Thursday, February 25, 2010

Study: HIV Math Models
Must Be 'Carefully Examined'


If UCLA math modeler Sally Blower were a responsible researcher, one who respectfully engages the communities affected by her studies, she would be having a public conversation with her critics. She isn't on both accounts.

I'd like to have the opportunity to ask her why she thinks other HIV math modelers should have their work "carefully examined," and why their models must be "transparent enough to permit policy makers to understand them." She's demanding something she herself hasn't delivered.

In a 7-page October 29, 2009, study debunking and criticizing a World Health Organization math model created by Dr. Ruben Granich, proposing widespread HIV testing and treatment could eliminate AIDS in the next decade, Blower and a colleague advance some surprising conclusions and recommendations.

Blower said, on page 6, emphases mine:

We conclude that the model developed by Granich et al., when used with realistic parameter values, does not show that HIV elimination is (theoretically) possible. We recommend that any modeling results that are used as a foundation for health policy decisions should always be carefully examined. [...]

In addition, assumptions that are made to construct health policy models should be made transparent enough to permit policy makers to understand them.

Modeling results should always be interpreted with caution. We recommend that models should never be used as the sole basis for making health policy decisions; many other, often more important factors that are not included in the modeling framework, need to be considered.

Well, since January, with Dr. Grant Colfax of SF DPH's HIV prevention division, we in San Francisco have been carefully examining Blower's dubious math model on alleged mini-epidemics of HIV strains resistant to cocktails.

She has steadfastly turned up her nose when asked for further discussion. Blower has also virulently resisted efforts by public policy makers, to make her convoluted model transparent and understandable.

When will Blower follow the advice she hands out to rival math modelers working on AIDS? Transparency and engagement: Two items missing from Blower's resume.

No comments: