The Social Medicine Portal

HIV Testing: Social Context is more important than risk factors

Hospital General, Guatemala City

We have just published an article in the International Journal of STD & AIDS entitled “Can a clinical prediction tool guide HIV-testing decisions? Experience at a national hospital in Guatemala.”  In this article we discuss how an HIV testing algorithm failed to predict which patients would have a positive HIV test.  We first present the abstract and then discuss the article’s finding in greater detail:

Summary: The USA and international recommendations no longer emphasize using risk factors to target groups for HIV-testing.  Using a Guatemalan database of HIV tests, we developed a clinical prediction rule to guide decisions on HIV-testing. Prior to HIV-testing, data were collected on demographics, risk factors and prior testing. Based on a theoretical construct incorporating demographics, known HIV risk factors and symptoms, we developed a logistic regression model to predict HIV seropositivity.  Between 2000 and 2005, 16,471 tests were performed, of which 19.8% were positive. The algorithm successfully predicted 1883
of 2489 HIV-positive tests (sensitivity 76%, likelihood ratio [LR]-positive 2.45) and 6282 of 9086 HIV-negative tests (specificity 69%, LR-negative 0.35). Although the model indices are robust, applying the model in a clinical setting would have little impact on improving selective testing practices. Our findings support current recommendations for universal HIV-testing, not selective testing based on risk factors. Before these recommendations can be adopted widely in Guatemala, treatment access needs to be assured and protections put in place for people diagnosed with HIV infection.

Commentary

For many years HIV testing was recommended primarily for people who had specific risk factors.   But a number of developments have led to an expansion of the criteria for HIV testing and since 2006 the Centers for Disease Control (CDC) has recommended that all persons in the US from ages 13 to 65 undergo HIV testing.  This change in recommendation was based on several developments.  We now have effective treatments for HIV disease, treatments which can also reduce the risk of HIV transmission.  Persons who are aware of their HIV diagnosis seem to decrease their risk behaviors.  Universal testing programs have been quite successful in reducing blood-borne and maternal to child HIV transmission.  On the other hand lots of experience with risk-based protocols shows they miss patients who are HIV infected.  The UNAIDS has also abandoned risk-based HIV testing, instead endorsing voluntary universal screening in places where HIV is prevalent, treatment is available and proper counseling can occur.

We had a database of over 16,000 HIV tests done at the Luis Angel Garcia Clinic, an HIV specialty clinic located at the Hospital San Juan de Dios in Guatemala City.  These tests had been done over a five year period (2000-2005) and about 1/5th were positive.  We had collected information on risk behaviors, demographics and symptoms before the tests.   With such a rich database, we wondered if we couldn’t develop an algorithm to predict who would be HIV infected and who would not.

We based our algorithm on three different sets of data:

1. Demographic data : income, education, ethnicity, and marital status.  Not surprisingly we found that lower income, lower education and Mayan ethnicity (all signs of social exclusion) were associated with HIV infection.  So was being married.

2. Risk behaviors:  Having an HIV-infected partner and being a man who has sex with men were the most strongly associated with HIV infection in our sample.  So were being involved in commercial sex work, having a history of STD’s, using IV drugs, and reporting more sexual partners.

3. Symptoms:  A variety of symptoms – fevers/sweats, weight loss, oral thrush, chronic cough, recurrent sores, and diarrhea were all highly associated with HIV infection.  In fact, diarrhea was the most predictive of any of the variables.

These findings confirm what we already know about HIV infection and show that our population is not all that different from other HIV populations.  But could we use these associations to predict the results of the HIV tests?

The answer is that we could not.  Our prediction algorithm failed quite miserably. The algorithm successfully predicted only 1883 of 2489 HIV-positive tests (sensitivity 76%, likelihood ratio [LR]-positive 2.45) and 6282 of 9086 HIV-negative tests (specificity 69%, LR-negative 0.35).

In simple English, many people with lots of risk factors were HIV negative.  And many people with no or few risk factors were HIV positive.

The algorithm did much better than chance alone, but not enough to be very useful.  (This is an excellent example of the distinction between statistical significance and clinical significance).

These findings would support recommendations for universal testing.  In the Bronx we are currently in the middle of a campaign to have everyone HIV tested.  But should everyone in Guatemala get HIV tested?

Despite our data we felt it was premature to call for universal testing of all Guatemalans.  There is no national protocol for counseling and testing, high quality tests are not universally available, there remains enormous stigma and discrimination against those with HIV, and treatment is not available everywhere. Perhaps the place to start is providing HIV testing and treatment to all pregnant woman.  This, however, is a challenge because many pregnant women in Guatemala do not get prenatal care.

We concluded that: “when considering public policy for HIV-testing, the risk profile of the individual seems less important than the social and medical context in which testing takes place.”

For a reprint of the article, please contact Matt Anderson.  To find out more about the work of the clinic please contact Dr. Eduardo Arathoon

Posted by Matt Anderson, MD