Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
The findings are amazing indeed!!
- Dr Fatai Kunle Salawu,FMCP,MWACP, Dr Laraba Ahmed Bello,Senior Registrar,Specialist Hospital,Maiduguri (30 July 2005)
-
Re: The findings are amazing indeed!!
- Charles DA Wolfe (1 August 2005)
-
Ethnicity definition issues
- Carole A Cull (19 August 2005)
-
Does the age of onset account for survival differences?
- Dougall McCorry (19 August 2005)
-
Re: Ethnicity definition issues
- Nigel C Smeeton (22 August 2005)
-
Re: Does the age of onset account for survival differences?
- Nigel C Smeeton (22 August 2005)
-
Challenging the hypothesis ethnicity affects survival after stroke.
- Dougall Mccorry (24 August 2005)
-
RE: Ethnicity and genetics
- John H. Lange (28 August 2005)
-
Re: Challenging the hypothesis ethnicity affects survival after stroke.
- Nigel C Smeeton (30 August 2005)
-
Survival of patients with stroke and Growth-Factors
- Salvador Vale (6 September 2005)
-
Are Black people more likely to survive a stroke than white people?
- Farzana Parveen Huq (1 July 2006)
-
Re: Are Black people more likely to survive a stroke than white people?
- Nigel C Smeeton (4 July 2006)
-
Challenging the post-stroke median survival times and the impact of stroke unit care on survival
- Farzana Parveen Huq (9 March 2007)
-
Prior Barthel <15
- C Lim (31 March 2009)
-
Re: Prior Barthel <15
- Nigel C Smeeton (1 April 2009)
|
|
|||
|
Dr Fatai Kunle Salawu,FMCP,MWACP, Consultant neurologist Neurology unit,Department of Medicine,Specialist Hospital,Maiduguri.Nigeria, Dr Laraba Ahmed Bello,Senior Registrar,Specialist Hospital,Maiduguri
Send response to journal:
|
We were really thrilled by your interesting findings.The black Africans used in the study were mostly to be more enlightened about the morbidity and probably the mortality of stroke than the average black African who has never travelled out of his fatherland. I would be interesting to compare how black Africans with first-ever stroke would fare compared to their white counterparts in Africa. Competing interests: ...Have you thought of comparing white stroke patients with black Africans living in Africa? |
|||
|
|
|||
|
Charles DA Wolfe, Professor of Public Health KCL, L ondon SE1 3QD
Send response to journal:
|
Thank you for your response. We have made estimates of the differences in survival in Barbados and London and found that the survival advantage in London is greater than in the Caribbean island. This may be, as we argue, because of a health migrant effect conferred on those that migrated to London. Charles Wolfe Competing interests: None declared |
|||
|
|
|||
|
Carole A Cull, Senior Medical Statistician Diabetes Trials Unit, OCDEM, Churchill Hosptial, Oxford OX3 7JL
Send response to journal:
|
You have classified patients as black including those of African origin together with those of 'other' origin. What were these other group? There is substantial evidence that those of Indian sub-continent origin have a different pattern of stroke occurrence from those of African sub- saharan origin, whether or not the latter have come via the Caribbean. What proportion of your black group were not African? Do the results vary if you take them out? Do you have any information about whether the patients are first or second generation UK residents? If so, does this make any difference to the results? Competing interests: None declared |
|||
|
|
|||
|
Dougall McCorry, research fellow neurology Walton centre, Liverpool L9 7LJ
Send response to journal:
|
With a mean age difference of 11 years for stroke onset between white and black groups there are clearly major differences between the two groups. The Kaplan-Meier survival estimates graph looks dramatic but is not corrected for age. When corrected for age there was no difference in survival in the patients aged 65 or less. The author comments there were modest differences in 65-74 age group and ‘dramatic differences’ in the 75+ age group. Could the author confirm how many black individuals this cohort consisted of? I presume this must be a small number. The mean age of the over 75s (plus 65-74 group) may be significantly lower in the black group and account for the difference in prognosis. Can the author provide these figures and reassure me this is not the case? Younger individuals are more likely to be managed on stroke units. The lower mean age of the black group may account for ‘better access’ for black individuals to stoke unit care. Age inequality may simply be a greater factor than ethnic inequality- comparison is meaningless without age matching. Competing interests: None declared |
|||
|
|
|||
|
Nigel C Smeeton, Lecturer in Medical Statistics KCL, London SE1 3QD
Send response to journal:
|
The 414 patients classified as Black comprised 270 Black Caribbeans, 135 Black Africans, with a further 8 either mixed white and Black Caribbean or mixed white and Black African. One patient stated their ethnicity as 'any other Black background'. Asians formed part of the 186 patients in the other groups/ missing category. Sample size/ power considerations meant that individual ethnic groups within this category (e.g. the 23 Bangladeshis) were too small for separate statistical analyses. Unfortunately, we do not have information on whether a patient is a first or second generation UK resident. Competing interests: None declared |
|||
|
|
|||
|
Nigel C Smeeton, Lecturer in Medical Statistics KCL, London SE1 3QD
Send response to journal:
|
Thank you for raising this important point. There were 76 blacks (mean age 81.9 years) and 886 whites (mean age 83.2 years) in the 75+ age group. I do not believe that such a modest difference in mean age could explain the wide difference in the survival curves. Competing interests: None declared |
|||
|
|
|||
|
Dougall Mccorry, research fellow neurology Walton Centre, Liverpool. L97LJ
Send response to journal:
|
If the hypothesis that ethnicity accounted for the survival differences after stroke is true then it would be expected to be seen across all age groups. The under 65 group had the largest number of patients and would be expected to show differences even in small; none were demonstrated. The hypothesis is based on a modest number of elderly blacks that survived better than a larger number of elderly whites. The mean age difference combined with differing case mix (small vessel versus large vessel disease) would be a more plausable hypothesis in my view. Secondly an elderly black population, many of whom are immigrants from developing nations, may not have accurate recording of date of birth. Were dates of birth verified? Competing interests: None declared |
|||
|
|
|||
|
John H. Lange, Environmental and Occupational Consultant Envirosafe Training and Consultants, P.O. Box 114022, Pittsburgh, PA 15239 (USA)
Send response to journal:
|
The paper by Wolfe et al. (1), further demonstrates the importance of ethnicity and disease. Numerous investigations have suggested that there is an important genetic influence that appears to follow ethnicity in the occurrence and outcome of disease (2,3). There are concerns that evaluation of such factors is expanding into the realm of eugenetics. However, this study (1) along with others (2,3) suggests that ethnicity and most likely other factors related to location of origin are more important in disease processes than previously considered. It is likely that these genetic “factors” will some day be considered highly relevant in the treatment and evaluating disease. It is important for us not to close our eyes to potential information that benefits man, even though some of the concepts have a painful past. References 1. Wolfe CD, Smeeton NC, Coshall C, Tilling K, Rudd AG. (2005). Survival differences after stroke in a multiethnic population: follow-up study with the South London stroke register. BMJ. 331:431, 2. Lanting LC, Joung IM, Machenbach JP, Lamberts SW, Bootsma AH. (2005). Ethnic differences in mortality, end-stage complications, and quality of care among diabetic patients: a review. Diabetes Care. 28:2280- 2288. 3. Nazer D, Thomas R, Tolia V. (2005). Ethnicity and gender related differences in extended intraesophageal pH monitoring parameters in infants: a retrospective study. BMC Pediatrics 5:24. Competing interests: None declared |
|||
|
|
|||
|
Nigel C Smeeton, Lecturer in Medical Statistics KCL, London SE1 3QD
Send response to journal:
|
We are not looking at ethnicity specifically but which of a multiplicity of factors that confound survival and indeed the label ethnicity. We have adjusted for all the factors you mention in the model and the results do show that Blacks do have a survival advantage that is NOT explained by age, case mix or other factors. Of course there is residual confounding that may explain this observation. The effect of ethnicity would not necessarily be seen across age groups and indeed we do discuss the effects of migration, which will have been seen in the older cohorts who may have a survival advantage over those born in the UK. The ethnicity effects will have age, period and cohort effects which will determine different risks for different subgroups. In Barbados and South London death certification is comparable and complete. Competing interests: None declared |
|||
|
|
|||
|
Salvador Vale, Senior Researcher Indesalud, Campeche, Mexico. CP 01420
Send response to journal:
|
Wolfe et al (1) found a residual unexplained overall survival advantage after a stroke in their black-ethnicity group. They propose that this difference requires a more detailed description of case mix and stroke subtype, including aetiological subtype. I wish to comment analogue data derived from cancer research. Prostate cancer is prevalent in black men, and part of the explanation is that they have a greater Insulin Growth Factor-I/Insulin- Growth Factor-Binding Protein-3 (IGF-1/IGFBP-3) ratio than white men across all ages, with an adjustment for height. Furthermore, age has an inverse correlation with IGF-I and IGFBP-3 levels in whites, but no such relationship in blacks. Consequently, the differences in IGF levels between blacks and whites may explain some of the racial disparity in prostate cancer risk (2). Thus, the increased activity of growth factors that can be a negative condition for black patients in relation to cancer may be an advantage in neuronal survival after stroke. Moreover, it has been shown that there is a growth factor - estradiol interaction, which may be the cause that estrogens play a central role in the regulation of cell proliferation and survival in numerous mammalian tissues and that they are implicated in the aetiology of several types of cancer (3). Consequently, at least in theory, locally produced estrogens (which can increase endothelial progenitor cells and also produce antiapoptotic effects leading to accelerated vascular repair and decreased neointima formation [4, 5[)can be upregulated even in the endothelium of men, by the augmented activity of IGF-1 plus the decreased activity of its binding protein. A better survival after neuronal insult may be the consequence. These observations may be relevant for the clinician if further research can disclose these molecular peculiarities in stroke survivors. 1.- Wolfe CD, Smeeton NC, Coshall C, Tilling K, Rudd AG. Survival differences after stroke in a multiethnic population: follow-up study with the South London stroke register. BMJ 2005;33:431. 2.- McGreevy K, Hoel B, Lipsitz S, Bissada N, Hoel D. Racial and anthropometric differences in plasma levels of insulin-like growth factor I and insulin-like growth factor binding protein-3 Urology 2005;66:587- 592. 3.- Avola R, Mignini F, Mazzone V, Fisichella A, Zaccheo D, Tomassoni D. Growth factor-estradiol interaction on DNA labeling and cytoskeletal protein expression in cultured rat astrocytes. Neurosci Lett. 2004;358:177-180. 4.- Strehlow K, Werner N, Berweiler J, Link A, Dirnagl U, Priller J, Laufs K, Ghaeni L, Milosevic M, Bohm M, Nickenig G. Estrogen increases bone marrow-derived endothelial progenitor cell production and diminishes neointima formation. Circulation. 2003 24;107:3059-3065. 5.- Imanishi T, Hano T, Nishio I. Estrogen reduces endothelial progenitor cell senescence through augmentation of telomerase activity. J Hypertens. 2005;23:1699-706. Competing interests: None declared |
|||
|
|
|||
|
Farzana Parveen Huq, Student Not working
Send response to journal:
|
Table 2 presents the effect of black ethnicity on post-stroke survival after adjustment for the effect of all other variables and interactions with age and prior Barthel score. I assume that the paragraph 7 of the results section relates to Table 2. It was mentioned in paragraph 7 that ethnicity was highly significant (hazard ratio 24.80, 95% confidence interval 4.70 to 130.87). From Table 2, I understand this result means that white patients had a substantial survival advantage over black patients, as the hazards ratio is greater than one. In the same paragraph, the authors go on writing that “but when we adjust for the interaction between ethnicity and age and prior Barthel score, black people aged more than 65 and with a prior Barthel score of at least 15 had a substantial survival advantage over white people.” I think the quoted hazards ratio (24.80) for black ethnicity, presented in Table 2, was after adjustment for these two interactions. If so, then black patients did not have survival advantage over white patients but the opposite. Can the authors explain it and reassure readers of this article that this is not the case. Competing interests: None declared |
|||
|
|
|||
|
Nigel C Smeeton, Lecturer in Medical Statistics KCL London SE1 3QD
Send response to journal:
|
The hazard ratio value of 24.80 for 'Black ethnicity' in Table 2 is obtained following adjustment for the remaining terms in the Table. It represents the nominal ethnicity effect for an individual aged zero (i.e. at birth). From the 'ethnicity by age' interaction term this nominal hazard ratio value for being Black should be multiplied by 0.97 for each year of life, which reduces it considerably. The key to the Black survival advantage, however, is from the 'ethnicity by prior Barthel' interaction term. For patients who are mobile prior to their stroke (the great majority) this reduces the Black ethnicity effect on the hazard ratio to less that 1. For these patients there is hence a survival advantage for those with Black ethnicity. Out of interest, inspecting our data set we found that as this model suggests, for the small number of young adult stroke cases there is no survival advantage for having Black ethnicity. Competing interests: None declared |
|||
|
|
|||
|
Farzana Parveen Huq, Student Not working
Send response to journal:
|
1. Some considered censored times were longer than the actual length: The survival time was measured from date of stroke to date of death. Patients with no record of death were censored at the end of 2003. I assume, those patients who died after the end of 2003 were considered censored at the end of 2003 as well. What about the loss to follow up patients? I am sure that a study with 8-years of follow-up will surely suffer from loss to follow up. It seems to me that patients were considered at risk of death (after being lost to follow up) until the end of 2003. It was a very liberal assumption, which generated some censored times longer than their actual lengths. It could be more appropriate to censor the lost to follow up patients at their last date of contact. 2. Mean and Median survival times were wrong! Authors reported mean (median) survival was 31.2 (20.0) months for white people and 39.3 (33.7) months for black people. Depending on whether the largest observed analysis time was censored or not, reported mean could be underestimated unless the mean survival time was calculated by exponentially extending the survival time to zero. It seems from presented K-M curves that there was more censoring among the blacks than whites. It does not seem to me that author used exponential extension for calculating mean. On the other hand, reported median survival times do not match with authors K-M curves and hence they must be wrong! The K-M curves show that the median survival should be at least 24 months for white people and at least 84 months for black people. Table 2 shows that there were significant interactions of ethnicity with age and prior barthel. Therefore, median survival times of black and white patients were expected according to age and prior barthel categories. 3. Residual confounding may partially explain the survival difference: Authors dichotomised continuous predictors prior Barthel and Glasgow coma score (GCS) in their Cox model in Table 2. However, recent study suggests that dichotomization of continuous data is unnecessary for statistical analysis and in particular should not be applied to explanatory variables in regression models as it may create a considerable loss of power and residual confounding (1). GCS is one of the most influential predictors of post-stroke survival (2). However, it was categorized and used as a stratifying variable rather than as a covariate in the Cox model. Therefore, the effect of GCS was not adjusted fully to see the independent effect of ethnicity. Table 1 shows that there were considerable missing values in most of the considered predictors. Authors seem to have done complete case analysis, in Table 2, without any missing data treatment. However, there is nothing wrong to do that if the missing data mechanism is completely at random (MCAR) (3). Authors did not present any results of sensitivity analysis to demonstrate that their data fulfil the MCAR assumption. I think that the values of the predictors corresponding to patients who died shortly after their stroke were more likely to be missing. If the percentage of black people with at least one of the predictors’ value missing was more than that of white, then the ethnicity effect was over estimated. 4. Stroke unit care was likely to be beneficial: Table 2 shows that patients who were not admitted had significant survival advantage than those who admitted to hospital. Did the non-admitted patients have milder strokes or they died directly after the stroke and hence did not get time to go to hospital? On the other hand, admission to a stroke unit had no impact on survival! I think, this was simply because authors combined non- admitted patients with non-stroke unit hospital patients in the reference category! If the comparison were with respect to non-stroke unit hospital patients only, stroke unit care was likely to be found as beneficial for survival. 5. Does atrial fibrillation influence survival? In Table 2, authors presented the p-values. However, for those variables with more than 2 categories, overall p-values were important in addition to the p-values of their categories. For example, from Table 2, we don’t know whether atrial fibrillation had any independent effect on the survival. References: 1. Royston R, Altman DG, Sauerbre W. (2006). Dichotomizing continuous predictors in multiple regression: a bad idea. Statistics in Medicine. 25: 127-141. 2. Handschu R, Haslbeck M, Hartmann A, Fellgiebel A, Rabas PK, Schneider D, Berrouschot J, Erbguth F and Reulbach U. (2005). Mortality prediction in critical care for acute stroke: Severity of illness-score or coma-scale? 252: 1249-1254. 3. http://www.lshtm.ac.uk/msu/missingdata/jargon_web/ node7.html Competing interests: None declared |
|||
|
|
|||
|
C Lim, student QMUL EC1A
Send response to journal:
|
In table 2, the interaction between ethnicity by prior Barthel <15 is 'HR 0.17'. Please indicate whether was the author meant to say Barthel score >15 instead? The maximum Barthel score is 20, score between 15-20 represent mild dependence to independent. If the statement (Barthel score <15) is true, we would see a Black person having a survival advantage if he has prior moderate to severe ADL's dependence. This would not correspond to what was stated in section 'Methods', paragraph '7', as it mentioned that 'black people aged >65 and with prior Barthel score of at least 15 had a substantial survival over white people'. Competing interests: None declared |
|||
|
|
|||
|
Nigel C Smeeton, Lecturer in Medical Statistics King's College London SE1 3QD
Send response to journal:
|
Thank you for pointing this out. The '<' symbol in the interaction variable should read '>='. The Barthel category main effect was originally intended to be published as "Prior Barthel >= 15" along with the reciprocal of the hazard ratio displayed in the paper. The reporting of the main effect was changed at a late stage but in the process both '>=' signs were amended. Competing interests: None declared |
|||