Diagnose and manage vaginal discharge and vaginitis using South African syndromic management

In one line

Vaginal discharge is three common conditions wearing one symptom — bacterial vaginosis, vulvovaginal candidiasis and trichomoniasis — and the consultant task is to separate the dysbiosis from the yeast from the sexually transmitted protozoan, treat the right one, and recognise that South Africa's nurse-led syndromic vaginal-discharge algorithm trades aetiological precision for reach, undertreating candida and overtreating those without an STI while it catches the women who matter most.

Mechanism & pathophysiology

The healthy vagina is an oestrogen-dependent ecosystem. Oestrogen thickens the squamous epithelium and loads it with glycogen; lactobacilli (chiefly L. crispatus, L. iners, L. gasseri, L. jensenii) metabolise that glycogen to lactic acid, holding the pH at 3.8–4.5, and several strains generate hydrogen peroxide and bacteriocins that suppress competitors. Not all lactobacillary states are equally protective: a L. crispatus-dominant community is the most stable, whereas L. iners dominance is a more fragile state that transitions more readily into dysbiosis — which is part of why BV is harder to eradicate durably than to treat acutely. This low pH is the single most informative bedside number in the topic, because each of the three diagnoses sits in a predictable relationship to it, and the oestrogen-dependence explains the conditions' age distribution: BV and candida are diseases of the reproductive years, and a complaint of "discharge" in a hypo-oestrogenic prepubertal or postmenopausal woman should redirect attention to atrophic vaginitis, a foreign body, or malignancy rather than to this triad.

Bacterial vaginosis is a dysbiosis, not an infection by a single organism. Protective lactobacilli are displaced by a polymicrobial overgrowth — Gardnerella vaginalis, Prevotella, Atopobium vaginae, Mobiluncus, Mycoplasma hominis and other anaerobes — that raises the pH above 4.5. Gardnerella is the keystone: it adheres to the epithelium and lays down a dense polymicrobial biofilm into which the other anaerobes are recruited. That biofilm is the mechanistic explanation for the two clinical signatures of BV — the high recurrence rate (antibiotics suppress planktonic bacteria but the adherent biofilm persists and reseeds) and the volatile amines (putrescine, cadaverine, trimethylamine) released by anaerobic metabolism, which volatilise and produce the fishy odour when alkalinised. The long-standing teaching that BV is simply an imbalance, not transmitted, is now under direct challenge: the demonstration that treating male partners reduces recurrence reframes BV as, at least in part, sexually transmitted, with the organisms exchanged and re-exchanged between partners.

Vulvovaginal candidiasis is an inflammatory response to commensal yeast. Candida albicans (80–90%) colonises the vagina asymptomatically in a large minority of women; disease occurs when the yeast transforms from blastospore to the invasive hyphal form and provokes a brisk host inflammatory response — the symptoms are immunopathology, not tissue destruction. This is why VVC produces vulvar burning, pruritus and dyspareunia out of proportion to any odour, and why the discharge is thick and white ("cottage cheese") rather than malodorous. Crucially, candida thrives at the normal acidic pH, so a pH below 4.5 with florid symptoms points to yeast and away from BV or trichomonas. Oestrogen, pregnancy, poorly controlled diabetes, recent antibiotics and immunosuppression (including HIV) drive symptomatic disease. Non-albicans species, principally C. glabrata and C. krusei, matter disproportionately: they are intrinsically less susceptible to azoles, cause more chronic low-grade symptoms, and are over-represented in recurrent and refractory disease and in immunocompromised women.

Trichomoniasis is genuinely a sexually transmitted infection. Trichomonas vaginalis is a flagellated protozoan that infects the squamous epithelium of the vagina and lower urinary tract, raising pH above 4.5 and producing a frothy, sometimes greenish, malodorous discharge with vulvovaginal inflammation; the "strawberry cervix" (colpitis macularis) of punctate haemorrhages is specific but seen in a minority. Because it is an STI, its presence mandates partner notification, HIV testing and a search for co-infections in a way the other two do not.

The reason any of this earns consultant attention beyond symptom relief is the downstream biology. A high vaginal pH and the loss of the lactobacillary barrier — the state shared by BV and trichomoniasis — amplifies HIV acquisition. The mucosal inflammation recruits activated CD4 target cells, the protective hydrogen-peroxide-producing flora is gone, the epithelial barrier is disrupted, and pooled epidemiological data put the increase in HIV-acquisition risk associated with BV at roughly 60%. There is biological reciprocity here: HIV-associated immunosuppression in turn predisposes to more frequent, more severe and more often non-albicans candidiasis, so the three vaginal conditions and HIV are not parallel problems but a connected system. In a country with South Africa's HIV incidence, a recurrent dysbiosis is not a nuisance diagnosis; it is a modifiable cofactor for the epidemic that dominates the rest of this domain — the HIV–gynaecology interface assumed from the Intermediate groundwork. BV and trichomoniasis are also linked to preterm birth and to post-procedural pelvic infection — untreated BV before instrumentation (termination, insertion of an intrauterine device, hysterectomy) raises the risk of post-operative endometritis and ascending infection, which is why they intersect with ascending pelvic infection and with the timing of elective gynaecological surgery.

Assessment

The history separates the three before any instrument is used. Odour (especially after intercourse, when alkaline semen volatilises the amines) with a thin grey discharge and minimal itch suggests BV; itch and burning with a thick white discharge and no odour suggests candida; a frothy, malodorous, irritating discharge, often with dysuria, raises trichomonas. None of these is reliable enough alone — symptom-based diagnosis is wrong often enough that office testing matters.

On speculum examination, look at the discharge (homogeneous grey film coating the walls in BV; white plaques adherent to an erythematous vagina in candida; frothy discharge with cervical petechiae in trichomonas) and at the cervix, because a mucopurulent cervix points past vaginitis to cervicitis and the STI cover that demands. The bedside tests, taken together, are diagnostic for most women:

• Vaginal pH (narrow-range paper against the lateral wall, away from cervical mucus and blood): >4.5 in BV and trichomonas, normal (<4.5) in candida. A normal pH effectively excludes BV.
• Whiff/amine test: a drop of 10% KOH on the discharge releases a fishy odour — positive in BV and frequently in trichomonas.
• Saline wet mount microscopy: clue cells (epithelial cells studded with adherent coccobacilli, borders obscured) for BV; motile, flagellated trichomonads the size of a leucocyte for trichomonas (sensitivity falls quickly as the slide cools and dries — examine it immediately).
• 10% KOH wet mount: lyses epithelial cells and reveals pseudohyphae and budding yeasts in candida.

These bedside findings are formalised in two scoring systems. Amsel criteria diagnose BV when three of four are present: thin homogeneous discharge, pH >4.5, positive whiff test, and ≥20% clue cells. The Nugent score is the Gram-stain laboratory reference standard, scoring the relative quantities of lactobacillary, Gardnerella/Bacteroides and curved gram-variable morphotypes from 0–10, with ≥7 diagnostic of BV (0–3 normal, 4–6 intermediate); the Hay–Ison system is a simpler Gram-stain alternative used in the same role. For trichomonas, wet-mount sensitivity is only 50–70% and falls further with delay, so where resources allow a nucleic acid amplification test (NAAT) is the most sensitive confirmation and is the modern standard in well-resourced settings; antigen point-of-care tests are an intermediate option, and the organism is sometimes reported incidentally on a routine cervical cytology slide, which counts as a real diagnosis but is too insensitive to use as a screen. Candidiasis deserves one caution at the microscope: up to half of symptomatic women have a negative KOH wet mount despite true infection, so a culture is warranted where symptoms are convincing, the wet mount is clear, or treatment has failed — both to confirm yeast and to speciate, since management diverges sharply for non-albicans disease.

The central tension in South African assessment is that most of this never happens. In the nurse-led primary-health-care clinic the woman is managed syndromically: she is treated for "vaginal discharge syndrome" on the symptom alone, frequently without a speculum, microscopy or pH paper. The trade-off has to be held in mind whenever the algorithm is applied. The syndromic approach maximises coverage and same-day treatment — it reaches women who would never get a laboratory diagnosis, treats presumptively for the STIs that carry the worst sequelae, and needs no microscope, which is why it is the chosen instrument of a service delivering care to millions through nurses. Its costs are systematic: it has poor specificity for the cause (most women with discharge in lower-prevalence settings do not have the STI being covered), it undertreats candida (antifungals are not in the standard discharge package), it cannot distinguish vaginitis from cervicitis without examination, and it drives antibiotic exposure and partner-notification consequences for women whose problem was a non-transmitted dysbiosis. The consultant position is not to reject syndromic management — at population scale in a high-STI-prevalence, resource-limited setting it is defensible, and it saves the tubes and the babies that aetiological delay would lose — but to know when to override it. The recurrent, treatment-failing, pregnant, HIV-positive or atypical patient is the one who warrants a speculum, a pH, a wet mount and an aetiological diagnosis, because in her the cost of a wrong presumptive label is high and the prior probability of the package being right is low.

Every woman with an STI-syndrome presentation is offered an HIV test — non-negotiable in South Africa — and, where trichomonas or cervicitis is found, partner notification and screening for co-infecting STIs.

Management

Treat the diagnosis, not the symptom, wherever an aetiological diagnosis exists; fall back to syndromic cover where it does not. The acute presentation and first-line treatment of each organism — the recognition, the wet-mount findings, the uncomplicated regimens — are the Intermediate-level groundwork (the candidiasis and genital-tract-infection chapters of the FCOG Intermediate course); the work here is the recurrent, refractory, pregnant and HIV-positive woman, and the choice between aetiological and syndromic strategy at scale.

Bacterial vaginosis. First-line is oral metronidazole 400 mg twice daily for 7 days (the South African dose; the CDC equivalent is 500 mg twice daily for 7 days), or intravaginal metronidazole 0.75% gel once daily for 5 days, or intravaginal clindamycin 2% cream for 7 days where metronidazole is not tolerated. Counsel on alcohol avoidance during and shortly after oral metronidazole (the disulfiram-like reaction) and that clindamycin cream weakens latex condoms. The defining clinical problem is not initial cure — that runs 65–85% — but recurrence, which approaches half of women within twelve months, driven by the persistent Gardnerella biofilm. The practical management ladder for recurrent BV is therefore distinct from the first episode: confirm the diagnosis on each occasion (recurrent "BV" is often misdiagnosed candida or trichomonas), re-treat, and for genuine frequent recurrence offer suppressive intravaginal metronidazole gel twice weekly for several months, accepting that the benefit wanes once it stops. Two newer strategies are addressed in the evidence section, but their headline is worth stating with the management: a live-biotherapeutic vaginal Lactobacillus crispatus product reduces recurrence after antibiotic cure, and treating the male partner reduces recurrence — the latter a genuine inversion of decades of teaching that partner treatment was useless.

Vulvovaginal candidiasis. The first clinical fork is uncomplicated versus complicated disease, because they are managed differently. Uncomplicated VVC — sporadic, mild-to-moderate, albicans, in a non-pregnant immunocompetent woman — responds equally to a short topical azole (clotrimazole pessary or cream over 1–7 days) or a single oral fluconazole 150 mg dose, the choice being patient preference. Complicated VVC — recurrent, severe, non-albicans, or in pregnancy, diabetes or immunosuppression — needs a longer course and a different mindset: a single dose under-treats it. Severe disease is given a second fluconazole 150 mg dose 72 hours after the first, or a 7–14-day topical course. The important blanket exception is pregnancy, where oral fluconazole is avoided: pregnant women are treated with a topical azole for 7 days (the longer course; single-day topical regimens are less effective in pregnancy), because high-dose and repeated oral fluconazole in the first trimester carries a teratogenicity signal and even a single 150 mg dose is now generally avoided antenatally. A woman with recurrent or stubbornly refractory candida should be checked for undiagnosed or poorly controlled diabetes and for HIV, since both unmask the disease and change its course. Recurrent VVC — defined as four or more culture-confirmed episodes in a year — is a maintenance problem, not a stronger-single-dose problem: induce remission (e.g. fluconazole 150 mg every 72 hours for three doses) then suppress, classically with weekly oral fluconazole for six months, or with an individualised degressive regimen that steps the dosing interval down as control holds. The patient must understand this controls rather than cures: relapse after stopping is the rule. Non-albicans disease (suspect it in azole failure and in HIV) needs a culture with speciation and a different agent — C. glabrata often responds to intravaginal boric acid or to nystatin rather than to more fluconazole, because azole resistance is the reason the standard treatment failed.

Trichomoniasis. This is an STI and is treated as one. First-line is oral metronidazole 400 mg twice daily for 7 days rather than the older single 2 g dose, because the 7-day course clears the organism more reliably (developed in the evidence section). Treat the partner(s) — concurrently and presumptively, the cardinal difference from BV and candida — and advise abstinence until both are treated and asymptomatic. Screen for co-infecting STIs and HIV. In HIV-positive women the failure rate is higher and reinfection common, so the multi-day regimen is used and re-testing (test-of-cure / rescreening) at around three months is recommended rather than assumed cure. In pregnancy, symptomatic trichomoniasis is treated for the woman's benefit, but the evidence does not support screening and treating asymptomatic trichomoniasis to prevent preterm birth — a counter-intuitive point developed below.

The South African syndromic regimen — what is actually given. A woman presenting with vaginal discharge to a nurse-led clinic is treated for vaginal discharge syndrome with combination cover: ceftriaxone (for N. gonorrhoeae), azithromycin or doxycycline (for C. trachomatis) and metronidazole (for trichomonas and BV), with the 7-day metronidazole course preferred over a single dose. Candida is not in the standard discharge package and is added — a topical or oral azole — only when curd-like discharge and vulvar itch make it clinically obvious or when first-line treatment fails. The combination and its blind spots run together: it presumptively treats cervical STIs in everyone with discharge (over-treating those without), it misses candida unless flagged, and it cannot separate vaginitis from cervicitis or PID without the examination it usually omits. The honest reconciliation is that the regimen is calibrated to the prior probability in the population it serves — in a setting where curable STIs are common and follow-up is uncertain, presumptive cover for the high-stakes infections is the right population trade, even though it is the wrong answer for the individual low-risk woman whose discharge was candida or a simple dysbiosis.

Guidelines compared

The four bodies a candidate should be able to contrast agree on the antimicrobials and diverge on philosophy and on the frontier questions.

Three divergences are worth holding. First, the diagnostic philosophy: South Africa's syndromic algorithm is a deliberate, defensible choice for a high-prevalence, nurse-delivered, laboratory-poor service, where the CDC and BASHH assume microscopy or NAAT in every consultation — transposing the aetiological CDC pathway onto a district clinic that has neither misreads the setting it is built for. Second, the trichomonas dose changed recently and in the same direction everywhere: the single 2 g dose, taught for years, has given way to the 7-day course on trial evidence, so single-dose metronidazole is no longer first-line for women anywhere. Third, BV partner treatment is the live frontier: every guideline above still says do not routinely treat partners, but all were written before the 2025 trial that reversed the evidence, so the defensible position is to state the current guideline and note that it is being actively reconsidered. These are not contradictions to memorise but a single coherent picture — the antimicrobials are agreed, the recent shifts (trichomonas dose, partner treatment) move every body in the same direction, and the only genuine philosophical split is aetiological versus syndromic, which is a resource question rather than a disagreement about the biology.

The evidence & the controversy

The most consequential recent development is that bacterial vaginosis may be sexually transmitted after all, and the trade-off here is between decades of microbiome dogma and a single high-quality trial. The Australian male-partner-treatment trial randomised couples to treat the woman alone or to treat both partners (the man receiving oral metronidazole plus topical clindamycin to penile skin) and was stopped early because treating the woman alone was inferior: 12-week recurrence fell from 63% to 35% when the male partner was treated. The mechanistic reading is that BV organisms colonise the male genital tract and reinfect the woman, so suppressing the partner reservoir lets her own flora recover — which, if it holds, rewrites BV from a non-transmissible dysbiosis into a sexually transmitted condition and makes partner treatment a legitimate tool for the woman with relentless recurrence. The defensible viva position is calibrated: this is practice-changing, biologically coherent and large in effect, but it is one open-label trial in monogamous heterosexual couples and the major guidelines have not yet incorporated it, so present it as the strongest current evidence for partner treatment in recurrent BV rather than as established universal practice.

A second strand attacks recurrence biologically rather than pharmacologically. Because the failure of standard treatment is the persistent Gardnerella biofilm and the failure to re-establish lactobacilli, replacing the protective flora is a rational adjunct: a live Lactobacillus crispatus (CTV-05/LACTIN-V) vaginal product given after metronidazole cure reduced 12-week recurrence from 45% to 30%. It is not yet a South African public-sector option, and it works as an adjunct to antibiotic cure rather than as a primary treatment, but it is the proof of concept that restoring the ecosystem — not just killing the overgrowth — is where durable BV control lies.

For trichomonas, the evidence settled a dosing question and unsettled a screening one. The dosing question: a multicentre randomised trial in non-pregnant women found the 7-day metronidazole course nearly halved test-of-cure positivity compared with the single 2 g dose (11% vs 19%), which is why first-line guidance for women moved to the multi-day regimen. The screening question is the more counter-intuitive. Trichomonas is epidemiologically associated with preterm birth, so treating it in pregnancy ought to help — but the landmark trial of treating asymptomatic trichomoniasis in pregnancy not only failed to prevent preterm birth, it was associated with more preterm delivery (19.0% vs 10.7%). The result killed routine screen-and-treat of asymptomatic pregnant women; symptomatic women are still treated for their own benefit, but the population strategy of screening every pregnant woman and treating to prevent prematurity has the evidence pointing the opposite way. The mechanism remains debated — possibly an inflammatory or endotoxin response to organism death, possibly reinfection from an untreated partner during the pregnancy — and it stands as a clean example of an epidemiological association that did not translate into a treatment benefit, the same pattern as the failed antibiotic trials for asymptomatic BV in pregnancy.

For recurrent candidiasis, the controversy is about chronicity and a new drug. Maintenance suppression works — weekly fluconazole holds the great majority of women symptom-free while they take it — but relapse on stopping is the rule, so the framing must be of a chronic relapsing condition managed long-term, not an infection cured in a course. The newest entrant, the long-acting oral CYP51 inhibitor oteseconazole, produced strikingly low recurrence rates over a year in phase-3 trials and is the first agent licensed specifically for recurrent VVC; its catch is teratogenicity and a very long half-life, restricting it to women not of childbearing potential or with reliable contraception, and it is not available in South African public practice — so it belongs in the answer as the direction of travel and an access caveat, not as a current district-level option. Threaded through all of this is the HIV interface, which is the South African current-events layer of the topic: BV and trichomoniasis both increase HIV-acquisition risk, recurrent and non-albicans candidiasis flag underlying immunosuppression, and the public-health argument for treating vaginal dysbiosis well — beyond symptom relief — is that it is a modifiable cofactor in HIV transmission in exactly the population the rest of this course concerns.

Landmark trials & key evidence

The clinical force of the BV-recurrence trials is best read as a comparison. LACTIN-V cut 12-week recurrence from 45% to 30% — an absolute reduction of 15 percentage points, a number needed to treat of about 1/0.15 ≈ 7. Partner treatment cut it from 63% to 35% — an absolute reduction of 28 points, NNT ≈ 1/0.28 ≈ 4. Two different levers on the same problem, both meaningful, but the partner-treatment effect is the larger, which is precisely why a single trial was able to overturn a long-standing teaching. Against that, the Klebanoff result is the cautionary counter-example: a plausible, epidemiologically grounded intervention that produced harm — an 8-percentage-point absolute increase in preterm birth — a reminder that association does not establish that treatment will help, and may reveal that it hurts.

Exam traps & red flags

• Calling everything "thrush." Empirical antifungals for any discharge miss BV and trichomonas — the two that raise HIV risk and demand partner treatment. The pH paper sorts them in seconds: normal pH points to candida, raised pH away from it.
• Recommending single-dose metronidazole for trichomoniasis in a woman. First-line is now the 7-day course; the single 2 g dose under-cures.
• Screening and treating asymptomatic trichomoniasis in pregnancy to prevent prematurity. The trial evidence shows no benefit and a signal of harm — do not do it; treat symptomatic women for their own benefit only.
• Giving fluconazole in pregnancy. Use a 7-day topical azole; high-dose oral fluconazole is teratogenic and even a single 150 mg dose is now avoided antenatally.
• Treating recurrent "BV" or "candida" without re-confirming the diagnosis. Recurrent symptoms are frequently the wrong organism or non-albicans candida; re-examine, re-microscopy, culture and speciate before escalating — more of a failed drug is not the answer when resistance is the reason it failed.
• Forgetting that trichomonas is an STI. It mandates partner treatment, HIV testing and STI co-screening; BV and candida do not (the partner-treatment evidence for recurrent BV is the nuanced exception, not a reason to treat partners routinely).
• Missing the HIV test. Every woman presenting with an STI syndrome in South Africa is offered HIV testing; BV and trichomonas are also markers of elevated acquisition risk.
• Mistaking cervicitis or PID for vaginitis. A mucopurulent cervix, cervical motion tenderness or pelvic pain is not vaginal-discharge syndrome — it is the ascending-infection pathway, and treating it as simple vaginitis misses tubal disease.
• Reciting the CDC/BASHH aetiological pathway in a nurse-led PHC setting as though microscopy and NAAT were available — or, conversely, defending syndromic management for the recurrent, pregnant, HIV-positive or treatment-failing woman who has earned a proper aetiological work-up.

Evidence anchors

• Male-partner treatment to prevent BV recurrence — Vodstrcil, Bradshaw et al., N Engl J Med 2025
• LACTIN-V (Lactobacillus crispatus CTV-05) to prevent BV recurrence — Cohen et al., N Engl J Med 2020
• Single-dose vs 7-day metronidazole for trichomoniasis in women — Kissinger et al., Lancet Infect Dis 2018
• Metronidazole for asymptomatic trichomoniasis in pregnancy — Klebanoff, Carey et al., N Engl J Med 2001
• Bacterial vaginosis and HIV acquisition meta-analysis — Atashili et al., AIDS 2008
• Maintenance fluconazole for recurrent VVC — Sobel et al., N Engl J Med 2004
• Individualised degressive maintenance fluconazole (ReCiDiF) — Donders et al., Am J Obstet Gynecol 2008
• Oteseconazole for recurrent VVC (VIOLET) — Sobel, Donders et al., NEJM Evidence 2022
• Southern African HIV Clinicians Society 2022 STI guideline — Peters et al., South Afr J HIV Med 2022
• CDC Sexually Transmitted Infections Treatment Guidelines 2021 — Bacterial Vaginosis
• South African National Department of Health Standard Treatment Guidelines & Essential Medicines List (Primary Health Care level) — vaginal discharge syndrome managed syndromically with ceftriaxone, azithromycin/doxycycline and metronidazole; candida added when clinically evident.
• South African National Consolidated Guidelines for HIV (2026) — HIV testing offered with every STI syndrome; PVT terminology.