INT J TUBERC LUNG DIS 14(11):1388–1394 © 2010 The Union

Screening for tuberculosis in asylum seekers: comparison of chest radiography with an interview-based system S. Schneeberger Geisler,* P. Helbling,* J. P. Zellweger,† E. S. Altpeter* * Division of Communicable Diseases, Federal Office of Public Health, Berne, † Swiss Lung Association, Berne, Switzerland SUMMARY SETTING:

Mandatory initial screening of asylum seekers for tuberculosis (TB) in Switzerland, 2004–2005 and 2007–2008. O B J E C T I V E : To compare the yield of screening by chest radiography with an individual assessment based on geographic origin, personal history and symptoms. D E S I G N : Cross-sectional retrospective comparison of two 2-year periods. R E S U LT S : The prevalence of detected TB cases was defined as the proportion of screenees starting antituberculosis treatment for culture-confirmed pulmonary TB within 90 days. TB prevalence was 14.3 per 10 000 asylum seekers screened (31/21 727) using chest radiography and 12.4 (29/23 402) using individual assessment. The sensitivity of radiography was 100% vs. 55% for individual assessment, but its specificity was lower (89.9%

vs. 96.0%, respectively). The higher sensitivity of radiography meant shorter delays between screening and start of treatment (median 6 vs. 25 days). Its lower specificity led to a larger proportion of screenees needing further investigations for suspicion of TB (12% vs. 4%). C O N C L U S I O N : The interview-based system initially missed more cases, but the ultimate 90-day yield was comparable for the two periods. The main difference is the delay until start of treatment, which potentially increases transmission and secondary cases. The radiograph system was more burdensome to both the health care system and the screenees, as more suspects required further investigations. K E Y W O R D S : tuberculosis; screening; immigrants; asylum seekers

ABOUT HALF of the European countries have a screening programme for tuberculosis (TB) among immigrants,1,2 yet little is known about the effectiveness of these programmes. Screening purports to reduce the time to diagnosis, and therefore the period of infectiousness, resulting in the prevention of secondary cases.3 As TB is only modestly contagious and most acquired infections never progress to disease stage, the effectiveness of screening is considered to be of limited value while its costs are high.4 Switzerland had an annual notification rate of six TB cases per 100 000 population in 2007.5 Two thirds of cases are currently of foreign origin. The incidence in the native population has decreased over many decades.6 The only high-prevalence group easily accessible to TB screening is the immigrant category of asylum seekers when they report to one of the four registration centres to claim asylum. Other groups of immigrants to Switzerland are not from highprevalence countries and/or are not easily accessible. Chest radiography was used until the end of 2005 as the basic method for screening for pulmonary TB, except for children aged 280 cases per 100 000 population. Age was classified into groups of 0–14, 15–34, 35–54 and >54 years. Double entries (more than one claim for asylum by the same person) as identified by soundex methods were excluded. All data management was performed using Microsoft Access 2003 for Windows XP (MicroSoft, Redmond, WA, USA). The overall yield was the number of cultureconfirmed pulmonary TB cases that had been started on anti-tuberculosis combination treatment within 90 days after screening in the two periods. The detailed comparison of the two periods also included coverage and the initial results of the screening tool measured as sensitivity, specificity and, as a summary measure for both sensitivity and specificity, the likelihood ratios with 95% confidence intervals (CIs): sensitivity/(1 − specificity) for the positive and (1 − sensitivity)/specificity for the negative ratio. The delay from screening to start of treatment was estimated using Kaplan-Meier survival analysis. Statistics were performed using R Version 2.7.1 for Windows XP (R Foundation for Statistical Computing, Vienna, Austria). All two-by-two tables were evaluated by Open Source Epidemiologic Statistics for Public Health. Approval by an ethical committee was not sought for this retrospective study, as the Federal Office of Public Health is legally entitled to collect and use the data of both databases for evaluation purposes.

RESULTS A total of 25 856 persons applied for asylum in Switzerland during the period from 2004 to 2005, and 27 450 in the period from 2007 to 2008 (Federal Office of Migration). During these periods, respectively 21 987 (coverage 84%) and 23 722 (coverage 85%) screening procedures were documented. Cases of

Figure 1 Culture-confirmed cases of PTB notified within 90 days after screening in the periods 2004–2005 (chest radiography) and 2007–2008 (screening based on origin, symptoms and history). PTB = pulmonary tuberculosis (culture-confirmed).

1390

The International Journal of Tuberculosis and Lung Disease

Table 1 Tuberculosis cases starting treatment within 90 days after screening. Comparison of the periods 2004–2005 and 2007–2008 2004–2005 (n = 21 727) Culture Site of disease Pulmonary Extra-pulmonary

2007–2008 (n = 23 402) Culture

Positive

Negative

Positive

Negative

31 8

32 4

29 8

8 4

repeat screening examinations (n = 260 and n = 320, respectively) were excluded from further analyses. During the period 2004–2005, 15% of the asylum seekers did not undergo radiological examination due to age < 15 years or pregnancy (Figure 1). In the period 2004–2005, 31 cases of cultureconfirmed pulmonary TB started anti-tuberculosis treatment within 90 days after the screening procedure. In the period 2007–2008, there were 29 such cases (Figure 1, Table 1). The prevalence of pulmonary TB was respectively 14.3 and 12.4 per 10 000 asylum seekers. The difference is not statistically significant. Radiography led to more diagnoses of pulmonary TB that remained unconfirmed by culture (Table 1), representing decisions to treat with combination anti-tuberculosis treatment based on clinical and radiological findings. During the period 2004–2005, all 31 cases of pulmonary TB had an abnormal chest radiograph at screening. During the period 2007–2008, only 16/29 cases (55%) were identified as TB suspects at screening.

Figure 2 Delay between screening and start of treatment, 2004–2005 and 2007–2008. During the period 2004–2005, time to treatment was shorter than during the period 2007–2008.

The 13 cases not detected by screening had scores below the threshold for which further investigations for TB were required. These cases also needed medical attention in the weeks following the screening procedure when they developed symptoms. Chest radiographic screening resulted in a faster identification of pulmonary TB. The median delay from screening to treatment was 6 days in 2004–2005 (range 0–79) and 25 days (range 0–85) in 2007–2008 (Table 2, Figure 2). The median delay in the subgroup

Table 2 Culture-confirmed PTB starting treatment within 90 days after screening. Comparison of the periods 2004–2005 and 2007–2008 2004–2005

Total Sex Men Women Age groups, years 0–14 15–34 35–54 >54 Tuberculosis prevalence in country of origin (/100 000 population)* 0–20 21–60 61–280 >280 Median time to start of treatment, days Suspicion of PTB at screening No Yes Screening not done

2007–2008

Difference between periods

PTB n

Screenees n

Prevalence /10 000 screenees

PTB n

Screenees n

31

21 727

14.3

29

23 402

12.4

−1.9

−8.6–4.9

25 6

16 154 5 573

15.5 10.8

25 4

17 648 5 754

14.2 7.0

−1.3 −3.8

−9.5–6.9 −14.8–7.2

0 27 4 0

2 780 15 263 3 393 291

0.0 17.7 11.8 0.0

0 25 4 0

3 040 16 907 3 145 310

0.0 14.8 12.7 0

0.0 −2.9 0.9 0.0

0.0–0.0 −11.7–5.9 −16.1–17.9 0.0–0.0

0 6 12 13

248 7 639 7 434 6 406

0.0 7.9 16.1 20.3

0 0 16 13

146 4 964 12 071 6 221

0.0 0.0 13.3 20.9

0.0 −7.9 −2.9 0.6

0.0–0.0 −14.1– −1.6 −14.1–8.3 −15.2–16.4

22 444 958 0

5.8 167.0 0

5.8 26.2 0

6 0 31 0

Prevalence Prevalence /10 000 /10 000 screenees screenees

25 16 263 2 202 3 262

* WHO report 2009.5 PTB = pulmonary tuberculosis; CI = confidence interval.

0.0 140.8 0.0

13 16 0

95%CI

Radiographic vs. symptom and history-based screening of immigrants

not identified by screening in 2007–2008 was 40 days (range 16–85). Men were more frequently affected than women. Asylum seekers between the ages of 15 and 54 years had a higher prevalence of pulmonary TB than other age groups. There were no significant differences as to age and sex between the two periods (Table 2). There were, however, differences between the two periods regarding the geographic origins of the cases (Table 2). In the group of asylum seekers originating from countries with an estimated TB prevalence of 21–60, the number of cases was 7.9/10 000 asylum seekers in 2004–2005 vs. 0 in 2007–2008, a statistically significant difference. The difference in the number of cases per 10 000 screenees found became smaller and statistically insignificant with increasing TB prevalence in countries of origin (Table 2). This was to be expected, as the screening procedure in 2007–2008 was designed to also assign score points based on TB prevalence in the country of origin. The higher the prevalence, the more likely was further investigation for TB. To estimate the sensitivity and specificity of radiographic screening (in 2004–2005), the denominator comprised both those with a ‘normal chest radiograph’ and those with ‘no chest radiograph taken’, giving a sensitivity of 100% and a specificity of 89.6% in identifying pulmonary TB. The positive likelihood ratio was 9.99 (95%CI 9.99–10.0) and the negative likelihood ratio was 0.00 (95%CI 0–∞). In 2007–2008, overall sensitivity was 55.2%, specificity 96.0%, and the positive and negative likelihood ratios 13.7 (95%CI 12.37–15.15) and 0.5 (95%CI 0.40–0.54), respectively. The items in the interview (except for smoking, a variable initially thought to serve as an alternative explanation for cough) and the assessment of the general health status had relatively high specificity, ranging from 85% to nearly 100% (Table 3). The sensitivities, however, were low, ranging from 7.4% to 38%. The three highest positive likelihood ratios were for subjects presenting with illness as judged by the nursing staff (21.3, 95%CI 3.22–141), mentioning previous anti-tuberculosis treatment (17.9, 95%CI 7.38– 43.50) or stating cough (3.4, 95%CI 2.83– 4.09). Table 3

1391

A major difference between the two systems lies in the number of further investigations necessary as a consequence of the screening procedure: 12% of all screenees in 2004–2005 vs. 4% in 2007–2008, with corresponding yields of respectively 1.4% and 1.7% (Figure 1). During 2004–2005, only 1% of the radiographs were classified as ‘highly suspicious for TB, to be immediately examined’ by the pulmonologists. The median delay from screening to start of TB treatment in the cases (n = 23) in this group was 4 days. Other radiological abnormalities (11%) were classified to be examined less urgently, and the delay in these cases (n = 8) was 34 days.

DISCUSSION Compared to radiographic screening, a screening system based on geographic origin, symptoms and history failed to rapidly identify half of all cultureconfirmed cases of pulmonary TB. However, the absolute numbers of pulmonary TB cases starting treatment within 90 days after screening were small and very similar in both periods. This demonstrates that when a screening system fails to identify suspects, access to care alone leads to diagnosis in many cases within a reasonable time frame. Compared to the system using radiography, time to treatment was delayed by a median of about 3 weeks in the interview-based system. Sensitivity was high for radiography and low in the interview-based system, while specificity was high in both systems. Previous findings led to the conclusion that chest radiography is the method of choice for active case finding in populations with an increased TB risk, provided that they can be reached, resources are available and treatment can be assured.2 Active case finding reduces delays in the treatment of cases that would otherwise also be diagnosed, although with some delay, provided that access to care is assured for spontaneously presenting patients.2 In the context of asylum procedures in Switzerland, access to care was assured. It compensated for the lower sensitivity of screening in the second period. Delays in diagnosis, however, may lead to more transmission and secondary cases. The principal purpose of active case finding

Findings at screening in cases of culture-confirmed pulmonary TB starting treatment within 90 days, 2007–2008

Smoking Cough Cough >3 weeks Productive cough Weight loss Night sweat Former treatment for TB Family history of TB Subjectively feeling ill Observed ill health

Sensitivity (n = 29)

Specificity (n = 23 373)

LR+ (95%CI)

LR− (95%CI)

0.34 0.38 0.14 0.21 0.34 0.17 0.24 0.07 0.31 0.17

0.67 0.89 0.97 0.91 0.85 0.93 0.99 0.97 0.90 0.99

1.03 (0.71–1.50) 3.40 (2.54–4.55) 4.81 (0.22–103.00) 2.33 (0.66–8.17) 2.35 (1.62–3.41) 2.54 (0.39–16.70) 17.91 (7.38–43.50) 2.25 (0.00–>1000) 3.21 (1.98–5.20) 21.32 (3.22–141)

0.98 (0.89–1.09) 0.70 (0.63–0.78) 0.89 (0.82–0.96) 0.87 (0.80–0.95) 0.77 (0.69–0.85) 0.89 (0.82–0.96) 0.77 (0.70–0.84) 0.96 (0.89–1.03) 0.76 (0.69–0.84) 0.83 (0.77–0.91)

TB = tuberculosis; LR+ = positive likelihood ratio; CI = confidence interval; LR− = negative likelihood ratio.

1392

The International Journal of Tuberculosis and Lung Disease

is thus a reduction in the number of infections of contacts and, consequently, secondary cases.13 How successfully this objective is met remains controversial. A model of transmission recently commissioned by the European Centre for Disease Prevention and Control suggests that, in a steady state with a well-functioning one-time screening procedure in steadily arriving new immigrants and over a time horizon of 10 years, the case load in the screened immigrant populations might be reduced by perhaps 10% compared to a ‘no screening’ strategy.14,15 This effect is mainly confined to the immigrant population, as there is little transmission to the native population.16–18 The methodology used in this study was based on practical considerations in a real-life situation. The effect of the new system cannot be determined accurately, as the two systems were not run in parallel. The geographic origins of asylum seekers changed over time. A concurrent cohort study was also not possible, as the subjects screened could not be individually followed up over time. However, both groups had the same opportunity to remain in the country for 90 days after claiming asylum and to be diagnosed with TB. Ninety days was chosen as the cut-off in the design phase as the effect of screening diminishes rapidly over time, any diagnosis related to screening should be made within a reasonable time frame, and an increasing proportion of subsequent cases are the result of newly emerging endogenous reactivation rather than persisting missed disease. The 13 cases not immediately detected at screening mentioned few symptoms at the time of screening. Communication problems, including differential conceptual representation of illness, and the belief that being ill might negatively affect the chances of being granted asylum, may play a role. Interestingly, most such cases originated from the Horn of Africa (data not shown). TB prevalence in the country of origin was an effect modifier: in the interview-based system, geographic origin was a component of the scoring system and more asylum seekers came from high-prevalence countries in 2007–2008. In practice, chest radiography remains the method of choice if effectiveness is considered alone. The new system, however, has advantages in terms of flexibility, cost and the burden in terms of medical procedures imposed on the screened population. The system is flexible as to the threshold chosen above which further investigations are required. It can thus be targeted to reach a defined proportion of screened subjects to be investigated. This has direct implications on the costs generated in the health care system: in the system using chest radiography, 12% of the screened population required further investigation, usually including a bacteriological examination, while in the new system only 4% required subsequent radiographic examination and even fewer required microbiological investigations. The proportion of abnormalities

found in this study (12%) is high compared to the 2.4% found in Netherlands data of 1998–2002.8 The vast majority were minor abnormalities compatible with residuals of previous TB. Their number had increased after the introduction of digital technology and centralised specialist reading. Nevertheless, this group of suspects comprised a quarter of the cases during this period and cannot therefore be ignored. Their detection is part of the 100% sensitivity found in the data at the expense of specificity. The costs of the screening system itself were also substantially lower (data not shown). Even taking the costs of additional secondary cases into account, screening for TB in immigrant populations always has net costs. In the very efficient Netherlands setting, costs per case prevented were estimated to amount to €30 000 to €50 000.19 The burden imposed on the screened population also varies according to the choice of screening strategy. In medical practice, bronchoscopy for more rapid diagnosis is often used to circumvent difficulties in communication to obtain an appropriate directly produced sputum specimen. Using interferon-gamma release assays would be an impractical screening method to reduce future case load, as all migrants with a positive test (estimated at 20% of the asylum seekers tested) would require further evaluation, while adherence to treatment for presumed latent infection with Mycobacterium tuberculosis is notoriously low.20,21 An interview-based system allows high coverage of the population targeted for screening, as no exclusion criteria, such as age or pregnancy for radiography, apply. It is also closer to the mainstream diagnosis of TB and it provides opportunities to address other medical needs. Despite the low sensitivity, the present screening system for TB in Switzerland can be improved only marginally. Adjusting score points for higher sensitivity also affects specificity. The system will therefore always miss cases that do not yet have specific enough symptoms or do not acknowledge them, or those who do not pass through the official system.22,23

CONCLUSION Methods of active case finding among asylum seekers differ as to their immediate short-term effects, but the effect of any screening method on the epidemiology of TB is likely to be very small and mostly limited to the population subjected to screening. The effect of a screening system based on origin, history and symptoms lies between the effect of a system relying on chest radiography and the effect of simply ensuring access to care without any screening. However, the total societal costs of TB in each of the three scenarios can also be ranked in this order, with very high marginal costs for each additional case prevented. Ensuring access to quality care is paramount. Screening

Radiographic vs. symptom and history-based screening of immigrants

of immigrants who can easily be reached is a small further step, with small gains for considerable effort. Acknowledgements The authors thank H Rieder of the International Union Against Tuberculosis and Lung Disease for his critical comments on the draft version of this article. They also acknowledge the daily effort of the nursing staff of ORS Service AG in assessing asylum seekers, entering data, and providing directly observed treatment to patients.

References 1 Coker R, Bell A, Pitman R, et al. Tuberculosis screening in migrants in selected European countries shows wide disparities. Eur Respir J 2006; 27: 801–807. 2 Rieder H L, Zellweger J P, Raviglione M C, Keizer S T, Migliori G B. Tuberculosis control in Europe and international migration. Eur Respir J 1994; 7: 1545–1553. 3 Verver S, Bwire R, Borgdorff M W. Screening for pulmonary tuberculosis among immigrants: estimated effect on severity of disease and duration of infectiousness. Int J Tuberc Lung Dis 2001; 5: 419–425. 4 Dasgupta K, Menzies D. Cost-effectiveness of tuberculosis control strategies among immigrants and refugees. Eur Respir J 2005; 25: 1107–1116. 5 World Health Organization. Global tuberculosis control: epidemiology, strategy, financing: WHO report 2009. WHO/HTM/ TB/2009.411. Geneva, Switzerland: WHO, 2009. 6 Office fédéral de la santé publique. La tuberculose en Suisse de 2001 à 2004. Bull OFSP 2006; 22: 428– 433. [French] 7 Office fédéral de la santé publique. Dépistage de la tuberculose chez les requérants d’asile de 2000 à 2003. Bull OFSP 2006; 1: 12–13. [French] 8 Erkens C, Slump E, Kamphorst M, et al. Coverage and yield of entry and follow-up screening for tuberculosis among new immigrants. Eur Respir J 2008; 32: 153–161. 9 Cohen R, Muzaffar S, Capellan J, Azar H, Chinikamwala M. The validity of classic symptoms and chest radiographic configuration in predicting pulmonary tuberculosis. Chest 1996; 109: 420–423. 10 Miller L G, Asch S M, Yu E I, Knowles L, Gelberg L, Davidson P. A population-based survey of tuberculosis symptoms: how atypical are atypical presentations? Clin Infect Dis 2000; 30: 293–299.

1393

11 Long R, Cowie R. Tuberculosis: 4. Pulmonary disease. CMAJ 1999; 160: 1344–1348. 12 Helbling P, Altpeter E, Raeber P A, Pfyffer G E, Zellweger J P. Surveillance of antituberculosis drug resistance in Switzerland 1995–1997: the central link. Eur Respir J 2000; 16: 200–202. 13 Verver S, van Soolingen D, Borgdorff M W. Effect of screening of immigrants on tuberculosis transmission. Int J Tuberc Lung Dis 2002; 6: 121–129. 14 Vos A M, Meima A, Verver S, et al. High incidence of pulmonary tuberculosis persists a decade after immigration, The Netherlands. Emerg Infect Dis 2004; 10: 736–739. 15 de Vlas S J, Meima A. Pulmonary tuberculosis incidence in migrants. Project ECDC.572, Part C. Rotterdam, The Netherlands: Erasmus Medical Centre, Department of Public Health, 2008. 16 Wolleswinkel-van den Bosch J H, Nagelkerke N J D, Broekmans J F, Borgdorff M W. The impact of immigration on the elimination of tuberculosis in The Netherlands: a model based approach. Int J Tuberc Lung Dis 2002; 6: 130–136. 17 Borgdorff M W, Nagelkerke N, Van Soolingen D, De Haas P E W, Veen J, Van Embden J D A. Analysis of tuberculosis transmission between nationalities in the Netherlands in the period 1993–1995 using DNA fingerprinting. Am J Epidemiol 1998; 147: 187–195. 18 Barniol J, Niemann S, Louis V R, et al. Transmission dynamics of pulmonary tuberculosis between autochthonous and immigrant sub-populations. BMC Infect Dis 2009; 9: 197. 19 Habbema J D, Meima A, Vos A M, de Vlas S J. Kosteneffectiviteit van huidige en alternatieve screeningsstrategieën van immigranten op tuberculose. Erasmus MC. Rotterdam, The Netherlands: Universitair Medisch Centrum Rotterdam, 2009. [Dutch] 20 Breuss E, Helbling P, Altpeter E, Zellweger J P. Screening and treatment for latent tuberculosis infection among asylum seekers entering Switzerland. Swiss Med Wkly 2002; 132: 197–200. 21 Hirsch-Moverman Y, Daftary A, Franks J, Colson P W. Adherence to treatment for latent tuberculosis infection: systematic review of studies in the US and Canada. Int J Tuberc Lung Dis 2008; 12: 1235–1254. 22 Monney M, Zellweger J P. Active and passive screening for tuberculosis in Vaud Canton, Switzerland. Swiss Med Wkly 2005; 135: 469– 474. 23 Mathez C, Bangala Y, Bady P, Zellweger J P. Active screening for pulmonary tuberculosis among immigrants by chest X-ray at the Swiss border. Swiss Med Wkly 2007; 137: 649–654.

RÉSUMÉ C O N T E X T E : Dépistage actif obligatoire de la tuberculose (TB) des demandeurs d’asile, Suisse, 2004–2005 et 2007–2008. O B J E C T I F : Comparaison entre le rendement d’un examen radiologique du thorax et une évaluation individuelle basée sur l’origine géographique, l’histoire personnelle et les symptômes. M É T H O D E : Comparaison rétrospective transversale de deux périodes de 2 ans. R É S U LTAT S : La prévalence des cas de TB détectés a été évaluée par la proportion de sujets débutant un traitement pour TB pulmonaire confirmée par culture dans les 90 jours après l’examen de dépistage. Elle était de 14,3 pour 10 000 demandeurs d’asile examinés (31/ 21 727) pour la radiographie et de 12,4 (29/23 402) pour l’évaluation individuelle. La sensibilité de la radiographie était de 100% contre 55% pour l’évaluation indivi-

duelle, mais sa spécificité était inférieure (89,9% contre 96,0%). La sensibilité plus élevée de la radiographie s’est traduite par des délais plus courts entre le dépistage et le début du traitement (médiane 6 contre 25 jours). Sa spécificité inférieure a eu pour conséquence qu’une proportion plus élevée de sujets a été soumise à des investigations additionnelles pour suspicion de TB (12% contre 4%). C O N C L U S I O N : Le système basé sur des interviews a initialement manqué plus de cas, mais le rendement à 90 jours était comparable dans les deux périodes. La différence principale est le retard subi jusqu’au début du traitement. Il augmente potentiellement la transmission et donc les cas secondaires. Le système radiographique était plus lourd pour le système de santé et pour les demandeurs d’asile, car un nombre plus grand de personnes suspectes de TB ont dû être examinées.

1394

The International Journal of Tuberculosis and Lung Disease

RESUMEN M A R C O D E R E F E R E N C I A : La detección sistemática obligatoria de la tuberculosis (TB) en las personas que solicitaron asilo político en Suiza, 2004–2005 y 2007–2008. O B J E T I V O : Comparar el rendimiento diagnóstico de la detección sistemática mediante radiografía de tórax y con la evaluación de la persona con base en la procedencia geográfica, los antecedentes y los síntomas. M É T O D O : Fue este un estudio transversal retrospectivo donde se compararon datos de dos períodos, de 2 años cada uno. R E S U LTA D O S : La prevalencia de casos de TB detectados se calculó según la proporción de personas evaluadas que comenzaron tratamiento contra la TB pulmonar confirmada por cultivo en un lapso de 90 días. Esta prevalencia fue 14,3 por 10 000 solicitantes de asilo que participaron en el tamizaje (31/21 727) con la radiografía de tórax y 12,4 (29/23 402) con la evaluación personal. La sensibilidad de la radiografía fue 100% en comparación

con 55% de la evaluación personal, pero su especificidad fue menor, 89,9% contra 96,0%. La sensibilidad más alta de la radiografía se tradujo en lapsos más cortos entre el tamizaje y el comienzo del tratamiento (mediana de 6 contra 25 días) y su especificidad más baja aumentó la proporción de personas tamizadas que necesitaron exámenes complementarios por presunción de tuberculosis (12% contra 4%). C O N C L U S I Ó N : Con la estrategia inicial basada en la entrevista se pasaron por alto más casos, pero el rendimiento final con el intervalo de 90 días fue comparable entre ambos períodos de estudio. La principal diferencia entre las estrategias es el retraso en el comienzo del tratamiento antituberculoso, que podría aumentar la tasa de transmisión y de casos secundarios. La estrategia basada en la radiografía es más gravosa para el sistema de salud y las personas estudiadas, pues un mayor número de ellas precisa exámenes complementarios.