Race and Ethnicity, or Socioeconomic Class, or Diet   (May 17, 2010)

A friend of mine sent me a link to a diabetes review article that had an interesting if somewhat lengthy title: Impact Of Race/Ethnicity On The Efficacy And Safety Of Commonly-Used Insulin Regimens: A Post-Hoc Analysis Of Clinicals Trials In Type 2 Diabetes. The article was published in Endocrine Today.

Before I started reading the study, I was intrigued by the concept: clearly, there have been concerns that race and ethnicity might be correlated with different diabetes outcomes, and I was looking forward to learning what impact (if any) that race and ethnicity might have on the usefulness of various insulin regimens.

What the authors did was to review results from 10 published clinical trials and one unpublished trial, which were conducted between 1995 and 2004. Patients were grouped into categories, based on the patient's self-selection on a standard questionnaire: "Caucasian (European, Mediterranean, and Middle Eastern), African-descent (Black), Asian (Burmese, Chinese, Japanese, Korean, Mongolian, Vietnamese, Pakistani, Indian sub-continent), and Latino/Hispanic (Mexican-American, Mexican, Central and South American)." A total of 1455* patients were included in this review, of which the overwhelming majority were Caucasian (1085); the rest were Latino/Hispanic (182), Asian (131), and African (57). Some of the included studies looked at basal insulin use (glargine or NPH), some at insulin premixes given either twice or three times daily. (*At one point in the article, there's an apparent typo, and the number of patients is given as 1445.)

I will not bore the reader with the results, except to mention that the authors concluded that racial groups "may differ." My conclusion is that the present study is severely flawed.

First, it's a retrospective study. It appears from the 10 study titles that no one planned up-front to study the effects of race and ethnicity, and hence there was no reason to set up the protocol so that these effects could be evaluated.

Second, as the authors point out, the data were pooled from studies that differed in length, population characteristics at baseline, and targets.

Third, too many of the patients in the analysis are white. And patients described as black were not living in Africa: indeed the table describing where these studies were run shows that none of the studies were run in Africa! And the majority of Asians came from Canada and India.

Next, as the authors point out, there's a total lack of socioeconomic data: were the whites poor and uneducated, and thus unable to afford medical care and unlikely to comply with the requirements of the protocols? Or, perhaps more likely, were the minority groups?

Of even more concern, there was a lack of race/ethnicity-related dietary assessment: and what you eat clearly has an impact on how your diabetes behaves.

Well, enough's enough. I've picked on the article enough to make my point. And to be fair, the authors point out many of these flaws and others, and describe their study as "exploratory."

Now what we need, is a prospective, global study, which would include clear information about the race and ethnicity of the subjects in the study, as well as controlling for socioeconomic status, education, dietary differences, and what-have-you. Of course, finding someone to fund such a study might be difficult. But if a prospective study were undertaken, then we can learn the true impact of race and ethnicity on the efficacy and safety of various insulin regimens.