Combined inter-observer and inter-method variation in bone histomorphometry

Manual methods for the measurement of bone biopsies have largely been superseded by semi-automatic computerised techniques. Histomorphometrists often use control data obtained by other observers using different methods, thus combining inter-observer and inter-method variation. We have examined the combined effect of inter-method and inter-observer variation on measurements of bone area, osteoid perimeter, and osteoid width in iliac crest biopsies from healthy subjects, one observer using the manual grid system and the other using a semi-automated technique. Inter-observer and inter-method variation were independently determined, and the proportion of each expressed as a percentage of combined error. Our results indicate that the combination of inter-method and inter-observer variation causes significant differences in the values obtained for osteoid perimeter, whereas inter-method variation is mainly responsible for differences in osteoid width values; differences in bone area are largely due to inherent sampling variation. These variations indicate that caution is required when comparison is made with control data from other sources, especially if different techniques are employed.