- Joined
- Mar 18, 2003
- Messages
- 2,271
- Reaction score
- 8
I'm sure many of you have heard about the article that supposedly found that the two black medical schools in the US, and another two medical schools in Mexico and the Philippines had the highest percentage of graduates who were disciplined compared to other medical schools in the US and abroad. Here's a letter I wrote to the editors addressing this article:
Dear Editors:
The Hartford Courant analysis of disciplinary actions against U.S. physicians, featured in Sunday's front-page article entitled "Med Schools: Four That Flunk," may have been beset by the specter of spurious correlation.
The article states that four medical schools?the Autonomous University of Guadalajara in Mexico, Howard University in Washington, Manila Central University in the Philippines, and Meharry Medical College in Tennessee "appeared in the bottom 5 percent" when databases of physician disciplinary records were ranked "according to the percentage of graduates from each school who had some sort of adverse event on their records." While the authors note that they only included larger, well-established medical schools in their analyses because "schools with small numbers of graduates could skew the results," there appears to be a weak statistical relationship between the percentage of disciplined physicians from the aforementioned medical schools and the percentage of disciplined physicians from other U.S. and foreign medical schools. This relationship is especially weak in the data on disciplined physicians in Ohio.
In order to examine the relation between medical school and percentage of disciplined graduates, I applied Fisher's Exact Test, a statistical test used to determine if there is a non-random association between two categorical variables. By convention, a statistically significant difference between two variables is defined as p < .05. P, or probability, provides a sense of the strength of the evidence against the null hypothesis. The lower the p-value, the stronger the evidence; the higher the p-value, the weaker the evidence. In this case, the null hypothesis is defined as no difference between medical school attended and percentage of disciplined graduates. Thus, if p > .05, the difference between the percentage of disciplined graduates at one medical school is not significantly different from the percentage of disciplined graduates at another medical school.
When Fisher's Exact Test is applied to the data on disciplined physicians in California (page A12), there is no difference between the percentage of disciplined graduates of the Manila Central University and the percentage of disciplined graduates of the University of Arkansas (p = .30), West Virginia University (p = .25), and University of California at Irvine (p = .15). There is also no difference between the percentage of disciplined graduates of the Autonomous University of Guadalajara and the percentage of disciplined graduates of the University of Vienna (p = .86), Korea University (p = .65), University of Arkansas (p = .40), and the University of California at Irvine (p = .07). For the sake of brevity, I have only included results from a handful of analyses, though the percentages of disciplined graduates from several other medical schools do not differ from percentages of disciplined graduates from the "Four That Flunk."
In Ohio, there is no difference between the percentage of disciplined graduates of Manila Central University and the percentage of disciplined graduates of the following schools: University of Arizona (p = 0.40), University of Arkansas (p = .33), University of Washington (p = .13), and University of Oklahoma (p = .13). There is also no difference between the percentage of disciplined graduates of Meharry Medical College and the percentage of disciplined graduates of the University of California at San Diego (p = .55), University of California at San Francisco (p = .35), University of Washington (p = .20), University of Oklahoma (p = .20), University of Tennessee (p = .09), Boston University (p = .08) and Yale University (p = .08). Percentages of disciplined graduates of the Autonomous University of Guadalajara and Howard University also did not differ consistently from percentages of disciplined graduates of other U.S. and foreign medical schools.
Nationally, the same pattern tends to persist. For example, there is no difference between the percentages of disciplined graduates of Manila Central University and University of California at Irvine (p = .15), among other medical schools.
A readily apparent confounding variable in these analyses is the number of graduates of each medical school included in this study. These numbers vary widely across the samples, and should have been adjusted for in the analyses.
While I think this research is promising with respect to exploring potential shortcomings associated with medical education in the U.S. and abroad, the researchers who conducted this study should pay closer attention to what the numbers are really saying. The "Four That Flunk" may actually be as bad as some of the best, including such esteemed medical schools as the University of California at San Francisco, and Yale.
Dear Editors:
The Hartford Courant analysis of disciplinary actions against U.S. physicians, featured in Sunday's front-page article entitled "Med Schools: Four That Flunk," may have been beset by the specter of spurious correlation.
The article states that four medical schools?the Autonomous University of Guadalajara in Mexico, Howard University in Washington, Manila Central University in the Philippines, and Meharry Medical College in Tennessee "appeared in the bottom 5 percent" when databases of physician disciplinary records were ranked "according to the percentage of graduates from each school who had some sort of adverse event on their records." While the authors note that they only included larger, well-established medical schools in their analyses because "schools with small numbers of graduates could skew the results," there appears to be a weak statistical relationship between the percentage of disciplined physicians from the aforementioned medical schools and the percentage of disciplined physicians from other U.S. and foreign medical schools. This relationship is especially weak in the data on disciplined physicians in Ohio.
In order to examine the relation between medical school and percentage of disciplined graduates, I applied Fisher's Exact Test, a statistical test used to determine if there is a non-random association between two categorical variables. By convention, a statistically significant difference between two variables is defined as p < .05. P, or probability, provides a sense of the strength of the evidence against the null hypothesis. The lower the p-value, the stronger the evidence; the higher the p-value, the weaker the evidence. In this case, the null hypothesis is defined as no difference between medical school attended and percentage of disciplined graduates. Thus, if p > .05, the difference between the percentage of disciplined graduates at one medical school is not significantly different from the percentage of disciplined graduates at another medical school.
When Fisher's Exact Test is applied to the data on disciplined physicians in California (page A12), there is no difference between the percentage of disciplined graduates of the Manila Central University and the percentage of disciplined graduates of the University of Arkansas (p = .30), West Virginia University (p = .25), and University of California at Irvine (p = .15). There is also no difference between the percentage of disciplined graduates of the Autonomous University of Guadalajara and the percentage of disciplined graduates of the University of Vienna (p = .86), Korea University (p = .65), University of Arkansas (p = .40), and the University of California at Irvine (p = .07). For the sake of brevity, I have only included results from a handful of analyses, though the percentages of disciplined graduates from several other medical schools do not differ from percentages of disciplined graduates from the "Four That Flunk."
In Ohio, there is no difference between the percentage of disciplined graduates of Manila Central University and the percentage of disciplined graduates of the following schools: University of Arizona (p = 0.40), University of Arkansas (p = .33), University of Washington (p = .13), and University of Oklahoma (p = .13). There is also no difference between the percentage of disciplined graduates of Meharry Medical College and the percentage of disciplined graduates of the University of California at San Diego (p = .55), University of California at San Francisco (p = .35), University of Washington (p = .20), University of Oklahoma (p = .20), University of Tennessee (p = .09), Boston University (p = .08) and Yale University (p = .08). Percentages of disciplined graduates of the Autonomous University of Guadalajara and Howard University also did not differ consistently from percentages of disciplined graduates of other U.S. and foreign medical schools.
Nationally, the same pattern tends to persist. For example, there is no difference between the percentages of disciplined graduates of Manila Central University and University of California at Irvine (p = .15), among other medical schools.
A readily apparent confounding variable in these analyses is the number of graduates of each medical school included in this study. These numbers vary widely across the samples, and should have been adjusted for in the analyses.
While I think this research is promising with respect to exploring potential shortcomings associated with medical education in the U.S. and abroad, the researchers who conducted this study should pay closer attention to what the numbers are really saying. The "Four That Flunk" may actually be as bad as some of the best, including such esteemed medical schools as the University of California at San Francisco, and Yale.
