nzcoops / r-code Goto Github PK

In Appendix B of the 4th report (NIH/NHLBI), the polynomial regression formulae are the same for SPB and DBP z-scores, with the exception of different beta coefficients e.g. the quartic polynomial in age has the form beta1*(age -10) + beta2*(age - 10)^2 + beta3*(age-10)^3 + beta4*(age-10)^4, where you provide the betas in in your table bplu.

This is in fact what you specify for female SBP , where each term has the form beta*(age-10)^x

    mu <- bplu$sbpf[1] +
    sum(bplu$sbpf[2]*(age[i]-10),bplu$sbpf[3]*(age[i]-10)^2,bplu$sbpf[4]*(age[i]-10)^3,bplu$sbpf[5]*(age[i]-10)^4) +
    sum(bplu$sbpf[6]*z_ht[i],bplu$sbpf[7]*z_ht[i]^2,bplu$sbpf[8]*z_ht[i]^3,bplu$sbpf[9]*z_ht[i]^4)
    sbp_z[i] <- (sbp[i] - mu)/bplu$sbpf[10]

However, when I look at the analogous formula for female DBP, you have written (beta*(age-10))^2 i.e. the beta is squared (and cubed and raised to the fourth power), which is a very different result.

    mu <- bplu$dbpf[1] +
    sum(bplu$dbpf[2]*(age[i]-10),(bplu$dbpf[3]*(age[i]-10))^2,(bplu$dbpf[4]*(age[i]-10))^3,(bplu$dbpf[5]*(age[i]-10))^4) +
    sum(bplu$dbpf[6]*z_ht[i],bplu$dbpf[7]*z_ht[i]^2,bplu$dbpf[8]*z_ht[i]^3,bplu$dbpf[9]*z_ht[i]^4)
    dbp_z[i] <- (dbp[i] - mu)/bplu$dbpf[10]

If I'm not mistaken, the beta should be outside the bracket, no?