Predicting rheumatic symptoms following acute diarrhoeal illness

Eric Järpe Department of statistics   Göteborg University

Boel Mörck Department of rheumatology and clinical immunology Göteborg University

Seminar paper 1996:12, Göteborg University, Göteborg.

Abstract
Signs of symptoms of locomotor system disorders following acute gastroenteritis seem much more common than previously known. Those symptoms could spring from other bacteria than the known arthritogenic. It would be of great value to detect the pathogenesis of postinfectious locomotor symptoms.

Whether the initial CRP value is pathological (CRP>10) or not, is too blunt an instrument for judging a development towards inflammatory joint disease. Perhaps a use of the longitudinal data could fix this but in that case there is a possibility that the mere development of arthritis itself directly causes higher CRP values.

For a new study we recommend:

• testing wheter classical reactive arthritogenic is predicted by very high initial CRP values (e.g. CRP>100) to get conditions for constructing judgement criteria regarding antibiotic treatment.
• measuring a sequence of initial CRP (for instance at days 1,2,3,...,10). Such a longitudinal material might make a major improvement of the basis for statistical analysis and the prediction of soft tissue disease.

Our study has merely closed in upon one possible predicting factor, that is the CRP level. Also other factors can be examined. Before any future attempts to longitudinally evaluate this statistical material, problems such as

• how to make the proper change in the time axis
• how to deal with different types of missing values
• what factors that are meaningful to be followed as predicting variables
• if any of the predicting variables, from a medical point of view, would be likely to be assumed as causing variables

must be satisfactorily solved. Until this has been properly done, any further longitudinal statistical conclusions drawn from the material must be regarded with the utmost caution. The snag with the sample size of only 68 patients showing locomotor problems, Of course still remains and so only large differences, within this group, can be verified.

If you want a gzipped copy of the whole work in english, get it here!