Article type
Year
Abstract
Background: Biased randomized controlled trials (RCTs) mislead clinical practice and adversely affect patients' outcomes. A number of methodological features serve as safeguards to limit the introduction of bias into RCTs.
Objectives: To systematically assess the descriptive characteristics, the methodological quality and their time trends in RCTs included in six systematic reviews of anticoagulation in cancer patients.
Methods: We conducted a comprehensive search, including several electronic databases (MEDLINE, EMBASE, ISI the Web of Science, and CENTRAL) until January 2007. We included RCTs in which the intervention and/or comparison consisted of: vitamin K antagonists, unfractionated heparin (UFH), low molecular weight heparin (LMWH), ximelagatran or fondaparinux. We used a standardized form to abstract study descriptive characteristics (publication year, funding, interventions, outcomes and outcome assessment methods) and rate methodological quality (blinding of patients, providers, outcome assessors and data analysts; intention to treat analysis; early stopping for benefit; and percentage follow-up). We performed descriptive analyses and assessed the association between the variables of interest and the year of publication categorized in 5-year intervals using chi-square test.
Results: The title and abstract screening of 3986 citations identified 73 eligible RCTs with 18707 participants. The publication year was in the 1980s, 1990s, and 2000s for 16%, 40% and 44% of studies respectively. English was the language of publication in 97% of studies. The funding agencies were governmental (18%), private not for profit (7%), and private for profit (52%); 34% did not report a funding source. Cancer patients constituted a study subgroup in 37% of studies. The most frequently studied medication class was LMWH (74%). The study of LMWH was associated with later years of publication (p=0.032). Studies reported data for cancer patients for the following outcomes: death (66%), DVT (55%; only 29% reported data on diagnosis triggered by clinical suspicion; the rest reported data on DVT screening), major bleeding (48%), minor bleeding (26%), and thrombocytopenia (21%). Assessment of venous thromboembolism and of major bleeding was associated with later years of publication (p=0.039 and 0.007 respectively). Forty nine percent of trials screened for DVT using at least one of the following tests: venography (22%), Doppler-ultrasound (11%) 125I-fibrinogen-uptake test (15%), impedance plethysmography (8%), or CT scan (1%). The use of the 125I-fibrinogen-uptake test, a test with inferior diagnostic properties, was associated with earlier years of publication (p<0.001). Eight per cent of trials screened for PE using ventilation-perfusion scan. 56% of studies assessed DVT using at least one of the following diagnostic tests: venography (62%), Doppler-ultrasound (33%) 125I-fibrinogen-uptake test (3%), impedance plethysmography (7%), and CT scan (1%). The use of Doppler was associated with later years of publication (p=0.001). 48% of studies assessed PE using at least one of the following diagnostic tests: angiogram (50%), CT scan (15%), ventilation-perfusion scan (55%), and autopsy (14%). The use of CT scan was associated with later years of publication (p=0.002). Descriptive analysis revealed the following features of methodological quality: adequate allocation concealment (59%), patients' blinding (34%), providers' blinding (32%), outcome assessors' blinding (55%), data analysts' blinding (26%), intention to treat analysis (56%), stopping early for benefit (1%). The mean follow-up rate was 95%. None of the quality criteria was associated with the year of publication.
Conclusions: RCTs of anticoagulation in cancer patients appear to use less rigorous outcome assessment methods and to have deficiencies in key methodological features with little improvement over the last 25 years. It is not clear whether this reflects a problem in the conduct or the reporting of these trials. Trialists need to pay closer attention to the design, implementation and reporting of RCTs of anticoagulation in patients with cancer.
Objectives: To systematically assess the descriptive characteristics, the methodological quality and their time trends in RCTs included in six systematic reviews of anticoagulation in cancer patients.
Methods: We conducted a comprehensive search, including several electronic databases (MEDLINE, EMBASE, ISI the Web of Science, and CENTRAL) until January 2007. We included RCTs in which the intervention and/or comparison consisted of: vitamin K antagonists, unfractionated heparin (UFH), low molecular weight heparin (LMWH), ximelagatran or fondaparinux. We used a standardized form to abstract study descriptive characteristics (publication year, funding, interventions, outcomes and outcome assessment methods) and rate methodological quality (blinding of patients, providers, outcome assessors and data analysts; intention to treat analysis; early stopping for benefit; and percentage follow-up). We performed descriptive analyses and assessed the association between the variables of interest and the year of publication categorized in 5-year intervals using chi-square test.
Results: The title and abstract screening of 3986 citations identified 73 eligible RCTs with 18707 participants. The publication year was in the 1980s, 1990s, and 2000s for 16%, 40% and 44% of studies respectively. English was the language of publication in 97% of studies. The funding agencies were governmental (18%), private not for profit (7%), and private for profit (52%); 34% did not report a funding source. Cancer patients constituted a study subgroup in 37% of studies. The most frequently studied medication class was LMWH (74%). The study of LMWH was associated with later years of publication (p=0.032). Studies reported data for cancer patients for the following outcomes: death (66%), DVT (55%; only 29% reported data on diagnosis triggered by clinical suspicion; the rest reported data on DVT screening), major bleeding (48%), minor bleeding (26%), and thrombocytopenia (21%). Assessment of venous thromboembolism and of major bleeding was associated with later years of publication (p=0.039 and 0.007 respectively). Forty nine percent of trials screened for DVT using at least one of the following tests: venography (22%), Doppler-ultrasound (11%) 125I-fibrinogen-uptake test (15%), impedance plethysmography (8%), or CT scan (1%). The use of the 125I-fibrinogen-uptake test, a test with inferior diagnostic properties, was associated with earlier years of publication (p<0.001). Eight per cent of trials screened for PE using ventilation-perfusion scan. 56% of studies assessed DVT using at least one of the following diagnostic tests: venography (62%), Doppler-ultrasound (33%) 125I-fibrinogen-uptake test (3%), impedance plethysmography (7%), and CT scan (1%). The use of Doppler was associated with later years of publication (p=0.001). 48% of studies assessed PE using at least one of the following diagnostic tests: angiogram (50%), CT scan (15%), ventilation-perfusion scan (55%), and autopsy (14%). The use of CT scan was associated with later years of publication (p=0.002). Descriptive analysis revealed the following features of methodological quality: adequate allocation concealment (59%), patients' blinding (34%), providers' blinding (32%), outcome assessors' blinding (55%), data analysts' blinding (26%), intention to treat analysis (56%), stopping early for benefit (1%). The mean follow-up rate was 95%. None of the quality criteria was associated with the year of publication.
Conclusions: RCTs of anticoagulation in cancer patients appear to use less rigorous outcome assessment methods and to have deficiencies in key methodological features with little improvement over the last 25 years. It is not clear whether this reflects a problem in the conduct or the reporting of these trials. Trialists need to pay closer attention to the design, implementation and reporting of RCTs of anticoagulation in patients with cancer.