We have to thank Ms. Mühlbauer and her coworkers for dedicating their review (1) to a topic of enormous relevance in daily practice. Their conclusion is—almost as expected—depressing. Due to the lack of evidence-based information about this problem, individual decisions have to be made on a case-by-case basis (1). However, we do know that in frail, older patients it rather does more good than harm to the patient if a medication can be avoided.
However—and this is my criticism—the effect measures listed in the extensive Tables 1a and 1b are exclusively stated as hazard ratios, i.e. as relative risks (RR), which is not very helpful.
Here an example: In Table 1a, a HR = 0.75 is stated under HYVET for stroke in the first row.
This can mean that in the placebo group 6.30% and in the active group 5.05% of patients experienced a stroke. Consequently, one would arrive at the following figures:
- Absolute risk reduction (ARR) = 1.25% (6.30 − 5.05)
- Number needed to treat (NNT) = 80 (100 : 1.25)
- Relative risk reduction (RRR) = 25% (1.25 based on 5.05)
- Relatives risk (RR) = 75%
- Hazard ratio (HR) = 0.75.
However, one would also arrive at HR = 0.75 if with placebo 63% and with active treatment 50.5% of patients had experienced a stroke (consequently: ARR = 12.5%, NNT = 8) or if with placebo only 0.63% and with active treatment 0.505% of patients had experienced a stroke (consequently: ARR = 0.125%, NNT = 800).
These considerations show that relative values (relative risks, hazard ratios, odds ratios, relative risk reductions) cannot reliably be interpreted and thus are unsuitable for the evaluation of effects.
Prof. Dr. med. Frank P. Meyer
|1.||Mühlbauer V, Dallmeier D, Brefka S, Bollig C, Voigt-Radloff S, Denkinger M: The pharmacological treatment of arterial hypertension in frail, older patients—a systematic review. Dtsch Arztebl Int 2019; 116: 23–30 VOLLTEXT|