Contact tracing for tuberculosis and treatment for latent infection in a low incidence country

a Division of Pulmonary Diseases; CHUV, Lausanne, Vaud, Switzerland b Department of Rehabilitation and Geriatrics; Geneva University Hospital, Geneva, Switzerland c Division of Clinical Epidemiology, HUG, Geneva University Hospital, Geneva, Switzerland d Division of Pulmonary Diseases; Geneva University Hospital, Geneva, Switzerland e Division of Pulmonary Diseases; Geneva University Hospital, Geneva, Switzerland


Introduction
No financial support declared.
The outpatient clinic of the Division of Pulmonary Diseases of Geneva University Hospital supervises all contact tracing procedures for TB in the Geneva area.All patients either treated for TB, or evaluated in contact tracing procedures, are entered in a database which stores information on gender, age, origin, exposure to index case, microbiological details of index case, BCG, Tu berculin skin test (TST; 2U of RT23 Tu berculin, Statens Serum Institute, Copenhagen, DK), co-morbidities, diagnosis of active, latent or history of TB and treatment prescribed.
This retrospective study includes all subjects evaluated over 10 years for contact with an index TB case with either smear positive and culture positive pulmonary tuberculosis (S+/C+TB), smear negative and culture positive pulmonary TB (S-/C+TB) or smear negative and culture negative pulmonary TB (S-/C-TB).Medical records of all contacts followed by our centre for whom a treatment for LT BI had been recommended were reviewed and acceptance to treatment, adherence (assessed by visit attendance and monthly urinary tests for isoniazid), tolerance (results of monthly ASAT: Aspartate amino-transferase; and ALAT: Alanine amino-transferase, reported side-effects), interruption of treatment and their causes were analyzed.
Tw o subgroups of subjects were not followed by our centre after initial screening and are not included in our analysis of treatment for LT BI: subjects known to be HIV-infected (referred to our HIV clinic), and children (referred to the Children's hospital or to their paediatrician).

Algorithms used for contact tracing
For subjects exposed to a S+/C+ TB, population screened included household contacts, close friends or relatives, home-care professionals and work or school contacts.For subjects exposed to a S-/C+ or S-/C-TB, household contacts only were screened (fig.1).TST was considered positive if induration was >10 mm, in agreement with national guidelines [1].For subjects screened, HIV testing was not mandatory and thus not recorded.For workplace contacts aged >35, unless identified as having a high exposure to the index case, screening consisted of a chest X-ray (CXR) -12 weeks after end of exposure.The default treatment prescribed for LT BI was isoniazid (INH) for six months, in accordance with the recommendations prevailing at that time in Switzerland and in the UK [1,19].
In this study, probable LT BI was defined as having either a TST induration >10 mm or a chest X-ray with images suggestive of prior TB.LT BI is reported as "probable" because of the lack of specificity of the TST and the chest X-ray.

Statistical analysis
Variables are reported as mean x (SD y).Exact Poisson confidence intervals were calculated for incidence of rare events.Comparison between groups was performed using unpaired t tests or chi-square tests when appropriate.Multiple logistic regression was performed to analyse the relationship between significant variables and occurrence of LT BI, probability of pursuing treatment for LT BI, or suffering from side effects.A p level of <0.05 was used for determining statistical significance.Statistical analyses were performed with Stata Statistical Software for PC computers (Version 9.  Between January 1993 and December 2002, 352 subjects were screened for contact with a patient with TB, 11 (4%) of whom were evaluated at our centre (table 1).The remaining 164 were evaluated either by private practitioners or by the Paediatric Department if aged under 16 (n = 356).Only one subject was known to be HIVinfected.

Materials and methods
Microbiological status of index case was S+/C+ TB for 4% of subjects screened, S-/C+ TB for 10%, S-/C-pulmonary TB for 2.3% and unknown for 3.%.Average number of contacts screened per case was 4.3.

Ta ble 1
Results of tuberculin skin tests (TST) according to exposure to index case and bacteriological status of index case.S+/C+ : smear positive and culture positive; S-/C+: smear negative and culture positive; S-/C-: smear negative and culture negative

Swiss
Foreign

Ta ble 2
Results of multiple logistic regression models for estimating probability of latent tuberculosis infection (lTBi) in contacts, probability of completing treatment for lTBi, and probability of treatment interruption due to side-effects of treatment for lTBi.OR: odds ratio; 95%Ci: 95% confidence interval.S+: smear positive; S-: smear negative; C+: culture positive; C-: culture negative.Age is expressed per 10 years.Adjusting for clustering was not possible due to lack of information regarding relationship between index cases and contacts in the original database.This may slightly underestimate confidence intervals.
tion, 3 (10%) had none.TST with chest X-ray when indicated were available in 231 subjects; 51 subjects either had no TST (chest X-ray only) or TST result was unavailable (no information from private practitioner or patient did not show up for reading of TST).TST induration was <5 mm in 56% of subjects (n = 1530); 5-10 mm in .3%(n = 22), and >10 mm in 36% (n = 94).The proportion of skin tests >10 mm was higher in foreign-born subjects than in Swiss nationals (40% versus 30%: table 1).
Combining chest X-ray and TST, a diagnosis of probable LT BI was made in 996 subjects (2% of subjects screened).

Predictors of LT BI
In a multiple logistic regression, male gender and age were significantly associated with a higher rate of LT BI (table 2).Conversely, Swiss nationality was associated with a lower rate of LT BI.Risk of LT BI was significantly related to exposure to the index case and to microbiological status of the index case (table 2).

Treatment for LT BI
Among the 996 subjects with LT BI, 05 were followed at our centre (fig.2).Of these, 1.4% (n = 10) had a history of treated TB, 0.6% (n = 4) had received previous treatment for LT BI, 0.1% (n = 1) had a history of probably untreated TB and .4% (n = 52) had a previously documented positive TST.For 6 (9.5%) subjects, treatment for LT BI was withheld for medical reasons (eg excessive alcohol consumption).Thus, treatment for LT BI was recommended in 51 subjects.Of these, 10 (19%) refused treatment or were lost to follow-up before treatment was initiated.Among the 463 (1%) who started treatment (aged 34.2 (11.5) years; range: 14-2, 21% Swiss), INH was prescribed to 9% (n = 449); 1.% (n = ) received rifampicin either alone or in combination with INH; 1.3% (n = 6) received other medications because of resistance of the index case to INH and rifampicin (MDR-TB); .6%(n = 3) were switched to rifampicin either because of intolerance (n = 1) or resistance to INH in the index case (n = 21).At 6 months, overall completion rate was 6% (n = 312); treatment was interrupted in 22% (n = 100) after a mean of 4 (49) days (range 3-206), either as a result of non-adherence (n = 6, 15%) or side effects (n = 32, %); 11% were lost to follow-up (n = 51).Neither age, gender, foreign origin nor intensity of exposure to index case was predictive of treatment completion by multiple logistic regression (table 2).
INH-related adverse effects and cases in which treatment was changed or interrupted because of adverse effects are shown in table 3.In multiple logistic regression, treatment interruption because of side effects was more likely in older patients.For example, the proportion of patients with ASAT or ALAT >5X upper normal limit was 2.5% for subjects aged 14-34 years, vs 4.9% for the 35-49 years age group and 13% for subjects aged >50.Among all subjects, 101 patients (22%) experienced one or more side effects, resulting in change or interruption of treatment in 45 subjects (10%).Hepatotoxicity accounted for change or interruption of treatment in 26 cases (5.6%).This study reports the yield of contact tracing over a 10-year period in a low incidence area for tuberculosis.Prevalence of LT BI in contacts was significantly related to exposure to the index case and to the contagiousness of the index case (table 2).The yield of contact screening for active TB cases was very low (n = cases, 0.2% of subjects screened; 95 % CI: 0.1-0.4%),when compared to previous reports in low-incidence countries [5, 6, 9-12, 20, 21] and far below that reported in developing countries [2,3].Among contacts with LT BI for whom treatment was recommended and who were followed at our centre, 1% started treatment (usually isoniazid), of which 6% completed treatment.Tr eatment was interrupted because of side effects in % of subjects and such interruptions were more prevalent in older patients.
Rate of TB detected through contact screening in Europe, USA or Canada ranges from 0.% to 4.9%, and that of LT BI from .6% to 4% [5- 12,14,20,22].In the present study, the rate of TST-based diagnosis of LT BI (36%) was similar to previous reports.Changing the threshold value for TST positivity to 5 mm, in agreement with recent AT S, CDC and IDSA (Infectious Disease Society of America) guidelines, would increase the infection rate by .3%(44%), ie in the upper range of previously reported data [23].Conversely, the number of active cases detected in this study was surprisingly low: the active TB cases represented less than 1% of the 3 cases of active TB reported in Geneva during the study period.The reason for such a low rate of TB is unclear.A plausible hypothesis would be underdetection of infected contacts [24].This seems unlikely however, because, over the study period, newly detected cases of TB were not related to previous cases, albeit for the above mentioned active cases.Indeed, DNA fingerprinting of mycobacteria in Switzerland shows a very low number of clusters of TB cases [25,26].The easy access to care in Geneva even for homeless subjects, illegal immigrants or other high risk groups may contribute to earlier detection of active cases, and thus decrease exposure of community members.
The major goal of contact-tracing is the detection and treatment of newly infected subjects with LT BI.Screening 352 contacts led to complete treatment for LT BI in at least 312 subjects (number of patients completing treatment and followed by private practitioners is unknown), thus theoretically preventing 11 cases of TB (based on the assumption of a 5% life-time risk of reactivation and a 0% protection by a 6 month regimen of INH) [2].Among subjects followed by our centre (fig.2), 32 subjects would be at lifetime risk of developing active TB (5% of the 63 patients for whom treatment for LT BI was indicated, including those for whom it was withheld for medical reasons).Therefore, one can assume that 34% (11/32) of future TB cases were prevented by contact tracing.Similar estimations were derived from a large multicentre evaluation of contact investigations in California [12].
Prevalence of positive TST increased significantly with age, exposure to index case and contagiousness of index case (S+/C+ vs S-/C+ or S-/C-).This contrasts with recent reports showing very low associations between exposure to index case and TST status when compared to g -IFN blood assays [2,29] (table 2).
Tr eatment for LT BI was completed in 55% of eligible subjects and 6% of those who started.A total of 40% of eligible subjects refused, stopped their treatment or were lost to follow-up (fig.2).Reported adherence to treatment for LT BI is highly variable.In a large INH preventive therapy programme in Atlanta, GA, 6% of subjects for whom preventive therapy was indicated started INH, but only 20% completed therapy.Foreign birth and age over 65 years were associated with higher rates of completion in multivariate analysis [].In recent US studies of contact investigations, 66-9% of subjects eligible for INH preventive therapy started their treatment, with completion rates of 44-64% [5,6,11,12,14,30].Tw o smaller Swiss studies reported 6-6% completion rates for patients treated for LT BI, most of whom were immigrants or asylum seekers [31,32].In our study, adherence was assessed by recording attendance to monthly visits and results of a qualitative urinary assay for INH.Age, gender, origin and importance of exposure to index case were not predictive of treatment completion (table 2).

Discussion
Increase in ASAT or ALAT above 5 times upper limit of normal (ULN), or above 3 times ULN with symptoms suggesting hepatotoxicity led to interruption of treatment, according to ATS/CDC guidelines [33].A moderate elevation of ASAT or ALAT is expected in 10-20% of patients treated with INH [34].As in previous studies, hepatotoxicity increased with age.Interruption of treatment because of liver toxicity occurred in 4.% of subjects with INH alone, which is slightly higher than previously reported (0.3-4%) [34].Rate of INH-related clinical hepatitis was similar to previous reports (0.6%) [33].Monthly liver function testing may lead to a higher detection of INH-induced hepatotoxicity, since it is frequently asymptomatic.Although not recommended by ATS/CDC guidelines nor by recently revised Swiss guidelines [16] (www.lung.ch),as stated by Fountain et al, the rationale for routine testing of ASAT and ALAT is that it probably leads to an earlier detection of liver toxicity.S ome of these patients would have probably become symptomatic, had INH been pursued [34].
In summary, this study performed in a low TB incidence area shows that contact tracing for TB in our area has a very low yield in terms of detection of secondary TB cases.Secondly it is associated with a moderate rate of acceptance for LT BI therapy and an average rate of treatment completion.Improving effectiveness of contact tracing thus relies on detecting LT BI with better specificity (ie through algorithms including the g -IFN assays as stated in the revised Swiss guidelines) and improving both initial acceptance to preventive therapy and completion rates.

Figure 2 Flow
Figure 2Flow chart of subjects screened for latent tuberculosis infection (lTBi) or TB and outcome of follow-up of patients treated for lTBi.*: Subjects with lTBi had TST >10 mm and/or chest X-ray suggestive of prior TB.§ : percentages refer to patients who started treatment.