algorithmic psychiatry – so little, so badly…

Posted on Saturday 28 May 2011

Who ever knows how history will be written in the future when it longer holds us in its grip. Will the Age of Corruption start with TMAP, maybe George W. Bush, Reagan’s Deregulation, Nixon’s Watergate, or maybe go back to LBJ’s decision to fabricating the Gulf of Tonkin? I doubt that TMAP and its spawn will deserve even a paragraph in this chapter [though the pharmaceutical invasion of psychiatry after 1980 may be a contender for honorable mention]. But from right here, it’s still a pretty big story what happened in Texas in 1996 with the Texas Medical Algorithm Project.

Dr. Rush arrived in Dallas around 1978 as a Cognitive Behavior expert. By the time Dr. Trivedi showed up [Trivedi’s first paper] in 1993, they were collaborating on a biology of depression paper. In 1996, they were both involved in setting up TMAP [Texas Medical Algorithm Project] for the State of Texas – a joint project of UT Southwestern, Pharmaceutical contributions, and the Texas Mental Health system. On the face of things, it seemed benign. The primary task was to create a Board to decide what medications would be approved for the vast public system in Texas [Medicare, Medicaid, Prisons, Mental Health Centers]. The twist was the idea to have treatment guidelines, or algorithms, to standardize and maybe optimize practice. Sounds reasonable. The program came at a time when there weren’t any options for generic SSRIs, but it was on the leading edge of the wave of atypical antipsychotics at a time when hopes were high for those drugs [as were prices] to replace the older neuroleptics. So it was no surprise that the atypicals appeared on the Texas State formulary. From Allen Jones:

    TMAP concluded that the Atypical antipsychotic medications Risperdal, produced by Janssen Pharmaceutica, Zyprexa produced by Eli Lilly, and Seroquel, produced by Astrazeneca, are the drugs of choice for all first, second, and third-line treatments of Schizophrenia.
    TMAP concluded that all newer, patented anti-depressants were superior to generics.
    TMAP concluded that the patented bi-polar drugs were superior to generic drugs.
    TMAP concluded that “Expert Consensus” established these drugs to be safer, more effective, better tolerated and relatively free of side effects when compared to the older, generic, medications.
    TMAP then formulated separate “algorithms” (flow charts) and drug menus for the treatment of schizophrenia, depression and bi-polar disorder. All of the new, patented drugs were incorporated into the TMAP algorithms.
    State doctors following the algorithms were and are required to use these drugs. The administrative decision of a State Mental Health Program to adopt TMAP brought with it the clinical decision to use the recommended drugs on all patients in the state system. A state doctor may choose which patented drug to use, but he may not choose to use a generic drug until at least two, often three, patented drugs have failed.

John Rush directed TMAP assisted by psychopharmacologist Steven Shon. Madhukar Trivedi headed the Depression section. They set about researching their treatment algorithm for Major Depressive Disorder first thing in 1998. Their study was something of a John Henry versus the machine affair – pitting public clinics using their complex algorithms [ALGO] against treatment as usual clinics [TAU]. This time, the machine [ALGO] won [though the results were disappointing overall]. They quickly designed a $35 M study for a similar enterprise through the NIMH using a modified algorithm. They published a sea of articles on TMAP, but the two definitive articles didn’t come out until after STAR*D [the NIMH study] was underway. One article [Trivedi] was about John JHenry’s defeat. The second explained the low response rates as being due to their chronic impoverished cohort of patients. STAR*D was aimed at a higher class clientele. Steven Shon was also on the move, jetting around the country and the world pitching TMAP [as it turned out, on the pharmaceutical dollar]. He engaged 16 states in adopting some version of the program. Drs. Rush and Trivedi were being well reimbursed for their active participation in pharmaceutical paid speakers bureaus, scientific advisory boards, and continuing education presentations. Meanwhile, the Texas public medications budget was nearing bankruptcy.

Once STAR*D was underway, Trivedi had time to start on an idea he’d been writing about for a while – computerizing his algorithms. So he started with the Major Depressive Disorder. He got $3M from the NIMH and began to churn out papers about his coming software and speculations about how to launch it. His study began recruitment in the TMAP clinic population at the end of 2003 [CDSS-D = Computerized Decision Support System for Depression][PPA-D = Paper and Pencil Algorithm for Depression][UC-D = Usual Care for Depression]:
    The patient outcomes in the CDSS-D, PPA-D, and UC-D will be evaluated in terms of clinical symptoms, function, and quality of life [aim 1] assessed by an evaluator, blind to treatment assignment, every 3 months for a total of 12 months. The primary aims of the study are to: [1] assess the of CDSS-D implementation in terms of patient outcomes [reduction of symptoms, changes in level of functioning, and quality of life measures]; [2] conduct an assessment of cost and utilization of the CDSS-D in a the public mental health sector; [3] directly assess the extent that the CDSS-D influences physician behavior; and [4] To compare the difference in patient outcomes with the difference in cost in the CDSS-D. Comparisons will be made between CDSS-D, PPA-D, and UC-D. 
Starting that project at that time in that place turned out to be as unwise as breaking ground for an up-scale housing development in early 2008. In the next year, Allen Jones had begun to investigate TMAP in Pennsylvania which had adopted the program and gotten fired . He had questioned the expenses and discovered that Steven Shon’s constant traveling was being financed by the pharmaceutical industry. Jones file a suit in Texas. And in 2006, the Attorney General of Texas had the State join the suit. TMAP was under siege and Shon was forced to resign. Trivedi’s recruitment was going badly. He continued to publish prequels. As they wound up the messy STAR*D study, they planned another – CO-MED – involving starting patients on two antidepressants up front in hopes of improving their success. The STAR*D papers kept coming.
When Senator Grassley began to investigate academic psychiatrists and their conflicts of interest in 2008, John Rush’s name was on the list.  He left UT and took a job with Duke in Singapore. Eric Nestler, Chairman at UT headed back east to Mt. Sinai. Trivedi wound down the ill-fated CO-MED study and his IMPACTS computer project, having never completed recruitment. There was a CDSS-D project in a few primary care clinics that did get finished [small, statistically significant, and irrelevant].

All told, these projects which produced no usable information cost the NIMH around $50 M, and the State of Texas much more [Billions?]. Trivedi’s IMPACTS [CDDSS-D] never made it off the ground. TMAP is long abandoned. CO-MED was a bust all around. And STAR*D would best be retracted as simply bad science with an unreadable outcome. As for John Rush, he’s in Singapore doing whatever he does there, but is involved in Brain Resources personalized medicine project. Trivedi is Principle Investigator in an NIMH study of the same thing – personalized medicine. It’s a new day with a new fad.

Why these two investigators, John Rush and Madhukar Trivedi, who have thrown away so much money doing so little so badly are still in the game is a mystery to me…
    May 28, 2011 | 9:10 PM

    It is no mystery to me. Charles Nemeroff and Tom Insell. Enough said.

    May 30, 2011 | 11:05 PM

    I’ve fallen way behind in this general discussion but I thought we solved Trivedi et al months ago! Isn’t he the twit who was in a huff because medical doctors didn’t behave like automatons in the “real” world?

Sorry, the comment form is closed at this time.