person time rate
- related: Biostats
- tags: #literature #pulmonary
You are the director of a post-COVID-19 surveillance program that has enrolled patients with confirmed mild disease not requiring hospitalization who then return to the clinic at least annually for scheduled testing to determine if they meet predefined criteria for pulmonary fibrosis in order to determine if this infection results in a higher than expected incidence of chronic lung disease. One thousand patients have been enrolled.
In year 1, 0 cases of fibrosis are identified in 920 patients; 80 patients are lost to follow-up.
In year 2, 20 cases of fibrosis are identified in 780 patients; 20 patients died (not of lung disease), and 100 were lost to follow-up.
In year 3, 30 new cases of fibrosis were identified in 600 patients; 20 more patients died (not of lung disease), and 110 more were lost to follow-up.
What should the person-time incidence rate of pulmonary fibrosis in this population be reported as?
The person-time rate presented in this question should be reported as the number of cases per a determined number of person-years, not as cases per person per year (choices B and D are incorrect). Because there are patients lost to follow-up or who die of causes unrelated to the end point being followed, an estimation needs to be made to account for these events in terms of person-years. Since information is typically lacking for patients lost to follow-up, an occurrence encountered in almost all longitudinal studies, treating these missing patients statistically is challenging. One generally accepted method is to assume that the loss to follow-up and death from unrelated causes, as reported here, are random events over the year and, thus, on average occurring at the half-year point. With this assumption, the following simple calculations can be made. During the first year, 920 patients contribute 920 person-years, and 80 patients lost to follow-up contribute 80 × 1/2 = 40 person-years, for a total of 960 person-years. During the second year, 780 disease-free patients contribute 780 person-years, and the 20 with newly diagnosed lung disease, the 20 who died, and the 100 lost to follow-up contribute 140 × 1/2 = 70 person-years, for a total of 850 person-years. In the third year, returning patients without lung disease contribute 600 years, and the 30 with lung disease, the 40 who died, and the 110 lost to follow-up contribute 90 person-years, for a total of 690 person-years. Thus, there were 50 cases identified in 2,500 person-years (960 + 850 + 690) (choice A is correct; choices B, C, and D are incorrect). The person-time rate of disease is often expressed per 1,000, 10,000, or 100,000 person-years; in this case, one could say there were 200 cases of pulmonary fibrosis per 10,000 person-years in patients after mild COVID-19. This value could then be compared with the incidence in other populations to determine the risk (or lack of risk) of mild COVID-19 for subsequent lung fibrosis.2