Desfechos propostos em registros de ensaios clínicos no tratamento do transtorno bipolar
Data
2023-03-30
Autores
Nascimento, Luiza Gomes de Campos 
Orientadores
Melo, Daniela Oliveira de
Tipo
Dissertação de mestrado
Título da Revista
ISSN da Revista
Título de Volume
Resumo
Introdução: O registro prospectivo de estudos clínicos possibilita melhorar a certeza da evidencia e a transparência na pesquisa, algo particularmente relevante na área de saúde mental – em que os desfechos subjetivos são frequentes e baseados em escalas e testes psicológicos - como ocorre com o transtorno bipolar (TB). Mas para que o registro de estudos clínicos tenham impacto, é necessário que as informações estejam bem descritas no registro, em especial com relação à apropriada descrição dos desfechos e modo de avaliá-los. Objetivo: Identificar, caracterizar e descrever os desfechos e a completude do seu relato em registros de ensaios clínicos de intervenções farmacológicas para o tratamento de pacientes com TB. Material e Métodos: Registros de estudos de fase III, cadastrados na plataforma ClinicalTrials.gov, foram recuperados, sem restrição de data, e categorizados de acordo com data de início do estudo, data do registro e se foram realizados de forma prospectiva, tipo de TB, idade dos participantes, fonte de financiamento, status do recrutamento, disponibilidade dos resultados na plataforma e quais as intervenções avaliadas (medicamento e sua classe farmacológica). Os desfechos foram extraídos e classificados de acordo com: aspecto avaliado (eficácia, segurança e outros); fase de manifestação do transtorno ou equivalente; tipo de variável, métrica empregada e o parâmetro de avaliação; o objeto de avaliação (detalhamento do que foi avaliado), tipo de desfecho (latente ou objetivo), qual a ferramenta utilizada na avaliação de desfechos latentes e o tempo de seguimento para cada desfecho. As informações coletadas serviram de base na avaliação da completude do relato dos desfechos. Resultados e Discussão: foram analisados 247 registros que incluíram 1.594 desfechos. O número de desfechos variou de zero a 51 por registro, com média 6,5 ± 7,2 desfechos/registro. A frequência daqueles relacionados à eficácia foi muito superior aos relacionados à segurança (1.131; 71% versus 332; 20,8%). A maioria das variáveis propostas eram do tipo latente (1.258; 78,9%) e avaliadas de maneira quantitativa contínua (1.095; 68,7%) – principalmente caracterizada por avaliações de mudança em relação ao valor basal (1.003; 62,9%). O objeto de avaliação mais frequente foi a melhora dos sintomas bipolares avaliados por escala (641; 40,2%), sendo a Clinical Global Impression Scale (CGI) a mais empregada (231;17,1%). Os desfechos foram avaliados em sua maioria (610; 38,3%) num curto prazo (3 a 8 semanas). Um total 43,9% (669) dos desfechos foi reportado de maneira completa, ao considerar apenas os desfechos primários, a proporção de relatos completos foi menor 33,6% (83). Não parece haver relação entre o número de desfechos no registro e a completude, pois a quantidade de desfechos não foi maior entre aqueles registros com nenhum desfecho completo. Conclusão: Na nossa amostra, de 1.594 desfechos, 669 (43,9%) tiveram a descrição classificada como completa; entre os 247 registros incluídos, apenas 74 registros (30,0%) tiveram 50% ou mais de seus desfechos descritos de forma completa e a maioria 143 (57,9%) não apresentou descrição completa para nenhum de seus desfechos, ainda que a mediana de desfechos propostos entre esses registros tenha sido 1. Apesar dos avanços no registro de estudos, a descrição incompleta ou parcial de como um desfecho será avaliado permanece um problema que pode favorecer o viés de relatos seletivo de desfechos e prejudicar a transparência na pesquisa.
Introduction: The prospective registration of clinical studies makes it possible to improve the certainty of evidence and transparency in the research, something particularly relevant in the area of mental health - where subjective outcomes are frequent and based on scales and psychological tests - as occurs with bipolar disorder (BD). But for the registration of clinical trials to have an impact, it is necessary that the information is well described in the registration, especially with regard to the appropriate description of outcomes and how to evaluate them. Objective: To identify, characterize and describe the outcomes and the completeness of their reporting in clinical trial registries of pharmacological interventions for the treatment of patients with TB. Material and Methods: Records of phase III studies, registered on the ClinicalTrials.gov platform, were retrieved, without date restriction, and categorized according to study start date, registration date and whether they were carried out prospectively, type of TB, participants' age, funding source, recruitment status, availability of results on the platform and which interventions were evaluated (drug and its pharmacological class). Outcomes were extracted and classified according to: evaluated aspect (efficacy, safety and others); disorder manifestation phase or equivalent; type of variable, metric used and evaluation parameter; the object of evaluation (detailing what was evaluated), type of outcome (latent or objective), which tool was used to assess latent outcomes and the follow-up time for each outcome. The information collected served as the basis for evaluating the completeness of the report of outcomes. Results and Discussion: it was analyzed 247 records including 1,594 outcomes. The number of outcomes ranged from zero to 51 per record, with a mean of 6.5 ± 7.2 outcomes/record. The frequency of those related to efficacy was much higher than those related to safety (1,131; 71% versus 332; 20.8%). Most of the planned variables were latent type (1,258; 78.9%) and evaluated in a continuous quantitative way (1,095; 68.7%) – mainly characterized by assessments of change related to the baseline value (1,003; 62.9%). The most frequent item of evaluation was the improvement of bipolar symptoms assessed by scale (641; 40.2%), with the Clinical Global Impression Scale (CGI) being the most used (231; 17.1%). Outcomes were mostly evaluated (610; 38.3%) in the short term (3 to 8 weeks). A total of 43.9% (669) of the outcomes were fully reported, when considering only the primary outcomes, the proportion of complete reports was less than 33.6% (83). There does not seem to be a relationship between the number of outcomes in the registries and completeness, as these number of outcomes was not greater among those studies with no complete outcome. Conclusion: In our sample, of 1,594 outcomes, 669 (43.9%) had the description classified as complete; among the 247 registries included, only 74 registries (30.0%) had 50% or more of their outcomes fully described and the majority 143 (57.9%) did not present a complete description for any of their outcomes, even though the median of proposed outcomes among these registries was 1. Despite advances in the registration of studies, the incomplete or partial description of how an outcome will be evaluated remains a problem that can favor the bias of selective reporting of outcomes and impair research transparency.
Introduction: The prospective registration of clinical studies makes it possible to improve the certainty of evidence and transparency in the research, something particularly relevant in the area of mental health - where subjective outcomes are frequent and based on scales and psychological tests - as occurs with bipolar disorder (BD). But for the registration of clinical trials to have an impact, it is necessary that the information is well described in the registration, especially with regard to the appropriate description of outcomes and how to evaluate them. Objective: To identify, characterize and describe the outcomes and the completeness of their reporting in clinical trial registries of pharmacological interventions for the treatment of patients with TB. Material and Methods: Records of phase III studies, registered on the ClinicalTrials.gov platform, were retrieved, without date restriction, and categorized according to study start date, registration date and whether they were carried out prospectively, type of TB, participants' age, funding source, recruitment status, availability of results on the platform and which interventions were evaluated (drug and its pharmacological class). Outcomes were extracted and classified according to: evaluated aspect (efficacy, safety and others); disorder manifestation phase or equivalent; type of variable, metric used and evaluation parameter; the object of evaluation (detailing what was evaluated), type of outcome (latent or objective), which tool was used to assess latent outcomes and the follow-up time for each outcome. The information collected served as the basis for evaluating the completeness of the report of outcomes. Results and Discussion: it was analyzed 247 records including 1,594 outcomes. The number of outcomes ranged from zero to 51 per record, with a mean of 6.5 ± 7.2 outcomes/record. The frequency of those related to efficacy was much higher than those related to safety (1,131; 71% versus 332; 20.8%). Most of the planned variables were latent type (1,258; 78.9%) and evaluated in a continuous quantitative way (1,095; 68.7%) – mainly characterized by assessments of change related to the baseline value (1,003; 62.9%). The most frequent item of evaluation was the improvement of bipolar symptoms assessed by scale (641; 40.2%), with the Clinical Global Impression Scale (CGI) being the most used (231; 17.1%). Outcomes were mostly evaluated (610; 38.3%) in the short term (3 to 8 weeks). A total of 43.9% (669) of the outcomes were fully reported, when considering only the primary outcomes, the proportion of complete reports was less than 33.6% (83). There does not seem to be a relationship between the number of outcomes in the registries and completeness, as these number of outcomes was not greater among those studies with no complete outcome. Conclusion: In our sample, of 1,594 outcomes, 669 (43.9%) had the description classified as complete; among the 247 registries included, only 74 registries (30.0%) had 50% or more of their outcomes fully described and the majority 143 (57.9%) did not present a complete description for any of their outcomes, even though the median of proposed outcomes among these registries was 1. Despite advances in the registration of studies, the incomplete or partial description of how an outcome will be evaluated remains a problem that can favor the bias of selective reporting of outcomes and impair research transparency.