Modelling study – self testing could lead to more HIV infections

This article is more than 10 years old. Click here for more recent articles on this topic

If gay men replace clinic-based HIV testing with less sensitive home tests, the prevalence and incidence of HIV could rise, American researchers report in the January edition of Sexually Transmitted Diseases. Because home tests have longer window periods than laboratory tests, there is a danger that men with recent HIV infection remain undiagnosed, the model suggests.

In 2012, American regulators approved the OraQuick In-Home HIV Test, giving a green light to HIV self-testing at home. In the United Kingdom, legal restrictions on selling such devices will be lifted in April.

systematic review of studies on self-testing has found it to be acceptable across a wide variety of populations, but identified few data on linkage to care after testing positive. One previous modelling study suggested that if inconsistent condom users used self-testing to screen sexual partners, it could lead to fewer instances of unprotected sex and HIV transmission.



In HIV testing, when the person testing collects their own sample and performs the whole test themselves, including reading and interpreting the result. 

window period

In HIV testing, the period of time after infection and before seroconversion during which markers of infection are still absent or too scarce to be detectable. All tests have a window period, the length of which depends on the marker of infection (HIV RNA, p24 antigen or HIV antibodies) and the specific test used. During the window period, a person can have a negative result on an HIV test despite having HIV.

mathematical models

A range of complex mathematical techniques which aim to simulate a sequence of likely future events, in order to estimate the impact of a health intervention or the spread of an infection.

linkage to care

Refers to an individual’s entry into specialist HIV care after being diagnosed with HIV. 

acute infection

The very first few weeks of infection, until the body has created antibodies against the infection. During acute HIV infection, HIV is highly infectious because the virus is multiplying at a very rapid rate. The symptoms of acute HIV infection can include fever, rash, chills, headache, fatigue, nausea, diarrhoea, sore throat, night sweats, appetite loss, mouth ulcers, swollen lymph nodes, muscle and joint aches – all of them symptoms of an acute inflammation (immune reaction).

The new modelling study was developed on the basis of the sexual and testing behaviour of gay men in Seattle. Of note, men in this setting already test quite frequently – the model assumed that at-risk men test on average once or twice a year.

Moreover, broadly in line with trial results of self-tests and data comparing different HIV tests, the researchers assumed that modern clinic-based laboratory tests have a window period of 15 days and detect 100% of infections after this time. In contrast, the OraQuick In-Home HIV Test was assumed to have a window period of three months and to detect 91.7% of infections after this time. While this is the window period stated by the manufacturer, it is quite a conservative assumption (some infections are likely to be detected prior to three months).


If all HIV testing continues to occur in the clinic, at the same frequency, then the model produced an estimate of HIV prevalence in gay men of 18.6%. This figure isn’t far off the observed prevalence in Seattle, which is around 15%.

However, the model found that HIV prevalence would go up if men replaced clinic tests with self tests. If half the men did so, prevalence would be over 23%, and if all men did so, it would rise as high as 27.5%. In terms of HIV incidence (new infections), it would rise from 1.2 to 1.8% a year if all men tested at home.

This is because a number of men who have acute (recent) HIV infection would remain undiagnosed and may continue to have unprotected sex while they have an exceptionally high viral load. 

If men tested more frequently and self tests were used, prevalence would not go up quite as much, but it wouldn't go down, regardless of how frequently testing occured. For example, if all men tested twice as often and half of them used self tests, prevalence would go up to 21%.

The researchers then tested other assumptions about the window period of self tests. As noted above, if all men tested at home and the window period is three months, prevalence would be 27.5%. If the window period is actually two months or one month, prevalence would be around 26% or 21% respectively – still above current levels.

Again, an increase in the frequency of testing could improve the situation, but for prevalence to remain at current levels, frequency would have to be multiplied by 2.6 (if the window period is two months) or 1.2 (if the window period is one month).

The modelling suggested that delays in linkage to care and to uptake of HIV treatment could cause additional rises in HIV prevalence, but the effect would not be as profound as that of the long window period. For example, if only half the home testers started treatment, prevalence would rise to 29%. If no home testers started treatment, prevalence would be over 31%.

But low uptake of treatment would have a strong impact on HIV incidence, which could rise from 1.2% annually to 2.1% or 3.1%.


“Our model suggests that replacement of clinic-based testing with home-use tests may increase HIV prevalence among Seattle MSM [men who have sex with men], even if home-use tests allow MSM to test more frequently," the researchers write. “This potential increase in prevalence seems to be driven primarily by the relatively long window period of the approved home-use test when compared with available laboratory based tests, such as NAAT and antigen-antibody combination assays.”

However, they do note some circumstances which their modelling did not take account of and which could lead to more encouraging scenarios. Firstly, if self-testing of sexual partners leads to people avoiding exposure during sex, it could have benefit. Secondly, if self testing leads to people who do not currently test at all to be tested, it could help lower prevalence. Finally, the long window periods associated with self tests would be less of an issue in settings where people test less frequently than in Seattle – with a longer gap between tests, the chances of a person with undiagnosed infection having acquired it very recently would be lower.


Katz DA et al. Replacing Clinic-Based Tests With Home-Use Tests May Increase HIV Prevalence Among Seattle Men Who Have Sex With Men: Evidence From a Mathematical Model. Sexually Transmitted Diseases 41: 2-9, 2014. (Abstract).

This news report is also available in Russian.