Background Non-tuberculous mycobacteria (NTM) are widely distributed in the environment and are difficult to diagnose and treat. Previous non-UK studies have reported increasing incidence and geographical variation in NTM isolates.1 We characterised the frequency and clinical relevance of positive NTM cultures in a large UK hospital and the effect of introducing a new liquid culture.

Methods We examined the notes of all patients from whom NTM had been isolated between July 1999 and September 2009. Diagnostic criteria for NTM disease published by the American Thoracic Society (ATS) were used to determine clinical relevance.1

Results NTM was isolated from 100 patients. 91 (91%) were respiratory tract samples. 14 (15%) of these met ATS criteria for NTM pulmonary disease. MAC (38%), M. xenopi (19%) and M. malmonese (11%) were most common. Of these, 14%; 6% and 30%, respectively, met the ATS criteria. The most clinically relevant species were M. bovis (1/1; 100%), M. simlae (1/2; 50%), M. kansasii (2/6; 33%) and M. malmonese (3/10; 30%). Cough (p=0.02) and night sweats (p=0.027) were associated with clinical relevance. Being asymptomatic was linked to not meeting ATS diagnostic criteria (p=0.029). Pre-existing pulmonary disease (p=0.012), ABPA (0.019) and dyspnoea (p=0.013) predicted having a second positive sputum.

The annual NTM isolates increased over 10 years. The liquid culture system was introduced in September 2006. Using a χ² comparison test there was no statistically significant difference in clinically relevant isolates pre and post September 2006 (p=0.633).

Conclusion This is the first study of NTM isolates in the north of England. Our study shows prevalence of clinically relevant disease, with isolates of M. kansasii and M. malmonese most likely meeting ATS criteria. MAC was most prevalent (38%). Since the introduction of a new liquid culture system the number of isolates increased, but the clinical relevance did not. The spread of species isolated differed from previous studies1 which highlights the geographical variation and the importance of regional data.