To routinely offer testing for HIV infection in all cases of tuberculosis: a rational clinical approach?

HIV infection affects the natural history of Mycobacterium tuberculosis (MTB) infection.1 Patients with HIV infection are more likely to develop active tuberculosis, develop disseminated disease,2 reactivate latent MTB infection,3 become re-infected with MTB,4 and present with extrapulmonary disease.5 Tuberculosis can occur at any CD4 count and may not be associated with clinical signs of immunodeficiency. MTB infection may accelerate HIV disease progression.6-8

HIV is estimated to account for 9% of all new tuberculosis cases worldwide.9 Tuberculosis may be the presenting feature of unrecognised HIV infection and is an AIDS-defining condition.10 Despite this relationship, HIV status is reported in less than half of the notified cases of tuberculosis in Australia.11 We present an argument for the routine offer of HIV testing to all people diagnosed with, or treated for, MTB infection in Australia.

The incidence of tuberculosis in Australia has remained fairly stable since 1985 (5.3 per 100 000 population), although there was an increase in extrapulmonary disease in New South Wales between 1975 and 1995.11,12 The contribution of HIV to this increase is unclear, as only 7% of reported patients had been tested for HIV infection. Between 1999 and 2005, HIV serostatus was reported to the tuberculosis register for only 21% of patients, although this proportion increased to 37% in 2005.11 This dataset does not capture whether patients were offered or declined HIV testing, and underreporting is possible. The prevalence of HIV infection between 1999 and 2005 was 4.67% (74/1583) in those tested, and therefore at least 1.0% (74/7374) overall.11

There are limited other data that can inform clinicians about the likelihood of HIV in tuberculosis patients in Australia. A small study in Sydney in the 1990s found that at least 16% of people with tuberculous lymphadenitis had HIV co-infection.13 The higher prevalence in the study may relate to the high prevalence of HIV infection in south-eastern Sydney or to selection of subjects with extrapulmonary disease.5 Interestingly, only a third of these patients were born overseas. The proportion of AIDS cases associated with tuberculosis increased fivefold between 1993 and 2000 (to 2.7% of cases).14 More recently, between 2003 and 2005, tuberculosis was reported as the AIDS-defining condition in 4.6% of AIDS patients.15 The absolute numbers have remained relatively stable, but such data rely on physician HIV testing and reporting, as AIDS will not be diagnosed if HIV status is unknown.

The current Australian National HIV Testing Policy 200616 advises HIV testing on the basis of epidemiological risk factors or patient request. The only recommended clinical indicators are pregnancy or a sexually transmissible infection. It does not specifically recommend testing in the context of HIV-associated clinical manifestations, although they have been well described.10 It does not specify that tuberculosis is an indication for HIV testing, although the prevalence of HIV infection in people with tuberculosis is at least the same as in people who inject drugs (one of the cited epidemiological risk factors).

The National Strategic Plan for TB Control in Australia Beyond 200017 identifies that the overlap between the tuberculosis-infected and the HIV-infected populations in Australia is numerically small. It recommends monitoring the incidence of MTB infection in the HIV seropositive population, but does not recommend HIV testing of all people with tuberculosis.

Where state or territory guidelines for clinicians could be found on an Internet search, they varied from always recommending testing to no recommendation or that clinicians “consider HIV infection in every case of tuberculosis infection”18 or ask about risk factors and have a “low threshold” for HIV testing.19

A range of developed countries recommend testing for HIV infection for all people with tuberculosis.20-23

North American data suggest that clinicians are unable to predict HIV infection in people with tuberculosis. A study of hospitalised patients with smear-positive tuberculosis revealed that systematic questioning did not identify risk factors for HIV infection in all HIV-infected tuberculosis patients.24 Disturbingly, they identified that more than half of those with HIV infection had no risk factors identified in the medical record. Another study found rates of HIV infection as high as 18% in people who were considered at low risk and that such patients were significantly less likely to be offered testing.25

In the United States, one study found that routine testing is cost-effective even when the prevalence of HIV infection is as low as 0.2%.26 It is likely that HIV testing of people with tuberculosis in Australia would be cost-effective, as the minimum prevalence is 1.0%.

The NSW Supreme Court found a general practitioner negligent in not diagnosing HIV infection in a man with hepatitis B virus infection who transmitted HIV infection to his partner.27 This precedent could provide legal argument for failure to diagnose HIV infection in the setting of a condition with a strong association with HIV infection (such as tuberculosis, which is an AIDS-defining condition).

HIV testing is appropriate for most people with tuberculosis regardless of identified risk factors. Clinicians are not able to reliably predict which people have HIV infection, as risk factor assessment alone is insufficient. Tuberculosis is an AIDS-defining condition and clinicians have a responsibility to ensure that HIV is excluded, as highly effective therapy for HIV infection is available28 and HIV transmission can be prevented.

We recommend that all patients with M. tuberculosis infection, whatever their perceived risk for HIV infection, be offered an HIV test, with appropriate pre-test discussion.