Natural history studies of Parkinson’s disease with adequate duration of follow up are scarce and fraught with difficulty due to selection bias and retrospective assessment in hospital series, confounding effects of comorbidity and problems of diagnostic accuracy.1 The pivotal study by Hoehn and Yahr2 on a cohort of 672 patients with “primary parkinsonism” came up with a rather bleak prognosis, with 61% of patients severely disabled or dead after 5 to 9 years of follow up, increasing to more than 80% of those who were followed up for more than 10 years. Overall mortality was increased to about threefold the expected rate in the general population. Such poor longterm outcome is thought to reflect the history of idiopathic Parkinson’s disease in the prelevodopa era with some added negative bias due to less stringent diagnostic criteria used in those days. Early postlevodopa mortality studies in Parkinson’s disease indeed found mortality ratios of 1.5 or less, rising again, however, with extended follow up, suggesting that levodopa reduces excess mortality early in the course of Parkinson’s disease but fails to prevent increased mortality in the long term.3

This general trend is also confirmed in the 10 year prospective follow up results on progression and mortality of the Sydney multicentre study of Parkinson’s disease now published by Hely et al (this issue, pp 300–7). Regular follow up of this cohort for a maximum of 13 years has provided valuable data on disease progression and mortality in those 126 patients in whom the original diagnosis could be upheld. By 10 years 38% had died, rising to 48% by last follow up, yielding a standard mortality ratio for the whole cohort of 1.58, which is similar to many of the previously published postlevodopa studies.3 Significant risk factors for increased mortality included old age at onset, rapid initial progression on the Hoehn and Yahr scale, and—surprisingly—initial randomisation to bromocriptine. Although this finding certainly does not support claims of possible neuroprotective effects of bromocriptine or dopamine agonists in general4 it is of limited relevance. Only very few patients originally randomised to bromocriptine continued such monotherapy for longer than 1 year and all patients taking bromocriptine had been switched to combined treatment with levodopa by year 5. So unfortunately the longterm outcome data of the Sydney study do not allow for conclusions about differential effects of levodopa monotherapy versus bromocriptine monotherapy versus combined treatment on longterm progression and prognosis.

The biggest surprise in the Sydney study, however, is that the percentages of patients severely disabled or dead after 10 years of follow up are very similar to the figures originally reported in the Hoehn and Yahr study in the prelevodopa area. Does this mean that dopaminergic replacement with levodopa, dopamine agonists, or combinations of both has not significantly altered the longterm outlook for people with Parkinson’s disease? Probably not. As the authors admit, their patients may have been undertreated due to the initial design of the study as a comparative trial of low dose levodopa versus low dose bromocriptine. Their outcome may not be representative for the treated parkinsonian population at large. By contrast the recent 9 year follow up results of the DATATOP cohort of patients showed supernormal life expectancy with a standard mortality ratio of 0.9.5Such discrepancies in outcome between prospective follow up studies over similar time periods are likely to reflect differences in baseline severity and comorbidity and possibly treatment strategies. Idiopathic Parkinson’s disease is not a prognostically uniform entity; elderly patients with comorbid dementia and cerebrovascular and heart disease face a high risk of significant disability or death after 10 years, contrasting with a near normal life expectancy in the younger onset patient without dementia or other significant comorbidity and optimal treatment under specialist supervision.