Why I Shouldn't Read Non-Systematic Review Articles: Special Pleadings and Undercover Authors

I usually resist looking at non-systematic review articles in medical journals, but because the title interested me, and things seem to be getting slow this holiday season, prompted by an update email from the American Journal of Medicine, I looked at Ram CVS. Beta-blockers in hypertension. Am J Cardiol 2010: 106: 1819-1825. (Link here.)

The Ram Article in Praise of Vasodilating Beta-Blockers

The article focused on the results of meta-analyses:

Concerns have also been raised by meta-analyses in which β blockers were reported to have a suboptimal effect on reducing stroke risk and increasing the risk for new-onset diabetes compared with other antihypertensive agents.

The article discussed several meta-analyses in which beta-blockers, [a specific class of blood pressure lowering drugs] but mostly atenolol (mostly sold generically), usually combined with a diuretic, were compared with other antihypertensive drugs, usually including angiotensin converting enzyme inhibitors (ACEIs) and calcium channel blockers. But after discussing these comparative results, the author jumped to descriptions of another group of drugs which were not included in any of these comparative studies. This was the category of vasodilating beta-blockers, consisting of labetolol (mostly sold generically), carvedilol (Coreg, GlaxoSmithKline [GSK]), and nebivolol (Bystolic, Forest Laboratories. Based on physiologic studies of these drugs and trials in which they were compared with placebo, but not on studies which directly compared clinical outcomes of patients given these drugs or other kinds of antihypertensives, Dr Ram reached conclusions that they were a better alternative:

The review of the evidence provided herein confirms that there are valid reasons to question the utility of certain β blockers in treating hypertension. However, many of the perceptions about β blockers are derived from data obtained from studies of traditional agents or combinations of diuretics and β blockers. Evidence suggests, and the guidelines concur, that there are intrinsic differences among members of the β-blocker class. Indeed, the vasodilatory β-blockers, which have generally not been included in comparative meta-analyses, lower blood pressure to a similar degree as other antihypertensive drugs, may provide better central aortic pressure reductions than traditional β blockers, and are associated with neutral or favorable metabolic effects.

Special Pleading

Dr Ram's questioned the old-fashioned beta-blocker atenolol (perhaps only when added to a diuretic)  based on the results of meta-analyses which attempted to compare it to other drugs. Such meta-analyses could suggest that atentolol (again, perhaps only in combination with a diuretic) might be in some way less preferable than the other drugs to which it was compared. However, the meta-analyses did not address the vasodilating beta-blockers at all. Dr Ram concluded that they were preferable based on different kinds of and probably less definitive evidence. He did not seek to compare such studies done on the vasodilating beta-blockers to similar studies done on the conventional beta-blockers.  Thus it seems his conclusions were based on a double standard.

In the vocabulary of logical fallacies, this was an example of a special pleading:

Special Pleading is a fallacy in which a person applies standards, principles, rules, etc. to others while taking herself (or those she has a special interest in) to be exempt, without providing adequate justification for the exemption.

Here was more reason not to bother reading a narrative review articles to learn how to better practice clinical medicine.  They often are based on idiosyncratic, if not biased evidence used to support illogical arguments.

But having read so far, I wondered why the authors of this article were so happy to use a double standard as the basis of their arguments.

Undercover Author

Those paying attention may also now be wondering why I referred to the authors in plural, when the citation lists only one author. The clue is at the end of the article:

Acknowledgment

I would like to thank Tamalette Loh, ProEd Communications, Inc. (Beachwood, Ohio), for her editorial assistance and literature validation in the preparation of this report.

Sensitized as I have been to the many recent discussions of ghost-writing, I immediately wondered who Ms Loh is, and what sort of "editorial assistance and literature validation" she provided. So first I looked at what her company, ProEd Communications does, wondering if its business is to help syntactically challenged academics and professionals write better sentences and more organized papers.

The ProEd Communications web-site states:

ProEd Communications balances diverse perspectives—clinical, regulatory, marketing, and customer—to create compelling messaging to maximize a product's potential.

So it appears that ProEd Communications does not provide independent editorial services. Instead, it is a medical education and communications company (MECC) which works mainly to market its clients' products, and further, its clients are essentially only pharmaceutical companies:

Of the world's 50 largest pharmaceutical companies, ProEd has worked with 18 in the past 3 years and currently supports projects for 7 of the top 10 largest pharmaceutical companies, as ranked by MedAdNews (2008).

Furthermore, a few minutes on Google reveals that Ms Loh seems to specialize in papers on anti-hypertensives, particularly Coreg (carvedilol), one of the three drugs in the vasodilating beta-blocker group favored by Dr Ram. In fact, two other papers whose authors she assisted sounded hauntingly familiar.

Let me first display the quote above from the Ram paper:

Concerns have also been raised by meta-analyses in which β blockers were reported to have a suboptimal effect on reducing stroke risk and increasing the risk for new-onset diabetes compared with other antihypertensive agents.

Now see this quote from Frishman WH, Henderson LS, Lukas MA. Controlled-release carvedilol in the management of systemic hypertension and myocardial dysfucntion. Vasc Health Risk Management 2008; 4: 1387-1400.  (Link here):

However, concerns have been raised recently from hypertension meta-analyses regarding suboptimal outcomes with use of beta-blockers, specifically atenolol, compared with outcomes for other antihypertensive drug classes.

Also see this relatively similar quote from McGill JB. Optimal use of beta-blockers in high-risk hypertension: a guide to dosing equivalence. Vasc Health Risk Management 2010; 6: 363-372. (Link here):

Concerns about the use of β-blockers as first-line agents for hypertension have been raised because of a 2005 metaanalysis that found β-blockers do not significantly reduce
cardiovascular events, especially stroke, compared with other antihypertensive drug classes

Furthermore, read this discussion of Coreg by Ram:

Carvedilol is a nonselective β blocker with α1 receptor–blocking activity and no intrinsic sympathomimetic activity. Clinical data suggest that carvedilol reduces systemic vascular resistance in patients with hypertension.

Here is Frishman et al:

Carvedilol is a third-generation, vasodilatory beta-blocker that nonselectively blocks both the beta 1- and beta 2-adrenergic receptors.... vasodilatory beta-blockers can lower blood pressure by reducing systemic vascular resistance (SVR)

Both Frishman et al and McGill are very positive about Coreg.  Although the emphases of the three articles are different, they have organizational similarities.  Again, all three were written with the assistance of Ms Loh.

At the end of the Frishman article we again find:

The authors would like to thank Tamalette Loh, PhD, ProEd Communications, Inc.®, for her medical editorial assistance with this manuscript.

At the end of the McGill article we find:

Editorial assistance, specifically revisions to the final draft, was provided by Tamalette Loh, PhD, at ProEd Communications, Inc.®, whose services were also funded by GlaxoSmithKline. Dr Loh’s revisions were reviewed and approved by Dr McGill.

Note that now it seems that Ms Loh has a doctoral degree, of unclear kind. Note also that now it seems that GSK, the manufacturer of Coreg, funded Dr Loh's work on the McGill article.

Dr McGill further disclosed:

Dr McGill is a consultant for GlaxoSmithKline and a speaker for AstraZeneca and Forest Pharmaceuticals. Financial support for medical editorial assistance was provided by GlaxoSmithKline, Philadelphia, Pennsylvania.

Finally, in another article, about angiotensin converting enzyme blockers combined with diuretics to treat hypertension (Egras AM, Ram CVS. Reduced cardiovascular risk and healthcare expenditures with angiotensin receptor blocker/ hydrochlorthiazide. Am J Pharmacy Benefits 2010; 2: 127-135. Lin here. ), Dr Ram acknowledged:

Dr Ram is in the speakers’ bureau pool of Cogenix, ProCom, and Genesis, which manage medical education programs for Bristol-Myers Squibb, GlaxoSmithKline, and Novartis.

The series of articles above demonstrated a phenomenon I have not seen explored before. All articles were written with some sort of assistance from an employee of a MECC, perhaps partly funded by a company which marketed a drug which was the subject of all the articles. The assistance was openly acknowledged.  The three articles had some remarkable similarities not explained by an overlap among their listed authors.  This suggested that the ostensible editorial assistant they had in common was substantively involved in the content of the papers, that is, was truly an author. Her presence was not ghost-like. However, the substance of her contribution may have been downplayed.  So let's call her an "undercover author."  (If someone has a better term, please leave a comment to that effect.)

Summary

So the reasons I rarely read narrative review articles except to make teaching points about health care dysfunction are:
- They often are based on idiosyncratic, if not biased selections of data
- They may employ logical fallacies to make their points
- They may be written by people with conflicts of interest, that is, with financial arrangements with companies seeking to market products, particularly drugs and devices.

My conclusions for health care professionals are: be very skeptical of non-systematic review articles, look for evidence that they are parts of stealth marketing campaigns, and do not assume all conflicts of interest are disclosed and all authorship roles revealed.

My conclusions for journal editors are: strictly demand more complete disclosure of conflicts of interest, or risk losing the trust of your readers.

My conclusions in general: until we start to sweep away the pervasive web of conflicts of interest that is draped over medicine and health care, expect further discombobulation.