Comparative statistics of illness prevalence from country to country provide potential clues into the aetiology of conditions. In many respects, such studies are the lifeblood of epidemiology, allowing associations to be made between environment and illness. Estimating the number of people worldwide who are affected with Parkinson’s is, at best, haphazard. Government supported registries for Parkinson’s are simply not available (Tanner, Brandabur & Dorsey, 2008) and therefore figures from country to country are difficult to compare. Inevitably worldwide figures make broad sweeping generalisations about prevalence from country to country. Typical figures quoted in the literature are “more than 6 million” , 6.3 million , to 7 – 10 million. The literature often assumes a prevalence of one in 250-300 for Parkinson’s.
It is possible however to get some idea of the relative prevalence of Parkinson’s in different countries. Using data from IMS or treatment of Parkinson’s, combined with demographic data from World Factbook, it is possible to draw some comparisons. IMS data on Parkinson’s treatment days is available. A treatment day is essentially one person treated for one day. So 100 treatment days could be one person treated for 100 days, 100 people treated for one day or 10 people treated for 10 days. But if we make an assumption that anybody treated at all is treated full-time, then it is possible to establish the number of people within that country who are estimated to have Parkinson’s. As well as assuming that anybody treated at all will be treated fully, it also makes the assumption that there is a link between diagnosis and treatment. In other words it assumes that patients diagnosed with Parkinson’s will be treated. This may well be incorrect, but we’ll get to that later. For the time being let’s call it treatment prevalence.
The treatment prevalence also takes no account of the treatments used. A patient who receives levodopa is counted exactly the same as a patient who receives a cocktail of levodopa, dopamine agonist, MAO-B inhibitor, COMT inhibitor and amantadine. So we have a simple estimate of the number of people with Parkinson’s in each country. With the knowledge of the population of the country (data widely available and fairly accurate), we can derive the prevalence of Parkinson’s in each country. The list is not exhaustive. I compared 16 countries, of widely differing economic status, ethnicity and industrialisation. In alphabetical order these were Argentina, Australia, Canada, China, the Dominican Republic, Hong Kong, Latvia, Pakistan, Puerto Rico, Romania, South Africa, Saudi Arabia, Sweden, the United Kingdom, the United States and Venezuela. To get some idea of health status in each country, I have also listed the life expectancy. The gross domestic product per capita gives a measure of economic status.
Table 1 (left) shows the data on treatment prevalence (per 1000 population), life expectancy (years) and gross domestic product per person (US dollar equivalents). Even a quick examination of the table reveals some striking inequalities among the countries listed. The lowest treatment prevalence of Parkinson’s, translating to a prevalence of one in 7123, is found in Saudi Arabia, while the highest, at a prevalence of 1 in 210, is Hong Kong. The highly westernised countries (USA, Canada, Australia, UK and Sweden) shows generally similar prevalences of 1 in ~260. So what do these data actually show us? Is there really a nearly 30 fold difference in the prevalence of Parkinson’s between Saudi Arabia and the USA? Is Parkinson’s underdiagnosed in Saudi Arabia or overdiagnosed in the United States? Are there ethnic, cultural or socioeconomic reasons that account for these differences? What factors might affect these statistics?
Firstly there is the obvious assumption made in translating the treatment days into numbers of people treated. This is probably not an unreasonable assumption since anyone who is treated in the first place is likely to stay on treatment and there is little therapeutic value in stopping and starting treatment. It is a bigger conceptual leap to assume that the number of patients treated is the same as the number of patients diagnosed. It’s highly possible that the costs of treatment may be such that patients may receive a diagnosis of Parkinson’s but be unable to pay for treatment. Even in the West, with highly subsidised healthcare or comprehensive insurance programs, these issues still remain. In poorer countries, it’s likely that their impact is more severe still. Comparison of the Hong Kong and China statistics is illuminating in this regard since there is obviously much common ethnicity between these populations but a 20 fold difference in treatment prevalence. It is a reasonable guess that economic circumstances are at least a partial contributor to this difference. If one factors personal economic circumstances into the equation, some of these differences can be explained.
Figure 1 (right) shows the treatment prevalence of Parkinson’s as a function of the GDP per person of each country. On the whole there is a reasonable correlation. Poorer countries have a lower prevalence of treatment for Parkinson’s. Wealthier countries show the reverse. However there are two striking anomalies even in this. Romania, by any standards a relatively poor country, has among the highest treatment prevalence for Parkinson’s noted. Saudi Arabia shows the opposite pattern. And these data points to the weakness of statistics taken in isolation. For instance, the Romanian data can be interpreted in a number of different ways. One possibility is that Parkinson’s is thoroughly diagnosed and treated. Alternatively, the data may suggest that there is a higher prevalence due to some environmental factor. Or both. Or neither. This is the danger of data taken in isolation. The same applies to the Saudi Arabian data which could, at one extreme, be taken to indicate a woeful failure of the diagnosis and treatment process or, at the other extreme, amount to a vindication of the Arabic diet and preventive health care. The likelihood is that neither provide an adequate simple explanation. To some extent, the effects of healthcare in each country can be examined by looking at life expectancy. Again this is little more than a crude measure since it can be affected not only by lifestyle but also by ethnicity. Nonetheless, as a crude measure, it suffices.
And if one excludes the data from Pakistan, where life expectancy is more than 15 years less than any other country examined, the data shows a reasonable correlation (figure 2). Countries where life expectancy is longer have a higher treatment prevalence of Parkinson’s. This may be a reflection of superior health care in this countries and/or the fact that Parkinson’s is still predominantly a disease of older age.
Above all the data show that one should be careful in drawing comparisons between countries. Although there is a degree of standardisation in the West in terms of diagnostic criteria and treatment algorithms, these are not universally accepted were implemented in other countries. In part these differences may be partly due to socioeconomic considerations. It is hard to pick out any meaningful epidemiological differences against this strong socioeconomic or geopolitical backdrop. Nevertheless, whether the differences are biological or political, there remains striking differences between different countries on the prevalence of treated Parkinson’s worldwide. Estimates of 6 to 10 million patients worldwide are little more than guesses. In any case, the burden of Parkinson’s is felt by the individual, not the world.