Cooking with Biomass Stoves and Tuberculosis: a Case Control Study Exposure to Biomass Smoke Was Significantly Higher in Cases than in Controls. Crude Odds Ratios for Tuberculosis and Biomass Smoke Exposure Were 5.2 (95%ci 3.1–8.9) for Current Exposure, 3.4 (95%ci 2.4– 5.0) for past or Present Expos

To search for an association between tuberculosis and use of biomass stoves found recently in a cross sectional study. D E S I G N : In a case-control study based in a chest referral hospital, the cases were 288 patients with active smear-positive or culture-positive tuberculosis, and the controls were 545 patients with ear nose and throat ailments with no evidence of chest disease studied at the same time as the cases. Exposure to present or previous biomass smoke by history of cooking with traditional wood stoves was assessed by positive or negative response. AN ASSOCIATION between self-reported tuberculosis and exposure to biomass smoke, as declared in interview, was recently described in India. 1 The study included 200 000 subjects, and those living in households where biomass was burnt reported tuberculosis more frequently than persons using cleaner fuels (odds ratio [OR] 3.6, 95% confidence interval [CI] 2.8–4.5). Adjustment for several socioeconomic factors reduced the odds ratio to 2.58; however the association remained significant for men and women and for rural or urban residence. The authors estimate that biomass smoke exposure can explain about 59% of rural and 23% of urban cases of tuberculosis in India, 1 given the high prevalence of exposure. We designed a hospital-based case-control study to seek an association between biomass smoke exposure and confirmed cases of tuberculosis. METHODS The study was conducted at the National Institute of Respiratory Diseases in Mexico City (INER), a refer-ral hospital and research institution funded by the health secretariat to provide service for the economically deprived and uninsured population of Mexico. Most patients reside in the Mexico City metropolitan area or in nearby states. Patients may be referred from other hospitals or doctors, or may be self-referred. All patients are screened by a brief interview with a pulmonary specialist who decides if tertiary care is required. In addition, patients are interviewed by a social worker to their assess socioeconomic level. Formal socioeconomic classification used at the INER includes seven categories (0, 1, 2, 3, 4, 5 and 6), and is a weighted composite of income per dependent (heaviest weight), household characteristics (building material, public services and location in rural, suburban or urban area), being the owner or tenant of the house, and the presence of other sick relatives. Those with the lowest socioeconomic level (level 0) receive free care, and as the level increases patients pay higher fees, although all …


S u m m ary
In a case-control study, fasting total homocysteinemia was deter mined in 208 consecutive outpatients who underwent phlebography be cause of the first episode of clinically suspected deep-vein thrombosis (DVT) of lower limbs. Contrast venography confirmed the clinical sus picion in 60 patients (28.8%). Hyperhomocysteinemia was detected in 15 of the 60 patients with DVT (25.0%), and in 17 of the 148 subjects without thrombosis (11.5%; p = 0.025). The OR for having an acute DVT in patients with hyperhomocysteinemia was 2.6 (95% Cl: 1. 1-5.9). It is concluded that high plasma homocysteine levels are sig nificantly associated with DVT in symptomatic patients. Further stud ies are needed to clarify the clinical implications of this association.
In tro d u c tio n In contrast to the extensive documentation available on the risk for arterial disease in patients affected by hyperhomocysteinemia (1)(2)(3)(4)(5), the association between high levels of homocysteine and occurrence of ve nous thromboembolism has remained unclear until recently (6)(7)(8). In two recent studies, the prevalence of hyperhomocysteinemia in young patients with venous thromboembolism was unexpectedly high, sug gesting that this disorder might have pathogenetic significance in pre mature venous thrombosis (9,10). In a case-control study, hyperhomo cysteinemia was found a risk factor for recurrent venous thromboembo lism (11). Finally, a recent case-control study has shown a strong as sociation of high plasma homocysteine levels with the occurrence of deep-vein thrombosis (DVT) in the general population (12). In this study, the control group was made of healthy volunteers.
We performed a case-control study in a wide series of consecutive symptomatic outpatients presenting with the first episode of clinically suspected DVT. All of them underwent contrast venography to confirm or refute the clinical suspicion of DVT. Cases and controls were inves tigated for the levels of homocysteine by applying sensitive methods. P atie n ts a n d M e th o d s in the study, provided that they were not pregnant, were free from active can cer, were not affected by liver or renal insufficiency, did not present contraindi cations to contrast phlebography, and gave their informed consent. Patients subsequently identified as carriers of antithrombin, protein C or S defect, APC resistance, and lupus-like anticoagulants were also excluded from the final evaluation. The study was approved by the local Institutional Review Board.
On admission, each patient was examined by a physician of the centre. For all patients a standard questionnaire was completed, detailing information on the eventual association of their leg complaints with known acquired risk fac tors for venous thrombosis, i.e. prolonged immobilization from any cause, re cent leg trauma, surgery or delivery, and oestrogen pill. At the same time, blood collection was performed as indicated below to determine the plasma concen tration of total homocysteine and the plasma concentration of cobalamine and folate. Methionine loading was not performed. Thereafter, to assess the pres ence or absence of DVT, all patients underwent contrast venography.
Laboratory tests and venography were performed and interpreted by inde pendent operators.

Blood Collection and Routine Coagulation Tests
Blood samples were collected in all patients after an overnight fast. Venous blood was collected by venipuncture with 21-gauge butterfly infusion sets into plastic syringe containing 3.8% (wt/vol) Na-citrate in a ratio 0.1:0.9 (vol/vol) anticoagulant to blood. Platelet poor plasma (PPP) was prepared by centrifuga tion at 2000 X g for 10 min and re-centrifugation of the harvested plasma under the same conditions. Aliquots of PPP were stored at -70° C until use.
Routine coagulation tests including prothrombin time, partial thromboplas tin time, antithrombin, protein C, protein S, plasminogen, fibrinogen, resistance to activated protein C, and lupus-like anticoagulants were performed as report ed elsewhere (13-15).
Cobalamine and folate plasma concentrations were determined with stan dard radio-immuno assays (Becton Dickinson, Orangeburg, NY). Criteria for the diagnosis of hyperhomocysteinemia were plasma levels above the 90th percentile, as assessed in 100 apparently healthy subjects, 50 women and 50 men whose age ranged from 18 to 65 years, belonging to the hospital staff (i.e., above 18.5 pM).

Contrast Venography
Contrast venography was performed using a technique previously described (17). Venograms were considered diagnostic for DVT if a persistent intralumi nal filling defect in at least two projections or non-filling of a vein segment af ter repeated injections of contrast material was observed. The venograms were interpreted by a panel of three independent and experienced observers. Patients with venographically confirmed DVT were regarded as cases, whereas patients with normal venograms were allocated to the control group.

Analysis
Data were organized into 2 X 2 contingency tables according to the DVT status and the presence or absence of hyperhomocysteinemia. Differences in the prevalence of hyperhomocysteinemia between cases and controls were test ed for significance by means of the chi-square test. Two-tailed p-values of 0.05 or less were considered statistically significant. Moreover, odds ratios (OR) and their 95% confidence intervals (Cl) were used to describe the association between DVT and hyperhomocysteinemia. An OR was considered to be statis tically significant when the lower limit of the 95% Cl was >1.0.

Results
Two hundred and twenty-seven potentially eligible patients were evaluated during the study period. Three patients were excluded from the primary analysis because of an inadequate phlebography. Sixteen Table 1 Clinical characteristics of patients with and without DVT  Fig. 1 Plasma total homocysteine concentrations in cases and controls.

Dashed line indicates upper limit of normal (18.5 pM)
was apparently spontaneous (idiopathic DVT: 46.7%). Among the 148 subjects without thrombosis, an alternative diagnosis was identified in 95 (64%): muscle strain or tear in 36, superficial thrombophlebitis in 24, venous reflux in 12, cellulitis in 8, lymphangitis or lymphatic obstruction in 5, rupture of Baker's cyst in 4, arthritis in 4, and swelling in a paralyzed leg in the remaining 2. Hyperhomocysteinemia was detected in 15 among the 60 patients with DVT (25.0%), and in 17 out of the 148 subjects with normal veno gram (11.5%) (p = 0.025). The OR for having an acute DVT was 2.6 (95% Cl: 1.1-5.9) for patients with hyperhomocysteinemia as com pared to those without hyperhomocysteinemia. The prevalence of hy perhomocysteinemia in patients with idiopathic DVT (8/28,28.6%) did not differ from that observed in patients with secondary DVT (7/32, 21.9%; p >0.2).
Graph depicted in Fig. 1 shows plasma homocysteine levels in pa tients with and without DVT. Table 2 shows the thrombosis risk for strata of homocysteine concentrations.
Cobalamine and/or folate deficiency was detected in 3 patients with hyperhomocysteinemia (in 1 case, and in 2 controls, respectively). The differences and analysis for trend between the control and thrombosis group remained unaltered when the analysis was restricted to patients with normal vitamin status.

D iscussion
th e results of our study demonstrate that a statistically significant association exists between the presence of high homocysteine levels and venous thrombosis (OR = 2.6; 95% Cl, 1.1 to 5.9). Hyperhomocys teinemia is equally distributed in patients with spontaneous DVT and in those with secondary DVT. Our results are in full agreement with those recently reported by den Heijer and associates in the general population (12). The higher prevalence of this abnormality among our patients (25.0%) than in those described by den Heijer (10.0%) is most likely at tributable to the definition of hyperhomocysteinemia (plasma levels above the 90th percentile, as assessed in healthy individuals, in our study, and above the 95th percentile in the Dutch study).
Since we determined only fasting homocysteine concentration, we could not establish the true prevalence of this abnormality in our pa tients (1). In their recent case-control study, den Heijer and associates were able to identify with the methionine-loading test a substantial number of carriers of this abnormality who would have gone undetect ed when relying only on the fasting determination (11). However, the odds ratio of having recurrent thromboembolism in carriers of post methionine hyperhomocysteinemia did not differ from that identified in carriers of fasting hyperhomocysteinemia. Therefore, our finding that mild hyperhomocysteinemia is associated with venous thrombosis most likely remains unaltered.
The underlying mechanism by which hyperhomocysteinemia may contribute to thrombosis is unknown. In tissue culture studies, high lev els of homocysteine contribute to endothelial damage, but the concen tration required for this effect is much higher than that usually observed in vivo (18)(19)(20). Recent observations suggest that homocysteine downregulates the endothelial thrombomodulin-protein C pathway and en hances factor V dependent activation of coagulation on the surface of endothelial cells (21,22). As epidemiologic evidence accumulates sup porting a role for homocysteine in circulatory diseases, there is a great er need for studies of the mechanism of action.
Hyperhomocysteinemia may be the consequence of a hereditary de fect in the enzymes involved in methionine metabolism or may result from vitamin deficiency (1). Since we did not perform enzymatic deter minations, and observed very few cases of vitamin deficiency (in 1 pa tient and 2 controls), our study does not help to clarify whether differ ent forms of hyperhomocysteinemia might be associated with a differ ent thrombotic risk.
The relatively high prevalence of hyperhomocysteinemia in the con trol group (11.5%) raises the question of the cut-off used to discrimi nate between patients with and without this abnormality. Up to now there is no consensus about reference values for plasma homocysteine concentrations. According to general guidelines, for the diagnosis of hyperhomocysteinemia we used plasma levels above the 90th percen tile, as assessed in 100 healthy subjects (i.e., 18.5 |iM). As shown in Fig. 1 and in Table 2, analyzing our data at different cut-off points (from 18.0 up to 20 |iM) would not have resulted in different odds ratios. Further studies aimed at identifying a standardized cut-off of plasma levels of homocysteine able to discriminate between people with and without this abnormality are urgently required.
We believe that the results of our study are valid and reflect the as sociation of hyperhomocysteinemia with acute venous thrombotic dis orders. Selection bias was prevented in our study by enrolling all con secutive patients referred to our Institution for clinically suspected DVT. Contrast venography, the golden standard for the diagnosis of DVT, was the only test used to select cases and controls. The decision to use as controls, subjects with clinically suspected DVT but with nor mal venogram was made because they are fully comparable with cases, being different only in the fact that they are free from thrombotic disor ders. The determination of homocysteinemia was performed according to sensitive methods, and the discriminant cut-off level was established according to recommended guidelines. Observation bias was prevented by having laboratory tests and venography performed and interpreted by independent operators. Potentially confounding factors were sup pressed by excluding from the analysis carriers of known thrombophil ic factors (antithrombin, protein C and S, APC resistance, and lupus like anticoagulants), as well as patients with active cancer, and those with recurrent thromboembolism.
We conclude that high homocysteine levels are significantly asso ciated with acute venous thrombosis in symptomatic patients. Further studies are required to determine if hyperhomocysteinemia is causally related to venous thrombosis, to establish the risk for recurrent throm boembolism in patients with hyperhomocysteinemia suffering an epi sode of venous thrombosis after the interruption of conventional anti coagulation, and to evaluate the role of homocysteine-lowering vitamin therapy -such as cobalamine, piridoxine, or folates administration -in the prevention of (recurrent) thrombotic events in carriers of this abnor mality (23)(24)(25).