HOW MANY AUSTRALIANS HAVE THYROID CONDITIONS
When my wife was diagnosed with Hashimoto's thyroiditis, one of my first points of interest as an actuary was what the incidence and prevalence rates were for the disease. (So that there is no confusion, incidence rate means the percentage of people who develop the disorder in a year, while prevalence rate means the percentage of people who have the disorder.) I have trawled through medical references and the internet for incidence and prevalence studies. I have found a number, but none which I regard as actuarially reliable for estimating population prevalence rates. I find this disturbing because it is impossible to develop effective public policy with respect to thyroid disorders without a knowledge of how many people are affected. How, for instance, can the Health Insurance Commission determine whether the amount of money spent on thyroid function tests is too high or too low without a clear idea of the number that should be ordered each year?
I decided to apply a little actuarial science to the studies that were available to try to estimate how many Australians have thyroid disorders - an actuarial doodle. The results are indicative rather than definitive.
The studies have generally used very small samples or samples not representative of the population. Some suffer from both problems. Most have only produced overall prevalence or incidence rates and have not determined the rates at specific ages. These results are not useful for estimating population prevalence even if they satisfy a medical purpose.
The other difficulty which besets this subject is the difficulty in deciding who actually has the condition. The problem is that irregularities in blood chemistry range from minor to significant and people with minor irregularities can have pronounced symptoms whilst others with significant irregularities can have less severe symptoms. Researchers have to draw the line somewhere. The problem is that they do not all choose the same definition. This obviously makes it very difficult to combine or compare results.
The only study that comes close to providing reliable results is the Whickham Survey1 which was conducted in England over 20 years from 1972. The results from this study with an original sample of 2,779 are still subject to large statistical errors. A sample some 10 to 20 times larger would have been needed to produce reliable results. Nonetheless, this is the best we have so what does it mean?
The Whickham study determined incidence rates by age for spontaneous (ie not caused by treatment) hypothyroidism and hyperthyroidism for women and overall incidence rates for men. The authors also reviewed a large number of earlier studies. I have taken these rates, smoothed the statistical fluctuations and applied them to the Australian population.2 The results for the prevalence of overt hypothyroidism and hyperthyroidism are given in the table.
As an example of what these results mean, consider the age group 51 to 60. In this age group some 107,000 women suffer from hypo- or hyperthyroidism. This represents some 11.25% of women in this age group. There are only some 25,000 men in the same age group with these conditions or 2.5%. There are therefore 4.2 times as many women with the conditions than there are men.
The very high rate for the over 80's must be treated with some caution as the incidence rates at these ages have large statistical errors. The errors in incidence rates for the younger ages also accumulate to produce bigger errors in the prevalence rate for the older ages.
The overall result is that some 850,000 Australians can be expected to be suffering hypo- or hyperthyroidism. This group represents some 7.5% of women and some 1.5% of men. Further analysis also indicates that just over 40,000 new cases will develop each year - approximately 35,000 women and 5,500 men.
There are other forms of thyroid disease which are not included in these results. Cancer is ignored as are goitres and nodules which are not accompanied by abnormal blood chemistry. People with sub-clinical conditions - ie their blood hormone levels do not fit the definition used - are also excluded. The total number of thyroid sufferers will therefore be bigger than these results indicate.
The results highlight the fact that thyroid disorders are predominantly female conditions - especially of the over 50's. The inevitable consequence of this is that the prevalence of thyroid disorders will rise markedly over the next 20 years as the Baby Boomer generation moves into this age group.
Another interesting point arises when these prevalence rates are compared to the results of the 1995 National Health Survey conducted by the Australian Bureau of Statistics.3 This survey reported that only 4% of women indicated that they had thyroid disorders against the 7.5% who are likely to have the disorders. This suggests that nearly half the thyroid sufferers are undiagnosed!
I also feel that I should comment on a prevalence study in Colorado which was published in February 2000.4 This study, which has received a lot of publicity on the internet, reported average prevalence rates of 9.5% for hypothyroidism and 2.2% for hyperthyroidism. The combined prevalence of 11.7% is much higher than earlier studies. This high result and the apparent statistical reliability of the study (25,000 people were tested) have been used to promote the cause that thyroid disease is much more widespread than medical authorities are prepared to admit.
The study cannot support this claim. The results show categorically that the population prevalence rate cannot be as high as 11.7%. The reason is very simply that although a large number of people were tested they represent an extremely biased sample. The results are, in fact, of no use for estimating population prevalence rates.
The age and sex distribution of the sample is a primary cause of bias. The sample was much older (median age 56) than the general population (median age 32). 56% of the sample were women whereas women represent only 51% of the population. Correcting for this bias reduces the combined prevalence rate to around 7%. This is still higher than other studies, but is still not meaningful because of the final source of bias.
The sample was not random. It was made up of attendees at a 1995 Colorado Health Fair who volunteered to have their thyroid status tested. This methodology is guaranteed to produce results which are too high. Unfortunately the level of bias cannot be calculated so the results cannot be adjusted and are effectively unusable. The only thing that the study tells us is that the average prevalence of thyroid disease is definitely less than 7% of the population.
I find it disappointing that there is no reliable estimate of incidence and prevalence rates for thyroid disease in Australia. It would be heartening to see a collaboration between medical and actuarial professionals to produce a meaningful set of statistics so that there is a proper basis for public health policy. We would need government or commercial support as a study over 2 to 3 years with some 25,000 subjects would not be cheap. I live in hopeful anticipation.
(The answer to the question regarding thyroid function tests is that Australia should be spending some $90m per year - at $60 per test.)
Alun Stevens MSc FIAA, is an actuary who manages his own consultancy.
(And for those who don't know, actuaries are the people who calculate rates of death etc., and then use these to fix the price for insurance - amongst other things.)
This article is published along with a number of other articles dealing with Hypothyroidism in our newsletter
Thyroid Flyer Volume 1 No 3, July 2000
and is available for download on our download page.
This article can be reproduced provided it is reproduced in full, acknowledges the source and is not sold for profit.
(c) Copyright 2000, Thyroid Australia Limited ABN 71 094 832 023
333 Waverley Road, Mount Waverley, 3149, Australia
Thyroid Australia Home-www.thyroid.org.au