I am wondering what is best/normal practice for reporting the results of an independent t-test of log-transformed data?

For context, I am comparing the value of household assets for children who are attending pre-primary school with children who are NOT attending pre-primary school in Uganda. The data was highly positively skewed & violated the t-test assumptions, after log10 transformation it meets all the assumptions so I am happy that transforming the data was appropriate.

The descriptive statistics of the transformed data
Attending: N=569, M=2.445, SD=0.684, SE=0.029
Not attending: N=947, M=2.123, SD=0.676, SE=0.022

The result of the t-test on the transformed data
t=8.925, df=1514, p=.000
Mean diff=0.321, SE Diff=0.036, 95%CIs=0.251, 0.392

If the data had not been transformed then I would report the following
An independent samples t-test revealed that children attending pre-primary school (M=2.45, SD = 0.684) had significantly higher household assets than children not attending (M=2.12, SD = 0.676) , M=0.321 95%CI [0.251, 0.392], t(1514)=8.925, p <.001, d=***,

I understand that I need to back-transform the means, and that this will give the geometric mean rather than the arithmetic - and I think this means SDs are no longer appropriate? But I am unclear about exactly how to calculate/report the t-test statistic, mean difference, CIs, Cohen's D etc?

Any help or advice on this would be amazing!

Thanks, Cordelia