Table 1.

Clusters of high and low HIV prevalence and VMMC uptake, Manicaland, east Zimbabwe.

A: HIV prevalence amongst all participants aged 15–54
	No. of villages	Cluster radius	Cluster population	HIV prevalence	Test for significance
No.a	N	(km)	N	(%)	aRRb	(95% CI)	p-valuec
High HIV prevalence clustersd
1	7	3.43	485	25.7	1.78	(1.53–2.07)	<0.001
2	1	0	98	46.0	2.50	(2.08–3.01)	0.003
Low HIV prevalence clustersd
3	39	15.05	1376	11.32	0.70	(0.60–0.82)	0.007
B: VMMC uptake amongst males aged 15–29
	No. of villages	Cluster radius	Cluster population	VMMC uptake	Test for significance
No.a	N	(km)	N	(%)	aRRb	(95% CI)	p-valuec
High VMMC coverage clusterse
4	24	7.14	174	10.3	3.26	(1.94–5.72)	0.082
Low VMMC coverage clusterse
5	38	10.25	332	0.90	0.19	(0.06–0.61)	0.043

^aThe cluster numbers correspond to the numbers indicated in Figure 2 and 3.

^bThe relative risk and confidence intervals for being HIV-positive (A) or medically circumcised (B) were calculated as the number of participants who were HIV-positive and were medically circumcised, respectively, within the cluster compared to the number not in the cluster (the reference category) adjusted for age delineated into 5-year age groups using the Cochran-Mantel-Haenszel method [29]. Note that the provided HIV prevalence and VMMC uptake statistics are not adjusted for age whilst the relative risks are.

^cLikelihood ratio test statistics are calculated and p-values are obtained through Monte Carlo hypothesis testing with 9999 iterations.

^dAreas of higher and lower than expected numbers of HIV-positive individuals.

^eAreas of higher and lower than expected numbers of medically circumcised males aged 15–29.

aRR, relative risk adjusted for age; CI, confidence interval.