I would like to determine if a particular screening modality changed the incidence rate of a disease and would like to compare the incidence rates before and after it was introduced.
Thank you.
Rates are comparable as opposed to absolute numbers. Rate has a numerator, denominator, time specifics and a constant. Thus incidence rates of diseases (given by new cases of disease divided by population at risk in a given time period multiplied by a constant) in two different eras can be derived and compared. One would not expect the screening modality in its self to cause any significant change in incidence of disease. It is not an intervention against the disease in question. Perhaps you want to know if there is any change (improvement) in its ability to detect new cases of the disease (sensitivity). Whatever the objective of the study, I think the parameter, annual percent change is more valuable than percent change.
Please clarify which disease you want to screen? Is it sample based or population based survey? Like cancer incidence is always calculated based on population based. Comparison of rates is a valid approach to find out the effect of screening. In case of sample based survey, proportion test or chi-square test is also possible.
Thank you so much for your answers.
Dr. Adogu - I agree that the screening modality will no change the incidence as it has no therapeutic capabilities, unlike colonoscopy. But, we were expecting more diagnosis with the screening modality as there exists no gold standard modality for diagnosis for the disease.
Dr. Takiar - It is a population based survey for cancer incidence.
Can you do a time trend analysis for variation in prevalence/incidence? It can be done through a polynomial regression.
Other option is use the invention of screening modality as an intervention (hypothetically) and do a quasi-experimental analysis for pre and post?
Dr. Khubchandani - I am not sure if I can do time trend analysis on the data available. The other option that you have suggested is something that I was trying to do. Can ou please elaborate on this?
If screening has effect, which should have, then immediate year or period after the diagnosis should show a rise in the number of cases or change in the incidence rate. If the assessment is after a long time of introduction of screening then time trend also play a role. Another approach could be as follows:
Calculate the standard error of the rate of the selected disease (if cancer) by the following formula: rate/(√cases). Find out the confidence limit using the formula rate+1.96 SE. If immediately after the screening, the rate fall beyond this limit, definitely, the screening has effect. otherwise not. We expect that the screening should increase awareness among the people thereby approach for diagnosis and thereby increase in the number of cases, to be reflected in the increase in the incidence rate. Allowing random variation, the rate should not change in one year or so from previous years by more than the confidence limit. I hope this should answer your question.
I go straight at your direct answer and I don't want suggest any other method because this is not professional. I would you recommend the percent change because it gives you the immediately impact of changing effect in you outcome, and compare also at annual percentage change for the total result in one year experiment. It means calculate the increasing or decreasing percentage from the baseline to the end, and also for the single period Tb to Ta, Tc to Tb, Te to Td etc. I should simple for showing the results for a great picture impact
Also, why not measure the difference from the different subtotal period by the T-Test performed by SPSS 24.0! You should see the significance increasing or not about the same treatment or different treatment too.
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology
- Biomolecules
- Nucleic Acids
- RNA