In this study, we apply both the dTDT and Mantel-Haenszel (MH) tests to the COGA data. The MH test was first proposed by Mantel and Haenszel in 1959 [78]. The method has been widely applied to analysis of contingency tables (normally 2 × 2) and comparison of results from different treatments. In case-control studies, a 2 × 2 table is typically used. The discrepancy between observed and expected values in each cell from the table is evaluated by χ 2 test with one degree of freedom. Comparatively, because the dTDT only takes account of values of b and c, the test can be constructed by a 1 × 2 table instead. To maximally benefit from all sample data and multiple studies, we extend the MH test to pool results on each SNP from these two contingency tables in both case-control study and dTDT. Calculation of each term in the MH test is shown in Table 7 . Having the Observed & Expected values and Variances from case-control study and dTDT, terms in MH test can be written as the sums of corresponding values these two tests. The null hypothesis assumes no association between markers and disease.
