OBJECTIVE: Quality control in the clinical pulmonary function laboratory has been well developed for spirometry and diffusing capacity but not for the measurement of TLC. The purpose of the present study was to test two approaches to this problem. First, we compared TLC by body plethysmography (TLCb) with a value predicted from TLC measured by multibreath helium dilution (TLCm). Concordance between the measured and predicted values would imply the validity of the measurements. Second, we measured the test-retest variability of TLCb, TLCm and TLC measured by single breath helium dilution (TLCs) to assess the consistency of the measurements. METHODOLOGY: We performed a prospective study of 815 veterans. RESULTS: The prediction of TLCb from TLCm improved (r2 increased from 0.44 to 0.64) when FEV1/FVC and the difference between TLCm and TLCs were added to the model. The coefficient of variation for test-retest of TLCs, TLCm and TLCb were 8.9, 7.1 and 5.4%, respectively. Of all tests, 5.9% were inconsistent based on pathophysiology or measurement error and attributed mostly to TLCm. CONCLUSIONS: Prediction of TLCb from TLCm was not sufficiently accurate as to be useful for quality control. Comparison of TLCs, TLCm and TLCb may be useful for determining the internal data validity.
OBJECTIVE: Quality control in the clinical pulmonary function laboratory has been well developed for spirometry and diffusing capacity but not for the measurement of TLC. The purpose of the present study was to test two approaches to this problem. First, we compared TLC by body plethysmography (TLCb) with a value predicted from TLC measured by multibreath helium dilution (TLCm). Concordance between the measured and predicted values would imply the validity of the measurements. Second, we measured the test-retest variability of TLCb, TLCm and TLC measured by single breath helium dilution (TLCs) to assess the consistency of the measurements. METHODOLOGY: We performed a prospective study of 815 veterans. RESULTS: The prediction of TLCb from TLCm improved (r2 increased from 0.44 to 0.64) when FEV1/FVC and the difference between TLCm and TLCs were added to the model. The coefficient of variation for test-retest of TLCs, TLCm and TLCb were 8.9, 7.1 and 5.4%, respectively. Of all tests, 5.9% were inconsistent based on pathophysiology or measurement error and attributed mostly to TLCm. CONCLUSIONS: Prediction of TLCb from TLCm was not sufficiently accurate as to be useful for quality control. Comparison of TLCs, TLCm and TLCb may be useful for determining the internal data validity.