Review Question - QID 218899

An orthopaedic surgery resident proposes doing a randomized clinical trial comparing open versus arthroscopic treatment for septic arthritis of the native knee. They perform a pilot study with 10 patients in each group and report their early findings using 95% confidence intervals. They found that arthroscopic treatment does not result in a significantly different hospital length of stay compared to open treatment. However, they proceed to include multiple other community hospitals in the same city and continue the study over the next two years with a final 100 patients in each group. Their final report published in a large journal states arthroscopic treatment results in a significantly shorter hospital length of stay compared to open treatment (Figure A) and formal comparative analysis shows a p-value of 0.04.

What is the most likely cause of the narrowed confidence intervals, and thus different results, in the final publication?