Can I use this method to estimate the standard error for confidence intervals of other effect measures (e.g., relative risk)?

While the general principle applies, the specific formula needs to be adjusted based on the effect measure and the distribution it follows. For relative risk, you'd need to use the natural logarithm of the relative risk and adjust the critical value (1.96) accordingly, based on the desired confidence level and the underlying distribution.