G. Stucki et al., Application of statistical graphics to facilitate selection of health status measures for clinical practice and evaluative research, CLIN RHEUMA, 18(2), 1999, pp. 101-105
Relatively little attention has been directed towards the study of the abil
ity of health status measures to demonstrate clinically meaningful change.
We examined the use of Tukey's stem and leaf plots and time-path diagrams t
o assess the sensitivity of a health status measure in showing change. We c
ompared the evaluative characteristics of the physical ability scales of th
e Sickness Impact Profile (PSIP), and the SF-36 (PSF-36), in 54 consecutive
patients undergoing elective primary or revision total hip replacement. To
validate time-path diagrams, we used patients' perception of change in phy
sical function (transition question) as an external standard. At baseline t
here was no clustering at the end of the scales. At 3 months, the distribut
ion of the PSIP showed a clustering at the upper end of the scale and a str
ong skewness towards lower scores, whereas the PSF-36 showed a wide distrib
ution of the scores. Seven out of 54 patients had perfectly normal scores o
n the PSIP at 3 months, whereas one patient had a perfect score on the PSF-
36 at 3 months. In time-path diagrams the response pattern of the PSIP was
similar to that of the PSF-36, with the exception of the patient-fifth with
the best initial health status. The overall impression of the time-path di
agram for these patients was that of no change. A closer examination reveal
ed that half of these patients had horizontal or downwards sloping lines, i
ndicating a worsening in physical function. However, all these patients per
ceived themselves as improved. The PSIP has a ceiling effect for patients w
ith relatively good physical health. It was concluded that stem and leaf pl
ots are useful in the assessment of health status measures, for discriminat
ive purposes, to differentiate patients' health cross-sectionally. Time-pat
h diagrams stratified by baseline health status with information on patient
s' perception or valuation of change provide useful information about the a
bility of an instrument to detect change over time.