Let $(P, \le)$ be a partially ordered set. A subset $A\subseteq P$ is a chain if any two of its elements are comparable. A subset $A\subseteq P$ is cofinal if every element of $P$ is less than or equal to some element of $A$.
Question: Suppose $P$ admits some cofinal chain. Does every cofinal subset $Q$ of $P$ also admit a cofinal chain?
If someone also has a reference, I'd be interested.
Some observations that could be useful to prove this:
If P has a maximum element, all cofinal subsets contain the degenerate chain consisting of just that element. That's the trivial case. So we can assume P has no maximum element.
A possibly useful lemma (easy to see): If $P$ has a cofinal chain, then every cofinal subset of $P$ is a directed set.
The related question Do all cofinal chains in a partially ordered set have the same cofinality? showed that all cofinal chains in $P$ (if they exist) have the same cofinality. This question is a little different, but some similar technique may help.
Yes, this is true:
Let $C$ be a cofinal chain in $P$ and $Q$ a cofinal subset of $P$.
Obviously, $P$ and $Q$ are upward directed. W.l.o.g, we may assume that $\emptyset \neq C$ has no maximum, hence $C$ contains a cofinal subset order-isomorphic to a regular cardinal. Hence, we may assume that $C = \{c_\alpha: \alpha < \kappa \}$ with $\kappa$ a regular cardinal and $(c_\alpha)_{\alpha < \kappa}$ strictly increasing.
By transfinite induction we define $(q_\alpha)_{\alpha < \kappa}$ strictly increasing, such that $c_\alpha \le q_\alpha \in Q$ for all $\alpha < \kappa$.
Hence, $\{q_\alpha: \alpha < \kappa\} \subset Q$ is a chain and cofinal in $P$.
I'm not aware of an explicit reference, but I'm pretty sure that this is considered to be folklore. You might find further information, perhaps including a reference, by searching for "cofinal types of directed sets". A very advanced paper is this one of S. Todorcevic. Might be one of the earliest paper on this subject, the 1940 paper of Tukey contains a reference?