Table 3. Multivariable analysis of factors associated with in-hospital mortality amongst patients aged ≥15 years with invasive pneumococcal disease (IPD) in South Africa from 2003–2008, with bacteremic pneumonia and meningitis separately.

Risk factor		Bacteremic pneumonia			Meningitis
		Case-fatality ratio n/N (%)	Adjusted odds ratio (95% confidence interval)	p	Case-fatality ratio n/N (%)	Adjusted odds ratio (95% confidence interval)	p
Age group (years)	15–24	23/225 (10)	Reference	Reference	54/141 (38)	Reference	Reference
	25–44	320/1582 (20)	2.7 (1.5–5.0)	0.009	421/770 (55)	1.8 (0.9–4.1)	0.110
	45–64	175/527 (33)	6.2 (3.1–12.7)	<0.001	145/226 (64)	7.1 (2.8–18.2)	<0.001
	≥65	58/150 (39)	11.7 (3.4–40.4)	<0.001	21/29 (72)	8.3 (0.6–108.4)	0.106
Pitt bacteremia score	<4				401/807 (50)	Reference
	≥4				240/359 (67)	2.4 (1.4–4.1)	<0.001
Prior antibiotic use (2 months)	Yes	27/95 (28)	2.3 (1.4–3.9)	0.001	22/34 (65)	5.0 (2.0–12.5)	<0.001
	No	200/1519 (13)	Reference	Reference	220/577 (38)	Reference	Reference
Appropriate antibiotic prescription	Yes	397/1976 (20)	Reference	Reference
	No	179/508 (35)	2.8 (2.0–3.8)	<0.001
HIV status	Positive				336/634 (53)	6.1 (2.1–17.6)	0.001
	Negative				19/73 (26)	Reference	Reference
Serotype	1	60/363 (17)	1.2 (0.6–2.4)	0.709	174/243 (72)	2.0 (0.8–4.6)	0.149
	19A	93/342 (27)	1.5 (0.8–3.1)	0.251	52/82 (63)	1.1 (0.4–3.2)	0.811
	4	29/177 (16)	Reference	Reference	51/95 (54)	Reference	Reference
	3	69/216 (32)	2.0 (1.0–4.2)	0.061	13/35 (37)	0.2 (0.02–1.5)	0.103
	6A	31/138 (22)	0.7 (1.3–2.0)	0.517	50/109 (46)	0.5 (0.2–1.2)	0.125
	14	55/184 (30)	2.2 (1.1–4.7)	0.039	20/47 (43)	0.5 (0.1–1.8)	0.267
	23F	27/124 (22)	1.5 (0.6–3.7)	0.352	59/106 (56)	0.9 (0.2–2.9)	0.765
	8	14/138 (10)	0.8 (0.3–2.0)	0.649	32/528 (55)	0.9 (0.3–3.1)	0.851
	19F	51/116 (44)	3.5 (1.6–8.0)	0.002	35/64 (55)	0.7 (0.2–2.5)	0.567
	6B	22/123 (18)	1.0 (0.4–2.5)	0.954	38/67 (57)	1.0 (0.4–3.0)	0.975
	12F	16/79 (20)	1.2 (0.4–3.3)	0.778	38/75 (51)	0.7 (0.2–2.6)	0.598
	9V	23/93 (25)	2.0 (0.8–5.2)	0.153	19/36 (53)	0.2 (0.01–1.4)	0.230
	16	19/68 (28)	0.6 (0.2–2.4)	0.508	12/30 (40)	0.2 (0.03–0.9)	0.040
	9N	17/66 (26)	1.1 (0.4–3.5)	0.863	6/16 (38)	0.4 (0.1–4.0)	0.444
	7F	10/56 (18)	0.9 (0.2–3.5)	0.868	3/14 (21)	0.4 (0.1–2.1)	0.271
	18C	6/26 (23)	1.5 (0.4–85.8)	0.583	10/30 (47)	1.4 (0.3–7.0)	0.716
	25	7/54 (13)	0.5 (0.1–2.6)	0.440	3/11 (27)	0.2 (0.01–2.0)	0.155
	22F	11/43 (26)	1.6 (0.4–5.6)	0.498	9/17 (53)	0.5 (0.1–3.0)	0.409
	13	7/31 (23)	1.6 (0.5–5.7)	0.435	11/27 (41)	1.1 (0.3–5.2)	0.851
	5	9/47 (19)	0.7 (0.2–3.6)	0.703	2/4 (50)	0.8 (0.02–36.0)	0.923

N-number, OR—Odds ratio, CI—confidence interval, HIV—Human immunodeficiency virus. Serotype was retained in the multivariate models a priori because it was the main variable of interest. Only factors statistically significant on multivariable analysis for the syndrome of interest are presented in the table. Additional factors evaluated for patients with bacteremic pneumonia were: (a) factors non-significant on univariate analysis: race, gender, province poverty level, appropriate antibiotic prescription (b) factors significant on univariate analysis but not on multivariable analysis: underlying medical conditions other than HIV, serotype, receipt of antibiotics in the 24 hours preceding culture, nosocomial infection, penicillin non-susceptibility, multidrug resistance. Additional factors evaluated for patients with meningitis were: (a) factors non-significant on univariate analysis: gender, province poverty level, receipt of antibiotics in the 24 hours preceding culture, appropriate antibiotic prescription, nosocomial infection, penicillin non-susceptibility, multidrug resistance (b) factors significant on univariate analysis but not on multivariable analysis: race, underlying medical conditions other than HIV, serotype.