Introduction
A recent increase in tuberculosis (TB) in the UK has been observed in certain areas and patient groups, such as ethnic groups in London1. The emergence of resistance to antituberculous drugs and patient non-adherence rates to this potentially life-threatening disease suggests that conventional medicine, defined by the biomedical model, has failed to resolve the problem of TB2. Misconceptions about TB are partly to blame for disease stigma and may thwart eradication3. Beliefs about illness and medication have been useful in predicting and preventing non-adherence to medication and poor treatment outcomes for other diseases4,5. A similar approach could be applied in TB management. However, this relies on the valid and reliable measurement of appropriate constructs. This study reports the reliabilities of scales used to measure perceptions of TB patients.

Method
Adult TB patients receiving treatment were recruited from three out-patient clinics in Northeast London. A questionnaire was completed including 1) six scales from the Revised Illness Perceptions Questionnaire (IPQ-R6), 2) the two scales from the Perceived Utility of Medicines scale (PUM7), and 3) various socio-cultural and medicines-behaviour indicators. The questionnaire was completed before or after patients attended clinics. Data were entered into SPSS® database. Factor analysis and internal reliability testing were performed.

Results
The questionnaire was completed by 118 patients (65.6% response). More men (58.3%) than women were recruited and the sample was ethnically diverse (Asian 47.2%, Black 29.6%, White 20.4%, and Mixed 2.8%) suggesting good representation.
Initial reliability testing of the six pre-defined scales in the IPQ-R suggested item discrepancies. Factor analysis was conducted to identify factor structure. After excluding a number of items due to poor factor loading, an eight factor solution was found to explain 67.8% variance. After testing for reliability, a two-item factor was excluded due to poor internal consistency. A number of items were returned to their original scales due to acceptable correlation and internal consistency in original scales. Characteristics of the final seven scales are shown in Table 1. This included a new 3-item 'Pessimism' scale. Factor analysis was also used to characterise one item to the two scales of the PUM. Internal reliabilities were acceptable (Table 1).

Conclusion
Some scales from the IPQ-R were adjusted according to the variance of the sample collected. Both IPQ-R and PUM scales demonstrated acceptable reliabilities for this diverse patient sample. These scales can now be used in further analysis investigating relationships between perceptions and 1) demographic/clinical variables, and 2) adherence