So, as predicted, an announcement is issued to the effect that the smoker bans were a good thing and many lives have been saved as a result.
Having seen the studies (remember that this Daily Mail story is just that: a story) from Piedmont, from Helena, from Rome, from France, from Ireland and from Scotland, and others, we can honestly state that they all have one thing in common: they are garbage. They were produced by those who desparately needed to prove to the politico's that their lies were true. They were produced by those outfits that had much to gain, financially, from studies such as these. The studies were needed to fan the flames of hatred and discrimination towards a section of society who are punitively taxed, and constantly hounded about their use of a legal product.
What follows is not my work. The letter written to Rebecca Camber of the Daily Mail, who authored this very badly researched piece, is not mine. The author has requested anonymity and in the best tradition of reporting I will protect my source. I can confirm that the author is a much respected statistician in his/her industry.
I will, however, publish the complete letter so that you can see the extent of the flaws in Rebecca's shoddy little piece.
Many, many thanks to the author of this letter.
"Smoking ban cuts the number of heart attacks by more than 40 per cent at some hospitals"
So screamed the headline to your article in Daily Mail Online today (14 June 2008).
Truly a miracle must have occurred during the period of 1 July 2007 to 31 March 2008, if the ban on smoking in public places has had such a dramatic effect on the health of the English population. Before cracking open the champagne, however, I thought it would be useful to pause and reflect on what you have discovered.
Given that the findings are not attributed to any particular medical or epidemiological research study, (indeed, you state that the figures were "obtained under the Freedom of Information Act by the Daily Mail"), I have decided to eschew my normal approach of trying to obtain an original copy of the quoted research, and to pose my specific questions to you directly. Can I ask you, please, to help me out here, Rebecca? Thank you.
Firstly, for the avoidance of any misunderstandings, let me expose my credentials. I have spent my whole career to date - over 30 years - working for the NHS, both as an employee and, more latterly, in a private sector consultancy capacity. I am a graduate Mathematician and have worked variously as a (senior) health statistician, and a manager of NHS Information Services. I have, with my own company, developed (commercial) computer software systems which have had widespread usage throughout the NHS in the capture, analysis and interpretation of health statistics at all levels from individual hospital/trusts to national performance management, and I have an intimate knowledge and understanding of healthcare statistics, the depth and breadth of which is shared by very few people in this country.
In other words it is fair to say that I am, to use a term so beloved of your own (journalistic) profession, something of an 'Expert' in this area.
Given my background, my first question to you, Rebecca, is: Who carried out the analysis of the data that you "obtained under the Freedom of Information Act"? Was it yourself, another journalist at The Mail, or did you use the services of a professionally-qualified health statistician? The question is not fatuous, nor is it intended to convey any trickery, I merely wish to establish the depth of the understanding of health data that was behind the 'figures' that you quote - the reason for this will become obvious later.
The remainder of my questions will no doubt best be answered by whosoever it was that did undertake the analysis in question. So, let's go...
1 Regarding the data obtained:
1.1 When exactly was this data obtained? April, May or June 2008?
1.2 Why was data obtained from only 116 NHS Trusts?
1.3 Was the data obtained directly from these Trusts, or through one of the Department of Health organisations, eg Secondary Uses Service (SUS), or Hospital Episode Statistics (HES), that have the complete and nationally responsibility for electronically collecting, collating and publishing all NHS inpatient episode data for England? Indeed, had you chosen to use the 'HES Online' service (free web service), are you not aware that you could have obtained such data for all NHS Trusts in England without recourse to the Freedom of Information Act? Of course you would then have had to wait until the data was officially available which, for July 1 2007 to March 31 2008 would have been much later this year (2008); which brings me to my next question...
1.4 Why did you not wait until the data was available on an official basis? Furthermore, why did you not wait until a whole year's data was available, ie the year to 30 June 2008? Anyone who undertakes serious research looking at year-on-year comparisons of health trends would know that it is bad form to just take a 9-month period. You also state that the numbers for this period were compared with those for "that period 12 months earlier", which suggests a comparison with figures for 1 July 2006 to 30 March 2007, yet you then, somewhat confusingly, refer to your baseline for comparison as "42,176 admissions to all the trusts from October 2006 to June 2007", which most certainly is not "that period 12 months earlier". If you were indeed comparing data for the 9-month period after the ban with the immediately preceding 9-month period, then your comparison is rendered completely invalid, as you are not comparing like-for-like at all. Consider the most obvious problem that the initial period excludes all of the main summer period, when the weather is at its best and people will be in a more relaxed mood - circumstances which will inevitably result in a reduced level of heart attacks in your second (1 July 2007 onwards) period, merely through a well-known (winter-predominant) seasonal variation in heart attacks. Clearly someone needs to address the question of flaws in reasoning here, never mind the problems of data quality, and to explain whether such an obvious bias in the numbers was understood. And I have to ask, how much was this driven by the need to publish a 'feel good' article to proclaim the success of the smoking ban in time for its first anniversary? Are journalistic deadlines more important than the truth?
1.5 Has anyone validated the data in terms of a) its completeness (are you certain that all of inpatient admission records for the period in question were counted?), and b) the accuracy of the recorded diagnoses (I assume that you used ICD-10 Codes I21 and I22 to identify Myocardial Infarctions - or Heart Attacks)? This is a vital question as you must surely know that the official processing of these data is undertaken by the organisations referred to above (SUS and HES) on a national basis, and that it is known to take many weeks, if not months, before the complete set of data is available, complete with diagnosis codes that have been checked and are assumed correct. Even then it is common to find that at least 10% of inpatient records have the wrong diagnosis! Time heals, as they say, and agencies such as HES would always advise that the completeness and accuracy of such data will improve as sufficient time is given to allow hospitals to complete their returns, and the agencies' data analysts to undertake thorough validation. I have serious reservations that the data used by The Mail would be anywhere near complete, nor yet proven to be accurate.
1.6 Were the 'heart attacks' in question diagnosed when the patients were first admitted? A strange question, maybe, but you should be aware that the actual diagnoses are usually not entered onto the (electronic) patient records until the patient is discharged. This of course makes it even more difficult to summarise and use such data so soon after the end of the period, not least because some patients who were admitted with heart attacks) in the period may still be in hospital after the period has ended - thus leading to a reduced count of heart attacks since, 'not discharged - no diagnosis recorded'!
1.7 Why did you not include data from other sources? Not all heart attacks will lead to a hospital inpatient admission you know - many poor souls will die before they get to a hospital, so you will only get a complete picture by including a count of death certificates for those who died in a residential or primary care setting.
Before moving on to my second set of questions, I would like you to take note of these cautionary words:
"Fluctuations in the data can occur for a number of reasons, eg organisational changes, reviews of best practice within the medical community, the adoption of new coding schemes and data quality problems that are often year specific. These variations can lead to false assumptions about trends. We advise users of time series data to carefully explore the relevant issues before drawing any conclusions about the reasons for year-on-year changes."
These are the words of the HESOnline service (ie the NHS) and can be found here. You may wish to contact HES yourself to determine whether their cautionary words may cause you to rethink your analysis. Anyway, let's plough on...
2 Regarding the summary statistics and their interpretation:
2.1 Why did your headline say that heart attacks were cut "by more than 40 per cent at some hospitals" when, according to your own words, the reduction was "over 40%", ie was 41%, in only one Trust, namely Shrewsbury and Telford Hospitals NHS Trust. Is the reference to 'some' hospitals merely a reflection of the fact that the word hospitals (plural) appears in the Trust's title. If so, don't you accept that this makes a somewhat misleading statement?
2.2 Given that the this Trust's reduction was so enormous, didn't it occur to you that this was almost certainly an erroneous figure (please refer to my narrative above regarding data, as well as the HES cautionary words)? Any self-respecting statistician would indeed have recognised such and consequently excluded such an obvious 'outlier' from his/her analysis.
2.3 Why did your headline scream "40 percent" when the actual reduction in heart attacks, across your (albeit limited) study set of Trusts was only 3%? A rhetorical question, I suppose, as a headline that screamed about cutting the "number of heart attacks by more than 3 per cent" doesn't sound quite so dramatic, does it? It would be more truthful, though.
2.4 Referring back to my question 2.2; when I recalculated the overall reduction in heart attacks, after excluding the data from Shrewsbury and Telford Trust, it was only just over 2%, or only two thirds of your real headline figure. Did you never see any fallacy in the fact that one single Trust, whose inpatient activity contributes a mere 1% (or less) of the whole of England's, should actual experience almost one third of the total drop in heart attacks of the whole country? Aren't journalists expected to be able to demonstrate a modicum of numeracy?
2.5 Given that we are now down to a drop in heart attack admissions of just over 2% - and this figure is itself of dubious provenance, given the known limitations or the use of incomplete and probably as-yet uncorrected data - can you tell me what was the change in heart attacks for the equivalent periods from 1/7/2006 to 30/3/2007, and from 1/7/2005 to 30/3/2006, and so on back in time? You see, a year-on-year change is unremarkable unless it is unexpected, as I am sure you are aware, and you really need to check out any pre-existing trends.
To help you out a little, Rebecca, I obtained data from HES Online (it only took 10 minutes with no FoI request, nor yet even a need to register!), and I can quote to you the following annual changes in the number of Inpatient Admissions, in England (yes, all of it)...
|1999/00 to 2000/01 ||-1.3%|
|2000/01 to 2001/02||+2.5%|
|2001/02 to 2002/03||+5.4%|
|2002/03 to 2003/04||-2.4%|
|2003/04 to 2004/05||+2.8%|
|2004/05 to 2005/06||-0.8%|
|2005/06 to 2006/07||-0.6%|
The first thing you should notice is that the numbers are pretty volatile - even given the scale of a relatively large population as that in England. Hence it is, apparently, not noteworthy that the number of heart attacks increased by over 5% from 2001/02 to 2002/03. How would you explain that, Rebecca? Yet in the following year (2003/04) the numbers fell by about the same percentage as in the ('latest') figures drawn from your sample. Correct me if I am wrong, but was there a smoking ban introduced in 2003/04? No, I thought not!
My (free) statistical advice to you is that in the final shake out, when all data is in, complete and validated, for the whole of England for the whole period from 1 July 2007 to 30 June 2008, you will see that the numbers will be very close to continuing the already downward trend of the past two years (2004/05 to 2005/06 to 2006/07). Even if they do turn out to be nearer to the current 'best-guess' of 2%+, or even close to your original erroneous figure of 3%, this will show nothing remarkable, given the nature of the fluctuations in preceding years.
3 Linking changes in numbers of heart attacks to smoking bans:
I feel that my next set of questions should now be seen as redundant, since there would appear to have been no extra-ordinary change in the number of heart attacks in England anyway, but I will persevere on the pretext that there does need to be an explanation for the change....
3.1 Of the 40,000+ poor souls who were admitted with heart attacks, how many of these were current smokers, ex-smokers and non-smokers? Of the smokers, what changes were there in their individual and average levels of tobacco consumption both prior to and post 1 July 2007? Of the non-smokers, how much exposure to second-hand smoke (SHS) did they experience both prior to and post 1 July 2007? You see, if you don't have this basic information available, you cannot conclude that the smoking ban had any effect on those that did have heart attacks, and thus, by inference, that the ban had any impact on those who did not have heart attacks. so my obvious next question is...
3.2 What evidence do you have that the smoking ban actually prevented one single heart attack? I put it to you that you have no such evidence. Unless you can answer my previous question. Do you understand the principle of cause and effect? Just because something happens in a time-span which is coincident with a change in prevailing circumstances (the smoking ban) it is, literally, impossible to conclude that the latter caused the former to happen - and yes, I do mean literally impossible! You would have to conduct much more detailed research into a) the relationship between your causal factors and the alleged outcome - which apparently you did not do - and also to identify, and eliminate from your statistical analysis, any other possible factors which could result in a similar outcome, (confounders). You clearly did not do this either if you chose to ignore the obvious seasonal variations in the outcome (heart attacks) between the before and after study periods. Hell, Rebecca, you could easily pick any other change that occurred around the time of the introduction of the smoking ban and link that. How about the succession of our dear current PM, Gordon Brown? Wasn't that about the same time. So why not draw the conclusion "Gordon brown's premiership cuts the number of heart attacks by more than 40 per cent at some hospitals"? Ludicrous? Of course it is, but it is seriously no less plausible than the conclusion that you chose to draw (although I have a suspicion that it would cause equal pleasure to your Editor, Mr Dacre!).
As a footnote to 3.2, I am well aware of the reluctance of The Daily Mail to publish any of the horror stories that are emerging all of the time from the licensed trade concerning the devastating effect that the smoking ban has had on pubs both in this country, and in others. It is very sad that the incredibly biased anti-smoking attitude of your editorial team should choose to just ignore the fact that around 1500 pubs have already closed, and still do so at a rate of 30 per week in England, with hundreds, if not thousands of jobs being lost and communities, and people, losing their social lives. Such is the callousness of the anti-smoker. However, I do have a less controversial point to make here. Given that the level of pub trade has dropped do alarmingly since 1 July 2007 (and it has - check it out - and that this drop is mostly due to the smoking ban, this is based on real information) isn't a more plausible explanation that it is the drop in alcohol consumption that is behind the drop (if any) in heart attacks? After all, whatever the risks to cardiac health due to smoking are, they are nowhere near as great as those due to excessive alcohol consumption. Think about it, you might find a more realistic story here!
3.3 Are you aware that there is no biological plausibility to the suggestion that a smoking ban would result in any immediate drop in heart attacks? Please talk to any cardiologist (who isn't blinded by anti-smoking dogma) and you will be reassured about this. It is well known that any effect on cardiac function that would result from smoking takes many years to develop, and it would not just magically disappear immediately because there was an indoor smoking ban. Even if any reduction in heart attacks did result, it would a) be very small (at best single-digit, or low double-digit numbers from your sample), and b) would be only visible after the ban had been in place for a considerable period of time. But you can no doubt tell me: what was the reduction in heart attacks on a month-by month basis across the 9-month period? Was it, as it should be, skewed towards a greater predominance in the later months? I note that, in the print version of your story, your 'Fact File' identified a "drop in cigarette sales of almost 7% in the year to 1 July 2007". Now this figure is highly debatable, since it would no doubt ignore illicit and cross-border sales, but, assuming some reduction did occur, wouldn't this be a much more likely explanation for a fall in heart attacks occurring some 12 months or so after the reduction? Well it makes infinitely more sense to me - but, then again, I don't have to satisfy an agenda, do I?
3.4 Why did you refer to "similar research in Scotland and Ireland that showed hospital admissions for heart attacks fell by 17 and 14 per cent respectively, in the year after the ban came in there during 2006" ? I will forgive you the obvious inaccuracy, in that the Irish ban came into force some 2 years previously, but what I cannot forgive is your failure to acknowledge that both of these 'studies' have shown to be completely fatuous and agenda-driven nonsense. In the example of Scotland, the purported 17% drop in heart attack admissions was announced in a press release to coincide with an antismoking conference in Edinburgh in September last, yet the research has never been published. Why? Even worse, it was only a few weeks later that the official (preliminary) statistics were published by the Scottish NHS which actually showed that the fall was an unsurprising 8%, large compared with England, maybe, but, as with England, totally consistent with such fluctuations in preceding years. Also, as with the English data, the Scottish data was also provisional (with cautionary words on the ISD Scotland website) so the final validated figure for Scotland will almost certainly be even lower than 8%! You can check out the real statistics here (see table AC5).
You can also read a good job of debunking of the Scottish 'study' here on the BBC's website by Michael Blastland. Remember this is the BBC, an organisation that isn't noted for being supportive of smokers' rights, is it?
But if that doesn't convince you, you could do worse than visit the on-line blog of Professor Michael Siegel. Michael is an MD and a long-time anti-smoking epidemiological researcher, who is an ardent supporter of smoking bans, and has lobbied - with much success - for these over many years. This is his take on the 'science by press-release' that is the Scottish study into heart attack reductions.
And while you are visiting Dr Siegel's blog, you should also check out his takes on the Irish smoking ban experiences, and also those in Italy, both of which you refer to, as well as equally egregious examples from North America. Do you recognize the emerging pattern here? Antismoking junk-science, heralded by lurid and massively exaggerated headlines, shown to be totally vapid research under minimal scrutiny. Seem familiar, Rebecca? Anyway, I would really urge you to acquaint yourself with Dr Siegel's blog, since I have a suspicion that your own name will be appearing there in the very near future!
I hope you have managed to stay with me through this rather lengthy interrogation, Rebecca, but I hope that you will now see from whence I come. I rather hope that you will answer all of my questions, but I have little expectation that you will do so in full, nor yet in the spirit of acknowledgment of civility and freedom from harassment that I seek, but is depressingly unavailable to an unreformed and unapologetic smoker such as myself. But let me offer you some friendly advice...
There is a crock of gold awaiting the first open-minded investigative journalist who is brave enough to look beneath the blanket of respectability that shrouds the antismoking movement. The organisations that are most active in the movement are highly critical of the Tobacco companies for their past misdemeanours, and yet they are now so corrupt and dishonest that they make 'Big Tobacco' seem like Benedictine Monks by comparison. These organisations are responsible, by their actions over the years, for probably more premature deaths than can ever be claimed for 'passive smoking', although, just as with passive smoking, non can be proven. If you can't guess who they are, then check back to your story and just see who has responded most enthusiastically to your 'findings', even though they are all well aware that your findings are just garbage!
The truth is out there, waiting to be released, and there are new groups springing up all over the world (such as Forces and Freedom2Choose) who are helping to get the truth out to a wider audience. And when the many, many millions of smokers (as well as ex-smokers and non-smokers) start to realise just how much they have been duped, for such a dishonest cause, I wouldn't want to be in the anti-smokers' shoes! The backlash will be very ugly.
I trust you will endeavour to correct your story in a future edition of the Daily Mail, Rebecca, in fact I look forward to seeing it.
Thank you for your time.
Go To Story »