1. Penetrance of hemo- chromatosis

11

Good: Genotyping of surviving Brusselton, Australia, cohort; potential selective mortality bias appears minimal. Small numbers.

17 y of clinical data for 10 screening-detected general population C282Y homozygotes illustrates variable disease expression and incomplete penetrance. Incomplete followup into older age where disease penetrance increases.

Fair: Genotyping of representative Danish cohort during third examination. Results are likely to be compromised by selective mortality bias due to 35% loss of followup. Even accounting for potential bias, disease penetrance about 60%.

Additional 23 screening-detected C282Y homozygotes from the general population also illustrates variable disease penetrance and variable patterns of iron accumulation. No liver biopsies to confirm iron overload or disease.

Fair to good: Studies compromised by frequent inclusion of already-identified C282Y homozygotes (not clearly screening-detected), by different standards for disease, and by potential selection bias due to non-protocol-based selection for further clinical work-up.

Estimates of disease in newly identified C282Y homozygotes at screening are too limited to provide confident estimates of penetrance.

2. Efficacy of phlebotomy treatment

5

Fair to poor: Studies compromised by selective samples, reporting on cases not clearly comparable to current diagnosis and treatment, incomplete followup on all cases, and failure to account for possible confounders in analyses.

Total number of reported cases is quite small and represents disease experience over 50 y. There are no data to determine the benefit of earlier treatment among screening-detected compared with contemporarily diagnosed clinical cases.

Fair: Possible recall bias in determining response to treatment.

Treatment is recalled to relieve some but not all symptoms in a survey of patients with hereditary hemochromatosis.

3. High-risk groups

7

Fair to good: Studies examined prevalence of C282Y homozygotes in various selective populations for possible targeted screening.

Patients selected on basis of certain signs and symptoms, in combination with phenotypic testing, may be at increased risk; data are still fairly limited.