Height at diagnosis of malignancies.

Studies of the presenting height of children with malignancies have produced conflicting results, from an excess of taller patients to an excess of shorter patients. The problems of measurement bias, inadequate comparison populations, small numbers of patients, subgroup analyses, and overreliance on simple significance tests are all possible reasons for the variation in results. To clarify this issue, we studied heights at diagnosis of 3657 children and adolescents aged under 18 years. Their malignancies included acute lymphoblastic leukaemia, non-Hodgkin's lymphoma, Hodgkin's disease, acute non-lymphoblastic leukaemia, osteosarcoma, retinoblastoma, neuroblastoma, Wilms' tumour, rhabdomyosarcoma, and Ewing's sarcoma. Compared with published standards for the heights of children in control populations, no significant deviation from population norms was found for patients in any of the 10 disease categories after proper adjustment for multiple significance testing.

Duke Scholars

Author Stephen L. George Biostatistics & Bioinformatics, Division of Biostatistics