Effectiveness of a telephone-based nursing intervention to reduce hospital utilization by COVID-19 patients.

This was a single-site retrospective comparison by test date, not a randomised trial, so groups could differ in ways not measured. The primary outcome was not statistically significant, and the post-hoc power analysis showed only 53% power, so the study may have been too small to detect a real difference. The intervention had several components, including symptom scoring, call frequency, appointments, paramedic visits, emergency department recommendations, isolation reminders, and retest confirmation. The contribution of any single component cannot be separated. Researchers could not fully separate hospital visits related to COVID-19 from unrelated visits, and they excluded only obviously unrelated cases. BMI was missing for some patients, so an initial model had 265 usable records. BMI was dropped to keep all 293 cases. Patients were assumed to use only the affiliated hospital for acute care because there was only one hospital in the county, and testing and travel were restricted during the study period. Generalizability is limited to a small county with testing capacity and strong hospital-public health collaboration. It may not work in large cities, areas with high COVID-19 activity, or settings with underserved populations and limited testing. Patient satisfaction was not measured, so the study cannot say whether the telephone support was acceptable or helpful to patients.

Do not conclude that the nurse-led telephone protocol reduced hospital use. The primary outcome did not reach statistical significance, the study had only 53% power, and the intervention included several components that cannot be separated.