- Research article
- Open Access
- Open Peer Review
A prospective longitudinal study of tuberculosis among household contacts of smear-positive tuberculosis cases in Lima, Peru
© The Author(s). 2016
- Received: 3 December 2015
- Accepted: 3 June 2016
- Published: 8 June 2016
Household contacts (HHCs) of TB cases are at increased risk for TB disease compared to the general population but the risk may be modified by individual or household factors. We conducted a study to determine incident TB among HHCs over two years after exposure and to identify individual and household level risk factors.
Adults newly diagnosed with a first episode of smear-positive pulmonary TB (index cases) between March 2010 and December 2011 in eastern Lima, were interviewed to identify their HHC and household characteristics. TB registers were reviewed for up to two years after the index case diagnosis and house visits were made to ascertain TB cases among HHC. The TB incidence rate ratio among HHCs as a function of risk factors was determined using generalized linear mixed models.
The 1178 index cases reported 5466 HHCs. In 402/1178 (34.1 %) households, at least one HHC had experienced a TB episode ever. The TB incidence among HHCs was 1918 (95%CI 1669–2194) per 100,000 person-years overall, and was 2392 (95%CI 2005–2833) and 1435 (95%CI 1139–1787) per 100,000 person-years in the first and second year, respectively. Incident TB occurred more than six months following the index case’s TB diagnosis in 121/205 (59.0 %) HHCs. In HHCs, bacillary load and time between symptoms and treatment initiation in the index case, as well as the relationship to the index case and the sex of the HHC all had a significant association with TB incidence in HHCs.
Incidence of TB among HHCs was more than ten times higher than in the general population. Certain HHC and households were at higher risk of TB, we recommend studies to compare HHC investigation to households at highest risk versus current practice, in terms of efficiency.
- Multilevel analysis
- Household contact investigation
- Longitudinal study
Household contacts (HHCs) of tuberculosis (TB) patients are at higher risk for TB infection and disease. In recent studies, 4.5 to 7.8 % of HHCs have been found to have active disease [1–3]. Investigating HHCs may facilitate timely diagnosis and treatment of active TB as well as of latent TB infection. However, contact tracing in general and HHC investigation in particular are resource-intensive activities, and their consistent implementation is a challenge in both high- and low-income countries. TB control programs in low-income high-incidence settings face competing priorities [4–6] but limited data suggest that screening HHCs of TB cases for active disease may be cost-effective in low- and middle-income countries . The World Health Organization recently extended its recommendations for HHC tracing to these countries .
TB risk in HHCs may vary by individual and household as well as by background community incidence, and investigating all HHCs in high-burden, low-resource countries may be inefficient or unfeasible. Estimates of risk should consider the correlation of risk factors of HHCs within a single household  to avoid overestimating effects when ignoring that correlation. To better inform TB programs on the precise burden of TB over time in HHCs, studies with sufficiently long follow-up and with comparable methods need to be conducted [1, 2, 10]. A better understanding of TB risk over time and risk factors would provide evidence to increase efficiency of HHC investigation guidelines in high TB burden settings  by strategies and timing following epidemiological data.
Studies have shown a high burden of TB among HHC of multidrug resistant (MDR) TB and drug susceptible cases in Peru [10–13]. We conducted a prospective study to determine the TB incidence over 2 years among HHC of all new TB cases in eastern Lima, and to identify individual and household level risk factors that determine this incidence.
Design, setting and study population
We conducted a longitudinal prospective study of TB incidence among HHCs of smear-positive cases in an eastern Lima district. The district is the most inhabited in Peru with 1,047,026 inhabitants in 2013 of which 27 % are living in poverty, higher than the 17.5 % for Lima . TB incidence in the district in 2012 was 175 cases of TB per 100,000 inhabitants, and among those patients tested for HIV 2.9 % were positive. The district has 34 public health facilities where clinics implementing the directly observed treatment short-course strategy (DOTS) operate. Consenting adults diagnosed with a first episode of smear-positive pulmonary TB between March 2010 and December 2011 (index cases) were enrolled in the present study.
Peruvian guidelines recommend evaluation of HHCs (symptoms screening, tuberculin skin test and chest X-ray) at diagnosis of the index case, and at the second and sixth month of the index case treatment. Isoniazid prophylaxis for TB is recommended to all HHCs under five years old and to HHC from 5 to 19 years old with latent tuberculosis infection documented with a positive tuberculin skin test. The present study did not interfere with routine HHC investigation by National TB Program health center staff. In 2010–2011, the study district reported that 74 % of HHCs had been evaluated and that prophylaxis had been prescribed to 54 % of eligible HHCs.
Baseline data collection and longitudinal detection of TB among HHCs
At enrollment, trained research field workers interviewed index cases to collect demographic and HHC data using a structured questionnaire. Index cases provided the names of their HHCs and any previous TB episodes among them. To determine incident TB for known HHCs, district TB registers were reviewed periodically for up to two years after the diagnosis of the index case. In each of these reviews, the names of newly registered TB cases were compared to the names of the HHC reported by index cases to capture TB diagnoses among HHCs. At the end of the study, the database of HHC reported by the index cases was matched with the TB registers in order to capture HHC registered as a TB case but missed by the periodic visual inspection. Queries were run to match combinations of given and family names and initials, thus reducing the risk of missing matches. To confirm true matches, two authors (LO and KV) subsequently verified all hits.
To capture TB diagnoses outside the study district or in a private health facility, a single household visit was conducted at the end of the two-year follow-up period. During that visit, the index case or the head of the household was asked if any of the HHC had developed TB since the index case diagnosis. Households where the index case had defaulted treatment or died did not receive the final follow-up visits. This was the case for 12.8 % (151/1178) of the study households.
A HHC was defined as a person sleeping under the same roof and sharing cooking facilities with the index case for at least three months before the case’s diagnosis. Past TB in an HHC was defined as a TB episode occurring more than two years before the diagnosis of the index case; recent TB was defined as TB diagnosed within the two years before the diagnosis in the index case. Incident TB in an HHC was defined as TB that came to diagnosis up to two years following the diagnosis of the index case.
Data were entered in an Access (Microsoft Redmond, WA, USA) database and analyzed with Stata v.12 (Stata Corp, 12.0, College Station, TX).
Index cases who lived in congregate settings such as shelters or rehabilitation centers for addicts and those who lived alone were excluded from the analysis. The contact’s, household and index case’s characteristics of HHC with incident TB were compared to those of HHC without an episode of TB. As our data source for incident TB episodes among HHC was the district TB register, which was checked up to two years after the inclusion of the last index case, complemented with a final household visit, we considered the two years follow up complete for all HHC. Only those that developed incident TB were censored.
Risk factors for acquiring TB, including those not measured directly, may be correlated at household/index case level. To account for this correlation, we used generalized linear mixed models (GLMM) with random effects for household, in Poisson regression analysis for count outcomes. A simple logistic regression model may overestimate the effects measured. The outcome was the number of HHCs with incident TB. To protect against the impact of model-misspecification, robust standard errors were used. Covariates that were significant (at p < 0.2) in the bivariate analysis as well as those that were a priori expected to have an influence on or to be confounding the incidence of TB in HHC were considered for inclusion in the model. Models were compared by backward selection. Variables with the weakest association were taken out until a significant difference between two models was found.
Characteristics of the index cases and household contacts
Age (median, IQR) in years
Presence of cough
Time from symptoms to treatment initiation (in days)
For drug susceptible TB
For MDR TB
Past TB episodes
Recent TB episodes
Persons per room
Household contact characteristics (n = 5466)
Age (median, IQR) in years
Relationship with the index case
TB burden in households and among HHCs
Past, recent and incident tuberculosis episodes among household contacts, per household
Number of tuberculosis episodes in household contacts
≥1 past case only
≥1 recent case only
≥1 incident case only
≥1 past + ≥1 recent case
≥1 past + ≥1 incident case
≥1 recent + ≥1 incident case
≥1 past + ≥1 recent + ≥ 1 incident case
Household contacts characteristics associated to incident TB in HHC
Incident tuberculosis in 5466 household contacts (HHC) in function of the HHC’s characteristics
Incident TB in the HHC
Crude odds ratio (95%CI)
Age in years
16 (2.4 %)
657 (97.6 %)
56 (3.8 %)
1407 (96.2 %)
100 (5.0 %)
1883 (95.0 %)
33 (2.5 %)
1314 (97.6 %)
88 (3.2 %)
2704 (96.9 %)
117 (4.4 %)
2557 (95.6 %)
84 (5.9 %)
1332 (94.1 %)
21 (2.2 %)
925 (97.8 %)
27 (6.2 %)
410 (93.8 %)
38 (3.7 %)
993 (96.3 %)
35 (2.1 %)
1601 (97.9 %)
14 (4.8 %)
278 (95.2 %)
191 (3.7 %)
4983 (96.3 %)
6 (4.4 %)
129 (95.6 %)
199 (3.7 %)
5132 (96.3 %)
Index case and household characteristics associated to incident TB in HHC
Incident tuberculosis among household contacts in 1178 households in function of the index case and household characteristics
Incident TB in the household
Crude odds ratio (95%CI)
Index case characteristics
86 (12.3 %)
611 (87.7 %)
77 (16.0 %)
404 (84.0 %)
96 (14.4 %)
571 (85.6 %)
34 (14.2 %)
206 (85.8 %)
11 (11.0 %)
89 (89.0 %)
22 (13.8 %)
149 (87.1 %)
160 (14.1 %)
975 (85.9 %)
3 (7.0 %)
40 (93.0 %)
Time from symptoms to treatmenta
50 (10.0 %)
448 (90.0 %)
113 (16.6 %)
567 (83.4 %)
Smear positivity at diagnosisb
1 (2.5 %)
39 (97.5 %)
110 (12.7 %)
755 (87.3 %)
52 (19.1 %)
221 (80.9 %)
159 (13.7 %)
1003 (86.3 %)
4 (25.0 %)
12 (75.0 %)
Past TB episodesc in the household
41 (19.0 %)
175 (81.0 %)
122 (12.7 %)
840 (87.3 %)
Recent TB episodesd in the household
23 (20.4 %)
90 (79.7 %)
140 (13.2 %)
925 (86.9 %)
Persons per room (median, IQR)
Multilevel analysis of index case, household and household contacts characteristics associated to incident TB in HHC
Index case, household and household contacts characteristics and their relation to incident tuberculosis among household contacts
Bivariate GLMM analysis
Multivariate GLMM analysis
Index case characteristics
Time from symptoms to treatmenta
Smear positivity at diagnosisb
Past TB episodesc in the household
Recent TB episodesd in the household
Persons per room
Household contact characteristics
Among the 287/5466 HHCs who had three of the above determinants –excluding male sex- associated to incident TB (siblings/partners, index case with a time from symptoms to treatment of > 30 days and a high bacillary load), 29 (10.1 %) had incident TB; among the 2419 HHC that had two of those factors, 159 (6.6 %) had incident TB.
The incidence of TB in HHC of new smear-positive TB cases in this district was 1918 per 100,000 person-years, ten times higher than that for the general population in the district. Among incident TB cases, 59 % occurred beyond six months of the index case’s diagnosis and would not have been identified as by current HHC investigation contracting procedures. TB burden in households was high, with 34 % of households having a HHC with past, recent or incident TB in addition to the index case. Being a sibling or a partner of the index case, and more than 30 days between symptoms treatment initiation or a high bacillary load in the index cases were associated with incident TB.
Index cases included in the study were representative of the new TB cases in the district, but this study has some limitations. Incident TB in HHC could have been diagnosed outside the district or in facilities not managed by the Ministry of Health and not appear in the TB register. We conducted household visits to ascertain these cases. However, households of index cases who defaulted treatment or died were not visited. This could lead to underestimating TB incidence in HHC. Furthermore, our definition of incidence was based on diagnosis and case notification. Since only individuals with symptoms are likely to present for diagnosis, true incidence may be underestimated. Past and recent TB episodes might also be underestimated as these were reported by the index case. Strengths of this study include its prospective design, the use of two complementary follow up approaches to ascertain incident TB in HHC and accounting for the correlation of data at household level in the analysis.
In a systematic review of 25 studies in low- and middle-income countries, the weighted estimated average of incident TB in HHC in the first year of follow-up was 478 (95%CI 897–2427) per 100,000 person-years and 831 (95%CI 624–1106) per 100,000 person-years in the second year . We found rates approximately two and five times higher, respectively. Some studies conducted in Lima have determined incident TB among HHCs of selected TB cases. In the first and in the second year, 2360 incident TB episodes per 100,000 person-years occurred among HHC of 80 MDR-TB index cases in southern Lima , while a more recent study found 2456 incident TB episodes per 100,000 person-years occurred among HHC of 213 multidrug resistant TB cases and 4351 incident TB episodes per 100,000 person-years among HHC of 487 drug susceptible index cases . A retrospective study that included 693 MDR and extensively drug resistant TB index cases found 3165 and 1092 incident TB cases in HHC per 100,000 person-years in the first and second year, respectively . Our data demonstrate similar high rates of incident TB among HHCs of a large sample of unselected new pulmonary TB patients.
Our findings confirm that siblings and partners [15, 16], male HHC [10, 16], and HHC exposed to index cases with higher sputum bacillary loads  or that are infectious for a longer period  are more likely to develop incident TB. Contrary to what is expected, HHC under six years old were not found to be at higher risk for incident TB than older age groups. However, it is possible that TB in children is under diagnosed in Peru. One explanation consistent with our findings is that there are important effects at the household/index case level on the incidence of TB in HHCs. This was also found in a multilevel study in Pakistan that looked into household and HHC determinants of TB infection . Characteristics at household level that may have influence should be considered in future studies, such as ventilation, sleeping arrangements, frequent exposure to non-household members as well as genetic predisposition of families. Measuring these potential risk factors might strengthen of the index case/household association on incidence rate ratio of TB among HHCs. Prioritizing individuals at highest risk for HHC investigation could reduce the number of HHC to be screened, but such selection may be logistically cumbersome. In addition, staff, patients and HHC themselves may not find it acceptable. However, it could be an option to use index case characteristics that put households at higher risk in order to select complete households in which to screen all contacts.
While HHCs of active TB patients may have a much higher risk of TB than the general population, TB cases in HHCs still account for only a small proportion of the total TB burden in a medium and high incidence area . For example, the 205 incident cases found in our study corresponded to approximately 5 % of the district’s TB notifications during the same period. Furthermore, in some high incidence areas, community transmission can be higher than household transmission [19, 20]. The background TB incidence at which full HHC contact investigation becomes cost-effective as compared to investigating only selected high-risk HHC or index cases/households would provide valuable information for National TB Programs decision-making.
Finally, National TB Programs typically attempt to evaluate HHC at the start of the index case’s TB treatment, at the change of their treatment phase and at the end of their treatment. However, this approach does not take into consideration the time that may elapse occur between TB infection and disease [1, 18, 20]. For that reason, even full compliance to Peruvian guidelines for HHC investigation of drug sensitive cases would have failed to identify 59 % of incident TB episodes as HHCs.
This study demonstrated that HHC of TB cases in Lima, Peru are at very high risk for incident TB. Studies comparing the feasibility and cost effectiveness of all-inclusive versus risk based HHC investigation strategies within National TB Programs are needed to inform policy formulation.
CI, confidence interval; GLMM, generalized linear mixed models; HHC, household contact; IRR, incidence rate ratio; TB, tuberculosis
We thank Dr. Andrés Lescano at Universidad Peruana Cayetano Heredia for his valuable advice in the interpretation of the statistical methods.
This study was funded by the Belgian Cooperation through a Framework Agreement project of institutional collaboration between the Institute of Tropical Medicine in Antwerp, Belgium and the Instituto de Medicina Tropical Alexander von Humboldt in Lima, Peru and through the Peru ICOHRTA Network for AIDS/TB Research Training National Institutes of Health (grant 5U2RTW007374-05). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Availability of data and materials
The data supporting findings of this manuscript will be shared upon request submitted to the corresponding author.
Conception and design of the work: LO, LS, EG, CS, PVDS. Study implementation: LO, CS, PVDS, Data analysis: LO, KV, PVDS, data interpretation, critical revision and important intelectual input: LO, LS, KV, TB, TB, EG, PVDS. First draft: LO. All authors approved the final version to be published. to the acquisition, analysis.
The authors declare that they have no competing interests.
Consent for publication
Ethics approval and consent to participate
The protocol was approved by the Institutional Ethics Committees from the Universidad Peruana Cayetano Heredia and from the University of Antwerp. All enrolled index cases provided written informed consent. The data was analyzed anonymously.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Fox GJ, Barry SE, Britton WJ, Marks GB. Contact investigation for tuberculosis: a systematic review and meta-analysis. Eur Respir J. 2013;41(1):140–56.View ArticlePubMedGoogle Scholar
- Morrison J, Pai M, Hopewell PC. Tuberculosis and latent tuberculosis infection in close contacts of people with pulmonary tuberculosis in low-income and middle-income countries: a systematic review and meta-analysis. Lancet Infect Dis. 2008;8(6):359–68.View ArticlePubMedGoogle Scholar
- Shah NS, Yuen CM, Heo M, Tolman AW, Becerra MC. Yield of contact investigations in households of patients with drug-resistant tuberculosis: systematic review and meta-analysis. Clin Infect Dis. 2014;58(3):381–91.View ArticlePubMedGoogle Scholar
- Golub JE, Dowdy DW. Screening for active tuberculosis: methodological challenges in implementation and evaluation. Int J Tuberc Lung Dis. 2013;17(7):856–65.View ArticlePubMedGoogle Scholar
- Thanh THI, Ngoc SD, Viet NN, Nguyen HV, Horby P, Cobelens FGJ, et al. A household survey on screening practices of household contacts of smear positive tuberculosis patients in Vietnam. BMC Public Health. 2014;14:713.View ArticlePubMedPubMed CentralGoogle Scholar
- Hwang TJ, Ottmani S, Uplekar M. A rapid assessment of prevailing policies on tuberculosis contact investigation. Int J Tuberc Lung Dis. 2011;15(12):1620–3.View ArticlePubMedGoogle Scholar
- Yadav RP, Nishikiori N, Satha P, Eang MT, Lubell Y. Cost-effectiveness of a tuberculosis active case finding program targeting household and neighborhood contacts in Cambodia. Am J Trop Med Hyg. 2014;90(5):866–72.View ArticlePubMedPubMed CentralGoogle Scholar
- World Health Organization. Recommendations for investigating contacts of persons with infectious tuberculosis in low- and middle-income countries. 2012. WHO/HTM/TB/2012.9Google Scholar
- Akhtar S, Rathi SK. Multilevel modeling of household contextual determinants of tuberculin skin test positivity among contacts of infectious tuberculosis patients, Umerkot. Pakistan Am J Trop Med Hyg. 2009;80(3):351–8.PubMedGoogle Scholar
- Grandjean L, Crossa A, Gilman RH, Herrera C, Bonilla C, Jave O, et al. Tuberculosis in household contacts of multidrug-resistant tuberculosis patients. Int J Tuberc Lung Dis. 2011;15(9):1164–9.View ArticlePubMedGoogle Scholar
- Bayona J, Chavez-Pachas AM, Palacios E, Llaro K, Sapag R, Becerra MC. Contact investigations as a means of detection and timely treatment of persons with infectious multidrug-resistant tuberculosis. Int J Tuberc Lung Dis. 2003;7(12 Suppl 3):S501–9.PubMedGoogle Scholar
- Becerra MC, Appleton SC, Franke MF, Chalco K, Arteaga F, Bayona J, et al. Tuberculosis burden in households of patients with multidrug-resistant and extensively drug-resistant tuberculosis: a retrospective cohort study. Lancet. 2011;377(9760):147–52.View ArticlePubMedGoogle Scholar
- Grandjean L, Gilman RH, Martin L, Soto E, Castro B, Lopez S, et al. Transmission of multidrug-resistant and drug-susceptible tuberculosis within households: a prospective cohort study. PLoS Med. 2015;12(6), e1001843. doi:10.1371/journal.pmed.1001843.View ArticlePubMedPubMed CentralGoogle Scholar
- Población y Vivienda. En: Instituto Nacional de Estadística en Informática. 2014. http://www.inei.gob.pe/estadisticas/indice-tematico/poblacion-y-vivienda/ Accessed 27 Sep 2015.
- Singh J, Sankar MM, Kumar S, Gopinath K, Singh N, Mani K, et al. Incidence and prevalence of tuberculosis among household contacts of pulmonary tuberculosis patients in a peri-urban population of South Delhi. India PLoS One. 2013;8(7), e69730. doi:10.1371/journal.pone.0069730.View ArticlePubMedGoogle Scholar
- Jia Z, Cheng S, Ma Y, Zangh T, Bai L, Xu W, et al. Tuberculosis burden in China: a high prevalence of pulmonary tuberculosis in household contacts with and without symptoms. BMC Infect Dis. 2014;14:64.View ArticlePubMedPubMed CentralGoogle Scholar
- Grzybowski S, Barnett GD, Styblo K. Contacts of cases of active pulmonary tuberculosis. Bull Int Union Tuberc. 1975;50(1):90–106.PubMedGoogle Scholar
- Chang KC, Leung CC, Tam CM. Household contact investigation of tuberculosis in low-income and middle-income countries: public-health impact. Lancet Infect Dis. 2009;9(1):3–4.View ArticlePubMedGoogle Scholar
- Verver S, Warren RM, Munch Z, Richardson M, van der Spui GD, Borgdorff MW, et al. Proportion of tuberculosis transmission that takes place in households in a high-incidence area. Lancet. 2004;363(9404):212–4.View ArticlePubMedGoogle Scholar
- Brooks-Pollock E, Becerra MC, Goldstein E, Cohen TB, Murray MB. Epidemiologic inference from the distribution of tuberculosis cases in households in Lima. Peru J Infect Dis. 2011;203(11):1582–9.View ArticlePubMedGoogle Scholar