Survival analysis was performed to assess the risk factors for acquiring malaria. Independent variables included age,

gender, country of origin, use of mosquito repellents, use of barrier clothing, compliance with chemoprophylaxis, smoking, consumption of alcohol, accommodation on different floors in the apartment buildings, and building of residence. Univariate analysis was performed by calculation of the incidence rate ratio for each exposure category. Statistical significance for this comparison was then assessed by the Fisher’s exact test, using COMPARE2 software in the WINPEPI statistical package. Kaplan–Meier survival curves were drawn for the effect of chemoprophylaxis Crizotinib mouse living in the ground floor and first floor (Figure 2). Variables that were associated with the risk of contracting malaria (by a significance level of 0.1) were included in the multivariate Cox proportional hazard regression model. To increase the power CP-868596 cell line of analysis, variables that had three categories in the univariate analysis were redivided into two categories. This model was constructed using the forward stepwise method. p Value <0.1 of the maximum likelihood ratio test was chosen as the cutoff value for exclusion of a variable from the model. Cox proportional hazard regression model analysis

was performed with the use of SPSS 17.0 software (SPSS Inc., Chicago, IL, USA). All 104 staff members residing in the hospital compound during November 2008 agreed to participate

in the study. Two workers developed symptoms and signs of malaria within 10 days of their arrival to Equatorial Guinea. Both workers were excluded from the study, as they were considered to be infected outside of the hospital grounds. Between September 2007 and December 2008 noncomplicated falciparum malaria was diagnosed in 13 workers (12.75%). An incidence rate of 15.29 cases/100 person-years was calculated [95% confidence interval (CI) = 6.46–23.14]. Surprisingly, all cases of malaria occurred in workers residing on either the ground floor or the first floor of all five buildings (Figure 3). Of the 13 people diagnosed as having acquired malaria, 10 were living on the ground floor and 3 on the first floor of different apartment buildings. No cases Glycogen branching enzyme occurred on the second and third floors. Survival curves describing acquisition of malaria in people who lived in the ground floor and first floor compared to those living in the second and third floors showed a statistically significant difference (Figure 2, pvalue = 0.006). There was no statistically significant difference in the incidence of malaria between all apartment buildings, and shorter distance of different buildings from the presumed mosquito breeding area was not associated with an increased risk of acquiring malaria (p = 0.204 on a Cox proportional hazard regression model, data not shown).