We propose a new methodology for the estimation of reference intervals
for data sets with small numbers of observations or for those with su
bstantial numbers of outliers. We propose a prediction interval that u
ses robust estimates of location and scale. The SAS software can be re
adily modified to do these calculations. We compared four reference in
terval procedures (nonparametric, transformed, robust with a nonparame
tric lower limit, and transformed robust) for sample sizes of 20, 40,
60, 80, 100, and 120 from chi(2) distributions of 1, 4, 7, and 10 df.
chi(2) distributions were chosen because they simulate the skewness of
distributions often found in clinical chemistry populations. We used
the root mean square error as the measure of performance and used comp
uter simulation to calculate this measure. The robust estimator showed
the best performance for small sample sizes. As the sample size incre
ased, the performance values converged. The robust method for calculat
ing upper reference interval values yields reasonable results. In two
examples using real data for haptoglobin and glucose, the robust estim
ator provides slightly smaller upper reference limits than the other p
rocedures. Lastly, the robust estimator was compared with the other pr
ocedures in a population where 5% of the values were multiplied by a f
actor of 5. The reference intervals were calculated with and without o
utlier detection. In this case, the robust approach consistently yield
ed upper reference interval values that were closer to those of the tr
ue underlying distributions. We propose that robust statistical analys
is can be of great use for determinations of reference intervals from
limited or possibly unreliable data.