Table 2.

Age- and geographic location-adjusted odds ratios for invasive bacterial death with homozygosity at specific microsatellite markers in multivariable models restricted to cases of Gram positive sepsis, gram negative sepsis or including all invasive bacterial deaths combined.

		Gram Negative		Gram Positive		Combined
Markers	Interaction	OR	95% CI	OR	95%CI	OR	95% CI
D7S486		1.02	0.16 – 6.53
D7S486	D16S423	40.7	4.28 – 387
D12S310		14.0	2.70 – 72.7	4.73	1.59–14.1	4.94	2.27 – 10.8
D13S158		6.11	1.45 – 25.8			4.66	1.92 – 11.3
D15S1007		7.28	1.89 – 28.1
D16S423		1.61	0.31 – 8.33			7.65	2.55 – 22.9
D9S164	infants			0.18	0.018 – 1.90
D9S164	children			1.65	0.249 – 10.9
D14S275				3.93	1.35 – 11.4	12.2	4.44 – 33.3
D14S275	D13S158					10.1	3.84 – 26.5
D16S3103				3.70	1.20 – 11.4	7.04	2.56 – 19.4
D16S3103	D16S423					10.2	4.10 – 25.2

The table shows odds ratios, adjusted for age in 6 strata and for geographical location in 8 strata, of death due to all invasive bacterial disease, of death due to Gram negative invasive bacterial disease and of death due to Gram positive invasive bacterial disease, for homozygosity at microsatellite markers either alone or paired (where interactions were noted with LRT p values < 0.05). Variables excluded, with LRT p values = 0.05, in addition to sex and SOH, were markers d12s352, d14s261, d14s275, d16s3103 and d18s452 in the gram negative model, markers d13s158, d14s261, d14s280 and d16s423 in the gram positive model and markers d12s352, d14s261, d14s280 and d15s1007 in the combined model. Variables and interaction terms in the final models all had LRT p values < 0.02.