Skip to main content

Archived Comments for: Efficiency of the Austrian disease management program for diabetes mellitus type 2: a historic cohort study based on health insurance provider’s routine data

Back to article

  1. Selection Bias in retrospective DMP-evaluation

    Andreas Sonnichsen, Paracelsus Medical University

    10 March 2013

    Unfortunately the study of Ostermann et al only adds to the montain of literature on disease management evaluation without conveying any additional information. As many authors before them, Ostermann et al come to the conclusion that we need randomized controlled trials to really prove the effectiveness of a DMP. Surprisingly the authors neither cite nore discuss the RCTs and metaanalyses that already exist on the topic.

    The Austrian Disease Management Programme "Therapie aktiv" Ostermann et al are targeting, has already been evaluated in an RCT (1) which showed some improvements in process quality but no effects on metabolic control, and thus most likely little or no effect on patient outcome. These findings are in line with the study of Olivarius et al that showed in a six year RCT that the Danish DMP improved surrogates but had no effect on outcome (2).

    The results presented by Ostermann et al are in line with many similar studies that have been published during the past two decades (and Ostermann et al cite some of that literature), and the effects they show are probably not due to the DMP but to the selection bias implicated by their study methodology. Which patients will be included in a DMP by general practicioners? Of course those who will most likely benefit from the programme, and those are not the comorbid and severely ill ones. If these patients are compared to non-DMP-patients without correcting for comorbidity and other confounders (which is not possible in a retrospective study with limited access to full sets of data), the non-DMP-patients (including patients with malignant disease, advanced stages of diabetes complications, limited life expectancy etc) will of course spend more days in the hospital, produce more costs, and will reveal a higher mortality. The striking difference in mortality between DMP-patients and non-participants has been shown by Miksch et al: They saw a more than 10% mortality reduction in the DMP-patients in only three years after inclusion in the programme, making DMP appear like a real "miracle drug", better than all diabetes therapy we know (3).

    Unfortunalely this wonderful effect is not due to the DMP, but to the comparison of apples and pears, i.e. the selection bias described above. But do we really need more RCTs? From the vast amount of studies evaluating DMPs we already know that the effects of DMP on outcome are small if at all measurable. This knowledge will most likely not be altered by any additional RCT, let alone by bias-prone studies like the one presented by Ostermann et al. It's time to move on: there are so many better concepts in chronic care like peer support, patient involvement, personalized care etc that really need our attention regarding further development, evaluation in randomized trials, and implementation on a larger scale. It would be wise to use our resources for these future concepts rather than for yet another mediocre study on something we already know.

    (1) Sonnichsen A, Winkler H, Flamm M et al. The effectiveness of the Austrian disease management programme for type 2 diabetes: a cluster-randomised controlled trial. BMC FamPract 2010;11:86.

    (2) Olivarius NF, Beck-Nielsen H, Andreasen AH, Horder M, Pedersen PA: Randomised controlled trial of structured personal care of type 2 diabetes mellitus. BMJ 2001, 323:970-975.

    (3) Miksch A, Laux G, Ose D, Joos S, Campbell S, Riens B, Szecsenyi J: Is there a survival benefit within a German primary care- ased disease management program? Am J Manag Care 2010, 16:49-54.

    Competing interests

    None declared

Advertisement