MSM in Hong Kong

MSM does not mean MainStream Media in this post.  Instead, it means Men Having Sex With Men.

The starting point is this Ming Pao ad taken out by the Truth & Light Society.  I found this from Miss Lee, Sinan and Seechuen.

Two sets of numbers are cited in this ad.  

The first set of numbers is derived from Lau, J.T., Lau, M. and Tsui, H.Y.  HIV related behaviors and attitudes among Chinese men who have sex with men in Hong Kong: a population-based on study.  Sex. Transm. Infect., 2004 Dec: 80(6), 459-465.  The data are summarized in the schematic diagram below.  Please be mindful that I have not read the original paper, and do not know if there are more complex sampling issues involved here.  I only have this diagram to go by.

The second set of numbers comes from a press release from the Virtual AIDS Office of Hong Kong.  The data are as follows:

A total of 268 HIV cases were reported to the Department of Health (DH) last year. ...  Of these cases, 107 involved heterosexual transmission and 65 occurred between men having sex with men.  "Under the voluntary reporting system, some 28 per cent of the reports were incomplete, making the determination of the route of infection impossible,'' Dr Lee said.

Now I will translate for you what the ad asserts:

A survey conducted in Hong Kong estimates that 4.6% of men had engaged in MSM.  In the past 6 months, about 2% had engaged in MSM.

According to the information provide by the government, during 2004 there were 172 cases of new HIV cases that occurred as a result of sexual contact.  Of these, 65 were infected via MSM behavior.

In other words, the male homosexuals account for only 2% of the population accounts for 37.7% of the new HIV cases in 2004.  Based upon these figures, the odds ratio of MSM to heterosexual behavior in causing HIV infection is 29:1.

 Let us do these numbers line by line.

• There were 14,985 in the LLT study, of whom 667 ever had MSM.  Thus, the percentage is 100 x 667 / 14985 = 4.45%.  (So maybe this is not the 4.6% listed, but so what?)  [Alternately, a reading suggested that the 4.6% came from J T F Lau, J H Kim and H-Y Tsui (2005)  Prevalence and factors of sexual problems in Chinese males and females having sex with the same-sex partner in Hong Kong: a population-based study, International Journal of Impotence Research (see abstract).  This was cross-sectional, anonymous telephone survey.  Would you discuss your sexual preferences and practices with a stranger over the telephone?]
   
• There were 14,985 in the LLT study, of whom 292 engaged in MSM.  Thus, the percentage is 100 x 292 / 14,985 = 1.95%.  (This is the 2% listed)
   
• There were 65 cases of MSM HIV cases in the DH statistics.
   
• Where did the 37.7% come from?  As far as I can see, 100 x 65 / (107 + 65) = 100 x 65 / 172 = 37.8%.  But what is it?  65 is the number of HIV infections via MSM whereas 172 is the number of HIV infections via sexual contact, both heterosexual and MSM.  Properly speaking, this is defined as the percent of MSM HIV infections out of all sexually transmitted HIV infections.
   
• Now we parse this most astonishing statement: the male homosexuals account for only 2% of the population but also account for 37.7% of the new HIV cases in 2004.
   
  This is wrong in several ways.
   
  If you want to preserve the numbers, you have to change the language as follows: the male homosexuals account for only 2% of the population but alsl account for 37.7% of the new sexually transmitted HIV cases in 2004.
   
  If you want to save the language, you will have to change the number.  There were 268 cases in the government's records, but some 28% have no route of infection.  Thus, the number of classificable cases = 268 x 0.72 = 193.  Of these, 65 were infected via MSM behavior.  100 x 65 / 193 = 33.7%.  Thus, the statement becomes: the male homosexuals account for only 2% of the population but also account for 33.7% of the new classifiable HIV cases in 2004.
   
  But now we hit the biggest stumbling block in this silly exercise.  
   
  What is the universe of the LLT study?  The title is "HIV related behaviors and attitudes among Chinese men ..."  So, the universe and the study sample are "Chinese men" (assuming that means males 18 years or older of Chinese descent).  
   
  What is the universe for the government's records?  You can read through the press release and you won't find any mention about Chinese men.  That universe is the total population.  Who is in the total population but not a Chinese man?  Babies who can receive perinatal infections, children under 18, the entire female population, European men, Southeast Asian men, Japanese men, and so on.  Within the 65 MSM HIV cases, there may be minors and non-Chinese men.  Within the 107 HIV cases involving heterosexual transmission, there also women, minors and non-Chinese men.  Within the 268 total HIV cases, there are infants, children, women and non-Chinese men.
   
  So what is the ratio of MSM HIV infections among all HIV infections within Chinese men?  I don't have a clue.  The government data here do not say so.  If you want to know, you'll have to go back and get that information from the government.  But I do know that the assertion in the ad is WRONG because the two parts have different universes.
   
  The actual number may be more, the same or less than 37.7%.  But the point here is that you cannot assert 37.7% on the basis of these two datasets.  You can assert it on the basis of personal belief or faith and that is your prerogative.  But you cannot say, "The data show ..."  The viability of the science of statistics depends on this.

The Truth & Light Society ad was signed by a whole lot of healthcare and medical professionals who placed their names onto a petition based upon scientific data and arguments.  The data may be scientific but the statistical inference is wrong.  And this is not exactly rocket science because the survey universes are clearly different.  These people must have put their signatures down without even looking at the statements.  Think about that kind of professionalism the next time that you visit a doctor.

Since this is paid advertisement and the information is wrong, it is false advertising.  Both the advertiser and the newspaper have liabilities as such.

Message to MSM (mainstream media in Hong Kong): This is scandalous.  Will you pick it up and report it?