Table 2 Sensitivity analysis: Spectrum-estimated syphilis prevalence in 132 countries, under varying scenarios of data included and modelling assumptions.

From: Syphilis prevalence trends in adult women in 132 countries – estimations using the Spectrum Sexually Transmitted Infections model

Scenario

Countries included

Median prevalence, 2012

Median prevalence, 2016

Countries with prevalence trend from 2012 to 2016

Substantive increase

Non-substantive increase

Non-substantive decrease

Substantive decrease

Default estimates

132

0.57%

0.88%

14 (11%)

64 (48%)

48 (36%)

6 (5%)

A. At least one data point from 2012 or later

129

0.55%

0.87%

14 (11%)

62 (48%)

47 (36%)

6 (5%)

B. More restrained time trends: extrapolate past the last year with national data, for 1 year instead of 2 years

132

0.57%

0.92%

13 (10%)

65 (49%)

48 (36%)

6 (5%)

C. Restrict data to 2005 and later (instead of 1990)

130

0.46%

0.93%

17 (13%)

61 (47%)

44 (34%)

8 (6%)

D. ANC (survey & routine screening) data only

131

0.56%

0.90%

18 (14%)

62 (47%)

43 (33%)

8 (6%)

E. Syphilis infections positive on both treponemal and non-treponemal tests

85

0.94%

0.86%

8 (9%)

32 (38%)

42 (49%)

3 (4%)

F. Countries not included in default but made eligible for trend analysis by adding blood donor data

34*

0.04%

0.05%

3 (9%)

21 (62%)

9 (26%)

1 (3%)

  1. Notes to Table 2. The presented medians, which are unweighted, should not be interpreted as indicative of global burden trends, as global trends depend on national population sizes. Scenario F presents 34 countries that were not included in the default analysis.
  2. In none of the 7 scenarios, none of the countries and none of the years between 2012–2016 was the 20% maximum value imposed on estimated prevalence ever reached, neither in the best estimate (i.e. the median of 400 bootstraps) nor in the upper-bound limit of each corresponding 95% confidence interval.
Back to article page