The analytic framework is a schematic outline used to define the population, preventive service, interventions, and health outcomes considered in the review (Appendix Figure 1). The arrows represent key questions that the evidence must answer to demonstrate the chain of logic from the preventive service (AAA screening) to improved health outcomes (reduced AAA-specific morbidity, mortality, or both). The key questions were determined in conjunction with the USPSTF liaisons.
We searched the topic of AAA in MEDLINE® (January 1994 to May 2004), the Cochrane Database of Systematic Reviews (2004, Issue 1), and the Cochrane Controlled Trials Register (January 1994 to May 2004) to identify studies about the effectiveness of AAA screening in population-based settings, screening harms, and harms of treatment for AAAs at least 5.5 cm. We included only data published in full-article form. In addition, we obtained articles from the reference lists of pertinent studies and reviews and from expert recommendations. The flow of articles is shown in Appendix Figure 2.
All searches included only English-language abstracts. Full search strategies may be found in the full review at http://www.preventiveservices.ahrq.gov. The search for key question 1a crossed terms for randomized, controlled trials with the terms abdominal aortic aneurysm and mass screening or screen. Terms for cohort, cross-sectional, epidemiologic, or longitudinal studies were crossed with mass screening and abdominal aortic aneurysm for the search for key question 1b. The search for key question 2 crossed psychological harms terms; false-positive or false-negative reactions; and the terms risk assessment, predictive value of tests, attitude to health, psychiatric status rating scales, health status, health status indicators, severity of illness index, and quality of life with the terms abdominal aortic aneurysm and mass screening or screen to locate articles pertaining to the harms of screening for AAA. The search for key question 4 used the randomized, controlled trial or review terms and the term abdominal aortic aneurysm with surgical terms, prognosis terms, or treatment outcome terms to locate articles relating to harms of AAA repair.
Two reviewers individually reviewed each abstract using the inclusion and exclusion criteria. A subset of full-text articles was selected for further review. We resolved disagreements on inclusion or exclusion of individual studies by consensus and by examining each study's relevance to the key question. One reviewer abstracted relevant information into an evidence table, and another reviewer checked the tables for accuracy.
The common inclusion criteria used for all key questions were English-language publication, presentation of original data, and use of human subjects. For key question 1a, only randomized population-based trials that compared screening with unscreened controls and reported AAA-related mortality were included. For key question 1b, we included followup or cohort studies that involved repeated scanning of a representative population. For key questions 2 and 4, we included studies of harms from randomized, controlled trials, or retrospective or prospective cohort studies with comparative data.
We assessed the quality of studies on the basis of USPSTF criteria26 to rate studies as "good," "fair," or "poor." Quality criteria for randomized, controlled trials included assembly of comparable groups, maintenance of comparable groups, important differential loss to followup or overall high loss to followup, valid and reliable measurements, clear definition of interventions, all important outcomes considered, and intention-to-treat analysis. In general, a good study meets all criteria, a fair study does not meet all criteria but is judged to have no fatal flaw, and a poor study contains a fatal flaw. The final quality rating was assigned by the consensus of the investigator team.
Data were abstracted from each study considered for inclusion in the review. Information abstracted fell into 3 categories: study quality criteria as listed earlier; study characteristics, such as study identification, participants, and intervention and control conditions; and study outcomes, including harms. Only studies receiving a fair or good rating were included in the evidence synthesis.