Electronic Letters to:
|
|
eLetters: Electronic letters published:
-
![[Read eLetter]](/icons/misc/sectionDOWN.gif)
Fallacies in Standardised Mortality Ratios in Anorexia nervosa
- Gerald F.M. Russell, Dr. Ward Anne, Psychiatrist in Psychotherapy, Maudsley Hospital, London. Anne.Ward@slam.nhs.uk (6 February 2009)
-
Re: Fallacies in Standardised Mortality Ratios in Anorexia nervosa
- Fotios C Papadopoulos, Anders Ekbom, Lena Brandt, Lisa Ekselius (19 March 2009)
|
|
|||
|
Gerald F.M. Russell, Emeritus Professor of Psychiatry Institute of Psychiatry, London, Dr. Ward Anne, Psychiatrist in Psychotherapy, Maudsley Hospital, London. Anne.Ward@slam.nhs.uk
Send letter to journal:
geraldrussell{at}btinternet.com Gerald F.M. Russell, et al.
|
The article by Papadopoulos et al adds to the evidence of high mortality rates in anorexia nervosa. An impressively large cohort was obtained through the Swedish Cause-of-Death Register which includes all Swedish persons who died since 1952. The crude mortality rate for 6009 females with at least one hospital admission for anorexia nervosa was 4.41 per cent over a mean follow-up of 13.4 years, (averaging 0.33 per cent per annum). This rate compares favourably with other studies (0.5-2.2 per cent per annum)2, yet the authors, after much manipulation of their data, conclude that the mortality rate in Swedish women was "astonishingly" high. We contest this finding based on misleading calculations of standardised mortality ratios (S.M.R.s). S.M.R.s are a means of comparing mortality in a specified patient population with a standard population. The S.M.R. value will exceed 1 in proportion to the risk of death from the disease under study. The authors have two different usages of S.M.R. The first is the customary one when the calculation is applied to a cohort of persons who have been given a specific diagnosis at the outset. In Table 3 this S.M.R. is given as 6.2 for the 6009 patients with anorexia nervosa among whom there occurred 265 deaths whereas the expected deaths were 42.6. So far so good. Their second approach was to count the number of deaths according to each specific cause of death, yielding a different kind of S.M.R. For example, there were 84 suicides yielding an S.M.R. of 13.6, signifying that suicide was 13.6 times commoner among the cohort of anorexia nervosa patients than generally expected. Similarly, the S.M.R. for deaths due to respiratory disease was 11.5. But the S.M.R. for anorexia nervosa as a cause of death was said to be 650.0 and it is this figure which leads the authors to conclude the death rate in their sample was astonishingly high. So it would be if it had clinical and statistical validity. The authors' errors arise from estimating the S.M.R. for a sub-group (39) of the original cohort using the fraction: observed number of deaths / number of expected deaths The numerator is given as 39 patients in whom anorexia nervosa was the main cause of death on the death certificate. It is the denominator which is elusive in its estimated value. It is given as 0.1 but the authors' own data suggest this is an approximation for 0.06, a very low figure which results in an inflated value for the S.M.R. (650) in this ambiguous sub- group of anorexia nervosa. We suggest that when an underlying cause of death (e.g. suicide, respiratory infection, etc.) was not identified, the certifier of the death entered anorexia nervosa on recognising a cachectic state, especially as malnutrition does not feature in the list of "underlying" causes. These objections do not apply to the first calculation of the S.M.R. in the full cohort of anorexia nervosa patients, whose value was found to be 6.2, by no means an astonishing death rate. References 1.Papadopoulos, F.L., Ekbom, A., Brandt, L., Ekselius, L. Excess mortality, causes of death and prognostic factors in anorexia nervosa. Br. J. Psychiatry 2009, 194, 10-17. 2.Ramsay, R., Ward, A., Treasure, J., Russell, G.F.M. Compulsory treatment in anorexia nervosa: short-term benefits and long-term mortality. Br. J. Psychiatry 1999, 173, 147-153. Authors: Professor Russell, Gerald. Emeritus Professor of Psychiatry, Institute of Psychiatry, London, SE5 8AF e-mail: geraldrussell@btinternet.com Dr. Ward, Anne. Consultant Psychiatrist in Psychotherapy Department of Psychotherapy, The Maudsley Hospital, Denmark Hill, London, SE5 8AZ e-mail: Anne.Ward@slam.nhs.uk |
|||
|
|
|||
|
Fotios C Papadopoulos, Resident in General Psychiatry Department of Neuroscience, Psychiatry,, Anders Ekbom, Lena Brandt, Lisa Ekselius
Send letter to journal:
fotis.papadopoulos{at}gmail.com Fotios C Papadopoulos, et al.
|
Professor Russell and Dr. Ward (1) raise the issue of the suspected erroneously inflated value for SMR (650) for the subgroup of women in whom anorexia nervosa was stated as the underlying cause of death on the death certificate in our paper (2). The expected number of deaths for this subgroup was indeed 0.06 (denominator) as the authors point out and it was presented with its one decimal approximation (0.1). The authors further suggest that the certifiers of the death would be prone to enter anorexia nervosa on the death certificate when a specific underlying cause of death could not be identified but a cachectic state was evident. We agree that this could be true, but we do not believe that such “misclassification” would be problematic if those women had an active anorexia nervosa at the time of death. On the contrary, it would be worrisome if women with other diagnoses that lead to cachectic states (other than anorexia nervosa) were misclassified as anorexia nervosa on death certificates, but our inclusion criteria were specifically selected in order to reduce this possibility. In addition, we believe that the estimation of the SMR value for this specific subgroup of patients does not confer more information than what common sense dictates, namely that those with a lifetime diagnosis of anorexia have a much higher risk of dying from it. Overall, women with anorexia nervosa in our cohort had a six-fold increased mortality compared with the general population. This excess mortality in anorexia nervosa is 2-3 times higher when compared to the excess mortality observed in mental disorders in general (3) and more specifically in schizophrenia (4), bipolar and unipolar disorder (5). Moreover, we would like to point out that we were most astonished by the persistence of this unfavourable outcome throughout lifetime, with high SMR for both natural and unnatural causes of death even 20 years or more after the first admission for anorexia nervosa. References 1. Russell G., Ward A. Fallacies in Standardised Mortality Ratios in Anorexia nervosa. 2. Papadopoulos FC, Ekbom A, Brandt L, Ekselius L. Excess mortality, causes of death and prognostic factors in anorexia nervosa. Br. J. Psychiatry 2009, 194, 10-17. 3. Valenti M, Necozione S, Busellu G, Borrelli G, Lepore AR, Madonna R, Altobelli E, Mattei A, Torchio P, Corrao G, Di Orio F. Mortality in psychiatric hospital patients: a cohort analysis of prognostic factors. Int J Epidemiol. 1997; 26:1227-35. 4. Saha S, Chant D, McGrath J. A systematic review of mortality in schizophrenia: is the differential mortality gap worsening over time? Arch Gen Psychiatry. 2007; 64:1123-31 5. Osby U, Brandt L, Correia N, Ekbom A, Sparén P. Excess mortality in bipolar and unipolar disorder in Sweden. Arch Gen Psychiatry. 2001;58:844-50. |
|||