Lawani Moulikatou Adouni, Valéra Béatriz, Fortier-Brochu Émilie, Légaré France, Carmichael Pierre-Hugues, Côté Luc, Voyer Philippe, Kröger Edeltraut, Witteman Holly, Rodriguez Charo, Giguere Anik M C
Department of Family and Emergency Medicine, Laval University, Pavillon Ferdinand-Vandry, Room 2881, 1050 Avenue de la Médecine, Quebec city, QC, G1V 0A6, Canada.
Quebec Excellence Center on Aging, St-Sacrement Hospital, Room L2, 1050, chemin Sainte-Foy, Quebec city, QC, G1S 4L8, Canada.
Syst Rev. 2017 Mar 15;6(1):56. doi: 10.1186/s13643-017-0446-2.
Decision support tools build upon comprehensive and timely syntheses of literature. Rapid reviews may allow supporting their development by omitting certain components of traditional systematic reviews. We thus aimed to describe a rapid review approach underlying the development of decision support tools, i.e., five decision boxes (DB) for shared decision-making between seniors living with dementia, their caregivers, and healthcare providers.
We included studies based on PICO questions (Participant, Intervention, Comparison, Outcome) describing each of the five specific decision. We gave priority to higher quality evidence (e.g., systematic reviews). For each DB, we first identified secondary sources of literature, namely, clinical summaries, clinical practice guidelines, and systematic reviews. After an initial extraction, we searched for primary studies in academic databases and grey literature to fill gaps in evidence. We extracted study designs, sample sizes, populations, and probabilities of benefits/harms of the health options. A single reviewer conducted the literature search and study selection. The data extracted by one reviewer was verified by a second experienced reviewer. Two reviewers assessed the quality of the evidence. We converted all probabilities into absolute risks for ease of understanding. Two to five experts validated the content of each DB. We conducted descriptive statistical analyses on the review processes and resources required.
The approach allowed screening of a limited number of references (range: 104 to 406/review). For each review, we included 15 to 26 studies, 2 to 10 health options, 11 to 62 health outcomes and we conducted 9 to 47 quality assessments. A team of ten reviewers with varying levels of expertise was supported at specific steps by an information specialist, a biostatistician, and a graphic designer. The time required to complete a rapid review varied from 7 to 31 weeks per review (mean ± SD, 19 ± 10 weeks). Data extraction required the most time (8 ± 6.8 weeks). The average estimated cost of a rapid review was C$11,646 (SD = C$10,914).
This approach enabled the development of clinical tools more rapidly than with a traditional systematic review. Future studies should evaluate the applicability of this approach to other teams/tools.
决策支持工具基于对文献的全面且及时的综合分析。快速综述可能通过省略传统系统综述的某些组成部分来支持其开发。因此，我们旨在描述一种用于开发决策支持工具的快速综述方法，即用于痴呆症患者、其照护者和医疗服务提供者之间共同决策的五个决策框（DB）。
我们纳入基于PICO问题（参与者、干预措施、对照、结局）的研究，这些问题描述了五个具体决策中的每一个。我们优先考虑高质量证据（例如系统综述）。对于每个决策框，我们首先确定文献的二级来源，即临床总结、临床实践指南和系统综述。在初步提取后，我们在学术数据库和灰色文献中搜索原始研究以填补证据空白。我们提取了研究设计、样本量、人群以及健康选项的受益/危害概率。由一名评审员进行文献检索和研究选择。由另一名经验丰富的评审员对一名评审员提取的数据进行核实。两名评审员评估证据质量。为便于理解，我们将所有概率转换为绝对风险。由两到五名专家对每个决策框的内容进行验证。我们对综述过程和所需资源进行了描述性统计分析。
该方法允许筛选有限数量的参考文献（范围：每次综述104至406篇）。每次综述我们纳入15至26项研究、2至10个健康选项、11至62个健康结局，并且我们进行了9至47次质量评估。一个由十名专业水平各异的评审员组成的团队在特定步骤得到了信息专家、生物统计学家和平面设计师的支持。完成一次快速综述所需的时间为每次综述7至31周（均值±标准差，19±10周）。数据提取耗时最长（8±6.8周）。一次快速综述的平均估计成本为11,646加元（标准差 = 10,914加元）。
与传统系统综述相比，这种方法能够更快地开发临床工具。未来的研究应评估该方法对其他团队/工具的适用性。
Zotero 插件