Recent eLetters
Displaying 1-10 letters out of 225 published
-
Validity of self-reported prevalent cases of stroke and acute myocardial infarction in the Spanish cohort of the EPIC study
Submit responseThe aim of the authors was to assess the validity and agreement of self-reported prevalent cases of stroke and AMI in the Spanish cohort of the European Prospective Investigation into Cancer and Nutrition (EPIC). They calculated sensitivity, specificity, positive predictive values and ? statistics. The sensitivity of self-reported prevalent cases of stroke was 81.3% and that for AMI was 97.7%. The positive predictive value was 22.2% and 60.7% for stroke and AMI, respectively. The agreement between self- report questionnaire results and medical records was substantial (?=0.75) for AMI but not for stroke (?=0.35).1 To scientifically assess the accuracy (validity) of a test, there are 7 estimations named Sensitivity, Specificity, Positive Predictive Value (PPV), Negative Predictive Value (NPV), Likelihood ratio positive, LR+ (true positive/false positive), Likelihood ratio negative, LR- (false negative/true negative) and finally Odds ratio, OR (true results /false results).2 Considering limitations of the first 4 estimations, preferably the last 3 estimations are being reported. However, due to the different range of these estimations [(LR+ from 1 to infinity; the higher, the better) (LR- from 0 to 1; the closer to the zero, the better) and OR greater than 50 indicates a valid test), usually two different tests are being evaluated compared to a gold standard. 2 Regarding agreement, to compute kappa value, just concordant cells are being considered, whereas discordant cells should also be taking into account in order to reach a correct estimation of agreement (Weighted kappa).2-4 It is crucial to know that there is no value of kappa that can be regarded universally as indication good agreement. Statistics cannot provide a simple substitute for clinical judgment. Two important weaknesses of k value to assess agreement of a qualitative variable are as follow: It depends upon the prevalence in each category and also depends upon the number of categories. So it is obvious that the less our categories, the higher will be our kappa value which can easily lead to misinterpretation.2-4
S.Sabour, MD, PhD
References: 1- Mach?n M, Arriola L, Larra?aga N, Amiano P, Moreno-Iribas C, Agudo A, Ardanaz E, Barricarte A, Buckland G, Chirlaque MD, Gavrila D, Huerta JM, Mart?nez C, Molina E, Navarro C, Quiros JR, Rodr?guez L, Sanchez MJ, Gonz?lez CA, Dorronsoro M. Validity of self-reported prevalent cases of stroke and acute myocardial infarction in the Spanish cohort of the EPIC study. J Epidemiol Community Health. 2012 May 10
2- Epidemiology, biostatistics and preventive medicine, Jeckel, 1st edition, 2008 3- Modern Epidemiology, K. Rothman, 3 rd edition, 2010 4- Clinical Epidemiology, D.E Grobbee, 1st edition, 2010
Conflict of Interest:
None declared
-
Health inequalities and IMR
Submit responseI was interested to read your letter/article in the Journal of Epidemiology and Community Health, and your conclusion that there were significant reductions in IMR. You wondered whether this might have been due to interventions such as Sure Start and the Health in Pregnancy grant. I would be surprised if the latter played any significant part, as it came far too late in pregnancy to do anything significant and, anecdotally at least, was often spent on items that would not contribute to health outcomes. As a midwife and health visitor, it seemed the most ill-thought- out piece of spending the government put in place, spending that could so easily have been better used earlier in pregnancy, if directed more specifically - maybe to provide maternal folic acid and Vitamin D freely to all pregnant women.
If we are seeing an improvement in inequalities in IMR, I would submit that higher breastfeeding initiation and continuation rates, and the investment that the government of the time put into supporting breastfeeding (largely withdrawn now), could well be a significant contributing factor. Recent DH data comparing admissions and breastfeeding rates show a significant reduction in admissions of infants to hospital for conditions such as chest infections, bronchiolitis and gastroenteritis in areas where breastfeeding rates are high, even where deprivation levels are also high. Modelling by Bartick and Reinhold (2010) in the US showed that, if 90% women followed the recommendations to breastfeeding exclusively for 6 months, over 900 excess infant deaths would be prevented each year, as well as $13 billion annually. I believe that similar modelling is being undertaken in the UK, and I would imagine it might well show similar results, even if on a smaller scale.
Conflict of Interest:
I am Infant Feeding Coordinator for a London borough, tasked with leading the borough to UNICEF Baby Friendly accreditation, to ensure that all mothers, however they choose to feed their babies, receive the information and support they need to do that appropriately and successfully.
-
"Green cities and mortality: is migration the answer?"
Submit responseNumerous studies conducted have found evidence for a positive relation between green space in peoples environment and self reported indicators of morbidity and mortality (Lee and Maheswaran, 2011). The authors in this study have found mortality from all causes to be higher in greener cities. Attempts have been made to adjust for factors which may act as confounders, however other community level factors have not been taken into account which influence mortality levels like health care access and crime rates in the cities. Part of the relation found in the study between green space and all causes mortality may be explained by indirect selection. Indirect selection takes place when people with certain characteristics related to well being (such as income) tend to live in a greener environment. Migration flows are related to sociodemographic characteristics such as age, income and education (Maas et al, 2009). Possible migration of the elderly post retirement to greener areas may most likely skew all cause mortality levels and which needs investigation. Moreover, no association was found between greenness and mortality from diseases of lung cancer, heart disease, diabetes and motor vehicle fatalities when analyzed individually. Analysis of the mortality data based on the city of residence may prove more helpful in understanding the association between greenness and health, given that many diseases develop temporally and exposure could have occurred in the past. Also, the authors have used the proportion of households without an automobile as a measure of automobile dependency for urban form and function. The number of household members and the number of cars per household have been ignored and which influence the measure of automobile dependency differently. References 1. Lee ACK and Maheswaran. The healthcare benefits of urban green spaces: a review of the evidence. J Public Health (2011) 33 (2); 212-222 2. Maas J, Verheij RA, Spreeuwenberg P, Groenewegen PP. Physical activity as a possible mechanism behind the relationship between green space and health: a multilevel analysis. BMC Public Health. 2008 Jun 10;8:206.
Conflict of Interest:
None declared
-
It might be Survival that determines Shopping !
Submit responseSir, I read with interest the article by Yu-Hung Chang et al.(1) Authors have articulated some limitations in their paper. However, the findings are derived from purpose of the Elderly Nutrition and Health Survey in Taiwan (1999-2000), done to assess the diet, nutrition and health of persons aged 65 and above in Taiwan. One of the common and often forgotten limitation of such surveys is Survivor Bias.(2) People who go to shopping might be those who are healthy and survived minor illness. Hence, Survival is a determinant of whether people remain active and whether they can go to shopping. Also, people who are debilitated, disabled and who have severe illness might not be able to go to shopping at all. It will be interesting to know whether authors have considered this limitation and if yes, why have they not chosen to discuss regarding this in their paper. References:- 1. Chang Y-H, Chen RC-Y, Wahlqvist ML, et al. J Epidemiol Community Health (2011). doi:10.1136/jech.2010.12669 2. Giridhara R Babu.Do you see an elephant or just its trunk? The need of learning Modern Epidemiologic Methods: An introduction. The Internet Journal of Epidemiology. 2011 Volume 10 Number 1 (Under Print)Conflict of Interest:
None declared
-
Coffee, hepatitis B and hepatocellular carcinoma: study exclusions and omissions are significant
Submit responseSir,
As a coffee-drinking virologist, I read Leung et al's report on coffee consumption and risk of hepatocellular carcinoma (HCC) with interest[1].
The authors excluded those "under medication for liver diseases". It is a reasonable assumption that this means that patients who were receiving antiviral therapy for hepatitis B virus (HBV), e.g. lamivudine, entecavir etc. were excluded. Because patients with a higher risk of developing HCC tend to receive drug therapy for HBV[2], there is a risk that the exclusion of HBV carriers on anti-HBV therapy may have introduced a bias into the non-HCC control group.
Healthy HBV carriers in whom antiviral therapy is not indicated may be fundamentally different from patients with HBV-related HCC. One would expect patients who do not meet criteria for initiating antiviral therapy to have a better prognosis i.e. are less likely to develop HCC.
Leung et al acknowledge some limitations of this study e.g. lack of HBV viral load data and alanine transaminase (ALT) level. However, patients' hepatitis B e antigen status, one measure of HBV infectivity and an independent predictor for progression to HCC[3], is also missing. I would suggest these limitations have been underplayed.
The absence of HBV viral load data, e antigen status and ALT levels are significant omissions and linked to my first point about excluding patients on "medication for liver disease". These variables are both prognostic indicators and indicators for the necessity for HBV treatment.
Finally, it would be of interest to know the hepatitis C (HCV) and HIV status of the patients studied, as these do affect the prognosis in hepatitis B carriers. I concede that both HCV and HIV seem to be relatively uncommon in Hong Kong[4,5]. As this information was not provided, yet more potential confounding variables have been left unaddressed.
The apparent exclusions and omissions detailed above leave considerable room for doubt about the comparability of the cases and controls and thus of the conclusions drawn.
The proposition that coffee consumption reduces the risk of HCC developing in HBV carriers is a seductive one. It would be an almost ideal public health intervention, as coffee is widely available, economical, given that the cost is borne by patients and there are few significant side effects with light or moderate consumption.
Unfortunately, it is difficult to concur with the author's conclusion that moderate coffee consumption significantly reduces the risk of HCC in HBV carriers based upon the findings of this study.
References:
1. Leung W W-M, Ho SC et al. Moderate coffee consumption reduces the risk of hepatocellular carcinoma in hepatitis B chronic carriers: a case- control study. J Epidemiol Community Health 2011;65:556-558
2. Sung JJ, Amarpurkar D et al. Treatment of chronic hepatitis B in Asia-Pacific countries: is the Asia-Pacific consensus statement being followed? Antivir Ther 2010;15(4):607-16
3. Yang HI, Lu SN, Liaw YF et al. Hepatitis B e antigen and the risk of hepatocellular carcinoma. N England J Med 2002;347(3):168-74
4. Hong Kong Department of Health. Viral Hepatitis Preventive Service website. http://www.info.gov.hk/hepatitis/english/hep_c_set.htm (accessed 18/7/11)
5. Hong Kong Department of Health. HIV/AIDS Situation in Hong Kong (2005) factsheet. http://www.info.gov.hk/aids/pdf/g154.pdf (accessed 18/7/11)
Conflict of Interest:
None declared
-
Income Inequality and the Prevalence of Mental illness: A Note of Caution
Submit responseDear Editor,
The correlation (r = 0.73) between income inequality and prevalence of mental illness reported by Pickett, James and Wilkinson (2006) was an intriguing finding, but we should be extremely cautious interpreting it.
First, it was admittedly only a preliminary analysis and hence the number of data points (countries) was small (n = 8). Consequently, the correlation estimate will lack precision and this is reflected in the confidence intervals for the correlation estimate, 95% CI [0.03, 0.95]. Hence, the correlation estimate could be substantially lower.
Second, referring back to the raw data that originated from the UNDP Human Development Report (2005) and Demyttenaere et al. (2004) I discovered that data from the four developing countries (Mexico, Columbia, Ukraine and Nigeria) was excluding from the original analysis. If we recomputed the correlational analysis using the original data supplemented by the data from the four developing countries we obtain a correlation of 0.07 (95% CI [-0.52, 0.62]).
Third, assuming that there are good reasons for excluding the developing countries, a high correlation between income inequality and prevalence of mental illness tells only part of the story. What is equally important is the unstandardized slope of the linear regression line (i.e., simple effect size). This is because, though income inequality and prevalence of mental illness may be highly correlated, the magnitude that the prevalence of mental illness increases in relation to the income inequality may be negligibly small. The unstandardized slope of the linear regression line for the original data from Pickett et al (2006) is 3 (95% CI [0.12, 5.87]). Therefore, as the income equality, measured in terms of the ratio of the top 20% to the bottom 20% incomes increases by 1 the prevalence of mental illness increases by 3%. But what we have to bear in mind is that a ratio increase of 1 actually represents a 100% increase in the difference between the top 20% and the bottom 20% incomes. Therefore, the prevalence of mental illness increases by only 3% when the difference between the top 20% and the bottom 20% incomes increases by 100%. Or in other words 1% increase in the difference between the top 20% and the bottom 20% incomes is associated with a 0.03% increase in the prevalence of mental illness. Hence, the impact of income equality on the prevalence of mental illness appears to be, as one would expect, relatively small.
References
Demyttenaere K, Bruffaerts R, Posada-Villa J, et al. (2004). Prevalence, severity, and unmet need for treatment of mental disorders in the World Health Organization world mental health surveys. JAMA, 291, 2581 -90.
Pickett, K.E., James, O.W., & Wilkinson R.G. (2006). Income inequality and the prevalence of mental illness: a preliminary international analysis. Journal of Epidemiology & Community Health, 60, 646-647.
United Nations Development Programme. (2005). Human Development Report. New York: Oxford University Press. Retrieved from: http://hdr.undp.org/en/media/HDR05_complete.pdf
Conflict of Interest:
None declared
-
Good fit, but still the wrong model? Understanding data generation matters more than likelihood-based model-fit statistics.
Submit responseWe read with interest the article describing methods for modelling count data with excess zeros compared to standard count distributions, such as Poisson1. This topic has been extensively discussed in the statistical and epidemiological literature2-3. Didactic messages given by statisticians can often lack an appreciation of the epidemiological context and sadly this article has the same shortcomings. The primary novelty is the context in which the issue is discussed: application to counts of Activities of Daily Living (ADL-s). It is laudable that methodological issues are tackled using practical examples, but our concern is that the novel context has been largely overlooked and an educational opportunity lost. The message is too narrow in scope and arguably misleading. The reader is told that the zero-inflated Negative Binomial is the most appropriate model for the ADL-s data, yet this is likely incorrect and could lead to inappropriate models being adopted in this domain. The Binomial distribution deals with bounded data and is extended to yield the Beta Binomial, which accommodates overdispersion. Both the Binomial and Beta Binomial distributions have zero-inflated extensions (e.g. the "zero-inflated Binomial") to account for an excess of zeros4. We were surprised to see no mention of these alternative distributions since the Poisson distribution exhibits an infinite tail and predicts counts in excess of six, albeit with low probabilities. Allowing for over-dispersion, the Negative Binomial accommodates a longer tail for the same mean. Both distributions (whether zero-inflated or not) thus possess properties that are incongruent with the nature of the data in this context. Any count distribution assumes an increment of one has the same meaning from one to two as from five to six; for zero-inflated extensions, the increment from zero to one may have a different meaning. As each ADL-s is unique, each is likely to have a different meaning. Hence, as individuals deteriorate in their condition, some activities may become a challenge before others. Consequently, an increment of one ADL may have different meaning along the scale and it is sensible to assume an ordinal outcome5, which also accommodates differences in the increment from zero to one. Good agreement between observed and predicted outcomes is necessary but not sufficient. Disparity between models with regard to predicted outcomes can often be negligible whilst models differ substantially in parameterisation and hence interpretation. Likelihood-based model-fit criteria are only one facet of model development; context validity and interpretability must also have a bearing and researchers must appreciate the context in which data are generated. For this reason the best model may not yet have been found. We have not investigated the dataset, nor do we feel the need to do so when proposing the ordinal model and only with no compelling evidence that increments differed might we instead propose the zero-inflated Binomial or Beta Binomial for parsimony. References 1. Zaninotto P, Falaschetti E. Comparison of methods for modelling a count outcome with excess zeros: application to Activities of Daily Living (ADL-s). J Epidemiol Community Health 2011; 65: 205-210. 2. Lambert D. Zero-inflated Poisson regression, with an application to defects in manufacturing. Technometrics 1992; 34:1–14. 3. Gilthorpe MS, Frydenberg M, Cheng Y, Baelum V. Modelling count data with excessive zeros: The need for class prediction in zero-inflated models and the issue of data generation in choosing between zero-inflated and generic mixture models for dental caries data. Statistics in Medicine 2009; 28: 3539-3553. 4. Vieira AMC, Hinde JP, Demetrio CGB. Zero-inflated proportion data models applied to a biological control assay. Journal of Applied Statistics 2000; 27:373–389. 5. Lall R, Campbell MJ, Walters SJ, Morgan K and MRC CFAS Co-operative. A review of ordinal regression models applied on health-related quality of life assessments. Stat Methods Med Res 2002; 11: 49-67. -
Not all fats the same
Submit responseThere is much evidence linking learning and behaviour problems in childhood to refined oils in the maternal diet during pregnancy and lactation, particularly linoleic acid in both cis and trans forms which may impair fetal brain development. A lower IQ in these children would not be surprising.
I have seen no evidence, however, linking fresh natural fats to any of these problems. Diets high in refined sugars tend to be high in refined oils as well -- the two go together -- so I think it's important to stress that "high fat" in this case refers only to processed oils, not the healthy fats naturally present in fresh whole foods.
I would be pleased to provide references re linoleic acid and effects on offspring to anyone wishing to contact me at the address above.
Thomas Anderson, Ph.D.
Conflict of Interest:
None declared
-
Re: UK newspapers' representations of the 2009-10 outbreak of swine flu: one health scare not over-hyped by the media?
Submit responseSir,
The authors conclusions that the media did not 'over-hype' the swine flu nondemic are reached as a result of a methodology that directed them towards that conclusion; an analysis solely of the UK print media.
I travelled to New Zealand on 11 May - two weeks after the first cases emerged - and arrived in Melbourne on the 5th June, a time when this was arguably the swine flu capital of the world. Yes, it was apparent that there was flu circulating and I fell victim to it, but the print media response was far more measured and balanced than it was here. The joys of the internet meant that I was following the print media in the UK closely, as well as reading the major daily newspapers in both Australia and New Zealand.
I suspect that an analysis of the UK print media v that of other countries may expose our response to this nondemic for the hysteria that it was.
Yours sincerely,
Sarah Jones
Conflict of Interest:
None declared
-
And what of political science?
Submit responseThere is much to like in the argument presented by Clare Bambra. Epidemiology (and public health generally) has much to learn from the social sciences and the reverse is also true. And Bambra is surely correct to raise concerns about the associated risks including the "engrained caution and purism of epidemiology" and the excessive deference to experimental research designs before acting. However, in order to effectively manage and profit from "the interface between social science and epidemiology" it is essential that epidemiologists and their public health colleagues engage with the broad range of social science including, and perhaps especially, political science. In many discussion of the linkage between epidemiology and the social sciences there are repeated references to the insights of psychology, economics and sociology but few if any direct references are made to political science. This is ironic given the sheer size of the political science enterprise. It is also tragic insofar as the goal of much of the writing on the social and economic determinants of health is to encourage action on these determinants often, if not primarily, by means of the state. Political science has much to say about the why, what and how of state action. Fore example, having highlighted the association between social inequality and poor health, while it is reasonable to call, as Bambra does, for income redistribution, this is but the start. Contemporary political science (including political theory) can offer a great deal of insight into why inequality is on the rise (e.g., Pierson, Paul, and Jacob S. Hacker. 2010. Winner-Take-All Politics: How Washington Made the Rich Richer--and Turned Its Back on the Middle Class. New York: Simon & Schuster.), why this varies between states, what alternative exist for ensuring more redistribution, and how to counter the inevitable arguments that this would be unfair or unwise (e.g., Cohen, G. A. 2008. Rescuing Justice and Equality. Cambridge, Mass.: Harvard University Press).Conflict of Interest:
None declared
Free sample
This recent issue is free to all users to allow everyone the opportunity to see the full scope and typical content of
JECH.
View free sample issue >>
Don't forget to sign up for content alerts so you keep up to date with all the articles as they are published.
