Misclassification problems / Heterosexual misclassifications?
Could you please comment on the study by Murphy, Mueller, and Whitman in J AIDS and Human Retrovirology 16: 122-126, 1997? In case you haven't been able to keep up with the latest literature, this article presented the results of a more detailed "look-back" (using other sources of information than are typically used) at the growing number of "hetero- sexual" cases in Chicago, and found that 85% could be reclassified into different transmission categories (nearly 70% were reassigned to categories that had nothing to do with "heterosexual" transmission).
- Do you think Chicago is unique with regard to this
- If this problem can be assumed to be more widespread
than just Chicago, could you comment on what this study
means with regard to messages we have been hearing about the "heterosexuals being the fastest growing group"? It would seem that much of this "increase" may really be just sloppy epidemiology! Thanks for any comments on this rather shocking study!
Thanks for taking the time to read my question, which is:
Considering the article above, would you predict that
similar reductions could be obtained if other large
metropolitan areas reanalyzed their surveillance data?
How does this significant problem with misclassification
affect the message that "heterosexuals are the fastest
growing transmission group", which is repeated over and over by "experts" and the media? Weren't the increases seen actually less than this 50% that was reclassified and
doesn't this mean that heterosexual transmission may actually be on the decline?
Thanks for any comments.
Thank you for your questions. When reading a paper such as this, one must be EXTREMELY cautious in how one interprets this data. This report can be very easily misinterpreted. This report does NOT mean that transmission through heterosexual contact is on the decline, NOR does it mean that a lot of our transmission information is incorrect. What it does show is that AIDS risk-factor reporting IN THIS ONE CITY was not being done correctly. There is NOTHING to suggest that this is a widespread problem. Let me review with you this report in further detail.
One of the advantages in HIV and AIDS reporting, is that we can use this data to look at the spread of HIV over time. If we see any dramatic or unusual changes in the patterns of spread of infection, this gives us an indication that there may be an unexpected problem occurring. This is essentially what happened in Chicago.
Prior to 1991, most heterosexual AIDS cases reported in Chicago, were among people having sex with Injecting Drug Users (IDUs). However, beginning in 1991, there was a dramatic change in the pattern of spread among heterosexuals acquiring HIV infection. Beginning in that year, reports indicated that a smaller and smaller percentage of infected heterosexuals were becoming infected from IDUs; more and more heterosexual cases were becoming infected by having sex with a person known to have HIV, but whose risk factor was unknown. The change in this pattern was so dramatic, that it was noticed by local health officials.
As a result of this dramatic and unusual change in reported AIDS cases, each of these cases was reviewed to make sure that they were categorized correctly. What was discovered is that this dramatic change was a result of misclassification, rather than any dramatic change in what populations of people were becoming infected.
They reviewed 395 cases to verify their classification status. 30% these cases (118 people) were reclassified into a category called "No Identifiable Risk". This is because, when these cases were reviewed, either the person moved and could not be found (64 cases), or they died (49 cases), or they refused to be interviewed (3 cases), or they did not report a commonly transmitted route of infection (2 cases). If a person is unable to be located, if they die, or if they refuse to cooperate, we cannot collect the necessary information as to how they became infected. Therefore these 118 cases were taken out of the "heterosexual" category, and put into the "No Identifiable Risk (NIR)" Category. This does NOT mean that HIV was transmitted in a new way. What it does mean is that the researchers were simply unable to collect the information to correctly categorize these cases. These cases may have indeed become infected through heterosexual sex. But because of the researchers inability to collect and verify this information in virtually all of these 118 people, the researchers could not assume their risk was heterosexual, and they had to be taken out of the "heterosexual" category, for statistical purposes. For more information about the NIR category, and about the 2 cases where the risk could not be determined, see the posting, Disturbing statistic.
Of the remaining 277 cases (395 cases minus the 118 cases described above), many cases (123) did indeed become infected through heterosexual contact. The remaining cases fell under other risk categories (most often "Injecting Drug User,") and to a lesser extent "Men Having Sex with Men" and other risk categories.
For diseases like HIV, and other STDs, when changes in transmission patterns are seen, we would expect gradual changes over time (as is the case in changes in transmission patterns nationally). When changes occur dramatically in a short period of time (as was the case in Chicago), this tells us that either something is rapidly changing, or there are problems with collection of information. When this occurs, we investigate why these changes are occurring (which is exactly what happened in Chicago).
Nationally, we have not seen a dramatic unexpected shift in transmission patterns. Changes that are being seen nationally (for example, changes in the number of heterosexuals becoming infected), have been gradual, which is exactly what we would expectto see, for the spread of diseases such as HIV. When changes in Chicago's statistics occurred so dramatically over such a short period of time, it was easily noticed.
I also want to point out that these statistics are for AIDS cases, not HIV cases. Because the period of time from infection with HIV, to full-blown AIDS, averages 10 years, AIDS risk-factor statistics (like those above) tell us how HIV was transmitted an average of 10 years ago, not today. Until HIV becomes reportable in all states, we must be very cautious in how we interpret AIDS data, and how those statistics apply to the spread of HIV today. It is also extremely important to remember that statistics like these can be very easily misinterpreted. This is why it is important to understand the "statistic behind the statistic", which is what I have tried to do here.
In summary, if we see unusual changes in the spread patterns of any disease, we follow up on that information to determine if those changes are due to differences in how we collect statistics, or actual changes in the spread of a disease. Based on all available data nationwide and worldwide, the misclassification problems reported here, appear to be an isolated incident, and do not represent what we see nationally, or internationally.
If you have any further questions, please feel free to call the Centers for Disease Control at 1.800.232.4636 (Nationwide).