Healthy worker survivor bias: implications of truncating follow-up at employment termination