicos randomizados (ECR) é fundamental para que se possa tomar decisões com base nas melhores evidências sobre cuidados de saúde. Alterações nos desfechos, no tamanho da amostra e nos critérios de inclusão e exclusão podem causar alterações dos resultados, levando a vieses e que levam a interpretações equivocadas sobre essas pesquisas. OBJETIVOS: Verificar a frequência de protocolos de pesquisa (PP) entre os ECR publicados e analisar as discrepâncias das informações dos estudos com as contidas em seus respectivos PP e no registro de ensaios. MÉTODO: Pesquisa metodológica de ECR em Ginecologia e Obstetrícia publicados na base de dados PubMed, no período de janeiro de 2013 a maio de 2019, com restrição aos dez periódicos com maior fator de impacto e categorizados como “Gynecology and Obstetrics”no Journal Citation Report 2018. Foram incluídos ECR, podendo ser estudos de eficácia/efetividade, de segurança ou estudos pilotos. Foram selecionados aleatoriamente ECR sem PP publicado na proporção de 1:1 em relação ao ECR com PP identificados. Calculamos um tamanho amostral de 380 ECRs para verificar a frequência PP. Os ECR para o estudo foram selecionados aleatoriamente. Os dados foram extraídos por dois autores de forma independente, com um terceiro autor para solucionar os conflitos sobre a extração. Foram verificadas as proporções de (i) ECR com PP publicado em relação ao total de ECR incluídos no estudo e (ii) discrepâncias (desfecho, declaração de financiamento e conflito de interesse, tamanho da amostra, critérios de inclusão e exclusão e finalização antecipada do estudo). Comparamos a proporção de discrepâncias entre o registro dos ECR e suas publicações de acordo com a publicação ou não do PP. Os resultados são apresentados em porcentagens e intervalos de confiança de 95%. Foi utilizado o teste de qui-quadrado para comparação das proporções, com correção de Pearson quando necessário e correção de Bonferroni para comparações múltiplas. Significância de 5%. RESULTADOS PRINCIPAIS: Dentre os 2546 estudos identificados pela estratégia de busca, selecionamos 384 ECR. A frequência de PP publicados foi de 6,3% (IC95% 4,2 – 9,1). Independentemente da publicação do PP, 8,3% (IC95% 6,0%-11,5%) dos artigos não relataram o número e base de registros de ECR. Estudos com PP publicado foram realizados mais frequentemente na Europa (87,4% vs 33,3%) e financiados em maior proporção por agências não associadas a indústria (79,2% vs 49,4%) quando comparados aos ECR sem publicação do PP. Com exceção da declaração sobre o financiamento da pesquisa e conflito de interesse, houve discrepância em todos os itens estudados nos ECR em relação a seus PP. Na avaliação das discrepâncias entre registro e ECR, alterações nos desfechos e nos critérios de inclusão e exclusão foram mais frequentes nos estudos sem PP (8,3% vs 33,3% e 16,7% vs 54,2%, respectivamente) CONCLUSÕES: A frequência de publicação de PP de ECR em Ginecologia e Obstetrícia é muito baixa. Há discrepância entre registro e ECR, independentemente da publicação do PP. Entretanto, algumas discrepâncias são mais frequentes entre os estudos sem PP. Melhor transparência pode ser alcançada quando os protocolos forem disponibilizados ao público.
INTRODUCTION: The report of randomized clinical trials (RCTs) is essential for making good decisions based on the best evidence on health care. Changes in outcomes, sample size and inclusion and exclusion criteria can cause bias, leading to research misinterpretations. OBJECTIVES: To verify the frequency of studies protocols (SP) between reported RCTs and to analyze the discrepancies between RCTs and SP and RCTs and trial register. METHOD: Systematic survey in RCT about Gynecology and Obstetrics published in the PubMed database, from January 2013 to May 2019, with restriction to the ten journals with the highest impact factor and categorized as “Gynecology and Obstetrics” in the Journal Citation Report 2018. RCTs, that were included, could be effectivity or efficacy, safety or pilot and feasibility studies. RCTs without published PP were randomly selected in the proportion of 1: 1 in relation to the RCT with published SP. 380 RCTs were needed to estimate the SP frequency. The RCTs were selected randomly between the RCTs identified in the PubMed database. The data were extracted by two authors independently, with a third author to resolve conflicts about selection of RCT and data extraction. The proportions of (i) RCTs with published PP were verified in relation to the total RCTs included in the study and (ii) discrepancies (outcome, financial sponsor and conflict of interest, sample size, inclusion and exclusion criteria and early completion of the study). We compared the proportion of discrepancies between the registration of RCTs and their publications according to published SP. The results are presented in percentages and 95% confidence intervals. We used chi-square test to compare proportions, with Pearson correction when necessary and Bonferroni correction for multiple comparisons. Significance of 5%. MAIN RESULTS: Among the 2546 studies identified by the search strategy, we selected randomly 384 ECR. The frequency of published SP was 6.3% (95% CI 4.2 - 9.1). Regardless of the published SP, we identified 8.3% (95% CI 6.0% -11.5%) with no report of the number and basis of RCT data register. Studies with published SP were carried out more frequently in Europe (87.4% vs 33.3%) and financed by sponsor agencies not associated with the industry (79.2% vs 49.4%) when compared to RCTs without published SP. Except the statement on research funding and conflict of interest, there was a discrepancy in all items studied in the RCTs in relation to their SP. In assessing discrepancies between registration and RCT, Discrepancies in outcomes and inclusion and exclusion criteria were more frequent in studies without PP (8.3% vs 33.3% and 16.7% vs 54.2%, respectively) when evaluated the RCT register. CONCLUSIONS: The frequency of published SP in Gynecology and Obstetrics is very low. There is a discrepancy between registration and RCT, regardless of the publication of the PP. However, some discrepancies are more frequent among studies without PP. Better transparency can be achieved when the protocols are made available to the public.
INTRODUCTION: The report of randomized clinical trials (RCTs) is essential for making good decisions based on the best evidence on health care. Changes in outcomes, sample size and inclusion and exclusion criteria can cause bias, leading to research misinterpretations. OBJECTIVES: To verify the frequency of studies protocols (SP) between reported RCTs and to analyze the discrepancies between RCTs and SP and RCTs and trial register. METHOD: Systematic survey in RCT about Gynecology and Obstetrics published in the PubMed database, from January 2013 to May 2019, with restriction to the ten journals with the highest impact factor and categorized as “Gynecology and Obstetrics” in the Journal Citation Report 2018. RCTs, that were included, could be effectivity or efficacy, safety or pilot and feasibility studies. RCTs without published PP were randomly selected in the proportion of 1: 1 in relation to the RCT with published SP. 380 RCTs were needed to estimate the SP frequency. The RCTs were selected randomly between the RCTs identified in the PubMed database. The data were extracted by two authors independently, with a third author to resolve conflicts about selection of RCT and data extraction. The proportions of (i) RCTs with published PP were verified in relation to the total RCTs included in the study and (ii) discrepancies (outcome, financial sponsor and conflict of interest, sample size, inclusion and exclusion criteria and early completion of the study). We compared the proportion of discrepancies between the registration of RCTs and their publications according to published SP. The results are presented in percentages and 95% confidence intervals. We used chi-square test to compare proportions, with Pearson correction when necessary and Bonferroni correction for multiple comparisons. Significance of 5%. MAIN RESULTS: Among the 2546 studies identified by the search strategy, we selected randomly 384 ECR. The frequency of published SP was 6.3% (95% CI 4.2 - 9.1). Regardless of the published SP, we identified 8.3% (95% CI 6.0% -11.5%) with no report of the number and basis of RCT data register. Studies with published SP were carried out more frequently in Europe (87.4% vs 33.3%) and financed by sponsor agencies not associated with the industry (79.2% vs 49.4%) when compared to RCTs without published SP. Except the statement on research funding and conflict of interest, there was a discrepancy in all items studied in the RCTs in relation to their SP. In assessing discrepancies between registration and RCT, Discrepancies in outcomes and inclusion and exclusion criteria were more frequent in studies without PP (8.3% vs 33.3% and 16.7% vs 54.2%, respectively) when evaluated the RCT register. CONCLUSIONS: The frequency of published SP in Gynecology and Obstetrics is very low. There is a discrepancy between registration and RCT, regardless of the publication of the PP. However, some discrepancies are more frequent among studies without PP. Better transparency can be achieved when the protocols are made available to the public.
Descrição
Palavras-chave
Ginecologia, Obstetrícia, Ensaio clínico controlado aleatório, Ensaio clínico randomizado, Protocolo de pesquisa, Projeto de pesquisa, Gynecology, Obstetrics, Randomized controlled trials, Clinical trial protocol, Clinical protocols