A Systematic Review of Clinical Diagnostic Systems Used in the Diagnosis of Tuberculosis in Children

Table 6

Studies that specified how many patients were coinfected with HIV.

Author	Year	Country	Total patients	Percent HIV positive	Findings
Madhi et al. [31]	1999	South Africa	130	40%	Did not attempt to validate scoring criteria
Kiwanuka et al. [42]	2001	Malawi	110	71% (of 102 tested)	Did not attempt to validate scoring criteria
Palme et al. [39]	2002	Ethiopia	517	11.2%	Did not attempt to validate scoring criteria
van Rheenen [18]	2002	Zambia	147	30%	Keith Edwards scoring system: sensitivity 88% and specificity 25% in this study. Most of the children with a false positive score were malnourished (48%) or had AIDS (31%)
Marais et al. [26]	2006	South Africa	428	8.8%	Sensitivity, specificity, and PPV all decreased significantly when HIV infected children included
Edwards et al. [47]	2007	Democratic Republic; of Congo	91	46%	Out of 8 scoring systems analyzed, 3/8 systems did not recommend treatment in 14% of HIV-infected children compared to 2% of noninfected children. Mean score tended to be higher for HIV-infected children, but only significant for Edwards score
Viani et al. [8]	2008	Mexico	13	100%	Applied Stegen-Toledo criteria retrospectively but without culture results: 77% had highly probable TB, 15% probable, and 8% suspicion of TB
Pedrozo et al. [23]	2009	Brazil	239	5%	Analyzed scoring system by looking at median scores of various groups: median score of 3a (TB+, HIV−) and 3b (TB+, HIV+) sig. higher than TB negative groups, median score of TB+ groups also was higher than the cutoff of 30

PPV: positive predictive value.