[Skip to Navigation]
March 2004

Visual Inspection of Age-Specific Incidence Rates : Don't Forget the Scaling!

Arch Dermatol. 2004;140(3):361-362. doi:10.1001/archderm.140.3.361-b

In a recent article, Hemminki et al1 reported time trends and familial risks in squamous cell carcinoma (SCC) of the skinbased on the Swedish Family-Cancer Database. In particular, they tried to address this question with reference to sun-exposed and covered sites of thebody. Figure 3 of their article displays the age-specific incidence ratesof invasive SCC of the skin on sun-exposed and covered body parts. They definedthat sun-exposed sites among males "included the head, neck, and arms, and for women, also legs. All other sites were considered nonexposed." From Figure3 they conclude, "the slopes (incidence/age) are quite different for sun-exposed and covered sites." We think that this conclusion is wrong. They displayedincidence rates by age group on a linear scale, which simply masks the slope of the age-specific incidence rate of the covered skin sites. If the authorswould have used a logarithmic scale, they would have found that the slopes of the age-specific incidence rates of sun-exposed and covered skin sitesare very similar. It is well known that using linear scales hampers the visual assessment of proportional rates of change. The use of a logarithmic scalefor the y-axis will result in a straight line when the rate of change is constant. Parallel lines on a logarithmic scale for 2 or more groups indicate the samerate of change.2

Add or change institution