I am currently conducting a study that follows patients over 30 days to see if they have an event.
Patients may have multiple events, meaning they may enter the study and be followed for 30 days then exit and then re-enter
In order to account for this I used multilevel survival analysis with panel data

I started with set up of the survival set:
stset failuretime, failure(failure==1) origin(origintime)

Then I setup panel data:
xtset idvariable

Then I ran an empty model;
xtstreg , d(weibull)

I got the results in picture 1, so I assumed a mixed effects model would be needed

When I run a weibull distribution using mestreg and a lognormal distribution I get completely opposite coeffecients so I realized the distribution really matters. (picture 2)

I followed the instructions in the stata manual (picture 3) to try to determine which distribution to use

In picture 4 are the results, the p-value of kappa is greater than 0.05 but the p-value for the Wald test is less than 0.05

I assumed the lognormal distribution would be the correct thing to use

Am I correct in my assessment and procedures I used?
How else can I determine which distribution to use?
Is mestreg appropriate here?
If I use the lognormal distribution can I get it to display hazards ratios?