SummaryWe carried out a meta-analysis of observational case-control studies published before May 2004 to assess the degree of association between Chlamydophila pneumoniae (Cp) infection and PAOD. A search of the Medline database was performed using atherosclerosis and "Chlamyd* pneumoniae" as keywords. Strict criteria were applied for the selection of case studies, which had to be studies of Cp seroprevalence or of Cp detection in patients versus controls. Forty-three published studies that met these criteria were selected. An association between PAOD and Cp was revealed by immunohistochemical analysis (OR=15.4, 95%CI=5.0–46.9) and nested PCR studies of arterial biopsies (OR=4.3, 95%CI=1.8–10), by PCR study of non-arterial samples (OR=2.9, 95%CI=1.2–7.0), by other direct-detection tests (OR=16.7, 95%CI=7.0–39.8), and by ELISA and MIF tests to detect high IgG (OR=2, 95%CI=1.1–3.5 and OR=1.7, 95%CI=1.0–2.9, respectively) and IgA (OR=1.9, 95%CI=1.1–3.4 and OR=1.5, 95%CI=1.1–2.0, respectively) titers. No significant association was found in simple PCR studies of arterial biopsies, MIF tests to detect low IgG titers or IgM, or ELISA studies to detect IgM. According to this review, the association between Cp infection and PAOD depends on the analytical method adopted. Establishing a relationship between Cp and PAOD will require a case-control study with an adequate number of cases and samples that uses a combination of direct and indirect techniques to identify the presence of the bacterium in different types of sample from the same subjects, correlating the results with the activity of the disease.
Estimation and selection of complex covariate effects in pooled nested case-control studies with heterogeneity
