Jump to:
Welcome
by Ruth Armstrong
Review
by Harvey Whiteford
Review
by Tony Jorm
Review
by Elmer Villanueva
Comments
by David Jayne, ABS
Comments
by Michael Dudley, FRANZCP
Comments
from the Editors - Ruth Armstrong
Response
from the Authors - Chris Cantor
Editors'
decision - Ruth Armstrong, 7 June 1999
22/2/99
| Date | Comment | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Mon Feb 22 09:44:01 1999 |
Welcome everyone and thank you for participating. We eagerly await your comments. Please take a look at the Instructions to contributors before proceeding. In particular, note that it is safe practice to keep a local copy of your comments. Ruth Armstrong Editorial registrar MJA
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Fri Feb 26 07:36:05 1999 |
This article is focussed, well written and topical.
It addresses an important public health issue and an issue of clinical
importance to mental health professionals and general practitioners. It
incorporates the first independent analysis of the 1997 ABS data which,
since it's release, has been of concern given that the rising rates of
male suicide for the 15-34 years group were thought to have, perhaps, been
plateauing. I have only a few comments as follow: Tables 1 and the last (unnumbered) table are the same. [This editorial error has now been corrected - Craig Bingham -1 March 99] As for par 38, I would argue for the text version (as I assume will a space conscious Editor), although the tables would be of some interest to readers wanting to know which countries are above and below Australian rates. This is however covered to some extent toward the end of the Discussion section. I am pleased the authors (who rightly emphasise "years of life lost") address the concern of the rates in the 75+ years group (par 44). Statistical analysis appears sound, although confirmation from the ABS may be worthwhile. Par 42. I am not aware State and Territory Coroners Act's encourage a "prohibition" of findings of suicide and the reference used to support this, rather controversial statement, is 'in press'. Nevertheless I would stand to be corrected if the authors can support their assertion. Certainly coronial reporting is an issue to be addressed in getting better quality data on suicide rates. Since the article has, as one aim, informing governments about priority target groups for suicide prevention, the authors should note that the current Federal government has announced a broadening of it's National Youth Suicide Prevention Strategy to a National Suicide Prevention Strategy in response to the very arguments the authors are making. This broadening was made on the advice of advocates and academics (including Dr Cantor) and before the 1997 ABS data became available. In the Acknowledgements (par 49) I do not know which Department of Public Health the authors are referring to. Do they mean the Queensland Department of Health? References are comprehensive. There are a few typos (e.g. line 7 par42; line 6 par 47). Dr Harvey Whiteford The World Bank Health, Nutrition and Population Human Development Network Room G3-060 1818 H Street, NW Washington DC 20433 USA | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Mon Mar 1 15:12:15 1999 |
Review of paper by Cantor et al. on "Australian
suicide trends 1964-1997-youth and beyond?" I think this is a very useful paper on a topic generating a lot of discussion in Australia today. It draws together Australian and overseas data in a readily accessible form. I particularly liked the way the paper dealt with suicide trends in all age groups, not just in youth. What has happened with older Australians has tended to get overlooked. I have the following specific suggestions for how the paper could be improved: 1. I wonder whether the authors are aware of the recent editorial by Goldney commenting on the fall in suicide in older Australians (Australasian Journal on Ageing, 1998, 17, 54-55). Goldney essentially makes the same point as the authors do in par 43, namely that the fall in suicide in older people might have something to teach us about how to reduce youth suicide. Rosenman also has an interesting comment on this issue on p. 151 of the same journal. I am not necessarily suggesting that the authors should cite these papers, only that they may find some of the ideas of relevant to their data. 2. In the formula in par 24, it is not clear why the Cx term is in the equation. A sentence explaining this would help. 3. The paper uses 95% CIs to determine whether period trends are significant. I am unclear about the rationale for this. In most studies, a sample is drawn from the population and used to make inferences about that population. CIs give an indication of the reliability of the estimate which is made about the population. However, in this paper the whole Australian population is being analyzed. It is not a case of using a sample to draw conclusions about a population. There is of course the issue of whether a yearly suicide rate is a temporary fluctuation or part of a stable trend. Time series analysis may be appropriate for making inferences about this (I don't have the expertise in time series analysis to say), but I don't think 95% CIs on the mortality rates are appropriate. At the very least, some rationale for the methods used needs to be given. 4. I am not convinced by the discussion in par 41 about an unfavourable youth influence offsetting a favourable trend for women in general. Looking at the Figures, the favourable trend is not for women specifically, but for older people in general. From age 45 onwards, the trends for men and women are very similar. The divergence in trends is only seen in younger people, where there is a rise for men and a static trend for women. There does not seem to be "a favourable female gender influence", but a general favourable effect on older people. 5. I am unclear what is meant by "these rates continued downwards at the 1997 endpoint" (par 43). I guess it means that the trend for the last few years up to 1997 was downwards. 6. Par 45 states that "service provision to the Australian elderly is less developed than for many nations with higher suicide rates". Which services are these and what is the evidence for this statement about relative service provision? Perhaps the authors are referring to psychogeriatric services, but these only deal with the very tip of the iceberg of mental health problems in the elderly. Primary medical services are very well developed in Australia, and community support services through the HACC program have increased dramatically over recent years ("Australia's Welfare 1997" published by Aust Institute of Health and Welfare"). 7. Par 47 suggests that the high young male suicide rate in many countries could be due to "cultural expectations that men should be tough and resilient". However, is there any reason to believe that countries with much lower young male suicide rates (England & Wales, Netherlands, Spain, Italy, Portugal, Greece) are any different in this regard? Also, is there any reason to believe that this cultural expectation has increased in Australia along with the increase in young male suicides? I raise these questions, because I doubt that this explanation accounts for the data reported in this paper. I think Tables 2 and 3, which the authors list as optional, should be included. These are very interesting data to have drawn together. Tony Jorm
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Thu Mar 4 05:24:09 1999 |
Greetings of Peace! Dr. Cantor and his colleagues certainly provide a focused and succinct view of an important subject area. My comments pertain to some methodologic aspects and various other sundry. 1. The authors use ABS data exclusively in the examination of suicide trends from 1964-1997. However, no mention is made as to the process ABS uses to: (1) define deaths as being due to suicide; (2) how this definition has changed over the period; or (3) how complete this case-ascertainment is performed over the same period. For example, there's a distinct peak occurring in the 1984-1985 rates. A possible administrative explanation for this jump may be that the definition of suicide was expanded to include deaths due to complications arising from failed attempts. A brief discussion of ABS data extraction methods and the completeness of the same is needed. 2. The authors explicitly mention the population used to standardize Australian rates from 1964 to 1997. However, no mention of standardization is made when comparing the mean suicide rates of 22 other countries. This concept of confounding by a third variable may explain the comparisons seen. Crude rates reflect actual rates in the population, but should be used with caution when making comparisons between populations, even if the authors' use of stratification is laudable. 3. Would it be possible to set the y-axis of the graphs to a standard scale? As they are currently displayed, the axes range from a low of 0-25 to a high of 0-50. The differences in the axes tend to inflate the perceived trends in rates, especially if one does not pay careful attention to detail. 4. I'd like to echo Tony Jorm's sentiment about the use of CI's in par19. The fact that this confusion is apparent is a call for a more detailed exposition of the statistical methods used. 5. The statement "Confidence interval (P<0.05) were..." in p34 is confusing. Do the authors mean 95% confidence intervals? 6. Often, one is interested NOT in whether increases or decreases in trends were simply "statistically significant" or not, but what the statistical estimate of this change was. The authors leave one asking such a question about the estimate of this change in par34. 7. The authors discuss age and (briefly) period effects in par40, par41, and par43, but fail to expound on the possibility of a cohort effect in action. 8. In drawing conclusions from data collected from large populations, one may be drawn to the so-called ecologic fallacy -- attributing within-population effects to between-population differences. The authors make such a claim when they discuss how Canada, the US, and New Zealand have similar characteristics (par 46), or that the cultural similarities of Hungary and Finland contribute to high suicide rates in these countries (par47). The authors should tread lightly and not overreach the extent of their data. 9. Finally, a comment on the DESIGN AND PARTICIPANTS section of the abstract (par 13). I think the second sentence should contain a mention that international comparisons were limited to 1990-1994 data. As it is currently written, one is led to believe that these comparisons were made for the entire 1964-1997 period. Elmer Virgil S. Villaneuva, MD, ScM Johns Hopkins School of Hygiene and Public Health Elmer Villanueva
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Tue Mar 16 10:59:17 1999 |
1. I have been requested to contribute to the review
on behalf of the ABS as Dr Cantor and his colleges have used ABS data
extensively for this article. [David Jayne is Manager of the Health and
Vital Statistics Unit of the Australian Bureau of Statistics - CB]
2. A "spot check" of the graphs have confirmed that the correct rates figures have been used for these. 3. An overall comment on the methodology used is that Dr Cantor and his colleagues included a brief description within the text of the article. However, the degree of detail was not sufficient to enable checking of methodology to be undertaken, but enough to cause confusion to some readers (judging by the responses received). The flow of the article might be improved by including separate technical notes at the end of the article and referring interested readers to these notes in the body of the text. 4. Para 14 (line 1) - should read 25 - 34 years not 24 - 35 years ? Para 14 - Disagree with the statement that rates for females in 15-24 and 25-34 year age groups showed no significant change between 1964 and 1997. For 15 - 24 year olds, the rate increased from 5 to 7 which was a 40% increase, admittedly from a small base. For 25 - 34 year olds, the rate decreased from 10 to 8. This 20% decrease may be considered significant by some readers. Maybe a statement such as " while the suicide rates for females in 15-24 and 25-34 year age groups varied quite significantly in percentage terms between 1964 and 1997, actual numbers remained reasonably consistent and still well below the corresponding rates for males in these age groups" or something similar could be used. Para 14 - Disagree with the statement that a number of older age/sex groups declined and were continuing to decline at the end of the period. Attached are age specific suicide rates for the past five years.
A statement such as " the suicide rates for older age/sex groups have remained fairly constant in recent years" would be more suitable. 5. Graphs - Legend on y-axis of graphs - rate should read per 100,000 estimated mean resident population? 6. Para 42. This para would indicate that ABS determines whether a death is a suicide or not. This is not the case. ABS are completely reliant on the findings of the coroner for classifying suicides. In uncertain cases, ABS will query coroners for further information or clarification. I have attached a copy of some relevant correspondence which may have been misunderstood. Processes relating to collection and compilation of data relating to suicide and suicide attempts. ABS bases its codes on query replies received from Coroners and Government Medical Officers (GMOs). All unknown external cause records are queried to ascertain how the injuries occurred, and almost all ask whether the death was due to an accident, suicide or homicide. Records of infants and young children (under the age of 10) are queried to ascertain how the injuries occurred, but the query does not specifically ask whether death was due to suicide. This approach is based on past response to our query action. However, ABS coding is based on the findings of the Coroner and if the Coroner states suicide, it would be coded as such. Para 42. Would disagree that the Coroners Acts would prohibit findings of suicide, rather they would tend to emphasise the importance of the coroner's role and the need to ensure findings are appropriate. Consequently, the Coroners tend to take a 'conservative' approach when considering cases of potential suicide. There are some cases where only the means of death (eg hanging) are reported, but which turn out to be suicide when queried. All such cases would be picked up by query action for those aged 10 years or over. 7. Para 43. Line 4 The statement on the decline in older age rates; namely "Mostly these rates continued downwards at the 1997 endpoint" is not confirmed by the figures. Most rates rose between 1996 and 1997. For age 75+ this rise was from 32 in 1996 to 36 in 1997. 1996 rates were also an increase on the 1995 figure of 29. Response to a comment from Tony Jorm Para 24 The Cx term in the equation is an adjustment factor used to standardise the population to (in this case 1991). Perhaps this should be included in a technical note, as mentioned above. Response to comments from Elmer Villanueva 1. Para 1 It is possible to provide information on ABS processing and classification of mortality data if required. This could be included as a technical note. ABS does not make value judgements, but processes the information as provided on the medical certificate. If the death has been referred to the coroner, then information is sought from the coroner to assist in classifying this death. While the reference period covers three different versions of the International Classification of Diseases (ICD - 7, 8 and 9), I am unaware of any changes to classifications or administrative procedures which would account for increases in suicide rates for 1984 and 1985. David Jayne
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Thu Mar 25 14:38:12 1999 |
Cantor and colleagues' review is very timely and
will assisting in defining the suicide prevention agenda in Australia in
the immediate future. 1. I am not sure why the YPLL formula used is not gender-specific, since the life expectancy for the sexes is different. 2. With Harvey Whiteford, I am also glad that the authors have attended to the over 75 year group. However, since the highest rates are supposedly found in those over 85 years (although the numbers must be small), the issue that Riaz Hassan raised about trends in this group has still not been addressed. 3. The reference to confidence intervals (para. 34) I think could appear in the methods section, with a fuller explanation (as suggested by other reviewers). 4. end para. 43. I echo Elmer Villeneuva's comments about the need to discuss cohort effects. 5. It would be helpful if reference could be made to the reliability of international comparisons of suicide data. 6. In view of the authors' and reviewers' comments about coroners' procedures in determining suicide (including with children), it would also be useful to briefly discuss or refer to how such verdicts are reached, and any variability in such verdicts. 7. para41. Perhaps the 'unfavourable youth influence' refers not to the immediate data, but to the 'rashness of youth', propensity to impulsivity, violence etc; and the 'favourable female influence' to the fact that women do not resort to violence as a rule, have less externalising psychiatric disorders, ask for help more readily than men, and have lower suicide rates than men. If so, perhaps the authors could clarify this. 8. I agree with Tony Jorm's suggestion that Tables 2 and 3 be included, as they convey more information. 9. There are a few gramm/typo problems (e.g. 'the changes in rate over time was calculated' para. 19, Scandanavia para. 45). Michael Dudley, FRANZCP
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Thu Apr 1 14:52:28 1999 |
Thank you for submitting your manuscript "Australian suicide trends 1964 -- 1997 - youth and beyond" for consideration by The Medical Journal of Australia and thank-you everyone for your participation in and patience with the review process. This paper has attracted many useful comments, not only from the three reviewers but also from two members of the consultant panel. The reviewers comments are generally supportive of publication, given suitable revision of the manuscript. It has also been considered by the editorial committee. I apologise for the slight delay in my reply, related to our collaborative editorial process, which can take time. From an editorial point of view, we would like you to make the following changes in addition to responding to the referees' comments: We would like to consider publishing your paper in the Journal under the banner of "medicine in the community" (retaining the current research format).Please amend your abstract to include a section on "main outcome measures" The tables (2 and 3) relating to international comparisons are bulky and do not add to the main message of the paper. They should be removed and the optional paragraph of text used. The idea of a technical note on the use of the ABS statistics a good one. This should be a succinct paragraph which we can put in a box, outlining how ABS defines/ collects suicide statistics and how the authors have gathered and used the statistics. The question about the appropriateness of confidence intervals is also an important one. You may wish to consult a statistician about the best way to present your results. If you believe you can revise the manuscript to address these concerns, we would be keen to see a revised version by Friday, May 14 1999. You need to address all the referee's main points and to outline how you have done so in a "covering letter". Changes to the manuscript should be bolded or otherwise highlighted on the electronic version. Ensure that you include a word count. Please also include raw data for any figures. We look forward to receiving the revised manuscript. Thanks once again to the referees and consultant panel. We look forward to your comments on the revised paper. Ruth Armstrong Ruth Armstrong
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Fri May 14 17:53:53 1999 |
14 May 1999 Dear Dr Armstrong Re: Australian Suicide Trends 1964-1997 - Youth and Beyond? cantor.html Thank you for your letter of 1 April requesting revisions in accord with the assessors' comments. The receipt of six detailed assessments combined with the journal's policy of brevity requires my changes to be expressed succinctly. Hopefully this has been achieved. Changes will be described according to the initials of the assessors. DJ from ABS and HW object/question our suggestion that some Coroners Acts prohibit findings of suicide. Note that in the NH&MRC material that the Queensland Coroner responded that Section 37 of the Registration of Births, Deaths and Marriages Act requires that "where an entry of Cause of Death is made in any Register of Deaths pursuant to this Act, and the death in question was self-inflicted, there shall not be added to the entry the word "suicide" or any other word or words expressly indicating that the death was self-inflicted". Please also note that coroners from South Australia, Tasmania, Victoria and Northern Territory avoid using the word "suicide". Coroners from the ACT use the word "suicide" only if it is absolutely clear that the death was by suicide. It is only coroners in NSW and Western Australia who use suicide verdicts explicitly. It is inconceivable that these influences do not impact in some way on interpretation by coroners and therefore in turn by ABS, no matter how diligent they may be. Also DJ reiterates previous correspondence (I believe to myself) in which he refers to ABS procedure of querying less than certain suicides - these queries being directed to coroners and Government Medical Officers. However, as DJ reiterates in his paragraph 6 "records of infants and young children (under the age of 10) are queried to ascertain how the injuries occurred, but the query does not specifically ask whether death was due to suicide." This is in contrast to queries relating to those over the age of 10 which do ask whether the death was suicide. This ABS procedural policy is likely, in my opinion, to underreport child suicides. As you can see, this topic has been considered by us in detail and could be a whole paper in itself - and still not resolve the issues. My preference would be to omit these issues. Might I suggest that if you still want to cover this issue the following summation may be reasonable. It could be added as a footnote/appendix or box, as you suggested, if you prefer: This issue has potential to divert readers from the thrust of the paper. "Coding cases as suicide relies on an interaction between the ABS and State and Territory Coroners and Government Medical Officers. The six States and two Territories each have different Coroners Acts. Most States/Territories discourage Coroners from formal pronouncements about suicide. Western Australia and New South Wales are the only states that routinely use suicide verdicts. ABS receive information from coroners that is generally sufficient for coding most cases. Cases that in reality were suicides but were not deemed as such by Coroners or ABS will most likely be coded as undetermined deaths or accidents involving methods commonly employed for suicide (eg. poisoning). There has been a rise in both in recent years. The impact of these uncertainties is unlikely to be sufficient to greatly alter the overall findings of the present study". Please note, in the first the paragraph of the discussion I have added "(it should also be noted that at the time of submission of this paper there were significant concerns about the validity of the 1997 figure)." Yours sincerely Dr Chris Cantor Senior Research Psychiatrist
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Mon Jun 7 18:13:56 1999 |
Australian suicide trends Editorial comment and decision on the revised manuscript Thanks once again everybody for your contribution to this paper. The manuscript was discussed at our June 3 Editorial Committee meeting and a decision was made that it is now of an acceptable standard for immediate publication on our web site. Print publication will follow (at a date yet to be decided). We will include the information on coding of suicides in a text box in the print version. Thanks again everybody for your interest and patience. Ruth Armstrong Assistant Editor MJA
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||