Jiguo Cao | PubMed Health

PubMed2006

An S-Plus function to calculate relative risks and adjusted means for regression models using natural splines.

We provide for generalized linear regression models that use natural cubic splines to model predictors an S-Plus function to calculate relative risks (RR), log relative risk (logRR), mean percent change (MPC) for continuous covariates modeled using a logarithmic link as well as adjusted means differences (MD) for the identity link. The function makes explicit use of the natural spline basis functions, the estimated coefficients for each natural spline basis function, and the fitted correlation matrix for the estimated coefficients and can thus accommodate any number of degrees of freedom. The main function produces a publication-quality graph of all of these quantities as compared to a user-specified reference value as well as the associated confidence limits. In another function, specific values of these statistics comparing a vector of values of the independent variable to the reference value can be calculated rather than plotted.

Regression Analysis↗

PubMed2006

Arterial versus capillary blood gases: a meta-analysis.

A meta-analysis determined whether capillary blood gases accurately reflect arterial blood samples. A mixed effects model was used on 29 relevant studies obtained from a PubMed/Medline search. From 664 and 222 paired samples obtained from the earlobe and fingertip, respectively, earlobe compared to fingertip sampling shows that the standard deviation of the difference is about 2.5x less (or the precision is 2.5x better) in resembling arterial PO(2) over a wide range of arterial PO(2)'s (21-155 mm Hg ). The lower the arterial PO(2), the more accurate it is when predicting arterial PO(2) from any capillary sample (p<0.05). However, while earlobe sampling predicts arterial PO(2) (adjusted r(2)=0.88, mean bias=3.8 mm Hg compared to arterial), fingertip sampling does not (adjusted r(2)=0.48, mean bias=11.5 mm Hg compared to arterial). Earlobe sampling is slightly more accurate compared to fingertip sampling in resembling arterial PCO(2) (arterial versus earlobe, adjusted r(2)=0.94, mean bias=1.9 mm Hg ; arterial versus fingertip, adjusted r(2)=0.95, mean bias=2.2 mm Hg compared to arterial) but both sites can closely reflect arterial PCO(2) (880 total paired samples, range 10-114 mm Hg ). No real difference between sampling from the earlobe or fingertip were found for pH as both sites accurately reflect arterial pH over a wide range of pH (587 total paired samples, range 6.77-7.74, adjusted r(2)=0.90-0.94, mean bias=0.02). In conclusion, sampling blood from the fingertip or earlobe (preferably) accurately reflects arterial PCO(2) and pH over a wide range of values. Sampling blood, too, from earlobe (but never the fingertip) may be appropriate as a replacement for arterial PO(2), unless precision is required as the residual standard error is 6 mm Hg when predicting arterial PO(2) from an earlobe capillary sample.

Arteries↗