To the Editor
We appreciate the interest that Professor Leibowitz and Ms Desmond have taken in our article, and we are grateful to be able to respond to their constructive criticisms.
First, we address the statement regarding the, “...unexpected findings that 70% of the HIV positive Medicaid enrollees were female... The fact that the infection rate is 3.7 times higher among men than among women casts doubt on the selection criteria used by Johnston et al.”
This statement is misleading because it assumes that HIV screening rates are equal between men and women in Medicaid. You must screen for HIV to detect it, and if HIV screenings are overwhelmingly given to women, it follows logically that women may account for a large proportion of the individuals in whom HIV is detected.
We identified 1,540,361 Medicaid enrollees screened for HIV. A total of 1,529,564 of these enrollees had nonmissing information on sex, of whom 1,364,597 (89.2%) were female, demonstrating higher screening rates among women. Further analyzing the data, 2,091/164,967 (1.27%) men were diagnosed as having HIV, whereas 4,733/1,364,597 (0.35%) women were diagnosed as having HIV. Thus, in our sample, the rate of HIV infection is 3.6 times higher among screened men than among screened women, which is strikingly close to the Centers for Disease Control and Prevention data that Leibowitz and Desmond suggest belie our findings.
Next, we address the analysis of California Medicaid claims data that Leibowitz and Desmond conducted. They found a subgroup of individuals—82% of whom were female—who “had HIV diagnoses coded only on the same day they were screened for HIV, and no evidence of a subsequent diagnosis from a confirmatory test.” They conclude that “many of these enrollees were receiving HIV screening tests in prenatal care or at delivery and did not have HIV disease... Many of the ‘HIV diagnoses’ in 2007 California Medicaid claims appeared to be ‘rule out HIV’ diagnoses.”
If our data were subject to a large number of “rule-out” diagnoses driven by prenatal care, we would expect to see a substantially higher rate of linkage to HIV expert care among men than among women. In fact, although in our study male participants had shorter durations of time to linkage to HIV expert care, the proportion of male participants who linked to HIV expert care was nearly identical to that of female participants (22.4% male and 21.1% female for “standard definition,” 9.3% male and 9.5% female for “restrictive definition”).
Finally, from 1,540,361 HIV screenings that we identified, 14,045 (0.9%) had an HIV diagnosis code. Thus, for 99% of HIV screening claims, no HIV diagnosis is recorded. It is reassuring that the practice of recording rule-out diagnoses, a concerning practice given the seriousness of HIV, does not seem to be widespread.
In conclusion, we once again acknowledge that there is a possibility for misclassification of HIV status; however, our additional analyses affirm the validity of our findings and demonstrate that the specific criticism and evidence presented by Leibowitz and Desmond in fact do not apply to our results.
Stephen S. Johnston, MA
Truven Health Analytics