Chapter 7 make their results dubious. For dCA parameters ICCs are below 0.75 and especially for ARI ICC is very low (~0.5) although these values still are considered to represent fair to good reproducibility 9. High ICC for mean ABP and mean CBFV and reduced ICC for dCA parameters deducted from ABP and CBFV recordings make clear that either dCA tests or analysis techniques or both need to be improved to increase reproducibility to allow clinical applicability of dCA tests. Due to low subject numbers between 10 and 20 in studies on reproducibility of dCA parameters confidence intervals for the reported ICC values are very wide. Much larger studies with over 200 subjects on reproducibility are needed to narrow down these confidence intervals of ICC values, but more important is finding ways to improve dCA test reproducibility. Poor reproducibility of dCA parameters might also be due to true variation of autoregulation. Recent studies in which dCA was evaluated at a one minute time resolution suggest dCA varies over this time range 22, 24, 26. Further research is needed to investigate what causes the variability of dCA parameters. Multivariate analysis By definition dCA simply involves the univariate relationship between ABP and CBFV, but in practice it is difficult to study this relation without interference of covariates. The influence of covariates may result in decreased coherence. To cope with this, either one has to record under conditions of stable covariates or these covariates need to be monitored and corrected for in the results. An important covariate is arterial CO2 level, since CO2 as a potent vasodilator largely influences CBFV. In our reproducibility study in chapter 2 we have recorded end-tidal CO2 level (etCO2) in part of the measurements. These recordings showed that etCO2 was decreased in 6/min breathing compared with normal breathing and dCA phase was increased in these measurements. These findings have also been reported by Reinhard et al using paced breathing 30 and are consistent with our results during CPB in chapter 6, where dCA phase decreases with increasing paCO2. This implies paCO2 always needs to be monitored. Better than monitoring etCO2 and explaining its influence on dCA results, is incorporating recorded etCO2 as extra input of the transfer function or mathematical model. Recently, a multivariate autoregressive moving average model has been demonstrated to simultaneously analyze the influence of ABP, 128
proefschrift gommer inhoud met kaft.indd
To see the actual publication please follow the link above