HIV Serosorting

Abstract

HIV serosorting is a form of risk reduction behavior: choosing partners, or whether to use a condom, by each person's perceived HIV serostatus, so that condomless sex is restricted to seroconcordant pairs. It emerged from gay and bisexual communities in San Francisco and Seattle in the early 2000s as a grassroots adaptation to condom fatigue, and it works only to the degree that the serostatus labels it relies on are true. Because roughly one in five to one in seven infections is undiagnosed, and because recent infection is both undetectable and maximally infectious, serosorting concentrates residual risk in the partnerships it labels safe. This article defines the seroadaptive family, formalises this accuracy-dependence problem, reviews the evidence placing serosorting between condoms and no strategy, and situates it in the era of pre-exposure prophylaxis (PrEP) and Undetectable = Untransmittable (U=U).

Keywords: serosorting, seroadaptation, risk reduction behavior, seroguessing, undiagnosed infection

Introduction

Serosorting is a decision rule applied to sexual partnering. Its logic is deceptively simple: condomless sex carries HIV risk only when one partner is infected and the other is not, so if two partners share the same serostatus the act is, in principle, safe. An HIV-negative person who restricts condomless sex to other HIV-negative partners cannot acquire the virus from them; two HIV-positive partners have, classically, nothing further to transmit. The behavior was not designed by public-health authorities but arose from within affected communities as a grassroots harm-reduction strategy, and epidemiologists first named and measured it only after noticing that condomless sex was rising while HIV incidence was not rising proportionally (Snowden et al., 2009; Cassels et al., 2009).

The Medical Subject Headings (MeSH) vocabulary indexes serosorting under the descriptor HIV Serosorting (D059147), filed simultaneously beneath Risk Reduction Behavior and Sexual Behavior. That dual placement is an indexing convenience, not a claim that the behavior is two different things: MeSH classifies a term wherever a literature searcher might reasonably look for it, so a single practice can hang from two branches of the tree. Reading the descriptor as a taxonomy of the behavior itself would be a category error; it records where the concept is shelved, not what it is.

The Seroadaptive Family of Behaviors

Serosorting is the best-known member of a cluster of behaviors that epidemiologists group under seroadaptation: deliberate adjustments to sexual practice that use serostatus to lower transmission risk without demanding abstinence or universal condom use (Snowden et al., 2009; Parsons et al., 2005). The family resolves into several distinct strategies, often combined within a single partnership.

Pure serosorting selects partners by concordant status and then abandons condoms within concordant pairs. Condom serosorting is weaker: condoms are used with partners of different or unknown status and dropped only with concordant ones — a hedge that keeps a barrier where uncertainty is highest. Strategic positioning holds serostatus constant but adjusts the sexual role: an HIV-negative partner takes the insertive position in condomless anal intercourse, because the per-act transmission probability for the insertive partner is several-fold lower than for the receptive partner. Seropositioning is its serodiscordant cousin, with the HIV-positive partner taking the lower-risk role. Withdrawal before ejaculation is sometimes folded into the seroadaptive repertoire as a last, weak layer (Parsons et al., 2005).

These strategies differ sharply in how much they depend on information that may be wrong. Strategic positioning depends only on a mechanical fact — who is insertive — that both partners can observe directly and cannot misremember. Serosorting depends on a label, and a label is only as good as the testing, disclosure, and honesty that produced it. That contrast is the key to understanding why the evidence on serosorting is so much more equivocal than its clean underlying logic would suggest.

The Accuracy-Dependence Problem

Serosorting's protection is not a biological constant; it is a function of how often the serostatus assumption is correct. Three mechanisms degrade that accuracy, and each is largest in exactly the partnerships serosorting treats as safe.

The first is undiagnosed infection. A person who believes they are HIV-negative because their last test was negative may have been infected since, or may never have tested. In the United States a substantial minority of people living with HIV — historically on the order of one in five, and even now roughly one in seven — do not know their status (Marks, Crepaz, & Janssen, 2006). Modelling by Marks and colleagues (2006) found that this undiagnosed minority accounts for a disproportionate share of new transmissions, precisely because unawareness removes every downstream protective behavior, serosorting included.

The second is the acute-infection window. In the weeks immediately after infection, standard antibody tests are still negative while viral load — and therefore infectiousness — is at its lifetime peak. A partner who tested last month, seroconverted last week, and reports himself negative in good faith is both labelled safe and maximally dangerous. Serosorting cannot see this state at all.

The third is seroguessing: acting on an assumed rather than a disclosed and recent serostatus. Zablotska and colleagues (2009) distinguished true serosorting, grounded in explicit disclosure of a recent test result, from the far commoner practice of inferring a partner's status from context, silence, venue, or appearance. Eaton and colleagues (2009) likewise found that the behaviors people call serosorting frequently rest on little more than an untested presumption of concordance. When the label is a guess, the strategy's protective premise simply fails.

Because all three mechanisms load risk onto the partnerships that serosorting permits condomless sex within, the strategy does not merely offer incomplete protection — it can, under the wrong parameters, actively mislead, converting a feeling of safety into behavior that would have been avoided under a simpler rule.

How Well Serosorting Works

The empirical question is not whether serosorting is perfect — no one claims it is — but where it falls relative to the two reference behaviors: consistent condom use and no risk-reduction strategy at all. The weight of the evidence places it squarely in between.

On the protective side, Golden and colleagues (2008), analysing Seattle men who have sex with men, found that serosorting carried roughly half the HIV risk of condomless sex with partners of unknown status, though clearly more risk than consistent condom use. Philip and colleagues (2010), using the EXPLORE study cohort, likewise found serosorting associated with a decreased risk of seroconversion relative to no strategy. A community-level analysis by Cassels and colleagues (2009) argued that the rise of serosorting in Seattle was consistent with the observed plateau in HIV incidence despite increasing condomless sex — a population-scale signal that the behavior was doing something.

On the cautionary side, Wilson and colleagues (2010) modelled conditions under which serosorting can increase HIV acquisition risk relative to consistent condom use, driven precisely by the undiagnosed and acute-infection leakage described above. Golden and colleagues (2012) reported that serosorting failed to protect African American men who have sex with men, among whom higher background prevalence and higher undiagnosed fractions tilt the arithmetic against the strategy — a stark illustration that serosorting's value is not fixed but depends on the epidemiological context of the population practising it. Jin and colleagues (2009), in an Australian cohort, found that seroadaptive behaviors reduced but did not eliminate incident infection. A prospective analysis by Vallabhaneni and colleagues (2012) linked specific seroadaptive practices to HIV acquisition among HIV-negative men, confirming that the protection they afford is partial and ranked: full serosorting outperformed condomless sex with partners of unknown status, but every seroadaptive practice carried more risk than consistent condom use.

Two systematic reviews settle the overall shape. Kennedy and colleagues (2013), reviewing the literature to inform World Health Organization guidelines, concluded that serosorting was less effective than consistent condom use but more effective than no strategy, while cautioning that the evidence base was observational and heterogeneous. Purcell and colleagues (2017) quantified both the harms and the benefits in a meta-analysis, confirming the intermediate position and emphasising that the net effect depends on local prevalence, testing frequency, and disclosure accuracy. The consistent conclusion across two decades is that serosorting is a genuine harm-reduction behavior whose protective value is real but conditional, and never a substitute for condoms or biomedical prevention where those are available.

Prevalence and Trends

Serosorting is common wherever it has been measured among men who have sex with men. Snowden and colleagues (2009) documented substantial seroadaptive practice in San Francisco as early as 2004, and Khosropour and colleagues (2016) tracked its prevalence over time in Seattle, finding that it remained a widespread strategy even as its measured association with HIV and other sexually transmitted infections shifted across the decade. Chen and colleagues (2016) examined the predictors of serosorting and the intention to serosort among San Francisco men who have sex with men, finding that the practice is patterned by testing history, perceived prevalence, and disclosure norms rather than adopted uniformly.

A recurring finding is the gap between self-reported serosorting and the explicit, test-grounded disclosure the strategy's logic requires. Much of what is counted as serosorting in surveys is, on closer inspection, the seroguessing Zablotska and colleagues (2009) warned about — which means population estimates of serosorting tend to overstate how much true, accuracy-grounded serosorting is occurring.

The PrEP and U=U Era

Two biomedical developments have changed the meaning of the serostatus labels serosorting depends on. Pre-exposure prophylaxis (PrEP) allows an HIV-negative person to remain negative despite condomless exposure, decoupling a negative serostatus from the behavioral caution it once implied. Undetectable = Untransmittable (U=U) — the finding that a person with HIV whose viral load is durably suppressed by antiretroviral therapy cannot transmit the virus sexually — makes a disclosed-positive, virally suppressed partner safer than an assumed-negative one of unknown testing history. Both facts invert parts of the classical serosorting calculus: the HIV-positive partner serosorting was built to avoid may now pose no transmission risk at all, while the reassuringly negative partner may be the undiagnosed one.

Grov and colleagues (2018), in a US nationwide survey, documented how serosorting now coexists with PrEP and other biomedical strategies rather than standing alone, and Goodreau and colleagues (2021) modelled a behavioral cascade of seroadaptation in the PrEP and U=U era, showing that serosorting is best understood as one decision node in a larger, shifting sequence of prevention choices. The strategy has not disappeared; it has been absorbed into a richer repertoire in which serostatus is one signal among several, and no longer the most reliable.

Worked Example

Consider an HIV-negative man choosing partners for condomless receptive anal intercourse from a population in which the HIV prevalence among potential partners is p = 20%. Suppose a fraction u = 20% of those infected are undiagnosed and so report themselves negative in good faith, and take the per-act transmission probability for unprotected receptive anal intercourse with an infected partner to be β = 0.0138 (about 138 infections per 10,000 acts).

No strategy. If he picks partners at random, the chance a given partner is infected is simply p = 20%, and his per-act acquisition risk is 0.20 × 0.0138 ≈ 0.00276, or 276 per 100,000 acts.

Serosorting. If instead he accepts only partners who report themselves negative, he still unknowingly accepts the undiagnosed positives. The fraction of partners who are positive and unaware is p × u = 0.20 × 0.20 = 0.04. The fraction who report negative — the pool he draws from — is the true negatives plus those hidden positives: (1 − p) + pu = 0.80 + 0.04 = 0.84. So the probability that an accepted partner is actually infected is 0.04 / 0.84 ≈ 4.76%, a 76% relative reduction from the 20% he faced picking at random. His per-act acquisition risk falls to 0.0476 × 0.0138 ≈ 66 per 100,000 acts.

Consistent condoms. If he instead used condoms with random partners, assuming condoms cut per-act risk by about 80%, his risk would be 0.20 × 0.0138 × 0.20 ≈ 55 per 100,000 acts.

In this base case serosorting (66) lands just above consistent condom use (55) and far below no strategy (276) — exactly the intermediate position the literature reports. But the result is fragile. Double the undiagnosed fraction to u = 40% and the probability an accepted partner is infected rises to (0.20 × 0.40) / (0.80 + 0.08) = 0.08 / 0.88 ≈ 9.09%, pushing serosorting's per-act risk to about 125 per 100,000 — now more than double the condom figure, which did not move. The single parameter the chooser cannot observe, the undiagnosed fraction, is the one that decides whether serosorting approximates condoms or falls well short of them.

Discussion

Serosorting is a revealing case for the psychology of risk reduction because its logic is sound and its failure is informational. Unlike a strategy that is simply weak, serosorting would work exactly as advertised if serostatus labels were always accurate; its shortfall is the gap between an assumed status and a true one. That makes it a natural experiment in how people reason under uncertainty about a hidden binary state — and the evidence suggests they systematically under-weight the probability that a reassuring label is wrong, especially when the label licenses a desired behavior.

The behavior also illustrates the difference between individual and population reasoning. For a single well-informed person in a low-prevalence, high-testing network who insists on explicit disclosure of a recent test, serosorting can approach the protection of condoms. For a population in which testing is irregular, disclosure is often guessed, and prevalence or the undiagnosed fraction is high, the same nominal behavior performs far worse — the Golden et al. (2012) finding among African American men who have sex with men is the sharpest demonstration. A public-health message cannot assume the favourable case, which is why guidance has consistently framed serosorting as better than nothing but weaker than condoms, never as a recommended primary strategy.

Current Directions

Three strands define the current research front. The first is the integration of serosorting into biomedical prevention. With PrEP and U=U established, researchers are mapping how serostatus-based partner selection interacts with, substitutes for, or undermines pharmacological protection; the Goodreau et al. (2021) behavioral-cascade framework is the leading model, treating serosorting as one conditional node rather than a standalone strategy.

The second is measurement validity. Because surveys cannot easily separate true, disclosure-grounded serosorting from seroguessing, work continues on instruments and designs that capture the accuracy of the serostatus information actually used, so that prevalence estimates reflect protective behavior rather than wishful inference (Chen et al., 2016; Zablotska et al., 2009).

The third is equity and context-dependence. The demonstration that serosorting's value varies with background prevalence and undiagnosed fractions has focused attention on the populations for whom it performs worst, and on whether promoting it as harm reduction could widen rather than narrow disparities in HIV incidence (Golden et al., 2012; Purcell et al., 2017).

Common Misconceptions

“Serosorting is as safe as using condoms.”

It is not. Across cohorts and two systematic reviews, serosorting reduces risk relative to no strategy but offers weaker protection than consistent condom use, and under high undiagnosed fractions it can perform substantially worse (Purcell et al., 2017).

“If my partner says he is negative, serosorting protects me.”

Only if that statement reflects a recent test and honest disclosure. An assumed or guessed status — seroguessing — carries the undiagnosed and acute-window infections the strategy is blind to (Zablotska et al., 2009).

“Serosorting and strategic positioning are the same thing.”

They are distinct seroadaptive strategies. Serosorting selects partners by serostatus label; strategic positioning keeps status fixed and adjusts the sexual role to exploit the lower per-act risk of the insertive position (Parsons et al., 2005).

“In the PrEP and U=U era serosorting is obsolete.”

It persists, but its calculus has inverted: a virally suppressed positive partner cannot transmit, while an assumed-negative partner may be the riskier choice. Serosorting now operates as one node in a larger prevention cascade (Goodreau et al., 2021).

Glossary

Acute-infection window.

The early period after HIV infection when antibody tests are still negative but viral load and infectiousness are at their peak.

Antiretroviral therapy (ART).

Combination medication that suppresses HIV replication; sustained use can drive viral load below the transmissible threshold.

Condomless anal intercourse.

Anal sex without a condom; the exposure route whose per-act transmission risk serosorting attempts to manage through partner selection.

Men who have sex with men (MSM).

The behaviorally defined population among whom serosorting arose and in which almost all of its epidemiological study has been conducted.

PrEP.

Pre-exposure prophylaxis; antiretroviral medication taken by an HIV-negative person to prevent acquisition of the virus.

Seroadaptation.

The family of behaviors that adjust sexual practice on the basis of HIV serostatus to reduce transmission risk.

Seroconcordant.

Describing two partners who share the same HIV serostatus (both negative or both positive).

Seroconversion.

The transition from HIV-negative to HIV-positive antibody status; the acute-infection window precedes it and is the period serosorting cannot detect.

Serodiscordant.

Describing a partnership in which one partner is HIV-positive and the other HIV-negative.

Seroguessing.

Acting on an assumed or inferred partner serostatus rather than a disclosed, recently tested one — the commonest failure of true serosorting.

Serosorting.

Choosing sexual partners, or condom use, on the basis of perceived HIV serostatus so as to restrict condomless sex to concordant pairs.

Strategic positioning.

A seroadaptive strategy in which the HIV-negative partner takes the insertive role, which carries a lower per-act transmission probability.

U=U.

Undetectable = Untransmittable; the finding that a person with HIV and a durably suppressed viral load cannot transmit it sexually.

Undiagnosed infection.

HIV infection in a person who does not know their status and so cannot report it accurately; the principal leak in serosorting.

Viral load.

The concentration of HIV in blood; a durably suppressed (undetectable) viral load underlies U=U, while a high viral load in acute infection drives peak infectiousness.

Key Researchers

Susan Cassels

is a professor of geography at the University of California, Santa Barbara, working on HIV transmission modelling and spatial epidemiology. Her analysis linked the rise of serosorting in Seattle to the population-level plateau in HIV incidence. UC Santa Barbara - Google Scholar

Lisa A. Eaton

is a professor of human development and family sciences at the University of Connecticut (InCHIP) whose research examined the implications of serosorting for HIV transmission and the gap between assumed and actual concordance. University of Connecticut - Google Scholar

Matthew R. Golden

is a professor of medicine and epidemiology at the University of Washington and directs the Public Health – Seattle & King County HIV/STD program. His cohort studies of Seattle men who have sex with men produced the foundational evidence on whether serosorting is safe and on its failure to protect African American men. University of Washington

Seth C. Kalichman

is a professor of psychological sciences at the University of Connecticut and editor of AIDS and Behavior. His research addressed the behavioral and cognitive dimensions of serosorting and partner-selection strategies. Wikipedia - Google Scholar

Christine M. Khosropour

is an associate professor of epidemiology at the University of Washington whose work tracked serosorting prevalence and its shifting association with HIV and sexually transmitted infections over time. University of Washington

Willi McFarland

is a professor of epidemiology and biostatistics at the University of California, San Francisco, and leads HIV surveillance at the San Francisco Department of Public Health. His surveys documented the prevalence and predictors of seroadaptive behaviors among San Francisco men who have sex with men. ORCID - UCSF Profile

Jeffrey T. Parsons

is a former distinguished professor of psychology at Hunter College (CUNY) and founding director of the Center for HIV/AIDS Educational Studies and Training (CHEST). His work mapped the sexual harm-reduction practices of HIV-positive men, including serosorting, strategic positioning, and withdrawal. ORCID - Wikipedia

Frequently Asked Questions

What is HIV serosorting?

It is the practice of choosing sexual partners, or deciding whether to use condoms, on the basis of each person's HIV serostatus, so that condomless sex is restricted to partners believed to share the same status. It arose as a grassroots harm-reduction strategy among gay and bisexual men in the early 2000s.

Is serosorting as effective as using condoms?

No. The consistent finding across cohort studies and two systematic reviews is that serosorting reduces HIV risk compared with having condomless sex with partners of unknown status, but provides weaker protection than consistent condom use.

Why can serosorting fail even when both partners are honest?

Because honesty is not accuracy. A partner may be undiagnosed, or may have been infected recently enough that antibody tests are still negative while infectiousness is high. In both cases a truthful “I am negative” is wrong, and the infection it hides falls inside a partnership labelled safe.

What is seroguessing?

Seroguessing is acting on an assumed partner serostatus — inferred from context, silence, or appearance — rather than a disclosed, recently tested one. Much of what surveys count as serosorting is really seroguessing, which lacks the accurate information the strategy depends on.

How is serosorting different from strategic positioning?

Serosorting selects partners by their serostatus label. Strategic positioning keeps serostatus fixed and instead adjusts the sexual role, with the HIV-negative partner taking the insertive position because it carries a lower per-act transmission probability. They are distinct seroadaptive strategies and are often combined.

Does serosorting still make sense with PrEP and U=U?

Its logic has shifted. A partner with HIV and a suppressed viral load cannot transmit it (U=U), and an HIV-negative partner on PrEP is protected regardless of a partner's status. So an assumed-negative, untested partner can now be riskier than a disclosed-positive, suppressed one, inverting part of the classical serosorting calculus.

Is serosorting more dangerous in some populations than others?

Yes. Its protective value falls as background HIV prevalence and the undiagnosed fraction rise. Research among African American men who have sex with men found that serosorting failed to protect them, because the local epidemiology tilted the arithmetic against the strategy.

Should public-health guidance recommend serosorting?

Guidance has consistently framed serosorting as better than no strategy but weaker than condoms, and never as a recommended primary method. It is treated as harm reduction for people who are not using condoms or biomedical prevention, with the caveat that its protection depends on accurate, recent, disclosed serostatus.

References