Download ex_8_Mwanza_RCT_kps

Exercise 8
MWANZA TRIAL (VM)
A randomised controlled trial was conducted to evaluate the impact of improved management
of sexually transmitted disease (STD) at primary health care level on the incidence of HIV
infection in the rural Mwanza region of Tanzania. (H. Grosskurth et al., Lancet (1995)
346:530-6) HIV incidence was compared in six intervention communities and six pairmatched comparison communities. A random cohort of about 1,000 adults aged 15-54 years
from each community was surveyed at baseline and at follow-up 2 years later. Intervention
consisted of establishment of an STD reference clinic, staff training, regular supply of drugs,
regular supervisory visits to health facilities, and health education about STDs.
12,537 individuals were recruited. Baseline HIV prevalences were 4% in both the
intervention and comparison communities. At follow-up, 8,845 (71%) of the cohort were
seen. Of those initially seronegative, the proportions seroconverting over 2 years were 48 of
4,149 in the intervention communities and 82 of 4,400 in the comparison communities.
Allowing for the effects of confounding factors, the adjusted risk ratio† was 0.59 (95% CI
0.42-0.79, p=0.007). No change in reported sexual behaviour was observed in either group.
† Adjusted for age, sex, travel during follow-up period, history of STD (ever) at baseline, and male
circumcision.
a) Describe an RCT. Is this a management or explanatory trial?
b) Why was it important to select a random sample from each village?
c) Calculate incidence, crude relative risk, risk difference and numbers needed to treat for
HIV in the intervention and control communities.
d) Interpret the values given for the adjusted risk ratio, confidence interval and p value.
e) Explain what is meant by the underlined text: "Allowing for the effects of confounding
factors, the adjusted risk ratio was 0.59 (95% CI 0.42-0.79, p=0.007)"
f) What is the significance of the fact that there was no change in reported sexual behaviour
observed in either group?
g) Do you think these findings are relevant either to Ireland or to your own home country?
(state which country)
Key points
A
management
B
Too expensive to do all villagers so use sample. Random sample needed to ensure
generalisability of sample. (Effect of intervention in the samples should then be same as the
effect in the whole village.)
C
1.2% v 1.9% etc
D
40% lower HIV incid in intervention communities, data consistent with reduction of
between 58% and 21%, stat signif.
E
define CV and explain why the adjusted-for factors are confounding.
F
change in incidence is therefore not due to behaviour change leading to lower STD or
HIV but to STD treatment per se; health ed didn't work.
G
greater impact in high prev. area/ in areas where STD detection and treatment currently
low
Incid in IVN 48/4149 = 1.15 = 1.2%
Incid in cntrl 82 / 4400 = 1.86 = 1.9%
Crude RR = 1.15/1.86 = 0.618 = .62
RD = -0.71%
NNT = 1/.71 *100 = 140
(1.2/1.9 = .63)
( -0.7%)