The actual paper is directly linked in this press release. It contains three threats to validity:

First, due to the small sample size, demographic factors (such as prior experience with D&D and COVID-19 experiences) could not be entered into the statistical models as control variables. […]

Second, the single-arm design is vulnerable to participant-related effects (participants responding to the demand characteristics of the research situation and placebo effects) and experimenter expectancy effects. […]

Third, due to the age of participants (mean age was around 28), we should be cautious when generalizing these findings to other groups, such as geriatric or paediatric populations.

Overall, this is not very strong evidence. The primary conclusion from the researchers seems to be “promising for further research”.

One curious note:

Interestingly, only two of the five outcome variables were found to change from T1 to T2. This could indicate that the positive effects of D&D take time to manifest or that a threshold of exposure is needed before positive effects begin to manifest.

They had 8 weekly 1h sessions. T1 is “before the first session” and T2 is the middle “after 4 of 8 sessions”. (They also measured T3 “after eight sessions” and T4 “one month after the last session”).

Seems like one shots won’t do it.