Now she has returned with a study of similar quality and integrity. This time the narrative—as reported in the worldwide press release (repeated verbatim by The Scotsman)—is as follows:
Child asthma admissions drop 18% per year since Scottish smoking ban
The rate of hospitalisations for children with asthma in Scotland has dropped by more than 18 per cent year-on-year since the introduction of the ban on smoking in public places in 2006, according to scientists.
In a study led by Professor Jill Pell in the Centre for Population Health Studies at the University of Glasgow, researchers analysed data on hospital admissions for asthma in Scotland from January 2000 through October 2009 among children younger than 15 years of age.
As with Pell's last travesty, the study has been published in the allegedly peer-reviewed New England Journal of Medicine. The key graph is shown below (click to enlarge). The smoking ban (March 26 2006) is marked with a grey line.
From this graph, the unwitting reader might get the impression that childhood asthma rates were rising gently before the smoking ban, peaked in the first year of the ban and then went into a steep decline. That, indeed, is how the study has been reported.
Before the smoking ban came into force, admissions for asthma were increasing at a mean rate of 5.2 per cent a year. After the ban, admissions decreased by 18.2 per cent per year, relative to the rate on March 26, 2006.
But did they? The graph above purports to show the average daily rate of hospital admissions for childhood asthma. But it also shows the number of admissions in each year, so we can easily work out what the average daily rates were (taking into account leap years and the fact that the final 'year' (2009) actually ends in October.
2000: 2391/366 = 6.53 per day
2001: 2142/365 = 5.87 per day
2002: 2034/365 = 5.57 per day
2003: 1803/365 = 4.94 per day
2004: 2621/366 = 7.16 per day
2005: 2103/365 = 5.76 per day
2006: 2633/365 = 7.21 per day
2007: 2056/365 = 5.63 per day
2008: 2235/366 = 6.11 per day
2009: 1397/304 = 4.59 per day
Plotting all this on a graph, we can see that childhood asthma rates were not rising before the ban and the only evidence for even a vague drop since the ban comes from the incomplete ten-month 'year' of 2009—several years after the ban came in. And, again, we can see that the peak year for asthma hospitalisations came in 2006—the very year that the smoking ban came into effect, which—by the logic of the study—should have seen a large drop in admissions.
I am at a loss to explain how Pell transformed the mundane, patternless hospital admissions data into the graph published in the NEJM study. The data simply do not fit the chart. She says only that she applied "smoothing" to the graph, but that alone cannot explain the discrepancies (surely?).
There is no indication in the text that these particular figures have been adjusted. Nor is there reason to think that using calendar years—rather than April to March—makes much of a difference. ISD Scotland only provides data for three financial years (using these are more appropriate since the ban took place at the end of the financial year), but, again, these data show a peak in the first year of the smoking ban and no real decline thereafter:
Hospital admissions for asthma for childhood under 15 years old
2005/06: 2,182 (2.5 per 1,000 population)
2006/07: 2,603 (3.0 per 1,000 population)
2007/08: 2,061 (2.4 per 1,000 population)
All of which raises some big questions...
How has she managed to make it look like the peak in admissions came in 2000 when it came in 2006?
How has she managed to make it look as if rates fell continuously from 2007 when the rate rose in 2008?
How can she claim that "there was a mean reduction in the rate of admissions of 18.2% per year relative to the rate on March 26 2006"? Over three years, that equates to a 45% decline, which is crazy and obviously didn't occur, so where does this figure come from?
How can she claim that rates were rising by 5.2% per year when the rate fell for four out of five years before the ban? How, indeed, can anyone look at this data and seriously claim that there was a "year-on-year" rise or decline at any stage over this period?
Accepting that the rate in 2009 was unusually low, how can this plausibly be attributed to a smoking ban which began years earlier? The first year of the ban saw the highest childhood asthma rate of the decade and the next two years were in line with the pre-ban average.
So, dear readers, it's over to you. Is this the worst piece of pro-smoking ban junk science yet, or is there a perfectly innocent explanation?