Thursday, January 03, 2008

Why Pharma Fails

... from the Medical Informatics perspective.

(I've previously written on this issue at "Drug industry officials see room to improve safety - or do they?" and at "FDA's hefty bonuses seek to retain workers.")

The following job ads appeared recently for a Business Analyst (BA), Clinical Research Information Services and a Business Analyst, Clinical Data Repository Services at a major pharmaceutical company. They are similar to ads that have appeared at least for the past several years from multiple pharmas.

First and foremost, conflation of the term "business analyst" with "clinical research information" suggests a profound lack of understanding of the difference between information technology (especially mercantile, manufacturing, and management computing applications) and information science (including scientific computing applications, of which clinical computing is a subspecialty area).

The mission of the Business Analyst (BA), Clinical Research Information Services is:

Working as a Business Analyst (BA) within the CRIS organization — Business Analysis & Product Management team, the incumbent is primarily responsible for collaborating with colleagues to develop business cases, defining requirements specifications, and assisting business colleagues with the adoption of new capabilities for Clinical Development. The primary clients for this area are Clinical Research, Clinical Operations, Data Management, and other areas responsible for the execution of clinical trials ... The primary clients for this area are BARDS, WCDMO, Clinical Research, Molecular Profiling, etc. The incumbent will work with business colleagues to define new capabilities for the CDR, inclusive of data deviations and the instantiation of business standards/rules.

There seems a major disconnect between the issue of developing "business cases" and the needs of clients "responsible for clinical trials." Exactly what is a "business case" regarding the conduct of a clinical trial, a scientific endeavor used to detect drug efficacy and hazard?

The requirements for the Business Analyst, Clinical Research Information Services:

Required:Minimum of 5 years of business analysis experience in an Information Technology or Business Process Re-engineering role.

Say What? business analysis IT or business process re-engineering experience? Isn't that once again saying that experience in "merchant" computing is a qualification for clinical trials computing? How does "business process re-engineering" figure in? Exactly what qualifies an IT person to "re-engineer" anything related to conduct of clinical trials?

Lean/Six Sigma Green-belt or Black-belt certification is strongly preferred. Demonstrated ability in the practical application of business analysis techniques and tools is required, e.g. Rational Unifed Process, Unified Modeling Language ... Leadership experience a strong plus, ability to serve as a mentor and/or process consultant is required ...


Six Sigma is a set of practices originally developed by Motorola to improve processes by eliminating defects, largely regarding manufacturing of widgets. How is that a key requirement for "adoption of new capabilities for Clinical Development?" Just as no management fad or "process fanaticism" can turn an amateur pianist into a Beethoven, "Six Sigma" is no substitute - not even close - for people who actually know at its core, at a fine-grained level, exactly what they're doing, and not only in general or "working knowledge" terms. "Six Sigma" in this domain is one of many "fad" solutions promoted by IT biomedical information science amateurs as solutions to the wrong issues. Put succinctly, a hundred generic workers following the finest of process will always be outperformed by one person who actually knows what they're doing:

Desired: Working knowledge of clinical development work processes is preferred. Education Requirements:B.S.degree is required. M.S. degree with concentration in Information Management/Information Technology or related scientific discipline is a plus.


M.S. degree with concentration in IM/IT a "plus?" "Related scientific discipline" such as ... what? Do scientists labor in the conduct of clinical trials so that bachelors-level or masters-level computer geeks can have good jobs?

Finally - and this is critically important - where is a mention of expertise in biomedical science/biomedical informatics?

It is nowhere, because the people hiring for this position likely know nothing of it themselves, nor do they think it important. I conjecture that this position description was likely written by person(s) of an IT background with little input from biomedical scientists, the people they are supposed to serve. I know as a medical informaticist and IT professional that I would have flunked a student in my informatics classes who came up with such a requirements list for this type of position.

A job description for 'Business Analyst, Clinical Data Repository Services" is not too different.

Working as a Business Analyst (BA) within the CRIS organization — Business Analysis & Product Management team, the incumbent is primarily responsible for collaborating with colleagues to develop business cases, defining requirements specifications, and assisting business colleagues with the adoption of new capabilities for patient data management. The incumbent will have a primary focus on the ongoing development of the Clinical Data Repository and the associated standards that define the data structures used within the repository. These capabilities will be driven by the needs for both clinical development and external partnerships involving the use of collected patient data.


There's that conflation of "business cases" with "patient data management" issue again. As far as "defining the data structures used within the [clinical data] repository" I wrote a decade ago that knowledge of biomedicine was an essential competence for defining data elements and structures in a biomedical database, NOT information technology or database experience.

The requirements for this Clinical Data Repository position:

Required:Minimum of 5 years of analysis experience in a Business Analyst and/or Systems Analyst role. Demonstrated ability in the practical application of business analysis techniques and tools is required. Minimum of 7 years of pharmaceutical industry experience, including a minimum of 5 years working in a Clinical Development Information Technology role is required. A working knowledge of industry standards for Clinical Development Information, e.g. CDISC is required.

Desired: Knowledge of ORACLE databases, data warehousing technologies and web technology, is highly desirable. Education Requirements:B.S.degree is required. M.S. degree with a concentration in Information Management/Information Technology or related scientific discipline is a plus.

That's perhaps slightly better, althought he focus is still on IT and specific experience in biomedicine (beyond "a working knowledge of industry standards for clinical development information" as opposed to first-hand expertise in biomedicine) is not mentioned.

As far as Oracle databases, in "Essential Value of Medical Informatics Expertise in High-Risk Areas: an Invasive Cardiology Example" I wrote that:

... the informaticist [a.k.a. me - ed.] jettisoned the traditional MIS [i.e., merchant computing - ed.] approach of viewing a person's skills in using a specific database application (e.g., Oracle) as a critical factor. The informaticist identified as crucial the data modeling process for the cardiologist's needs. This process has very little to do with software or computer science and much to do with medical informatics. The finest technical expert in the world in database development systems such as Oracle, client-server tools, etc. is not very useful in such a function, since it is a high-level cognitive function requiring clinical experience combined with medical data modeling expertise, not computer or MIS expertise. The lack of recognition of the need to partition strategic, high-level, cognitive informatics functions such as healthcare data modeling from more mundane, low-level programming and implementation tasks in clinical projects inhibits progress in healthcare and biomedical IT. It is with amazement that informaticists such as the one in this story observe a blindness to this issue in biomedicine, including healthcare and the pharmaceutical industry. The highest levels of informatics expertise should be sought for any clinical initiative in busy clinical settings. In such settings, the data development and customization process is an essential competence.


I imagine support for clinical trials is not considered very high risk:

... When [two companies] delayed results of a clinical trial aimed at showing the benefits of Vytorin, a top cholesterol drug, they said more time was needed to ensure the validity of the data.

But for years, cardiologists questioned the study, wondering if it could ever prove the drug worked as marketed. Does the drug reduce heart attacks as the companies say it does? Maybe, experts say. The study just may not be able to prove it.

The study, called ENHANCE and started in June 2002, intended to prove that adding Zetia to Zocor caused less heart-attack-causing plaque to build up in patients' arteries than Zocor alone. The combination of Zetia and Zocor is marketed as Vytorin and represents sales of $3 billion. Zetia generates another $2 billion for the companies. But though treatment of patients in the trial ended in 2006, the results are not available 20 months later.

The companies say they have been legitimately engaged in ensuring the quality of the data from the study, raising concerns with consultants as early as April 2006 and struggling to get things right since. But from the design to the technology it uses, ENHANCE has long been riddled with problems, cardiologists say. A [company]spokesman acknowledged that the study "sets a high hurdle."

A big flaw, says Michael Davidson, director of preventative cardiology at the University of Chicago's Pritzker School of Medicine and an adviser to [the companies]: Patients in the study had been treated with cholesterol drugs like Zocor for so long it might be difficult to remove plaque from their arteries. "I was not really a big fan of the design in the first place," says Davidson ...


I was formerly a pharmaceutical company employee at the company that posted these ads. I lost my position during a mass layoff of 4,400 in 2003, and now my local community's economy is imperiled via the company's misfortunes, as they are the major employer in the area. While there I tried to reinforce the value of cross-disciplinary Medical Informatics expertise and provided considerable examples of its value.

The lessons from all this are as follows:

1. Pharma is far too much an information technology-dominated industry, with suboptimal results - the results that should be, after all, expected by the wise (but not the merely clever) when undue influence over vital scientific and clinical matters is held by computer geeks and process zealots with no knowledge of biomedicine - usually due to conflation of "IT expertise" with biomedical information science/informatics expertise. (E.g., the current CIO of the company that posted the above job ads came from a paper company and based on his corporate bio, his knowledge and experience in biomedicine are nil);

2. Pharma is still largely oblivous to the IOM's recommendations (either passively or deliberately, i.e., for reasons of IT reaction and territorial protectionism) on Medical Informatics expertise. Roles such as the above should be specifically calling for such experience;

3. Many pharma problems are not due to evil people, but to backwardness and stupidity;

4. The executives/board of pharma companies should consider downsizing the people responsible for such job ads - for failure to stay current with advancements in information technology and biomedical information science, and/or incompetence. These issues result in suboptimal staffing in research IT and other areas, harming the company.

Perhaps the IT-led debacle called the "CRISP" clinical trials database in the late 1990's at one major company was insufficient learning material (even though its victims were referred to as "crispy critters"):

In the 1990s, [this company] struggled to erect a modernized clinical data system known as CRISP (Clinical and Regulatory Information Strategic Program), a project that current and former information systems workers came to regard as a $100 million fiasco.


5. Personnel in pharma - or perhaps in healthcare overall - who can comprehend these issues are rare, as most still have their heads stuck in 1940's IBM card punch-plug board paradigms of "data processing." This brings to mind the "process recommendation" by the author of the Effect Measure Blog
in "Who will protect us from our protectors", namely:

Step One: Remove head from up your [slang term for anatomic rear end].

Finally:

6. These issues affect not just clinical trials and data management but also R&D and the pipeline. Pharma has not yet properly learned the wise lesson, told to me by an elderly jewelry, coin and stamp dealer in business for decades who has just about every date, mintmark and condition of coin ever asked for in stock in multiple quantities, that "You can't do business from an empty wagon."

-- SS

Addendum: the last pithy quote from the coinshop becomes even more relevant considering the findings of this article on pharma marketing vs. R&D
expenditures, as commented on at this HCRENEWAL post and its links to other blogs.

3 comments:

Anonymous said...

There is no question that our industry fails in information science, that those resources are under-invested in every company, and patients pay the price. But in terms of these job defs, you have no idea what you are talking about. The roles these positions are intended to fill in every large organization are not responsible for what you are suggesting needs stronger support. To say that pharma is too IT led is laughable -- IT is at a standstill in many of these firms. The fact that they do technology POORLY (not led by it, but buried by it) is the reason why these roles have been defined. In reality, these roles mark a positive shift towards something a lot more focused on medical value and a lot less focused on individual scientists' data whims and pet projects. Having said that, true enough that if I were hiring these resources, the optional job requirements would not be optional.

Just for context: I don't work in these roles, I've never worked for Merck, and I'm a scientist by training.

MedInformaticsMD said...

But in terms of these job defs, you have no idea what you are talking about ... In reality, these roles mark a positive shift towards something a lot more focused on medical value and a lot less focused on individual scientists' data whims and pet projects.

Dear Anonymous Scientist:

It seems you only superficially read the post and did not examine the empirical and academic evidence backing my claims, in the form of hyperlinks to other material that itself contains links to yet additional material. I leave it to you to explore these links.

By the way, ad hominem ("you have no idea what you are talking about") is not appreciated and future ad hominem will be deleted.

MedInformaticsMD said...

Addendum 1/2010:

Could "Anonymous" be yet another industry blog-heckling -- and in this case, scientist-denigrating -- sock puppet?

-- SS