HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

Correspondence to:

ARTICLES: Eric J. Sorenson, Jayawant Mandrekar, Brian Crum, and J. Clarke Stevens Effect of referral bias on assessing survival in ALS Neurology 2007; 68: 600-602 [Abstract] [Full text] [PDF]

Correspondence: Submit a response to this article

Correspondence published:

Effect of referral bias on assessing survival in ALS

Giancarlo Logroscino, Ettore Beghi, Orla Hardiman, Adriano Chio, J. Douglas Mitchell, Robert J. Swingler, Brian Traynor for EURALS   (24 May 2007)

Reply from the Authors

Eric J. Sorenson, Jay Mandrekar, Brian Crum, J. Clarke Stevens   (24 May 2007)

Effect of referral bias on assessing survival in ALS 24 May 2007
Giancarlo Logroscino, Department of Epidemiology, Harvard School of Public Health 677 Hungtington Avenue, Boston, MA 02115, Ettore Beghi, Orla Hardiman, Adriano Chio, J. Douglas Mitchell, Robert J. Swingler, Brian Traynor for EURALS Send Correspondence to journal: Re: Effect of referral bias on assessing survival in ALS glogrosc{at}hsph.harvard.edu Giancarlo Logroscino, et al.	We read with interest the article by Sorenson et al describing the effect of referral bias on assessing survival in amyotrophic lateral sclerosis (ALS). [1] The authors state that referral bias is the cause of misleading results in both cohort and case-control observational studies. However, we believe that the Mayo Clinic data presented in this paper is skewed. Sorenson et al report that the population-based cohort from Olmsted County have a shorter survival time from diagnosis (18 months) than the referral cohort of patients residing in the wider 150-mile area surrounding the Mayo Clinic (29 months). However, population-based studies have consistently shown that the median survival time from diagnosis within the general ALS population is closer to thirty months.[2-5] There are several reasons why the population-based median survival estimates presented by Sorenson et al conflict with the previous data. The population-based registries of Europe utilize multiple sources of information to ascertain cases within much larger reference populations (around 4 million).[5] Population-based studies that ascertaining all ALS cases within a large geographically defined area are more likely to be representative of the whole spectrum of disease severity and clinical course. In contrast, a single source of information (i.e. the Mayo ALS clinic) was used to ascertain cases within Olmstead County which has population of only 124,000. Secondly, it is unclear whether the 42 patients that make up the local population cohort of Olmstead county represent incident cases that were newly diagnosed cases during the three-year study period or whether data from prevalent cases diagnosed were also included. Based on the low incidence of ALS (2.1 per 100,000) and the small population of Olmstead County, the expected number of new cases during a three year period is only 7.8. We strongly suspect that prevalent cases were included in the local population cohort. If this is the case, the “population-based” survival estimate of this study would be furthermore biased. Based on these observations, we believe that the magnitude of the bias in the evaluation of survival times observed by Sorenson et al may be specific of Olmsted County and reflect the special referral characteristics of Mayo Clinic. Population-based research in larger reference populations is needed to address the issues of referral bias in a quantitative way because this is relevant not only for analytic research but for the design of future clinical trials for ALS. References 1. Sorenson EJ, Mandrekar J, Crum B, Stevens JC. Effect of referral bias on assessing survival in ALS. Neurology. 2007;68:600-602. 2. Traynor BJ, Codd MB, Corr B, Forde C, Frost E, Hardiman OM. Clinical features of amyotrophic lateral sclerosis according to the El Escorial and Airlie House diagnostic criteria: A population-based study. Archives of Neurology. 2000;57:1171-1176. 3. Chio A, Mora G, Leone M, et al. Early symptom progression rate is related to ALS outcome: a prospective population-based study. Neurology. 2002;59:99-103. 4. Zoccolella S, Beghi E, Palagano G, et al. Prognosis of amyotrophic lateral sclerosis in Southern Italy: results from a population-based registry, SLAP (Sclerosi Laterale Amiotrofica-Puglia). Neurology. 2006;66(5; Suppl 2):A231. 5. Logroscino G, Traynor BJ, Hardiman O, et al. EURALS Descriptive epidemiology of amyotrophic lateral sclerosis: new evidence and unsolved issues. JNNP. In press Disclosure: The authors report no conflicts of interest.
Reply from the Authors 24 May 2007
Eric J. Sorenson, Mayo Clinic Department of Neurology, 200 1st St. SW, Rochester, MN 55905, Jay Mandrekar, Brian Crum, J. Clarke Stevens Send Correspondence to journal: Re: Reply from the Authors sorenson.eric{at}mayo.edu Eric J. Sorenson, et al.	We thank the correspondents for their comments. They raise a number of important points which do convey the limitations of our study. They are correct in assuming this was a prevalence sample. However, after reviewing the correspondence, the subject numbers are high even for a prevalence sample. I have reviewed our data set and the cases extend over a longer period than that specified. The dates of diagnosis range from 1990 for the Olmsted County population and from 1998 for the referral population. A number of cases were added after peer review on methodological suggestion by the reviewers. This included a number of Olmsted County patients from the longer time period. Their comments that published survival time from diagnosis is consistently 30 months is not accurate. There are several population-based studies demonstrating a median survival similar to our population’s median of 18 months. The articles they cite supporting this assertion do not show a median survival from diagnosis of 30 months. The two cited Italian populations have a published median survival time from diagnosis of 19 and 18 months respectively.[3,4] The third cited article reports a median survival of 27 months but this is from symptom onset not diagnosis.[2] A subsequent study from the same investigators and same population (Irish ALS register) reported a median survival of 14 months with the use of riluzole and 10 months without.[6] The fourth study cited is currently in press and I have no access to it. Additionally, from the Scottish population, the median survival from diagnosis is approximately 15 months (median survival from 1st symptom was 25 months, with median time from 1st symptoms to diagnosis of 10 months).[7] From Harris County, Texas there is a reported median survival of 21 months.[8] We agree that our data, derived entirely from one center, may limit its generalizability to the broader ALS population. However a similar study published from Harris County, Texas found similar results.[8] That study reported a 5-year survival rate of 4% for the local patients (compared to 12% in the Olmsted County population), and 32% in their referral population (compared to 30% in our referral population) suggesting that referral bias is not limited to the Mayo Clinic. We concur with the need for testing the potential effects of referral bias in larger reference populations. References 6. Traynor BJ, Alexander M, Corr B, Frost E, Hardiman O. An outcome study of riluzole in amyotrophic lateral sclerosis: A population-based study in Ireland, 1996-2000. Journal of Neurology. 2003; 250: 473-479. 7. Forbes RB, Colville S, Cran GW, Swingler RJ. Unexpected decline in survival from amyotrophic lateral sclerosis/motor neurone disease. Journal of Neurology Neurosurgery and Psychiatry. 2004; 75: 1753-1755. 8. Lee JR, Annegers JF, Appel SH. Prognosis of amyotrophic lateral sclerosis and the effect of referral selection. Journal of the Neurological Sciences. 1995; 132: 207-215. Disclosure: The authors report no conflicts of interest.