Sample Size for Prevalence in Wild Bird Epidemiology

[The Filipino Vet Student]

Hi! I am a fifth year veterinary student at Aklan State University, Banga, Aklan. Currently, I am working on a draft for a thesis proposal (though I am still at the RRL phase) which is about the intestinal helminths of Eurasian Tree Sparrows in Panay, Philippines. Sadly, publications about the parasitology of the aforementioned species, I think, are scarcely available in the Far East. But I somehow managed to find few yet very useful literatures. However, there is one thing that I am not able to do, the STATISTICS. I have poor knowledge when it comes to the statistical aspect of research. And EPIDEMIOLOGY, may it be human or veterinary, is basically STATISTICS.

In my thesis proposal, I chose 'point prevalence' as a measurement mainly because I find it simple, quick, cost-effective, and I have not found a single study about wild bird helminthology in my locality. Thus, using a prevalence study may be a good start. Anyhow, I presented my proposed study to my instructor. And I was asked to work out the statistical justifications for the sample size.

Now my question is: what is the required sample size for point prevalence? And what is the appropriate data analysis? Thank you for your help!

---

Members don't see this ad.

 

[sErISaNo]

Hi! I am a fifth year veterinary student at Aklan State University, Banga, Aklan. Currently, I am working on a draft for a thesis proposal (though I am still at the RRL phase) which is about the intestinal helminths of Eurasian Tree Sparrows in Panay, Philippines. Sadly, publications about the parasitology of the aforementioned species, I think, are scarcely available in the Far East. But I somehow managed to find few yet very useful literatures. However, there is one thing that I am not able to do, the STATISTICS. I have poor knowledge when it comes to the statistical aspect of research. And EPIDEMIOLOGY, may it be human or veterinary, is basically STATISTICS.

In my thesis proposal, I chose 'point prevalence' as a measurement mainly because I find it simple, quick, cost-effective, and I have not found a single study about wild bird helminthology in my locality. Thus, using a prevalence study may be a good start. Anyhow, I presented my proposed study to my instructor. And I was asked to work out the statistical justifications for the sample size.

Now my question is: what is the required sample size for point prevalence? And what is the appropriate data analysis? Thank you for your help!

Sample size ( n ) for a study examining prevalence may be calculated using the formula: n=(Z^2 * P(1 - P))/d^2 where Z = Z statistic at your confidence level (1.96 for the usual 95% confidence), P = estimated/expected prevalence, and d = precision. P may be determined by a pilot study, using a smaller sample size, or prior works. In your case, if there is no research available and a pilot study is not feasible, you may consider an estimated P based on other, similar studies conducted in different regions or with different species. This is ultimately a decision you would need to make based on your lit review and knowledge of the topic. Once you determine a reasonable estimated P, you can easily calculate a sample size using the formula above.

As for appropriate analysis, if your measurement is point prevalence, then I would expect a confidence interval at least. What other analysis you do depends on what question you are looking to answer, past the obvious "what is the prevalence?".

 

[The Filipino Vet Student]

Sample size ( n ) for a study examining prevalence may be calculated using the formula: n=(Z^2 * P(1 - P))/d^2 where Z = Z statistic at your confidence level (1.96 for the usual 95% confidence), P = estimated/expected prevalence, and d = precision. P may be determined by a pilot study, using a smaller sample size, or prior works. In your case, if there is no research available and a pilot study is not feasible, you may consider an estimated P based on other, similar studies conducted in different regions or with different species. This is ultimately a decision you would need to make based on your lit review and knowledge of the topic. Once you determine a reasonable estimated P, you can easily calculate a sample size using the formula above.

As for appropriate analysis, if your measurement is point prevalence, then I would expect a confidence interval at least. What other analysis you do depends on what question you are looking to answer, past the obvious "what is the prevalence?".

Thanks! Unfortunately, after careful considerations. I have resorted to conduct a disease detection study on SOIL TRANSMITTED HELMINTHS of EURASIAN TREE SPARROWS instead.

Do you happen to know the formulae or calculating the required sample size for detecting presence of disease with an imperfect test (i.e., sensitivity and specificity= <100%) on an unknown or infinite population?

The only sample size formulae for disease detection I can find from Thrusfield & Christley (2018) Veterinary Epidemiology (4th ed.) and Dahoo et al. (2003) Veterinary Epidemiologic Research are applicable only to finite population with perfect tests. Alhough the former does include imperfect tests, but only provides a formula that considers sensitivity of <100% and not specificity (see image below).

Anyhow, the method I wil use for my study is flotation technique. And according to Setiawan et al. (2020) (https://www.researchgate.net/public...on_of_Worm_Eggssoil_Transmitted_Helminths_STH), ZnSO4 solution seems to be a good choice (sensitivity=71.42% & specificity 94.44%) for detecting soil transmitted helminths.

 

[sErISaNo]

I don't think you would need to change anything about your sample size calculation, since the calculated prevalence will take into account the sensitivity and specify. You might want to take a look at this paper - SANDER GREENLAND, Basic Methods for Sensitivity Analysis of Biases, International Journal of Epidemiology, Volume 25, Issue 6, December 1996, Pages 1107–1116, https://doi.org/10.1093/ije/25.6.1107-a