Tl;dr: The small portion of Sanders supporters that crossed over to Trump was enough to change the outcome in MI, WI, and PA, and therefore the election. However the bigger take away is that Democrats don't need to focus on this crowd - and from what I have seen Sander's messaging has been more focused on energizing voters historically more likely to vote D, but also more likely to sit out, which is wise. Clinton badly underperformed Obama in two states with very small margins of victory, directly leading to her defeat - two states in which Bernie Sanders beat her in the Democratic primaries. Sanders is at the top of the D field vs Trump in PA and is best positioned to bring new voters into the fold, which is probably the single biggest key there given Trump's strength. To say he isn't as well positioned to deliver WI, MI, PA as other candidates is incorrect. Full disclosure, I am going to vote for Elizabeth Warren unless the writing is just on the wall by the time TX votes. So I am not saying this as a Sanders supporter.
If you have free time:
I will concede the data is limited (WaPo notes three surveys), but the data that does exist indicates the number of Sanders voters that went for Trump over Clinton in the general was greater than Trump's margin in WI, MI, and PA. The same surveys also show that these voters actually hold a lot of the opinions associated with middle aged, uneducated white voters (I haven't seen any reporting on age): Low approval of Obama, don't believe whites have an inherent advantage, high opposition to free trade.
This is your "economic anxiety" crowd. A moderate is not the way to reach them. Structural economic and political reform is.
That is still secondary to the fact that untapped votes in the historically low participation/disenfranchised groups outnumber the unicorn swing voting white moderate in the midwest and everywhere else. Sander's messaging has been predominantly about getting these groups involved and that's for good reason.
Even if every single Trump voter shows up for him again, in these states where he is underwater in approval (-10,-12, -2 respectively in WI, MI, and PA per Morning Consult):
Trump won Michigan by 10k votes. Wayne County (Detroit - 66 percent D) turnout decreased by 40-45k from 2012 and Clinton underperformed Obama by 75k votes. Trump got about 15k more votes than Romney. Assume that those 40k missing voters also vote at a 66 percent D clip (in reality they'd likely vote D at a higher rate because nonvoters trend young and minority) and that flips the state.
Milwaukee County, WI - Clinton got 50k fewer votes than Obama. You'd think, okay, well yeah she lost the state and he won; but Trump got 30k fewer votes than Romney, so we have a turnout decrease of ballpark 65-70k when accounting for third parties. Trump won Wisconsin by a little under 25k votes. Those 65-70k missing voters voting at Milwaukee County's 65-29 percent D clip flip the state. Further, Trump won but also underperformed Romney in the counties adjacent to Milwaukee - the suburbs. Clinton just underperformed Obama too. Trump didn't run up the score with Obama voters, Democrats just stayed home.
So this wasn't some Obama-Trump voter/middle aged white moderate wave that flipped WI and MI. Or rather, if it made the difference, it wasn't because it was a ton of them, but because Democratic voters stayed home. There are Obama voters on the sideline, and knowing what we know about the Obama coalition, they likely trend younger and more racially diverse. This crowd is Sander's strongest bloc.
I'll also concede I don't see a similar trend in PA just by eyeballing for this post. Turnout was up across the board and generally the state appeared to consolidate along urban vs rural lines. Trump got 2.9mm votes statewide, about 65k more than Obama 2012 and over 750k more than Romney. In the other two states, Trump's numbers basically tracked Romney's while Clinton just underperformed Obama terribly. Between that and the approval rating gap between PA and the other two, I have to admit I think I need to reevaluate my previous stance that PA is essentially a Democratic lock. It almost looks like WI and MI are lower hanging fruit and PA is where the work is really going to be cut out for Ds.
Still between the 2018 midterms, where Democrats absolutely wiped the floor with the GOP statewide, and the fact that the areas driving growth in PA are Democratic and trending more so, I think there's a reasonable case to be made that Trump's performance in the state probably peaked in 2016. I'm not sure there are many new Trump voters waiting in the wings. New voters in PA are more likely to lean D, and in a state where the margin of victory was 45k votes, that matters inherently. Sanders is the leading candidate among young voters, which is among the biggest untapped Democratic voting blocs. Sanders also outperforms every Democrat not named Joe Biden in polls vs Trump in PA. Democrats are going to have to fight for the state, but Bernie is as well positioned as anyone to pull it off.