Adult mortality and household food security in rural South Africa: Does AIDS represent a unique mortality shock?

Figures & data

Table 1: Mean household characteristics and relative frequency of the roles of the deceased in household food provisioning, by recent household mortality experience

Variable	HIV death	Non-HIV death	No death
Household characteristics
Household size	5.0    (3.2)^a	5.5    (2.9)^a	4.5    (2.4)^a
SES score: 2003	–0.24 (2.38) ^a	1.09 (2.16) ^b	0.04 (2.20) ^a
SES score: 2005	–0.24 (2.17) ^a	0.91 (2.12) ^b	0.14 (1.88) ^a
Mortality variables
Time since death (months)	18.5    (6.6)^a	18.9    (5.2)^a	–
Total deaths over sample period	1.7    (1.0)^a	1.5    (0.6)^a	–
Role of the deceased in food provisioning
Contributed income	59.6%	70.8%	–
Tended a food garden and/ or field	59.6%	64.6%	–
Collected natural resources	51.9%	46.2%	–
Note: Standard deviation in brackets. Values which do not share a common superscript letter within a row are significantly different (p < 0.05) and highlighted in bold.

Figure 1: Frequency profile of household SES quintiles (1 = poorest, 5 = wealthiest) in 2005, by recent household mortality experience (n = 290)

Table 2: Summary of significant linear and logistic regression models for measures of food security against household size and SES score

		Model parameters
Dependent variable	Independent variable	Beta coefficient	p value	R^2
Dietary diversity
Food groups in last week	Household size	0.18	<0.05	0.03
	SES	0.24	<0.001	0.06
Total foods in last week	SES	0.34	<0.001	0.12
Total foods in last month	SES	0.35	<0.001	0.12
Total foods in last year	SES	0.27	<0.001	0.08
Prop. dietary diversity	SES	0.35	<0.001	0.13
Experience of hunger		Odds ratio	p value	Pseudo R ^2
Went hungry	SES	0.86	<0.05	0.02
Coping strategies
Skipped meals	SES	0.83	<0.01	0.03

Table 3: Raw and adjusted mean dietary diversity (controlling for household size and SES score), by household mortality experience

Parameter	Food groups in last week	Total foods in last week	Total foods in last month	Total foods in last year	Prop. dietary diversity
Range	4–10	5–50	10–64	36–86	0.07–0.52
Raw means
HIV death	7.8^a	21.9 ^a	33.2^a	57.9 ^a	0.25 ^a
Non-HIV death	7.9^a	24.2^ab	35.7^a	61.7 ^b	0.28^ab
No death	7.8^a	24.9 ^b	36.1^a	61.5 ^b	0.28 ^b
Adjusted means, controlling for household size
HIV death	7.7^a	21.9^a	33.2^a	57.8 ^a	0.26^a
Non-HIV death	7.9^a	24.1^a	35.5^a	61.7 ^a	0.27^a
No death	7.8^a	25.0^a	36.2^a	61.5 ^a	0.28^a
Adjusted means, controlling for socioeconomic (SES) score
HIV death	7.8^a	22.6^a	33.9^a	58.5^a	0.25^a
Non-HIV death	7.9^a	23.3^a	34.5^a	60.8^a	0.27^a
No death	7.9^a	24.9^a	36.1^a	61.5^a	0.28^a
Note: Means that do not share common superscripts within a comparison are significantly different (p < 0.05) and are highlighted in bold.

Table 4: Logistic regression models comparing the likelihood of experience of hunger in HIV-impacted households with households experiencing a non-HIV death or no death (1), and controlling for household size and SES score (2)

	Worried about food		Ran out of food		Went hungry		All three experiences
Parameter	(1)	(2)	(1)	(2)	(1)	(2)	(1)	(2)
Odds ratios
Non-HIV death	1.54	1.79	1.80	2.26	1.20	1.56	1.33	1.65
No death	0.62	0.69	0.78	0.87	0.50	0.59	0.42	0.50
Household size	–	1.05	–	1.02	–	1.02	–	1.04
SES	–	0.89	–	0.86	–	0.84	–	0.87
Model
LR chi^2	8.86	12.47	7.46	13.61	8.14	14.20	11.96	15.06
p value	<0.05	<0.05	<0.05	<0.01	<0.05	<0.01	<0.01	<0.01
Note: Statistically significant odds ratios are highlighted in bold (p < 0.05).

Table 5: Logistic regression models comparing the likelihood of food-related coping strategies in HIV-impacted households with households experiencing a non-HIV death or no death (1), and also controlling for household size and SES score (2)

	Ate unpleasant food		Asked neighbours for food		Skipped meals		All three strategies
Parameter	(1)	(2)	(1)	(2)	(1)	(2)	(1)	(2)
Odds ratios
Non-HIV death	1.47	1.69	1.47	1.66	1.13	1.56	0.87	1.04
No death	0.56*	0.61	0.73	0.86	0.49	0.62	0.37	0.45
Household size		1.04		1.08		1.07		1.11
SES	–	0.92	–	0.90	–	0.79	–	0.85
Model
LR chi^2	9.43	11.55	3.56	7.29	6.23	15.99	5.82	10.50
p value	<0.01	<0.05	0.169	0.121	<0.05	<0.01	0.055	<0.05
Note: Statistically significant odds ratios are highlighted in bold (p < 0.05). *p = 0.05.