The Einstein-Hilbert Lagrangian is:
$$\mathcal{L}_{EH}=\sqrt{-g} R$$
where $g={\rm Det}[g_{\mu\nu}]$ and $R$ is the Ricci scalar. In linearized gravity $g_{\mu\nu}=\eta_{\mu\nu}+h_{\mu\nu}$ and
$$\sqrt{-g}=1+\frac{h}{2}+\mathcal{O}(h^2)$$
and
$$R=\partial_\alpha(\partial_\mu h^{\mu\alpha}-\partial^\alpha h)+\mathcal{O}(h^2),$$
where $h=h^\mu_{~\mu}$.
Then
$$\mathcal{L}_{EH}=\partial_{\alpha}(\partial_\mu h^{\mu\alpha}-\partial^\alpha h)+\mathcal{O}(h^2)$$.
I am comparing this to this paper: https://arxiv.org/abs/hep-th/9411092 . Apparently they don't get an $\mathcal{O}(h) $ term (cf. their eqs. (2.15)-(2.18)). The first non-vanishing term they get is $\mathcal{O}(h^2)$. They say in the paper that they use the De Donder gauge which is $\partial_\mu h^{\mu\alpha}-\frac{1}{2}\partial^\alpha h=0$. This is very similar to the lowerst order term, EXCEPT the factor of $\frac{1}{2}$.
I am pretty sure I did the expansion of the Ricci scalar correct, since I find the same result in Carroll's book. I checked that the De Donder gauge condition usually has the factor $\frac{1}{2}$... So I really don't see why the first order term in the Einstein-Hilbert Lagrangian should vanish?