Monday, August 04, 2008

IN MY VIEW: Another Reason Why Conclusion of Scottish Smoking Ban Study is Invalid - Effect May Have Been Due to Diagnostic and Treatment Changes

I have already explained why I think the conclusion of the study which reported a 17% decline in acute coronary syndrome admissions in 9 Scottish hospitals following implementation of the national smoking ban is invalid. In that previous post, I focused on the failure of the study to establish a valid baseline of acute coronary syndrome admission in these 9 hospitals, and therefore, its inability to be able to comment on whether the observed 17% reduction is something different from that seen in the past.

The study reported a decline in acute coronary syndrome admissions from 3235 to 2684, comparing the ten-month period preceding the smoking ban with the corresponding period after the smoking ban. Now suppose that the number of acute coronary syndrome admissions in the corresponding period of the previous year was about 3300. Then, it would be clear that this 17% reduction represents something very different - there was basically no decline in the previous year.

However, suppose the number of acute coronary syndrome admissions in the previous year was about 3800. Then, the observed 17% reduction associated with the smoking ban would no longer be so impressive, since there was a 14% reduction during the previous year.

Well - which of these scenarios is the true one? The validity of the study conclusions rests entirely on the answer.

Unfortunately, the answer is unknown. The study does not report the number of admissions for acute coronary syndrome - using the same diagnostic criterion of an elevated cardiac troponin level - during the previous year. Thus, there is no way to know whether the 17% reduction is something new and different, or whether there already was a trend of sharply declining admissions.

As a result, the study findings simply are not enough to support the study conclusions. The research can surely conclude that there was a sharp decline in acute coronary syndrome admissions associated with the smoking ban, but the research cannot conclude that the decline was attributable to the smoking ban.

Here, I present another reason why the study conclusion is not valid: there is a very plausible alternative explanation for the observed decline in acute coronary syndrome admissions.

The Rest of the Story

What seems to have gone largely unnoticed in the article and the news coverage around it is the fact that another major heart disease intervention took place in these 9 Scottish hospitals around the time of the smoking ban: these hospitals used a relatively new system for diagnosing heart disease -- cardiac troponin assays.

Previously, the diagnosis of unstable angina and silent heart attacks (those without any symptoms) was problematic because of the lack of a sensitive laboratory test. Creatine kinase (CK) - the test previously used - is not sensitive enough to be able to pick up all cases of unstable angina and it may not pick up silent heart attacks that occurred in the recent past. However, cardiac troponin is a much more sensitive test, and it is able both to pick up very small areas of myocardial ischemic injury as well as injury that occurred in the recent past. For this reason, the use of cardiac troponin brings with it the ability to diagnose many more cases of severe cardiac disease that might not have been recognized in the past.

When physicians know about the presence of ischemic injury, they can take a more aggressive treatment approach. Ultimately, this is expected to reduce the incidence of future heart attacks in these patients.

This is all explained rather nicely in this article on Emax Health:

"Patients who test positive for the presence of a specific biochemical marker of heart cell death in their blood but who do not exhibit other risk factors for future heart attack should be treated as higher risk patients, according to a new analysis by cardiologists at Duke University Medical Center.

The researchers found that troponin, a protein that is released into the bloodstream as heart muscle cells die, can be a reliable indicator of future risk even when other traditional measures of heart health are negative. This is important, the researchers say, in light of the recent recasting of the definition of heart attack by the major cardiology organizations to place more emphasis on the results of troponin testing, in addition to the presence of chest pain and electrocardiogram abnormalities.

When a patient comes to the emergency room as a possible heart attack victim, physicians typically measure the heart's electrical activity (EKG) and also look for chemicals in the blood that might indicate if heart muscle is damaged or dead. For years, they have measured the levels of creatine kinase-MB (CK-MB) and, more recently, they also have been testing for troponin. Because the test for troponin can detect even small amounts of heart muscle damage, patients previously testing positive for troponin but negative for CK-MB were not always treated as being at high-risk.

'Our analysis shows that patients who test positive for troponin but not for CK-MB should still be treated as if they tested positive for both,' said Duke cardiology fellow Dr. Sunil Rao. 'This is important because in the past, these patients would usually not be treated aggressively.'"

An article by Drs. Brian Go and H. Vernon Anderson nicely explains how the use of cardiac troponin results in the earlier identification of patients with unstable angina, which can lead to more aggressive treatment. As these cardiologists explain, an elevation of troponin is a risk factor for future heart attacks. Thus, these represent patients who previously may not have been treated as aggressively and therefore would be at high risk for heart attack. With more aggressive treatment, heart attack risk can be reduced, at least in the short-term.

The bottom line is that a plausible explanation for the observed 17% reduction in acute coronary syndrome admissions in the 9 Scottish hospitals in the study is that the intervention (that is, the use of cardiac troponin testing for all patients with chest pain or suspected heart disease), rather than the smoking ban, caused this reduction.

Without knowing the effect on acute coronary syndrome admissions of using this new diagnostic system, it is impossible to conclude that the 17% reduction was due to the smoking ban, rather than to this important diagnostic and therapeutic change.

It is not clear exactly when the intervention was implemented in these 9 hospitals. Presumably, it was not in place in all 9 hospitals until 10 months prior to the smoking ban - this is presumably why the researchers included only data for this 10 month period in the study. If the intervention was in place in all 9 hospitals well prior to that, then the obvious question would become: why does the paper not present the data on admissions for acute coronary syndrome during the entire period in which the intervention was in place in these hospitals?

It is important to emphasize that I am not questioning here the validity of the study's conclusion that there was a 17% decline in acute coronary syndrome admissions. What I am questioning is the attribution of that decline - in its entirety - to the smoking ban, rather than to improved diagnosis and treatment, which is precisely the reason that the cardiac troponin system was put in place in these hospitals.

No comments: