- The American Conservative - http://www.theamericanconservative.com -

Robert Caro and the Lost History of the 1960s—Plus a Vioxx Note

The recent publication of the fourth long volume of Robert Caro’s biography of Lyndon Johnson demonstrates how much even the relatively recent printed past has almost totally disappeared from current consciousness.

Consider the 1958-1964 period covered by Caro’s current narrative, an era which might reasonably be called the political peak of Cold War liberalism, in which Caro focuses on the political maneuvers leading to Kennedy’s nomination and Johnson’s difficult years in the vice presidency. Many people have argued that the major political decisions made during the 1960s largely shaped modern America, but it is equally true that the political decisions described in Caro’s volume largely shaped those same 1960s. Yet what determined the political tide of those years and which media narratives shaped those decisions?

Obviously, there is no one answer to this important question. Certainly television was coming into its own during exactly this period, and many historians describe the very tight 1960 Kennedy-Nixon race as being the first whose outcome was decided by television, with a sweating Nixon losing the debate for those who watched it on television, while simultaneously winning the same debate for those who only heard it on radio. But even far more than today, the electronic media back then was largely derivative in its ideas, merely amplifying and broadcasting nationwide the thoughts of its writers and narrators, whose own views were almost always shaped by the printed word.

The daily newspapers played a huge role, but while their numbers and circulation were far greater than today, they remained almost entirely regional or local, except when delivered days or weeks late to the small out-of-town newsstands which have now almost totally vanished from our physical world. The New York Times and its local rivals might help shape the opinions of the broadcast industry workers who lived within their delivery area, but they were not national newspapers, and unless occasionally reprinted on the wire services, their daily stories would have remained almost completely unknown to perhaps 95% of America’s educated population. Nationally-syndicated columnists often carried huge political weight, but their word length was strictly limited, and rarely afforded enough space to substantially change minds on major topics.

Then as now, major books often were crucial, with Rachel Carson’s 1962 “Silent Spring” [1] helping to create the environmentalist movement while Michael Harrington’s “The Other America,” [2] published that same year, played a similar role in launching the fated War on Poverty. But people discovered these books by reading about them, and on the national level, the main way they read about them was in reviews or discussions published in national magazines.

With a special exception here and there—notably the New York Times—magazines largely determined the fate of books, but they did far more than that as well. A lengthy magazine article, powerfully written and perhaps backed by the visual impact of numerous gripping photographs, might easily match the psychological impact of a transitory one-time television broadcast, and have just as much national reach, but with far greater intellectual depth; few other media sources were remotely competitive. When we chart the political and ideological history of the 1950s and 1960s, a remarkable fraction of the intellectual branch-points seem determined by the impact of crucial articles in magazines, a situation which might seem totally alien to today’s overflowing media landscape of endless talking-head cable-news shows and Internet information gushers. The ideational world back then was ruled by Print and shaped by information scarcity, and to a considerable extent, the magazines dominated it.

But which magazines? Certainly the Big Three of Time, Newsweek, and US News had among the largest circulations, with tens of millions in total combined readership. However, they also tended toward the sort of straight-weekly-news summary reporting indicated by a name such as “Newsweek,” and it is difficult to recall too many landmark articles which they published from that era. At the other end of the spectrum, there were the small intellectual magazines from all different points across the ideological compass, including those at the extremes, but these publications usually had tiny circulations, and are perhaps better considered as being part of the “meta-media”—the media of the media—than the media itself. So in understanding the shaping influences of the Cold War liberalism of the late 1950s and early 1960s, there are surely worse places to begin than by considering the broader circulation magazines whose coverage extended well beyond straight news summarizing.

Consider the various possibilities. National Review, then as now, was probably the most influential conservative magazine, sometimes sparring with the harder-right American Mercury, while the Freeman largely focused on free market topics. But while Buckley’s magazine might prod and poke and denounce the dominant liberalism of that era, it can hardly be said to have regularly shaped it. Meanwhile, such venerable political voices as The Nation and The New Republic were both quite far to the Left in those years, largely outside the Cold War consensus, and the same was certainly true of IF Stone’s popular weekly newsletter. And a prominent general interest periodical such as The Atlantic Monthly appeared just twelve times each year, tending to lack the immediacy of something with shorter publication cycles and able to cover developing events.

So this simple process of elimination leaves us with two plausible candidates for inclusion in the small handful of media organs which most shaped this American era. Their names were once powerful in elite circles, but today they are almost totally forgotten and unfamilar even to most well-educated Americans: The Saturday Review [3] and The Reporter [4]. The Wikipedia entry [5] for the first of these is brief, almost skeletal, amounting to just a couple of substantive paragraphs summarizing the basic facts of a fifty year history, while the latter merits only the merest stub [6] with four short sentences. Yet a prominent scholar personally confirmed to me that for decades in America the two most important potential book reviews were those appearing in the New York Times and in the Saturday Review, while the Reporter’s Wikipedia stub actually mentions that in 1962 it ranked only behind the three bland newsweeklies as the most prominent referenced source for working journalists. Imagine if future individuals a few decades from now were attempting to understand the history of America during the 1990s and 2000s but had never even heard of the New York Times or the Wall Street Journal.

Since the era in question is before my time, my reconstruction is hardly conclusive, but a cursory examination of Caro’s index tends to support this analysis. Although the Reporter vanished over four decades ago and is almost forgotten, while its archives were completely unavailable in any electronic form when Caro was producing his book, it still warrants more appearances in his text than the combined total for the New Republic, the Congressional Record, the Congressional Quarterly, the Atlantic Monthly, Harpers, National Review, and the Nation, with several of these latter publications being completely absent from Caro’s 700 pages of text. Reading the pages of the Reporter may not necessarily provide us an accurate view of the reality of those years, but it should provide us with a reasonable sense of what leading journalists and other mainstream elites perceived to be the reality, which is a very useful thing to know.

In fact, I suspect that individuals seeking to gain a deep understanding or a feel for the years covered by Caro’s history could do far worse than simply browsing the pages and reading some of the articles and columns of those two magazines, now conveniently available online at UNZ.org [7]. In particular, “time-slicing” allows us to easily browse the hundreds of source articles appearing around the time of a major event, such as the Cuban Missile Crisis of October 1962 [8]. Given the exceptional thoroughness of Caro’s research methods, I would not be surprised if he won’t at least lightly skim those particular archives almost cover-to-cover during his ongoing research for the final volume of the Johnson history.

Others are gradually discovering the periodical archives as well, with the conservative-libertarian Acton Institute recently highlighting the research value of the system in an item very generously entitled “Unz’s Amazing Archive,” [9] which notes that spending ten seconds online is far much more convenient than a long drive to the Library of Congress for most researchers.



Meanwhile, on a completely different matter, the recent column in The Week by Alex Cockburn focusing on the implications of my Vioxx mortality analysis was redistributed by ReaderSupportedNews.org, a prominent liberal website, as its lead story [10] and currently ranks as the most popular item, having generated over 800 Facebook Likes, many dozens of Tweets, and all the other signifiers of social-networking success, as well as a vast outpouring of comments. Most of the latter were enthusiastically favorable, but one particularly agitated commenter, “pschaeffer”, vigorously disputed my analysis at exhaustive length, both there and also on various other websites, so perhaps I should briefly respond to some of his arguments.

First, I wish to emphasize that my Vioxx analysis [11] was entirely circumstantial. Vioxx had been marketed to the elderly and was withdrawn in 2004 because it had been scientifically proven to greatly increase the risk of death from cardiovascular causes, such as heart attacks and strokes. I noted that in 2004 the American death rate had experienced its largest drop in 60 years, a drop otherwise totally unexplained and mysterious, and that this drop had been overwhelmingly concentrated in the elderly, driven by a sharp decline in cardiovascular fatalities. I also noted that when Vioxx had been introduced in 1999, that year had seen the sharpest rise in American mortality rates during the last fifteen years, and once again, this change was overwhelmingly concentrated in the elderly and driven by a rise in cardiovascular deaths. This was all presented in my original article and I do not believe that any of these basic facts are under dispute.

The counter-argument advanced by “pschaeffer” is that these large mortality shifts could not possibly have been due to Vioxx, based on various detailed and rather complex analyses of the underlying mortality statistics and Vioxx prescription records. Readers are welcome to examine these arguments for themselves at length. In particular, it is suggested that there is a slight mismatch between the timing of Vioxx availability and the remarkable changes in 65+ mortality rates. Furthermore, those large shifts in 65+ mortality were largely confined to the years 1999 and 2004, despite the heavy Vioxx use during the intervening period.

The difficulty with these arguments is that they seem to amount to the counter-claim that the dramatic changes in American mortality rates noted above were purely coincidental and based on random fluctations, despite their extremely close match with the years, target population, and mortality risks proven to be associated with Vioxx. Speaking as someone with a scientific background, the likelihood of such a pattern of remarkably consistent random statistical coincidences seem quite implausible to me.

The precise number of likely Vioxx-related fatalities is not something I claim to know and the hypothetical possible total of 500,000 I originally advanced was hedged with huge uncertainty. Meanwhile, the disputatious commenter claims the official figure of 30-60,000 is quite plausible. But when we consider that Vioxx was withdrawn in the middle of 2004 and the number of American deaths for that one year alone unexpectedly fell by 50,000, I suspect my estimate is far closer to the truth. Furthermore, well over two-thirds of Vioxx use occurred overseas, so the total premature death toll worldwide from that one drug could easily top the million mark.

The word “premature” is absolutely crucial. I have never claimed that Vioxx killed people who otherwise would not have died. After all everyone dies, and elderly people in poor health much sooner than most. It seems plausible that Vioxx usually took just a few years off the lives of most of its victims, and in effect “time-shifted” their deaths forward a bit. Under such assumptions, any large impact on the American death rate would occur at only two points, namely when Vioxx was introduced in 1999 and when it was withdrawn in 2004, and this is exactly what we find in the data.

Furthermore, since my examination was purely statistical, I make no claims whatsoever about the precise mechanism behind the sharp swings in mortality, simply that there are clearly associated with widespread Vioxx use. Consider, for example, that Vioxx was marketed as an aspirin-substitute and Merck spent half a billion dollars promoting that highly lucrative drug against its ridiculously cheap rival. Over the years, there have been various research claims that regular aspirin use can act as a powerful protective agent against heart disease and strokes. If true, then perhaps much of the increased mortality pattern associated with Vioxx came less from direct Vioxx harms than from the loss of aspirin benefits, and this might account for the slight timing mismatch between the Vioxx and the mortality shifts. Or perhaps the main problem was some deadly combination of Vioxx together with various other drugs or treatments—I have absolutely no idea. But the remarkably close correspondence of American shifts in Vioxx use and shifts in mortality rates seems almost certainly more than pure coincidence.

(cross-posted at www.ronunz.org [12])

Comments Disabled (Open | Close)

Comments Disabled To "Robert Caro and the Lost History of the 1960s—Plus a Vioxx Note"

#1 Comment By peter_schaeffer On May 15, 2012 @ 6:38 pm


From 1998 to 1999 Vioxx prescriptions rose from zero to
4.845 million and death rates rose. From 1999 to 2000 they rose by 15.785
million and death rates fell. Given that the larger change in Vioxx
prescriptions was associated with a decline in death rates, this would tend to
indicate the Vioxx reduced mortality.


From 2003 to 2004 Vioxx prescriptions fell by 5.965
million and death rates fell. From 2004 to 2004 Vioxx prescriptions fell by
13.994 million (to zero) and death rates rose. Given that the larger change in
Vioxx prescriptions was associated with an increase in death rates, this would
tend to indicate the Vioxx reduced mortality.


Of course, it is unlikely (but not inconceivable) that
Vioxx reduced mortality while it was on the market and the withdrawal of Vioxx
increased mortality. A more likely explanation for the observed data is that
other factors (much) more than offset the impact of Vioxx either


However, it turns out that more sensitive analysis of
the data is possible. Vioxx was introduced in late May of 1999 and withdrawn on
September 30th of 2004. The CDC publishes monthly mortality statistics that can
be used to compare the actual 12 months before and after Vioxx was introduced
and the 12 months before and after Vioxx was recalled.


It turns out that the average monthly death rate was
almost identical (70.575 versus 70.599) for the 12 months before and after Vioxx
was introduced. This is a better test than using calendar years because Vioxx
was only sold for 7 months in calendar 1999. By comparing 1998/06 – 1999/05 to
1996/06 – 2000/05 we are testing 12 months of no Vioxx (some Vioxx in late May
1999 is possible) versus 12 months of full marketing.


The same approach can be used to analyze the Vioxx
withdrawal. For the period from 2003/10 – 2004/09 the average monthly death rate
was 69.444. For the 12 months after Vioxx as recalled, the average monthly death
rate was 68.708. Net, the average monthly death rate declined by 0.736. By
contrast, the average monthly death rate for calendar 2003 was 70.175 versus
68.107 for calendar 2004 (a net decline of  2.068). Using actual 12 month
periods is a better test because Vioxx was prescribed for 9 months in 2004
making comparisons with 2003 difficult. Stated differently the death rate
decline, using actual 12 month periods was 35.58% of the decline using calendar
year periods.


Was the 0.736 decline in average monthly mortality a
consequence of the Vioxx recall? Perhaps, perhaps not. Most of 0.736 decline
appears to be related to a large P&I (Pneumonia and Influenza) epidemic in
late 2003 / early 2004. Of course, the data is predictably more complex. The
monthly death rate in March of 2005 was higher than the death rate in March of
2004. Once again, Vioxx appears to have been a secondary (or lesser) influence
on mortality.

#2 Comment By peter_schaeffer On May 15, 2012 @ 8:01 pm

Please note that the death rates in the note below are raw death rates, not age-adjusted. The CDC monthly death data is not age-adjusted. It would appear that the age-adjusted death rate fell in the 12 months after Vioxx was introduced, compared to the prior 12 month period. This is not a claim that Vioxx “saved lives”. More that whatever adverse impact it had, was not sufficient to offset other larger influences on mortality.

 Note that the statistics in the TAC article appear to be raw death rates, rather than age-adjusted death rates.