On 2014 Oct 31, R Y Seedat commented:

Thank you for the positive comments regarding the study.

Lesotho is a different country, with different demographics and burden of disease and a different healthcare system with differences in accessibility, availability of services and referral protocols. It would thus not make sense to combine the data.

I don't think that the calculation of confidence intervals would be a valid statistical test since the study was not based on a population samples but on entire populations. While the incidences calculated are an underestimate, the calculation of confidence intervals would not assist in determining the degree to which the incidence is underestimated.

On 2014 Oct 29, Farrel Buchinsky commented:

Nicely done study and well reported. I have seen several incidence reports but almost always from North America, or Denmark (or another developed country) and never from anywhere in Africa. The best-done North American studies show incidence of about 1 per 100000 or below (around 0.2-1). As the author states, The data described in the Free State and Lesotho is probably an underestimate since pediatric hoarseness (sans dyspnea or stridor) may not be diagnosed amongst those with less access to health care.

• If all the cases from Lesotho are referred to Bloemfontein why not combine the statistics?<br>
• Also, one could provide 95% confidence intervals (Poisson model probably) around the best estimates and see where the estimates from other reports fall within that range.

On 2014 Oct 29, Farrel Buchinsky commented:

Nicely done study and well reported. I have seen several incidence reports but almost always from North America, or Denmark (or another developed country) and never from anywhere in Africa. The best-done North American studies show incidence of about 1 per 100000 or below (around 0.2-1). As the author states, The data described in the Free State and Lesotho is probably an underestimate since pediatric hoarseness (sans dyspnea or stridor) may not be diagnosed amongst those with less access to health care.

• If all the cases from Lesotho are referred to Bloemfontein why not combine the statistics?<br>
• Also, one could provide 95% confidence intervals (Poisson model probably) around the best estimates and see where the estimates from other reports fall within that range.

On 2014 Oct 31, R Y Seedat commented:

Thank you for the positive comments regarding the study.

Lesotho is a different country, with different demographics and burden of disease and a different healthcare system with differences in accessibility, availability of services and referral protocols. It would thus not make sense to combine the data.

I don't think that the calculation of confidence intervals would be a valid statistical test since the study was not based on a population samples but on entire populations. While the incidences calculated are an underestimate, the calculation of confidence intervals would not assist in determining the degree to which the incidence is underestimated.