This review was registered and is in accord with the standardised written protocol (systematic review registration with the International Prospective Register of Systematic Reviews (PROSPERO) database PROSPERO CRD42018104052) that followed the PRISMA-P (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Protocols) statement guidelines [32]. Additional file 1 shows the PRISMA checklist. The published protocol can be accessed on https://systematicreviewsjournal.biomedcentral.com/articles/10.1186/s13643-019-1137-y. Quality of included studies was assessed by Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) [33]. Institutional ethical review approval was not needed for this review.
Strategy
Electronic searches
Search terms ("tuberculosis", mycobacterium tuberculosis, extrapulmonary tuberculosis, pulmonary tuberculosis, paediatric tuberculosis), "Real-time polymerase chain reaction", real-time pcr, real-time pcr assay, "rt-pcr", "Nucleic Acid Amplification Test", "NAAT", "culture-based media", culture-based assay, "liquid media", "solid media", “paediatric”, “paediatrics”, “children”) were used to generate a list of primary studies in any language with no restriction on date of publication, and publication status (see Additional file 2 for search terms). There was no restriction regarding the language, date of publication, publication status. Studies that recruited children less than 16 years of age being investigated for MTB infection using RT-PCR assay accompanied by mycobacteriological culture investigation as the reference standard were included to achieve a more reliable estimate of diagnostic accuracy which is important to ensure that the process of identifying studies is as thorough and unbiased as possible.
Two investigators (EB, BC) independently and systematically carried out the search. Searches using electronic bibliographic databases (MEDLINE via PubMed, EMBASE, LILACS, BIOSIS Citation Index, Web of Science, SCOPUS, ISI Web of Knowledge, Cochrane Infectious Diseases Group Specialised Register (CIDG SR), Cochrane Registry of Diagnostic Studies, National Institute for Health Research, PROSPERO, Google Scholar Turning Research into Practice (TRIP) took place in August 2019 and was updated in November 2020. The MEDLINE search strategy is outlined in Additional file 2. The MEDLINE search was imported to EMBASE, Cochrane Infectious Diseases Group Specialised Register and other databases to identify additional records [34, 35]. The search strategy for each database was validated by a librarian information specialist familiar with the topic.
Attempts were made to avoid missing relevant studies by searching other sources such as reference lists of relevant reviews, selected studies, portal of the WHO International Clinical Trials Registry Platform (www.who.int/trialsearch) to identify ongoing trials, as well as StopTB Partnership’s New Diagnostics Working Group (www.stoptb.org/wg/new_diagnostics/), the World Health Organization and Centers for Disease Control and Prevention websites, and proceedings of the International Union Against Tuberculosis and Lung disease (UNION) conference. A search of grey literature including conference proceedings (Conference Proceedings Citation Index–Science (CPCI-S)), Dissertations & Theses (www.proquest.com), and expert information was sought and added to our resource material.
Besides full articles, abstracts, and letters to the editor with sample sizes >20 was also considered for inclusion. There was no language limitation to the search. Abstracts or articles in languages other than English were screened using ‘Google Translator’.
Inclusion and exclusion criteria
Study designs such as observational, cross-sectional studies, cohort studies (prospective and retrospective) and case-control designs for the detection of MTB from paediatrics clinical samples of age < 16 years were eligible for inclusion if the studies (1) compared RT-PCR based assay to a reference/gold standard method— MTB culture-based (either liquid or solid) assay, (2) described original research, (3) reported total number of patients tested and positive/negative results that allowed calculation of true positives (TP), true negatives (TN), false positives (FP) and false negatives (FN). Studies were excluded if (1) RT-PCR assay was not used in the study, (2) if age of participants is > 16 years, (3) all samples were not tested by reference/gold standard test—MTB culture-based (either liquid or solid) assay, (4) reference test was a combination of greater than one diagnostic test, (5) it included animal studies, (6) RT-PCR based assay was used for detecting non-tuberculosis mycobacteria, (7) RT-PCR based assay was used for detecting MTB from clinical isolates and not the pathological specimens/samples and (8) possible duplicate publication, when an author published more than one study. The existence of overlapping study populations was ascertained by checking sample recruitment sites and/or periods. The article reporting on the largest number of samples was included in our study.
Selection of studies
Full-text articles were screened independently (by EB and BC), using a PRISMA flow chart [32] for eligibility for use in the study to minimise bias in selection. Any disagreements were resolved through discussion and where needed, by a third reviewer (BO). Any rejected studies were documented.
Data extraction
Data extraction were independently carried out by EB and BC from each selected study using a predetermined list of categories/characteristics: participants/population, country, index test, reference test, disease and target sequence for detection of MTB DNA in PaeTB (Table 1).
Table 1
Characteristics of the included studies
Author Year
(n)
|
Country
|
Study design
|
Total number of samples (N)
|
Reference test:
Culture
|
Index test:
RT-PCR
|
Target sequence
|
PTB
|
EPTB
|
Bates et al., (2013) [44]
|
Zambia-(L)
|
Prospective-Descriptive
|
142
|
|
Liquid culture (MGIT)
|
RT-PCR Xpert MTB/RIF
|
rpoB probe
|
Bates et al., (2013*) [44]
|
Zambia-(L)
|
Prospective-Descriptive
|
|
788
|
Liquid culture (MGIT)
|
RT-PCR Xpert MTB/RIF
|
rpoB probe
|
Chipinduro et al., (2017) [45]
|
Zimbabwe-(L)
|
A cross-sectional
|
|
222 (stool)
|
LJ
|
RT-PCR Xpert MTB/RIF
|
rpoB probe
|
El Khechine et al., (2009)
[46]
|
France-(U)
|
Diagnostic case-control
|
-
|
134
|
BACTEC 9000 MB
LJ
|
RT-PCR(MX3000)
|
IS6110 gene
|
Gous et al., (2015) [47]
|
South Africa-(U)
|
Prospective
|
345
|
-
|
Liquid culture (MGIT)
|
RT-PCR Xpert MTB/RIF
|
rpoB probe
|
LaCourse et al., (2014) [48]
|
Malawi-(L)
|
Cross sectional study
|
300
|
-
|
Bactec MGIT, BD)
|
RT-PCR Xpert MTB/RIF
|
rpoB probe
|
Memon et al., (2018) [49]
|
India-(L)
|
Diagnostic Accuracy study
|
-
|
100
|
Bactec MGIT 960
|
RT-PCR Xpert MTB/RIF
|
rpoB probe
|
Mesman et al., (2019) [50]
|
Peru-(U)
|
Cohort study
|
|
259(stool)
|
BACTEC 9000 MB
LJ
|
TruTip Mtb DNA
|
IS6110 real-time PCR
|
Nhu et al., (2013) [51]
|
Vietnam-(L)
|
Prospective
|
|
96
|
MGIT, Becton Dickinson)
|
RT-PCR Xpert MTB/RIF
|
rpoB probe
|
Nicol et al., (2011) [52]
|
South Africa-(U)
|
Prospective-Descriptive
|
452
|
-
|
Liquid culture
|
RT-PCR Xpert MTB/RIF
|
rpoB probe
|
Nicol et al., (2013) [53]
|
South Africa-(U)
|
Prospective
|
-
|
115
|
Bactec MGIT 960
|
RT-PCR Xpert MTB/RIF
|
rpoB probe
|
Nicol et al., (2018) [54]
|
South Africa-(U)
|
Cohort Study
|
367
|
-
|
MGIT, Becton Dickinson)
|
RT-PCR Xpert MTB/RIF
|
rpoB probe
|
Oberhelman et al., (2010) [55]
|
Peru-(U)
|
Prospective Case-Control Study
|
|
218 (stool, GA, etc.)
|
LJ culture
|
hemi-nested IS6110 PCR
|
IS6110 PCR
|
Qing-Qin Yin et al., (2014) [56]
|
China-(U)
|
Prospective
|
255
|
|
Solid (LJ) and Liquid culture (Bactec MGIT 960
|
RT-PCR Xpert MTB/RIF
|
rpoB probe
|
Rachow et al., (2012) [57]
|
Tanzania-(L)
|
Prospective Cohort Study
|
164
|
-
|
Solid (LJ) and Liquid culture (Bactec MGIT 960
|
RT-PCR Xpert MTB/RIF
|
rpoB probe
|
Sekadde et al., (2013) [58]
|
Uganda-(L)
|
Cross-sectional diagnostic study
|
235
|
-
|
Solid (LJ) and Liquid culture (Bactec MGIT 960
|
RT-PCR Xpert MTB/RIF
|
rpoB probe
|
Walters et al., (2017) [59]
|
South Africa-(U)
|
Prospective
|
|
379 (stool)
|
BACTEC 9000 MB
|
RT-PCR Xpert MTB/RIF
|
rpoB probe
|
Wang et al., (2013) [60]
|
China-(U)
|
Retrospective
|
30
|
-
|
Bact/Alert 3D
|
LightCycler® 480 (Roche)
|
|
Wolf et al., (2008) [61]
|
Peru-(U)
|
Diagnostic Accuracy study
|
-
|
16 (6+) (stool)
|
|
hemi-nested IS6110 PCR
|
IS6110 PCR
|
Zar et al., (2012) [62]
|
South Africa-(U)
|
Prospective
|
535
|
-
|
Liquid culture (MGIT)
|
RT-PCR Xpert MTB/RIF
|
rpoB probe
|
Zar et al., (2013) [63]
|
South Africa-(U)
|
Prospective study
|
384
|
-
|
Bactec MGIT 960
|
RT-PCR Xpert MTB/RIF
|
rpoB probe
|
Key: LJ Löwenstein-Jensen, Middlebrook 7H9 Broth Liquid growth medium, Middlebrook 7H11 Solid medium, MGIT Mycobacterium Growth Indicator Tube, PTB Pulmonary TB, EPTB Extra-pulmonary TB, [n] reference list number, (L) Lower and middle-income countries, (U) Upper middle-income countries
Table 2
Summary of statistical results for pulmonary tuberculosis (PTB) and extra-pulmonary tuberculosis (EPTB) clinical samples
Test property
|
Summary of measure test accuracy* (95%)
|
Test of heterogeneity
|
PTB
(n= 11; †3,209)
AUC= 0.98
|
|
X2
(d.f.=10)
|
l2
|
p value
|
Sensitivity
|
56(51-62)
|
151.22
|
93.4
|
<0.001
|
Specificity
|
97(96-98)
|
277.67
|
96.4
|
<0.001
|
Positive likelihood ratio (PLR)
|
70.73 (8.55-585.40)
|
205.09
|
95.1
|
<0.001
|
Negative likelihood ratio (PLR)
|
0.43 (0.28-0.66)
|
99.77
|
90.0
|
<0.001
|
Diagnostic odd ratio (DOR)
|
193.06 (51.21- 727.83)
|
36.66
|
72.7
|
<0.001
|
EPTB
(n= 10; †2,327)
AUC=0.99
|
|
X2
(d.f.=9)
|
l2
|
p value
|
Sensitivity
|
87(82-91)
|
47.45
|
81.00
|
<0.001
|
Specificity
|
100(99-100)
|
19.19
|
53.10
|
0.0236
|
Positive likelihood ratio (PLR)
|
111.91(53.97-232.04)
|
11.74
|
23.40
|
0.2282
|
Negative likelihood ratio (PLR)
|
0.15 (0.07-0.30)
|
29.44
|
69.40
|
0.0005
|
Diagnostic odd ratio (DOR)
|
1337.84 (441.92- 4050.12)
|
13.02
|
30.90
|
0.1610
|
*: Random effects model; (X2): chi-squared; d.f.: degree of freedom; (I2): I-squared; †: number of specimens; n: number of studies; CI: confidence interval; AUC: area under receiver operating characteristics curve; PTB: pulmonary tuberculosis; EPTB: extra-pulmonary tuberculosis.
Assessment of study quality
The methodological quality for the included studies was assessed independently (EB and BC) according to the four domains (patient selection, index test, reference standard and flow and timing) of the QUADAS-2 tool [32]. The study QUADAS-2 quality criteria are given in Additional file 3.
Data synthesis and meta-analysis
We computed measures of test accuracy for each of the included studies using standard methods recommended for meta-analysis of diagnostic studies: sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic odds ratio (DOR) and 95% confidence intervals (CI) [36]. The 2x2 data (TP, FP, TN and FN) were extracted directly from the included studies. Where this information was not available, values were calculated from the data provided in the article. We used a DOR using the DerSimonian-Laird random-effect model to calculate and assess the overall accuracy. This model accounts for both within-study variability (random error) and between-study variability (heterogeneity) along with the area under the summary receiver operating characteristic (SROC) curve using the bivariate model [37, 38]. The bivariate model considers potential threshold effects and the correlation between binary tests (sensitivity and specificity). These measures were pooled using the random-effects model [37, 38]. Each of the included studies used in the meta-analysis contributed a pair of numbers: sensitivity and specificity. Since these measures are correlated, we summarised their joint distribution using a SROC curve. The SROC curve presents a global summary of test performance and shows the trade-off between sensitivity and specificity. A symmetric curve suggests that the variability in accuracy between studies is explained, in part, by differences in thresholds used by the studies. The area under the SROC curve is a global measure of overall performance of the test. An area under the curve value of 1 indicates perfect discriminatory ability of the test, while an area under the curve value of 0.5 means that the test does not have discriminating ability [37, 38].
Data were analysed using Meta-DiSC (version 1.4), Reviewing Manager ver. 5.4 (Cochrane Collaboration, Oxford, UK) [38, 39]. The data were displayed graphically on forest plots and SROC plots. The SROC curve was fitted using the Littenberg-Moses method [40].
We did not evaluate the publication bias because this is not usually recommended in the meta-analysis for diagnostic test accuracy [41]. Generally, a diagnostic accuracy study does not test a hypothesis; therefore, there is no p value for authors and publishers that may influence decisions about publication which are based on the statistical significance of the results [41, 42].
Investigations of heterogeneity
We investigated heterogeneity because of its critical importance (1) to understand the possible factors that influence accuracy estimates and (2) to evaluate the appropriateness of statistical pooling of accuracy estimates using random-effects meta-analysis to generate sensitivity and specificity with 95% CIs from various studies [42].
The heterogeneities among the included studies were assessed visually using forest plots and SROC curves with 95% prediction regions and statistically with chi-squared (χ2) and using I-squared (I2) statistics with the following interpretation: I2 = 0, no heterogeneity; 0 < I2 < 25, mild heterogeneity; 25 ≤ I2 < 50, moderate heterogeneity; 50 ≤ I2 < 75, strong heterogeneity; 75 ≤ I2 < 90, considerable heterogeneity and 90 ≤ I2, extreme heterogeneity [41, 43].
Source of heterogeneity was investigated using stratified (subgroup) analyses. The following factors were specified a priori as potential sources of heterogeneity: Impact of RT-PCR based assays on lower- and middle -income countries (LMICs) versus Upper middle-income countries (UMICs).
Table 3
Subgroup analyses by Impact of RT-PCR based assays on lower- and middle -income countries (LMICs) versus Upper middle-income countries (UMICs). We will assess sources of data to these graders
Test property
|
Summary of measure test accuracy* (95%)
|
Test of heterogeneity
|
Lower-MICs
(n= 8; †2,047)
AUC= 0.99
|
|
X2
(d.f.=7)
|
l2
|
p value
|
Sensitivity
|
65 (58- 72)
|
44.28
|
84.20
|
<0.001
|
Specificity
|
99 (99 – 99
|
10.46
|
33.10
|
0.1639
|
Positive likelihood ratio (PLR)
|
86.61 (46.72-160.53)
|
5.97
|
0.0
|
0.5432
|
Negative likelihood ratio (NLR)
|
0.367 0.233 - 0.578
|
29.54
|
76.30
|
<0.001
|
Diagnostic odd ratio (DOR)
|
311.43 (106.76 - 908.51)
|
11.25
|
37.8
|
0.1280
|
Upper-MICs
(n= 12; †3,487)
AUC=0.99
|
|
X2
(d.f.=11)
|
l2
|
p value
|
Sensitivity
|
68 (63-73)
|
197.71
|
94.40
|
<0.001
|
Specificity
|
97 (96 – 98)
|
291.40
|
96.20
|
<0.001
|
Positive likelihood ratio (PLR)
|
80.90 (10.31- 634.9)
|
247.81
|
95.6
|
<0.001
|
Negative likelihood ratio (NLR)
|
0.20 (0.09 - 0.42)
|
228.53
|
95.2
|
<0.001
|
Diagnostic odd ratio (DOR)
|
522.72 (107.04- 2552.8)
|
50.80
|
78.30
|
<0.001
|
*: Random effects model; (X2): chi-squared; d.f.: degree of freedom; (I2): I-squared; †: number of specimens; n: number of studies; CI: confidence interval; AUC: area under receiver operating characteristics curve; PTB: pulmonary tuberculosis; EPTB: extra-pulmonary tuberculosis.