Punjabi reported test reproducibility for DLCO was best assessed by absolute differences of 2.0 to 2.5 DLCO units between tests (Chest 2003; 123:1082-89). This recommendation has been criticized because it was based on only one instrument manufacturer and one laboratory.
Using a retrospective design to avoid influencing laboratory procedure, 16 laboratories from the ATS lab registry program submitted complete data for two DLCO trials (DLCO, VI, VA) and patient demographic data from 50 consecutive laboratory reports (792 records). Data were obtained from 22 instruments representing nine different models and four manufacturers. The figureillustrates percent and absolute differences between the two DLCO measurements by average DLCO.
There is strong nonlinear relationship between percent difference and average DLCO and a trivial linear relationship between absolute difference and average DLCO. After excluding four outliers, absolute differences for 70% of the tests were < 1 DLCO unit, 91.3% were < 2.0 units and 95.5% were < 2.5 units. These data confirm the Punjabi’s findings in 16 laboratories using a wide variety of instruments.
Acceptable DLCO variability should be assessed as absolute differences less than 2 or 2.5 DLCO units.
Doing so simplifies DLCO testing and may reduce the number of tests required to meet acceptability criteria.
R.O. Crapo, None.