Page 28 - Epidemiological studies on tuberculosis control and respiratory viruses

Chapter 6
present, as Cp values are inversely correlated with the number of target DNA or RNA molecules present in the sample (in this study denoted as viral load). A difference in Cp values of 3 approximately represents a 10-fold difference in the levels of target nucleic acids. A sample was considered positive if it passed the internal positive controls and if Cp value ≤40, and negative if the value was above 40.
Statistical analysis
The study population was divided in groups according to symptom status and health care use, which was used as an approximation of illness severity, ranging from (1) asymptomatic and (2) mildly symptomatic HELIUS participants, reflective of the general population who did not seek hospital care, to three groups of symptomatic patients seeking hospital care: (3) patients presenting at general outpatient clinics, (4) patients admitted to the hospital (but not to intensive care units [ICUs]), and (5) patients admitted to ICUs. Patients presenting at the Emergency Department were excluded due to the expected high variability in severity at time of presentation within this group of patients.
Analyses of the complete study population were done to compare characteristics of individuals and prevalence of respiratory viral infections between groups (Table 1), and to assess the association between PCR-positivity and approximated illness severity based on symptom status and health care use (Table S1). In addition, among individuals with positive PCR results, we compared the distribution of viral species between groups (Table 2), and analysed possible relationships between viral load and approximated illness severity (Table 3, Figure 1, Table S2). Analyses are described in more detail below.
Differences in categorical variables were assessed using the Chi-squared test or Fisher exact test; for continuous variables Mann-Whitney U test or Kruskal-Wallis were used. Three comparisons were made in order to assess differences (1) between the symptomatic and asymptomatic HELIUS groups reflecting the general population not seeking hospital care, (2) between the three groups presenting at or admitted to the hospital, and (3) between all five groups.
Associations between groups and PCR positivity were assessed using logistic regression analysis. Because the number of positive PCR results among HELIUS
98