Table 1. Risk of measles in infants during an outbreak, South Africa, 2009–2011.
| Parameter | Multivariate analysis (coverage)a |
Multivariate analysis (immunization)b |
|||
|---|---|---|---|---|---|
| RRc (95% BCId) | RRe (95% CI) | RRc (95% BCId) | RRe (95% CI) | ||
| District parameter | |||||
| Low- or medium-population-density nonmetropolitan area | 1 (ref.) | NA | 1 (ref.) | NA | |
| High-population-density nonmetropolitan area | 1.23 (0.64–2.43) | NA | 1.24 (0.65–2.54) | NA | |
| High-population-density metropolitan area | 4.37 (2.27–8.56) | NA | 4.42 (1.60–9.54) | NA | |
| Proportion of infants not covered by vaccinationf in preceding yearg | 7.84 (5.19–13.2) | NA | NA | NA | |
| Proportion of infants not immunizedh in preceding yearg | NA | NA | 12.88 (6.52–32.43) | NA | |
| Interaction between infant population susceptible to measles and population densityi | |||||
| Proportion of infants not covered by vaccinationf in the preceding year × high population density | NA | 66.20 (40.49–108.25) | NA | NA | |
| Proportion of infants not covered by vaccinationf in the preceding year × low or medium population density | NA | 7.45 (4.72–11.77) | NA | NA | |
| Proportion of infants not immunizedh in the preceding year × high population density | NA | NA | NA | 89.35 (50.65–157.60) | |
| Proportion of infants not immunizedh in the preceding year × low or medium population density | NA | NA | NA | 8.89 (5.14–15.38) | |
BCI, Bayesian credibility interval; CI, confidence interval; NA, not applicable; RR, risk ratio.
a The analysis considered infants aged under 1 year who were not covered by vaccination in the preceding year (i.e. they did not receive their first measles vaccine dose).
b The analysis considered infants aged under 1 year who were not immunized in the preceding year (i.e. they were not covered by vaccination or failed to seroconvert following the first vaccine dose).
c The multivariate association between parameters and the incidence of measles among infants was evaluated using a Bayesian Poisson conditional autoregressive convolution model.
d The 95% Bayesian credibility interval is analogous to the conventional 95% confidence interval.
e The multivariate association between parameters and the incidence of measles among infants was evaluated using a conventional Poisson approach.
f The proportion of infants that did not receive the first measles vaccine dose.
g The risk ratio is for a 1% increase, on a continuous scale, in the proportion of infants who were covered by vaccination or immunized.
h The formula used to calculate this proportion is described in the methods.
i Since there was a strong, significant correlation between population density and metropolitan area (i.e. Spearman ρ = 0.64, P < 0.001), the interaction between metropolitan area and high population density was not included.