A Study Using Categorical Data

People who are HIV+ should inform their sexual partners of their HIV status. Does the probability of doing this vary depending on mode of transmission? Researchers at Boston University School of Public Health sought to answer this question and reported their findings in the paper below.

Sexual ethics. Disclosure of HIV-positive status to partners. Stein MD, Freedberg KA, Sullivan LM, et al.: Arch Intern Med. 1998;158(3):253-7.


"OBJECTIVE: To determine factors associated with disclosure of human immunodeficiency virus (HIV)-positive status to sexual partners.

METHODS: We interviewed 203 consecutive patients presenting for primary care for HIV at 2 urban hospitals. One hundred twenty-nine reported having sexual partners during the previous 6 months. The primary outcome of interest was whether patients had told all the sexual partners they had been with over the past 6 months that they were HIV positive. We analyzed the relationships between sociodemographic, alcohol and drug use, social support, sexual practice, and clinical variables; and whether patients had told their partners that they were HIV positive was analyzed by using multiple logistic regression.

RESULTS: Study patients were black (46%), Latino (23%), white (27%), and the majority were men (69%). Regarding risk of transmission, 41% were injection drug users, 20% were homosexual or bisexual men, and 39% were heterosexually infected. Sixty percent had disclosed their HIV status to all sexual partners. Of the 40% who had not disclosed, half had not disclosed to their one and only partner. Among patients who did not disclose, 57% used condoms less than all the time. In multiple logistic regression analysis, the odds that an individual with 1 sexual partner disclosed was 3.2 times the odds that a person with multiple sexual partners disclosed. The odds that an individual with high spousal support disclosed was 2.8 times the odds of individuals without high support, and the odds that whites or Latinos disclosed was 3.1 times the odds that blacks disclosed.

CONCLUSIONS: Many HIV-infected individuals do not disclose their status to sexual partners. Nondisclosers are not more likely to regularly use condoms than disclosers. Sexual partners of HIV-infected persons continue to be at risk for HIV transmission."

In the methods section of the paper the authors noted that they used the two independent sample t-test to test for differences in continuous variables and used two-tailed tests. They used chi-square analyses to test for differences in discrete (i.e., categorical) variables.

[Link to full article: https://jamanetwork.com/journals/jamainternalmedicine/fullarticle/191291 ]

The table below shows part of Table 1 from the paper, presenting factors associated with disclosure of HIV+ status.

Characteristic (n) % Disclosed p-value
  • Male (89)
  • Female (40




  • High school grad. (77)
  • Non graduate (52)




Injection drug use
  • Yes (55)
  • No (74)




Alcohol problem (CAGE score)
  • 0-1 (67)
  • ≥2 (62)




  • Black (59)
  • White (35)
  • Latino (30)
  • Other (4)




Transmission risk
  • Injection drug use (52)
  • Homosexual (25)
  • Heterosexual (50)




The outcome is dichotomous (disclosed or did not), and the "exposure" categories are either dichotomous (two groups) or categorical (more than two categories). All of the p-values shown were computed using a chi-square test of independence.

In order to explain the chi-square (χ2) test, we will focus on the last row in the table above to determine whether the frequency of disclosing one's HIV+ status differs among injection drug users, homosexuals, and heterosexuals.