Skip to main content

Table 3 Sources of heterogeneity assessment based on meta-regression analyses (93 geo-referenced locations)

From: A systematic review of changing malaria disease burden in sub-Saharan Africa since 2000: comparing model predictions and empirical observations

Factors

Summary statistics

Pooled correlation (rho)

95% CI

p value

Residual I2 (%)

Percentage change in I2 (%)

Geographical region, n (%)

 Eastern Africa

42 (45.2%)

0.43

0.18, 0.62

0.01

67.7

2.8

 Southern Africa

23 (24.7%)

0.33

0.07, 0.54

 Western Africa

22 (23.7%)

0.73

0.53, 0.86

 Central Africa

4 (4.3%)

0.30

− 0.42, 0.79

 Horn of Africa

2 (2.1%)

0.84

0.60, 0.94

Quality of the study, n (%)

 High risk of bias

28 (30.1%)

0.54

0.30, 0.71

< 0.001

63.2

9.3

 Moderate risk of bias

41 (44.1%)

0.25

0.02, 0.46

 Low risk of bias

24 (25.8%)

0.78

0.63, 0.87

Data source, n (%)

 In-patient

34 (36.6%)

0.36

0.07, 0.59

0.01

68.1

2.3

 In-patient and out-patient

17 (18.3%)

0.38

0.03, 0.65

 Out-patient

42 (45.1%)

0.67

0.52, 0.78

Measure of malaria, n (%)

 Number of cases reported

30 (32.3%)

0.26

− 0.01, 0.49

0.001

66.2

4.9

 Test positive rate

47 (50.5%)

0.69

0.54, 0.79

 Incidence rate

16 (17.2%)

0.30

− 0.06, 0.59

Average starting parasite prevalence in children aged 2–10 years

 < 10%

19 (20.4%)

0.33

− 0.08, 0.64

0.002

66.2

4.9

 10–50%

55 (59.2%)

0.64

0.50, 0.75

 > 50%

19 (20.4%)

0.11

− 0.20, 0.41

Sample size, median (IQR)

15,779 (3957, 35,981)

0.01

− 0.09, 0.11

0.83

71.8

− 3.1

Sum of residuals of empirical clinical incidence/test positivity rate, median (IQR)

0.94 (0.06, 2.05)

0.07

− 0.03, 0.17

0.07

69.3

0.5

Sum of residuals of MAP modelled clinical incidence, median (IQR)

0.12 (0.06, 0.30)

− 0.02

− 0.65, 0.62

0.24

69.8

− 0.3

Slope of empirical clinical incidence/test positivity rate, median (IQR)

− 0.003 (− 0.16, 0.001)

− 0.989

− 0.998, − 0.939

< 0.001

58.8

15.5

Slope of modelled MAP clinical incidence, median (IQR)

− 0.02 (− 0.03, 0.0001)

− 0.99995

− 1.00, − 0.997

0.003

67.5

3.1

  1. Summary unadjusted I2 = 69.63%. Percentage change in I2 computed as (summary unadjusted I2 − residual I2)/summary unadjusted I2 × 100