Hello everyone,
I would greatly appreciate it if someone could assist me in understanding the steps required to estimate the sample size for a longitudinal study. Or advise bibliography.
I am investigating a bacteria which colonizes the upper respiratory tract of individuals. However, it's not consistently present since it can be cleared by the immune system.
Previously, we observed in a 6-month period follow-up a prevalence of carriage of about 30% in adults requiring a minimum of 6 repeated measures/individual and a sample size of 70.
For this upcoming study, we plan to conduct 12 measures per individual (each time assessing whether the individual is colonized or not) over a 6-month follow-up period among 30 families, each with an average of 6 members, giving us a total number of 30X6X12 samples to look for the bacteria. We intend to study transmission between members but I think it would be simple to estimate sample size based on a prevalence.
I am aware of the formulas used to calculate sample size for prevalence estimates and than correct for a one-stage cluster cross-sectional study with the design effect, DEFF(below).
However, in a longitudinal follow-up, individuals may become colonized at different time points (e.g., at t2 or t3), potentially requiring fewer individuals to reach the 30% prevalence threshold.
Would someone be able to guide me on how to adjust the sample size calculation for this longitudinal study?
For cross-sectional studies in one-stage cluster sampling:
DEFF=1+(m-1) X ICC
m: number of family members, ICC: intra-cluster correlation
n=(Z^2 X P X (1-P) X DEFF) /E^2
Where Z z-score for the 95% normal percentile and P the prevalence I want to estimate, E margin of error.
Thank you in advance for your assistance.
Best regards,
- Badges
- Science topic
- Similar topics
- Medicine
- Disease
- Infectious Diseases