The Contribution of Chest Radiography to the Clinical Management of Children Exposed to Tuberculosis

Abstract

Rationale

Although World Health Organization guidelines emphasize contact investigation for tuberculosis (TB)-exposed children, data that support chest radiography as a useful tool are lacking.

Objectives

We evaluated the diagnostic and prognostic information of chest radiography in children exposed to TB and measured the efficacy of isoniazid preventive therapy (IPT) in those with relevant radiographic abnormalities.

Methods

Between September 2009 and August 2012, we enrolled 4,468 TB-exposed children who were screened by tuberculin skin testing, symptom assessment, and chest radiography. Those negative for TB disease were followed for 1 year for the occurrence of new TB diagnoses. We assessed the protective efficacy of IPT in children with and without abnormal chest radiographs.

Measurements and Main Results

Compared with asymptomatic children with normal chest films, asymptomatic children with abnormal radiographs were 25.1-fold more likely to have coprevalent TB (95% confidence interval [CI], 1.02–613.76) and 26.7-fold more likely to be diagnosed with incident TB disease during follow-up (95% CI, 10.44–68.30). Among the 29 symptom-negative and CXR-abnormal child contacts, 20% (3/15) of the isoniazid recipients developed incident TB, compared with 57% (8/14) of those who did not receive IPT (82% IPT efficacy).

Conclusions

Our results strongly support the use of chest radiography as a routine screening tool for the evaluation of child TB contacts, which is readily available. Radiographic abnormalities not usually considered suggestive of TB may indicate incipient or subclinical disease, although TB preventive treatment is adequate in most cases.

At a Glance Commentary

Scientific Knowledge on the Subject

Although World Health Organization guidelines emphasize contact investigation for children exposed to tuberculosis (TB) in their households, it is unclear whether chest radiography contributes more to the clinical management of children exposed at home to TB compared with symptom screening alone.

What This Study Adds to the Field

Our study suggests that the addition of radiography for child TB contacts may not only allow for TB diagnosis earlier than would be accomplished through symptom screening alone but may also identify a subgroup of child contacts at high risk of later TB progression. We also show that nonspecific CXR findings in child TB contacts may indicate incipient or subclinical TB disease that requires formal treatment. However, access to tuberculosis preventive therapy should not be constrained because it remains highly effective in asymptomatic contacts, even in those with CXR abnormalities.

Tuberculosis (TB) remains a major threat to child health worldwide (1). Interventions to reduce the burden of TB in children include the systematic screening of those who have close contact with a known patient with TB and the provision of TB preventive therapy to those who do not have coprevalent TB disease (2). The 2021 World Health Organization (WHO) guidelines now recommend that these children undergo TB-related symptom screening, chest radiography (CXR), or both to detect undiagnosed TB disease and to identify individuals who would benefit from TB preventive therapy (TPT) once TB disease is ruled out (2). Although the guidelines state that this is a strong recommendation, they acknowledge “low certainty of evidence” in the accuracy of these screening tests and note that, as of yet, there is no standard screening approach for child contacts.

The clinical management of TB-exposed children involves two different decision points. First, making a definitive diagnosis of TB disease in children is challenging because they are much less likely than adults to have microbiologically-confirmed disease, and it is, therefore, necessary to rely on “clinical diagnosis” using a combination of findings from history, physical exam, radiologic imaging, and TB immune-based testing (e.g., tuberculin skin test [TST] and interferon-γ release assays), all of which have their limitations (3). Clinical manifestations of intrathoracic TB such as cough, fever, and failure to thrive may be present in a host of other infectious and noninfectious diseases of childhood (2). Radiologic abnormalities found in intrathoracic TB in children are also nonspecific; access to radiography may be limited; the films are sometimes of poor quality; and interpretation is often difficult (2, 4, 5). TB immune-based tests are not always available in resource-limited settings and have suboptimal sensitivity in detecting TB infection and even lower sensitivity in detecting TB disease (6, 7). Second, clinicians are often reluctant to initiate TPT for TB infection in children with clinical or radiologic abnormalities compatible with intrathoracic TB (but for which they have low suspicion that TB is indeed the etiology) because of the concern that a child with undetected TB disease who is inadvertently treated with preventive monotherapy might develop drug resistance (8–10). Thus, there is an urgent need for evidence regarding the benefit of adding CXR to routine screening in child contacts, both for the identification of those with clinical and subclinical TB disease and for guidance on the provision of TPT to those in whom a TB clinical diagnosis is uncertain because of a nonspecific CXR or lack of access to CXR screening (11).

Here, we leveraged data from a longitudinal study of TB outcomes in a cohort of children (age ⩽15 yr) exposed at home to TB. We sought to assess 1) whether CXR screening provides additional diagnostic information compared with symptoms screening alone; 2) the association between the CXR findings and the risk of subsequent TB in children; and 3) the impact of TPT in children with CXR abnormalities at baseline.

Some of the results of this study have been previously reported in the form of abstracts (12, 13).

Methods

Study Setting and Design

Between September 2009 and August 2012, we conducted a longitudinal cohort study in Lima, Peru, of household contacts of index patients with pulmonary TB. The detailed design and methods of this study have been previously described and are summarized in the online supplement (14). In addition to screening for symptoms of TB, the study protocol also specified that CXRs should be obtained for child HHCs (age ⩽15 yr) with a positive TST and for TST-negative child contacts if the evaluating health center physician requested one. District health clinic staff then made a determination as to whether household contacts (HHCs) had TB disease and initiated TB treatment as they deemed appropriate. Children who were not diagnosed with TB disease at baseline were offered isoniazid preventive therapy (IPT) in accordance with the National Tuberculosis Program guidelines, which specified that IPT should be offered to all TB-exposed HHCs aged ⩽5 years, TST-positive HHCs aged between 5 and 19 years, and adults with specified comorbidities (15, 16).

Study Outcomes

We considered two distinct study outcomes: coprevalent TB at study enrollment and incident TB detected over a 12-month follow-up in individuals in whom coprevalent TB was ruled out at enrollment. The incident TB diagnoses were made by the health center providers on the basis of their expertise and National TB Guidelines (15).

Retrospective Assessment of Chest Radiographs

We conducted a retrospective reading of the digitalized anteroposterior films by a pulmonologist (Q.T.). To minimize the incorporation bias, the reader was blinded to the symptom findings, TST results, and clinical outcomes of the participants. We defined a CXR to be “abnormal” in the presence of any of the findings that could be compatible with TB disease in children (Table 1) (17–19). On the basis of the CXR-specific features, we further categorized the abnormal CXR findings into TB-suggestive and non–TB-suggestive abnormal CXR findings (20–22) (Table 1). All the abnormal CXR findings were verified by an independent radiologist (H.X.).

Table 1.

Definition of Chest Radiography Abnormality

Abnormalities that are compatible with TB disease

Hilar and mediastinal lymph node enlargement and calcifications, collapse (atelectasis), hyperinflation, noncalcified nodules, calcified nodules (Ghon focus), diffuse micronodules (miliary pattern), lobar and diffuse consolidations, reticular or reticulonodular opacities (fibrosis), tracheal compression or displacement, cavities, pleural effusion, pleural thickening, pleural calcification, pericardial effusion, and spinal abnormalities.

TB-suggestive abnormalities

For all children: bilateral diffused micronodules (military pattern) and unilateral pleural effusion; for infants and children <10 yr old: lymphadenopathy, lymphadenopathy with parenchymal consolidation, or lymphadenopathy with atelectasis; for children ⩾10 yr old: parenchymal consolidation or cavity.

Definition of abbreviation: TB = tuberculosis.

Statistical Analyses

We restricted the analyses to child contacts exposed at home to a microbiologically confirmed patient with TB. We first investigated the impact of screening algorithms on the diagnosis of coprevalent TB using a modified Poisson generalized estimating equation. Then, we used the same method to examine the risk of coprevalent TB among the child contacts who had symptoms and the CXR results available. We categorized children by increasing degrees of clinical suspicion of TB disease: those with no TB symptoms and a normal CXR (symptom-negative and CXR-normal); those with TB symptoms and normal CXR (symptom-positive and CXR-normal); those with no TB symptoms but with an abnormal CXR (symptom-negative and CXR-abnormal); and those with both TB symptoms and an abnormal CXR (symptom-positive and CXR-abnormal). Because most of those who received a CXR at enrollment were TST-positive at baseline, we restricted our analyses to TST-positive participants only.

We used a Cox frailty proportional model to examine the association between baseline TB screening results and the subsequent risk of incident TB during the 12-month follow-up period in children who were not diagnosed with coprevalent TB.

We conducted two further analyses to evaluate whether the efficacy of IPT varied by the radiologic findings. First, we repeated our initial analysis, including IPT status in our model. Second, we included an interaction term for IPT and degrees of clinical suspicion of TB to obtain the efficacy of IPT at each degree. Our previous work demonstrated that the protective efficacy of isoniazid (INH) for child contacts did not vary by the isoniazid resistance profile of their index cases, so all child contacts were included in the analyses, regardless of the isoniazid resistance profile of their index cases (16).

Results

We enrolled 4,468 child HHCs who had been exposed at home to an index patient with TB with a microbiologically confirmed disease. The prevalence of HIV in the child contacts was <0.1%. Sixteen children had developed TB disease before enrollment, so they were excluded from the analysis (Figure E1 in the online supplement. Of the remaining 4,452 child contacts, 4,448 (∼100%) were screened for TB symptoms, 908 (20%) of whom were positive; 4,280 (96%) had a TST result (of whom 1,124 [25%] were positive); 1,012 (23%) had a CXR (79 [2%] of which were abnormal); and 2,320 (54%) received IPT. Ninety-five (2%) were diagnosed with coprevalent TB (of whom 42 [44%] were microbiologically confirmed), and 95 (2%) were diagnosed with incident TB during the first 12 months of follow-up (Figure E1 and Tables 2 and E1–E3). Among the 46 child contacts who were asymptomatic and had an abnormal CXR at enrollment, 28 (61%) (of whom 2 reported <14 days coughing) were diagnosed either with coprevalent or incident TB.

Table 2.

Baseline Characteristics of Child Contacts (N = 4,452)

Characteristics	n, (%)
Age (yr)
0–5	1,961 (44)
6–10	1,261 (28)
11–15	1,230 (28)
Female sex	2,227 (50)
With BCG vaccination history	3,514 (81)
TB symptom screening
No TB symptoms	3,540 (80)
With TB symptoms	908 (20)
Did not receive	4 (0)
TB infection status screening
Negative	3,156 (71)
Positive	1,124 (25)
Not performed	172 (4)
IPT
No	2,109 (47)
Yes	2,341 (53)
Missing	2 (0)
CXR screening
Normal	933 (20)
TB-suggestive abnormal*	26 (1)
Non–TB-suggestive abnormal*	53 (1)
Not performed	3,440 (76)

Definition of abbreviations: BCG = bacillus Calmette-Guérin; CXR = chest radiography; IPT = isoniazid preventive therapy; TB = tuberculosis.

^*All 79 abnormal CXRs identified by the pulmonologist (Q.T.) were confirmed by the independent radiologist (H.X.).

Coprevalent TB

Symptomatic children were more likely to be diagnosed with coprevalent TB than those without TB symptoms (adjusted risk ratio [aRR], 4.71; 95% confidence interval [CI], 2.93–7.59) (Table E4). Among the 855 TST-positive contacts who were screened for TB symptoms and CXR, 627 (73%) were asymptomatic with a normal CXR, 157 (18%) were symptomatic with a normal CXR, 49 (6%) were asymptomatic with an abnormal CXR, and 25 (3%) had both TB symptoms and an abnormal CXR (Figure 1 and Table E5). Sixty-two of these children were diagnosed with coprevalent TB, of whom 29 (47%) were asymptomatic, and 36 (62%) had an abnormal CXR. Compared with children who were symptom-negative and CXR-normal, children who were symptom-positive and CXR-normal were 4.21-fold as likely to have coprevalent TB (aRR, 4.21; 95% CI, 1.28–13.86); children who were symptom-negative and CXR-abnormal were 25.06-fold more likely to have coprevalent TB (aRR, 25.06; 95% CI, 1.02–613.76); and children who were symptom-positive and CXR-abnormal were 34.23-fold more likely to have coprevalent TB (aRR, 34.23; 95% CI, 7.58–154.68) (Tables 3 and E5).

Figure 1.

Flow chart of 855 TST+ child contacts who underwent both symptom and chest radiography screening at enrollment. CXR− = chest X-ray normal; CXR+ = chest X-ray abnormal; SYM− = symptom negative; SYM+ = symptom positive; TB = tuberculosis; TST− = tuberculin skin test negative; TST+ = tuberculin skin test positive.

Table 3.

Association between Symptom and Chest Radiography Findings and Coprevalent Tuberculosis Among Tuberculin Skin Test-positive Child Contacts

		Univariate (N = 855)*
	Coprevalent TB, n (%)	Risk Ratio (95% CI)	P Values
SYM− CXR−†	12 (1.9)	Ref	—
SYM+ CXR−	14 (8.9)	4.21 (1.28–13.86)	0.018
SYM− CXR+	17 (37.0)	25.06 (1.02–613.76)	<0.048
SYM+ CXR+	19 (76.0)	34.23 (7.58–154.68)	<0.001

Definition of abbreviations: CI = confidence interval; CXR− = chest X-ray normal; CXR+ = chest X-ray abnormal; SYM− = symptom negative; SYM+ = symptom positive; TB = tuberculosis.

^*All variables were dropped during the backward stepwise selection algorithms.

^† TB symptom was defined as a cough for more than 14 d or any of the following symptoms: hemoptysis, productive cough, fever, weight loss, and night sweats.

Incident TB

Ninety-five child contacts were diagnosed with incident TB, 73 (77%) of whom were asymptomatic at enrollment (Figure E1). The risk of a subsequent incident TB diagnosis was not significantly different between symptomatic and asymptomatic children (adjusted hazard ratio [aHR], 1.28; 95% CI, 0.53–3.08) (Table E6). Figure 2 shows the results of baseline screening tests and the risk of incident TB among the 793 TST-positive contacts who underwent both symptom and CXR screening. Of the 35 child contacts diagnosed with TB during the follow-up, 23 (66%) were microbiologically confirmed, and 25 (71%) did not receive IPT. After adjusting for bacillus Calmette-Guérin (BCG) vaccination and IPT status, we found that compared with children who were symptom-negative and CXR-normal, children who were symptom-negative and CXR-abnormal were 26.71 times more likely to be diagnosed with incident TB disease (aHR, 26.71; 95% CI, 10.44–68.30), and those who were symptom-positive and CXR-abnormal were 26 times more likely to be diagnosed with incident TB (aHR, 25.94; 95% CI, 4.1–164.28) (Tables 4 and E7). Among the 129 TST-negative children who received both TB symptom screening and a CXR, 4 developed incident TB; all 4 had been symptom-positive and CXR-abnormal at baseline (Figure E2).

Figure 2.

Tuberculosis (TB) incidence among 793 TST+ child contacts who received both TB symptom and radiographic screening at enrollment, stratified by isoniazid preventive therapy status. CXR = chest X-ray normal; CXR+ = chest X-ray abnormal; IPT− = did not receive isoniazid; IPT+ = received isoniazid; SYM− = symptom negative; SYM+ = symptom positive; TST+ = tuberculin skin test positive.

Table 4.

Association between Symptom and Chest Radiography Findings and Incident Tuberculosis Among Tuberculin Skin Test-positive Child Contacts

		Univariate (N = 793)		Multivariate (N = 793)*
	Incident TB, n (%)	Hazard Ratio (95% CI)	P Values	Hazard Ratio (95% CI)	P Values
SYM− CXR−†	15 (2.44)	Ref	—	Ref	—
SYM+ CXR−	7 (4.90)	1.91 (0.75–4.83)	0.175	1.83 (0.72–4.66)	0.13
SYM− CXR+	11 (37.93)	25.76 (10.1–65.69)	<0.001	26.71 (10.44–68.3)	<0.001
SYM+ CXR+	2 (33.33)	27.31 (4.42–53.88)	<0.001	25.94 (4.1–164.28)	0.001

For definition of abbreviations, see Table 3.

^*Adjusted for the history of bacillus Calmette-Guérin vaccination and isoniazid preventive therapy.

^† TB symptom was defined as a cough for more than 14 d or any of the following symptoms: hemoptysis, productive cough, fever, weight loss, and night sweats.

Among the 10 IPT recipients who developed TB, 2 took IPT for more than 3 months, 7 took IPT for ⩽3 months (4 of whom stopped IPT because they were diagnosed with TB), and 1 had IPT adherence data missing. Among the 4 children (out of the 10) who were culture-positive on TB diagnosis, none developed breakthrough isoniazid-resistant TB during IPT, as shown by their isoniazid-resistant profiles (2 INH-sensitive and 2 INH-resistant), all of which matched the resistance profile of their respective index case. Among the 29 symptom-negative and CXR-abnormal child contacts, 15 received IPT (of whom 3 [20%] developed incident TB) and 14 did not (of whom 8 [57%] developed incident TB). The efficacy of IPT in symptom-negative and CXR-abnormal children was 82% (aHR, 0.18; 95% CI, 0.04–0.95). Among 12 symptom-positive and CXR-normal children, 2 of the 5 (40%) children who did not receive IPT were diagnosed with incident TB, while there were no incident cases among the 7 children who received IPT (Tables 5 and E8).

Table 5.

Effect of Isoniazid Preventive Therapy on Incident Tuberculosis Among Tuberculin Skin Test-positive Child Contacts, Stratified by Symptom and Chest Radiography Findings

	Without IPT	With IPT	Multivariate (N = 793)*
	Incident TB, n (%)	Incident TB, n (%)	Hazard Ratio (95% CI)	P Values
SYM− CXR−†	10 (3.85)	5 (1.41)	0.38 (0.13–1.16)	0.089
SYM+ CXR−	5 (7.69)	2 (2.56)	0.35 (0.06–1.92)	0.224
SYM− CXR+	8 (57.14)	3 (20)	0.18 (0.04–0.92)	0.039
SYM+ CXR+	2 (66.66)	0 (0)	0 (0–Inf)	—

Definition of abbreviations: CI = confidence interval; CXR− = chest X-ray normal; CXR+ = chest X-ray abnormal; IPT = isoniazid preventive therapy; SYM− = symptom negative; SYM+ = symptom positive; TB = tuberculosis.

^*Adjusted for the history of bacillus Calmette-Guérin vaccination.

^† TB symptom was defined as a cough for more than 14 d or any of the following symptoms: hemoptysis, productive cough, fever, weight loss, and night sweats.

When we examined the types of CXR findings most likely to subsequently predict incident TB, we found that among 35 TST-positive and CXR-abnormal child contacts, 28 (80%) had CXR-abnormal findings that were not considered to be suggestive of TB disease (Figure 3). Figure 3 shows that over half (8/14) of the children in the group that did not receive IPT were diagnosed with incident TB, while only 1 of 14 who did receive IPT (7%) was diagnosed with incident TB. The efficacy of IPT in this subgroup was 93% (95% CI, 59–99%).

Figure 3.

Tuberculosis (TB) incidence among 35 child contacts who were TST+ and had abnormal chest radiography at enrollment. CXR = chest X-ray normal; CXR+ = chest X-ray abnormal; IPT− = did not receive isoniazid; IPT+ = received isoniazid; SYM− = symptom negative; SYM+ = symptom positive; TST+ = tuberculin skin test positive.

Discussion

Here, we found that in child contacts of patients with TB, baseline CXRs contributed to the diagnosis of coprevalent TB beyond symptom screening and that CXRs also predicted the diagnosis of incident TB in children in whom TB had been ruled out at baseline, suggesting that an abnormal baseline CXR might be an indicator of subclinical or incipient disease. We also showed that IPT provides more than 80% protection against the subsequent diagnosis of incident TB even in children with an abnormal CXR at enrollment, regardless of the presence or absence of TB symptoms.

Our finding that a substantial proportion of child contacts diagnosed with coprevalent TB was asymptomatic is consistent with the results of similar studies from Brazil, South Africa, India, Spain, Cameroon, Central African Republic, and Uganda (23–31). Although these studies varied by age group and ethnicity of the study population, the definition of coprevalent TB, and the method of screening, they all reported that a sizeable portion of child contacts diagnosed with TB was asymptomatic at the time of baseline screening, with proportions ranging from 22–38%. These findings suggest that a substantial proportion of coprevalent TB will be missed if the presence of TB symptoms is used as the only screening criteria. Compared with these reports, our study found a moderately higher proportion of asymptomatic coprevalent TB disease (46%), which may reflect differences in how clinicians diagnosed coprevalent TB in these different settings. Notably, three of the studies mentioned above provided data on both symptoms and CXR results among those diagnosed with coprevalent TB; these reported that 88–100% of the asymptomatic coprevalent patients with TB had an abnormal CXR (23, 24, 27). These results suggest that routine CXRs in child contacts provide additional diagnostic information compared with symptoms screening alone.

Our finding that an abnormal CXR predicts a subsequent diagnosis of TB disease is consistent with several previous studies. A randomized clinical trial conducted in the 1950s followed 2,750 children with asymptomatic primary TB to evaluate the preventive efficacy of isoniazid. Two-thirds of children had abnormal CXRs at enrollment. In the placebo arm, those with hilar/paratracheal and parenchymal lesions had, respectively, 3.6 times and 12.1 times higher risk of TB incidence during a 1-year follow-up compared with those with a normal CXR (32). In a study from Hong Kong, researchers found that among 127 adults with an abnormal CXR compatible with TB and at least 5 negative sputum smear examinations before enrollment, 93 (53%) progressed to microbiologically confirmed TB within 30 months of enrollment (33). A mass districtwide CXR screening among adults in Czechoslovakia between 1961 and 1963 found that 44% of patients diagnosed with incident TB within the following year had fibrotic lesions on CXR before diagnosis (34). Similarly, among 152 TST-positive and sputum smear-negative individuals with apical lung lesions in South Africa, 88 (58%) developed microbiologically-confirmed pulmonary tuberculosis within 5 years (35). Our finding that in the absence of preventive therapy, 59% of child contacts with abnormal CXRs were diagnosed with incident TB within 1 year is highly consistent with these results. Furthermore, we also demonstrated that even when the evaluating clinician did not consider CXR abnormalities to be suggestive of TB disease, CXRs nevertheless strongly predicted subsequent incident TB.

Our finding that IPT reduces the occurrence of incident TB disease in children with CXR abnormalities is also consistent with the few previous studies that have examined this question. In the same clinical trial conducted in the 1950s mentioned above, children in the IPT arm had half of the risk of incident TB observed in those children who received a placebo during the 1-year follow-up (32). Another study to estimate the effect of TPT in child contacts with abnormal baseline CXRs (24) found no incident cases of TB among 11 child contacts under 5 years of age. These findings suggest that TPT may be an option when clinical diagnosis is uncertain because of a nonspecific CXR finding, as it may prevent those with incipient or subclinical TB from further disease progression. However, because a considerable proportion (in this case, 20%) of symptom-negative and CXR-abnormal child contacts who do get TPT develop incident TB, close follow-up is indicated to detect these TPT failures as early as possible. Because the current recommendation of 3HP (3 months of weekly isoniazid plus rifapentine) or 3HR (3 months of daily isoniazid plus rifampicin) for treatment of latent TB infection and 2HRZ + 2HR (2 months of daily isoniazid plus rifampin plus pyrazinamide followed by 2 months of daily isoniazid/rifampin) for treatment of nonsevere disease are not very different, an alternative strategy is to use 3HP or 3HR for treating latent TB infection and subclinical TB (36).

One issue that complicates the use of CXRs in the diagnosis of TB in children is that radiologic abnormalities of intrathoracic TB are often nonspecific, and interrater agreement on TB-suggestive radiographic findings is poor (37–39). Previous studies that compare CXR to symptom screening have yielded inconsistent conclusions. For example, one systematic review found that most children without TB-suggestive symptoms did not have TB-suggestive CXR abnormalities and concluded that the value of adding CXR to symptom screening in TB-exposed children is low (40). In contrast, another systematic review identified CXR as the most accurate screening test in child contacts (compared with a composite reference standard) (41). While the first study focused only on “TB-suggestive abnormalities”, the second used the same approach we employed here, examining the association between a more expansive list of abnormal CXR findings and TB occurrence (30). One recent study of patients with clinically manifested and microbiologically confirmed intrathoracic TB had key abnormal features with high specificity to support the disease (42). The results of these studies suggest that although CXRs can provide diagnostic and prognostic information, the criteria with which readers interpret them as TB-relevant may need to be reassessed on the basis of empiric data on the association between CXR findings and TB diagnoses (including subclinical disease) that depend on other diagnostic inputs.

The high incidence of subsequent TB disease among individuals with TB nonsuggestive radiographic abnormalities suggests that CXR screening may also help identify a subgroup of child contacts at risk of being diagnosed with TB in the near future. It is now widely accepted that TB is a dynamic continuum of states that includes recent TB infection at one extreme and clinically manifest disease at the other, but it also includes incipient and subclinical TB as earlier or intermediate states that may be amenable to therapeutic interventions that prevent further disease progression (43–46). Our results suggest that abnormalities on CXR cannot only help diagnose child contacts with coprevalent TB but can also be used to identify incipient or subclinical TB not captured through symptom screening. In resource-limited settings in which CXR may not be currently available for screening contacts, the use of TPT for TST-positive, asymptomatic child contacts becomes crucial as our results indicate that TPT is likely to protect child contacts from incident TB even if they have lesions that are not detected.

Our study has several limitations. First, even though the parent study cohort includes more than 14,000 HHCs, the number of study outcomes within the subcohort of child HHCs was small, and confidence intervals were consequently wide. However, it will be difficult to design a well-powered study in the future to estimate the value of adding CXR to symptom screening for preventing incident TB because TPT provides more than 80% protection against the later diagnosis of TB even in children with an abnormal CXR. Second, although our findings indicate that IPT may protect children with incipient or subclinical TB from progressing to disease within 1 year, we did not follow the cohort long-term to determine whether they have an increased risk of TB after completing IPT or if they develop drug-resistant forms of the infection. Therefore, in child contacts with a radiologic abnormality compatible with intrathoracic TB, the possible diagnosis of subclinical TB disease should be seriously considered, and an appropriate TB therapeutic intervention should be initiated in a timely manner as delayed diagnosis and treatment may lead to complications (47). Third, although health center providers were expected to make the clinical diagnosis according to Peruvian National Tuberculosis Program guidelines, we cannot rule out the possibility that different providers may have used different criteria for TB diagnosis. The lack of a consistent set of criteria for the determination of clinical TB may have led to nondifferential misclassification and driven the results toward the null. Fourth, the X-ray readings were blinded for the first reader, and abnormal findings were confirmed by a second reader who was not blinded. We, therefore, cannot rule out the possibility that the X-ray findings may be misclassified. Fifth, the set of TB symptoms that we used for screening was more restrictive than those proposed in the WHO symptom-based screening approach, and it is possible that this could reduce the generalizability of our results (11). Sixth, because CXR was used as one of the tools for clinical diagnosis, we may have overestimated the contribution of CXR to clinical diagnosis because of incorporation bias. Finally, it is not clear whether the protective effect of the 6-month INH preventive therapy regimen used in our study can be generalized to other different TPT regimens.

Conclusions

Our results strongly support the inclusion of CXR in routine screening of child TB contacts when possible. Nonspecific CXR findings may indicate incipient or subclinical TB disease that requires formal treatment. However, access to TPT should not be constrained because it remains highly effective in asymptomatic contacts, even in those with CXR abnormalities.