This was a prospective, multi-centre trial conducted over the course of three years in two chronic haemodialysis outpatient units affiliated with the University Hospital of Patras and a third unit in the Nephrology Department of the General State Hospital in Patras. The study protocol was approved by the ethical committee of the medical school of the University of Patras. Patients files were searched for factors, that might have influenced the risk of developing TB, like: age, primary renal disease, TB history (disease, contact, vaccination), time on dialysis (time elapsed since undergoing haemodialysis), medication (vitamin D supplements and immunosuppressive drugs like corticosteroids, cyclophosphamide, calcineurin inhibitors, azathioprine and mycophenolate mofetil), and laboratory data like serum albumin and zinc levels. Dialysis efficiency was measured using equilibrated dialysis dose (Kt/V) and urea reduction ratio (URR). Body Mass Index (BMI) and serum albumin, were used as markers of the patients' nutritional status.
Entering the study, participants were evaluated, and active TB was ruled out by history, physical examination, chest radiography, and when indicated, bacteriological studies. Following patients were tested with tuberculin. TB incidence and the subsequent relative risk (RR) for TB was then determined over a 36 month follow-up.
All efforts were made for bacteriological and/or histological confirmation. In patients with clinical and/or radiographic findings suggestive of TB, respiratory samples consistent of three sputum specimens (or three gastric washings in the absence of of productive cough) were stained with Ziehl-Neelsen stain and cultured in Lowenstein-Jensen medium for M. tuberculosis. A Bronchoalveolaren lavage was also performed for each patient with negative cultures, but highly suggestive features of TB. HDPs with Lymphadenitis had superficial or deep lymph node biopsies. In case of no lymph node extrapulmonary involvement, TB was identified on the basis of clinical findings and positive biopsies and cultures from one or more clinical specimens.
To identify the RR for TB in haemodialysis patients accurately, on a population basis, the Prefecture's Health Service registries, where all TB cases diagnosed are mandatory reported, were also reviewed. From these, the rate of TB, for the age-matched control group was estimated. Participants provided informed consent.
Of the 272 end stage renal disease patients who were included in the study, 193 were male and 79 female. Their median age was 52.7 years, ranging from 25 to 77 years. They were treated with dialysis for a mean of 3.9 years (range 1 week to 18.4 years). Primary renal disease was classified in eight categories: glomerulonephritis (48), nephroangiosclerosis (46), diabetic nephropathy (75), polycystic kidneys (22), interstitial nephritis-reflux nephropathy or nephritis due to analgesic abuse (42), infection (15), renal carcinoma (2) and unknown origin (22). Five patients had a history of active pulmonary TB, but all had been treated adequately. Close, non-hospital contacts of persons with infectious TB were reported in seven patients. As Greece is among the countries where BCG vaccination is still in use, most patients have been vaccinated at the age of six years. However, only 173 patients remembered or had scars of vaccination with BCG. Patients on corticosteroids or other immunosuppressive therapy were not included in this study because of the rather small number among the patients dialysed in our units. We did not perform a routine assessment for HIV infection, as in our area prevalence of HIV associated TB is insignificant.
Tuberculin Skin Testing (TST)
TST was carried out by the intradermal (Mantoux) method. It was administered by injecting 0.1 ml of commercially available tuberculin (PPD-Merieux, 5 U/0.1 ml dose), intradermally into the anterior surface of the forearm, and was read 48-72 hours later. A positive TST was a skin induration ≥10 mm (5-9 mm weak-positive and 0-4 mm negative). A second booster injection with 10 U/0.1 ml dose, was given within a 10-day interval to those HDPs responding with a <10 mm induration to the first test. The pen method was used for the measurement of skin tuberculin reaction as previously described [7, 8].
Entering the study, all patients had a chest X-ray, evaluated by two independent physicians, (radiologist and pulmonologist). Radiographic findings that could be attributed to previous healed TB were described as: 1) Dense pulmonary nodules, with (group-A, n = 51) or without visible calcification (group-B, n = 36), in the hilar area or upper lobes, 2) smaller nodules with or without fibrotic scars in the upper lobes, accompanied with upper-lobe volume loss (group-C, n = 32) and 3) bronchiectasis of the upper lobe and/or pleural scarring (group-D, n = 37). Nodules and fibrotic lesions of previous, healed TB were prescribed as hard-well-demarcated with sharp margins.
TB prevalence was defined as the number of TB cases per a hypothetical population at risk of 100. Analysis of variance (ANOVA) was used to assess the statistical significance of differences. For multivariate analysis, only factors that were significantly different (P < 0.05) in univariate comparison were retained. The independent effects of the significant factors, was determined using a multivariate analytical method the Cox's proportional hazard model.  Trends in prevalence and risk for TB through the different studied factors were assessed by Spearmans's product moment correlation coefficient (r), a dimensionless index that ranges from -1.0 to 1.0, reflecting the extent of the relationship between data sets.