Diagnose and manage pelvic inflammatory disease and tubo-ovarian abscess

In one line

Pelvic inflammatory disease is ascending polymicrobial infection of the upper genital tract whose diagnosis is clinical and deliberately over-sensitive, because the cost of a missed case is measured in tubes; treat empirically on minimal criteria with broad-spectrum cover that includes anaerobes, and escalate to admission and drainage when a tubo-ovarian abscess declares itself.

Mechanism & pathophysiology

The disease begins at the endocervix and travels upward. Organisms breach the cervical barrier — often around menses, instrumentation or recent intrauterine-device insertion, when the protective mucus plug and the local immunity are disturbed — and ascend the endometrial cavity to the tubes, ovaries and peritoneum. The named stations of that journey are the diagnoses: endometritis, salpingitis, oophoritis, parametritis, and pelvic peritonitis, with a tubo-ovarian abscess as the end-stage walled-off collection.

It is polymicrobial, and that fact dictates the antibiotics. The classical sexually transmitted initiators are Neisseria gonorrhoeae and Chlamydia trachomatis; chlamydia is the single commonest identified cause, found in roughly 14–35% of cases, and in South Africa, where gonococcal prevalence is high, N. gonorrhoeae is a more frequent contributor than in low-prevalence European settings. But the cervical breach is followed by an ascent of the vaginal flora, so anaerobes (Prevotella, Atopobium, Bacteroides), Gardnerella vaginalis and other bacterial-vaginosis-associated organisms colonise the upper tract alongside — which is why a regimen that covers only the gonococcus and chlamydia under-treats. Mycoplasma genitalium is now recognised as a genuine cause of upper-tract infection rather than a bystander, and a substantial proportion of PID is pathogen-negative on lower-tract testing, meaning a negative swab never excludes the disease.

The damage is immunopathological. The host inflammatory response to chlamydia in particular — the cell-mediated reaction to persistent infection — scars the tubal mucosa, flattens the cilia and obliterates the lumen. The mucosal cilia are not merely passengers: their loss is what converts a tube from a transport organ into a blind pouch, so even a tube that remains anatomically patent on a hysterosalpingogram can be functionally useless for ovum pickup. That scarring is the substrate of every long-term sequela: tubal-factor infertility, ectopic pregnancy (a damaged but patent tube traps the conceptus), chronic pelvic pain from dense pelvic adhesions, and recurrent PID, because a once-injured tube defends itself less well against the next ascent. The injury accumulates with each episode and with severity, which is the mechanistic reason early, adequate treatment matters and why repeat infection is so much worse than a single bout — the tubal damage is broadly stepwise with successive episodes.

The microbiology shifts as the disease progresses, and this drives the antibiotic logic over time. The initiating event is frequently the sexually transmitted organism breaching the cervix, but once the upper tract is inflamed and the tissue planes are disrupted, the established and abscess-forming infection becomes increasingly anaerobic — a tubo-ovarian abscess yields a luxuriant mixed anaerobic flora regardless of whether gonorrhoea or chlamydia opened the door. This is why a regimen that covers only the initiating STI is adequate for the earliest endometritis but progressively inadequate as salpingitis matures into an abscess, and why anaerobic cover becomes non-negotiable in severe disease. The groundwork on acute pelvic infection — the cervical defence, the ascending route, the lower-tract microbiology — is assumed here; revise it in the Intermediate chapter on acute pelvic infection and on acute pelvic pain pathophysiology. The consultant layer is what follows: how to diagnose against a low threshold, how to read the patient who is not improving, and how to defend a regimen.

Two perihepatic and structural consequences deserve naming. Fitz-Hugh–Curtis syndrome is perihepatitis — right-upper-quadrant pain from "violin-string" adhesions between the liver capsule and the anterior abdominal wall — occurring in a minority of women with PID, especially chlamydial, and treated as ordinary PID rather than with any additional intervention. The tubo-ovarian abscess is the other end of the spectrum: a confluent inflammatory mass of tube, ovary and sometimes bowel and omentum, which behaves as a surgical collection — it can rupture into the peritoneum and precipitate septic shock, and it does not reliably resolve on oral antibiotics.

Assessment

The diagnostic posture is the single most examinable idea: treat on a low threshold. Because clinical signs are insensitive and non-specific (the positive predictive value of a clinical diagnosis against laparoscopy is only 65–90%) and because every delayed case risks a tube, empirical treatment is started on minimal criteria rather than withheld pending confirmation.

• Minimum criteria (CDC): a sexually active young woman, or a woman at risk of STIs, with pelvic or lower abdominal pain, no other identifiable cause, and at least one of — cervical motion tenderness, uterine tenderness, or adnexal tenderness on bimanual examination. One of the three is enough. Requiring all three loses too many cases.
• Supportive findings raise specificity but are not required: oral temperature >38.3°C, abnormal cervical or vaginal mucopurulent discharge, abundant white cells on saline microscopy of vaginal fluid, raised ESR or CRP, and laboratory documentation of gonorrhoea or chlamydia. The absence of pus cells on a Gram-stained vaginal smear has a good negative predictive value (around 95%), so a genuinely clean wet prep should make you reconsider the diagnosis; their presence is non-specific.
• History targets risk: age under 25, a new or multiple partners, no barrier contraception, recent IUD insertion (the excess risk is confined to the first 4–6 weeks after insertion), and a previous episode.

Exclude the mimics before you commit — these are the cases that kill or castrate.

• Ectopic pregnancy is the non-negotiable exclusion: a urine or serum βhCG in every woman of reproductive age, because a ruptured ectopic masquerading as PID is lethal and the SA syndromic approach explicitly flags lower abdominal pain in a young woman as ectopic until proven otherwise.
• Acute appendicitis overlaps closely; nausea and vomiting occur in most appendicitis but only about half of PID, and cervical motion tenderness appears in roughly a quarter of appendicitis, so neither symptom settles it.
• Ovarian torsion or a ruptured/haemorrhagic cyst present more suddenly and are surgical.
• Endometriosis, urinary tract infection, irritable bowel and functional pain round out the differential of lower abdominal pain in a young woman.

Investigations earn their place by changing management.

• Pregnancy test — always, to exclude ectopic and because pregnancy mandates admission.
• NAAT for N. gonorrhoeae and C. trachomatis (endocervical or vulvovaginal swab, or first-void urine) — a positive result confirms the aetiology and directs partner management, and where available, NAAT for M. genitalium guides therapy because it carries specific resistance implications.
• HIV testing in every patient — mandatory in the South African setting, both because the STI consultation is the entry point to HIV care and because HIV status frames the whole encounter; a negative test is repeated after the window period.
• Imaging is for complications, not for uncomplicated PID. Transvaginal ultrasound is the first-line test when a tubo-ovarian abscess is suspected — a systemically unwell woman, a palpable adnexal mass, or failure to respond to antibiotics — showing a complex multiloculated adnexal collection. CT is valuable where the differential includes appendicitis or diverticular abscess and to map a collection for drainage; MRI gives the best soft-tissue resolution and avoids ionising radiation but is access-limited. In a SA district setting, ultrasound and clinical judgement carry the load, with referral upward for cross-sectional imaging.
• Bloods — FBC and CRP to gauge severity (usually only abnormal in moderate-to-severe disease), and U&E and a venous lactate where sepsis from a leaking abscess is a concern.

Severity stratification changes the disposition, not just the wording. Mild-to-moderate disease is the woman who is systemically well, tolerating oral intake, without a mass or peritonism — she is an outpatient. Moderate-to-severe disease is marked by high fever, vomiting that defeats oral therapy, signs of pelvic peritonitis (rebound and guarding, which the SA syndromic algorithm flags as a referral trigger), or a palpable adnexal mass — she is admitted for intravenous therapy. The presence of a tubo-ovarian abscess, pregnancy, or an unexcludable surgical abdomen overrides everything else towards admission. Reading severity correctly at the first contact is what prevents both the under-treated abscess managed at home and the well woman occupying a scarce bed.

The host is part of the assessment in South Africa. HIV status frames tolerance and severity — an immunosuppressed woman may present more floridly — and the same consultation should quantify the rest of her reserve: anaemia, nutritional state, and whether she is already on antiretrovirals. None of this changes the antibiotic, but it changes the threshold for admission and the index of suspicion for a complication or a confounding diagnosis.

The harder consultant task is not making the diagnosis but reading the patient who is not getting better. Failure to defervesce or improve within 48–72 hours of appropriate antibiotics is the trigger to re-image for a tubo-ovarian abscess, to reconsider the diagnosis (appendix, a missed surgical abdomen, an ovarian accident), to question adherence and whether the partner was treated (re-infection mimics treatment failure), and — in the SA context — to confirm HIV status and exclude genital tuberculosis as a confounding, paucisymptomatic upper-tract process that ordinary PID antibiotics will not touch.

Management

Structure the answer immediate → ongoing → long-term, and let the antibiotic choice fall out of the microbiology: broad-spectrum cover that reaches the gonococcus, chlamydia AND the anaerobes, started empirically.

Immediate — empirical broad-spectrum antibiotics. The current backbone, harmonised across CDC 2021, BASHH and the South African syndromic approach, is a third-generation cephalosporin for the gonococcus, doxycycline (or azithromycin) for chlamydia, and metronidazole for anaerobes.

Several details are load-bearing. Metronidazole is no longer optional in the recommended regimen — its addition is the 2021 change, driven by trial evidence that it eradicates endometrial anaerobes and reduces residual M. genitalium and pelvic tenderness. In the SA pathway, the cephalosporin dose is ceftriaxone 500 mg IM (the older 250 mg dose has been superseded as gonococcal MICs have crept up), and azithromycin rather than doxycycline carries the chlamydial cover in the syndromic algorithm. Quinolones (ofloxacin, levofloxacin, moxifloxacin) are effective alternatives but are avoided where gonococcal risk is high because of quinolone-resistant gonorrhoea — which makes them a poor empirical first choice in a high-prevalence setting like South Africa.

The admission decision separates the woman who can be managed at home from the one who cannot. Admit for parenteral therapy and observation if: a surgical emergency cannot be excluded; there is a tubo-ovarian abscess; she is pregnant; she has severe disease (high fever, vomiting, signs of pelvic peritonitis); she cannot tolerate or has failed oral therapy; or there is no clinical response within 72 hours of outpatient treatment. Everyone else can be treated as an outpatient with planned review.

When the patient improves on intravenous therapy — afebrile and clinically better for 24–48 hours — step down to oral doxycycline plus metronidazole to complete the full 14 days. The 14-day course is the standard duration across regimens; stopping at clinical improvement under-treats and invites recurrence. Where a tubo-ovarian abscess was present, the oral tail must retain anaerobic cover, so clindamycin or metronidazole continues alongside the doxycycline to 14 days rather than doxycycline alone.

Tubo-ovarian abscess is a distinct management problem and the point at which PID becomes surgical territory. It demands admission and intravenous broad-spectrum antibiotics with secure anaerobic cover, and at least 24 hours of inpatient observation watching for rupture or sepsis. Many smaller abscesses (conventionally those under about 7–9 cm) resolve on antibiotics alone, but a large collection, a failure to improve over 48–72 hours of adequate intravenous therapy, or any sign of rupture mandates drainage. The least-morbid effective route is image-guided drainage — transvaginal or percutaneous under ultrasound or CT — which both controls the source and yields fluid for culture to refine the antibiotics; it has displaced upfront laparotomy as the first interventional step for an accessible unilocular collection. Surgery (laparoscopic or open drainage and washout, occasionally unilateral salpingo-oophorectomy for a destroyed adnexa) is reserved for an abscess not amenable to or failing image-guided drainage, a ruptured abscess with generalised peritonitis, or septic deterioration. A ruptured tubo-ovarian abscess is a surgical emergency with a real mortality — generalised peritonitis and septic shock — and is managed with resuscitation, urgent laparotomy and source control, not with antibiotics alone. Fertility-preserving conservatism (drainage rather than oophorectomy, unilateral rather than bilateral surgery) is the goal in the young woman wherever the abscess and her stability allow it, because the whole point of treating PID promptly is the tube.

The intrauterine device. A diagnosis of PID is not an automatic indication to remove the IUD. In mild-to-moderate disease the device may be left in situ while antibiotics are started, with review at 48–72 hours and removal only if there is no clinical improvement. The decision to remove must be weighed against pregnancy risk — if the woman has had unprotected intercourse in the preceding seven days, removal exposes her to conception, so offer emergency contraception in that situation. Severe disease or a tubo-ovarian abscess tilts the balance towards removal once antibiotic cover is established.

PID in pregnancy is uncommon (the cervical mucus plug and decidua are protective) but, when it occurs, mandates admission and intravenous therapy because of the risk to mother and pregnancy, and forces a deliberate change of drugs: doxycycline is avoided (tetracycline effects on fetal bone and teeth) and quinolones are not used, so the regimen is rebuilt around a cephalosporin with a pregnancy-acceptable macrolide and metronidazole. Confirming the diagnosis is also harder because the gravid uterus and the differential of obstetric causes complicate the picture — a low threshold for imaging and senior involvement applies.

Ongoing — supportive care and source control. Adequate analgesia, antipyretics, and rest in severe disease; abstinence from intercourse until the woman and her partner(s) have completed treatment. Reassess at 72 hours: improvement confirms the diagnosis and the regimen; failure to improve re-opens the diagnosis and triggers imaging for an abscess.

Long-term — partners, prevention and sequelae. Partner notification and empirical treatment is integral, not optional: untreated partners re-infect, and re-infection multiplies the risk of tubal damage. Male partners are treated empirically — the SA syndromic approach treats the index and the contact, and contemporary guidance favours doxycycline for male partners to limit macrolide exposure and the selection of M. genitalium resistance. Test-of-cure is not routinely required for chlamydia or gonorrhoea after a recommended regimen, but is advised where adherence is doubtful, symptoms persist, in pregnancy, and specifically for M. genitalium (where treatment failure is common) and for gonorrhoea treated with a non-first-line regimen; re-testing for re-infection at 3 months is good practice. Counsel explicitly on the long-term implications — that fertility is usually preserved after a single well-treated episode but that future infertility, ectopic pregnancy and chronic pelvic pain remain risks that rise with each recurrence — and reinforce barrier contraception.

Guidelines compared

The major bodies agree on the principle — low threshold, broad spectrum, anaerobic cover — and diverge mainly on the first-line agent and the handling of M. genitalium.

The instructive contrast is syndromic versus aetiological management. The UK and US treat against a laboratory result wherever possible and increasingly tailor therapy to M. genitalium resistance; South Africa, like much of the WHO's membership, runs a syndromic algorithm — the woman with lower abdominal pain is treated for the likely pathogens immediately, without waiting for (or often having) a NAAT, because the laboratory turnaround and access realities would otherwise leave her untreated. The trade-off is real: syndromic management treats some women who do not need it and misses asymptomatic upper-tract infection, but in a high-prevalence, access-limited service it captures and treats disease that aetiological pathways would lose. A defensible SA answer names the syndromic regimen, adds the NAAT and HIV test where available, and treats the partner.

The evidence & the controversy

Outpatient treatment is as good as admission for mild-to-moderate disease. This is settled by a single large trial and it reorganises the service: most women with PID do not need a hospital bed, which matters enormously for a system short of beds. The corollary, often missed, is that the trial studied mild-to-moderate disease — it does not license treating a tubo-ovarian abscess, a pregnant woman, or a septic patient at home.

Anaerobes matter, and metronidazole earns its place. For years the standard regimen of a cephalosporin plus doxycycline carried only weak anaerobic activity, and whether broader cover changed outcomes was genuinely unsettled. A double-blind placebo-controlled trial answered it: adding metronidazole eradicated endometrial anaerobes, suppressed residual M. genitalium, and reduced persistent pelvic tenderness, without a tolerability penalty — which is precisely why the 2021 CDC update made metronidazole standard rather than discretionary. The remaining nuance, preserved in the UK guideline, is that the anaerobic contribution is greatest in severe disease, so metronidazole may be dropped in mild disease only if it genuinely cannot be tolerated.

Mycoplasma genitalium is the live controversy, and the South African position is distinctive. Globally, M. genitalium is developing alarming antimicrobial resistance: contemporary pooled estimates put macrolide resistance around a third of isolates and fluoroquinolone resistance over one in ten, with dual-class resistance emerging and rising fastest in Europe and the Western Pacific. That trajectory threatens the macrolide-based syndromic approach and is the argument for resistance-guided therapy and test-of-cure where laboratories permit. South Africa, however, has so far been spared the worst of it — surveillance in local cohorts has found macrolide-resistance mutations to be uncommon, even rare, despite azithromycin's use in syndromic management since 2015. That divergence is the SA-context point: the macrolide-based syndromic regimen remains effective here in a way it no longer reliably is in high-resistance settings, but the gap is a window, not a guarantee — continued surveillance is the price of keeping it, because imported resistance or selection under syndromic pressure could close it. A candidate who simply parrots "moxifloxacin for resistant M. genitalium" without the local resistance picture has missed the point: the right answer is contingent on where you are practising.

Prevention is moving upstream, and the newest controversy is doxycycline post-exposure prophylaxis (doxyPEP) — a dose of doxycycline taken after condomless sex to prevent bacterial STIs and, downstream, the ascending infection that becomes PID. Trials in men who have sex with men have shown large reductions in chlamydia and syphilis and a smaller reduction in gonorrhoea, and the strategy is entering some international guidelines for that population. The South African reading is pointedly more cautious for two reasons that are exactly the kind of resource-and-resistance argument the topic demands. First, a trial in young women did not show efficacy, with adherence the likely barrier — so the population most affected by PID in SA is the one in whom doxyPEP has not been shown to work. Second, a large majority of South African gonococci are already tetracycline-resistant, so doxycycline does little against the local gonococcus and population-wide use risks selecting further resistance. The defensible position is therefore that doxyPEP is not a general-population or routine women's intervention in SA, and any use sits within controlled settings with resistance surveillance — a clean example of why an intervention that is "evidence-based" elsewhere is not automatically the right answer here.

The sequelae are the reason any of this matters, and they are the strongest argument for the deliberately over-sensitive diagnostic threshold. Long-term follow-up of laparoscopically verified PID shows that the damage is real and rises with the number and severity of episodes — failure to conceive (16% vs under 3% in controls) and confirmed tubal-factor infertility (about 11% vs none in controls) are markedly higher after PID, and the ectopic-pregnancy rate is several-fold raised (around 9% vs 1–2%). The clinical force is preventive: each episode prevented, and each episode treated early and adequately, is a tube saved.

Landmark trials & key evidence

A short piece of arithmetic from the resistance data sharpens the SA point: if roughly a third of M. genitalium worldwide is now macrolide-resistant, then in a high-resistance setting an azithromycin-based syndromic regimen fails its M. genitalium infections about one time in three — which is why those services have moved to resistance-guided moxifloxacin. The SA cohort finding of essentially zero macrolide resistance is therefore not a footnote: it is the empirical justification for a different national regimen, valid for exactly as long as surveillance keeps confirming it.

Exam traps & red flags

• Waiting for confirmation before treating. PID is a clinical diagnosis treated on minimal criteria; demanding all three of cervical motion, uterine and adnexal tenderness, or a positive swab, before starting antibiotics loses tubes. One minimum criterion in an at-risk woman is enough.
• Missing the ectopic. Every woman of reproductive age with lower abdominal pain gets a pregnancy test. A ruptured ectopic treated as PID is a preventable death.
• Omitting anaerobic cover. A cephalosporin plus doxycycline alone under-treats; metronidazole is part of the standard regimen, not an optional extra — its omission is now a clear error in anything but the mild, metronidazole-intolerant patient.
• Treating a tubo-ovarian abscess as ordinary PID. A complex adnexal mass, a systemically unwell woman, or failure to respond at 72 hours means image and admit; a large or non-resolving abscess needs drainage, and a ruptured one is a surgical emergency. Outpatient oral therapy here is undertreatment.
• Reflexly removing the IUD. The device may stay in mild–moderate PID with review at 48–72 hours; removing it without considering recent unprotected intercourse can leave the woman pregnant.
• Forgetting the partner. Untreated partners re-infect and compound tubal damage; partner notification and empirical treatment is part of managing the index case, not an administrative afterthought.
• Skipping the HIV test. In the SA setting the STI consultation is a gateway to HIV diagnosis and care; HIV-positive women may have more severe symptoms but respond to standard antibiotics — no dose change is needed, and HIV is never a reason to under-treat.
• Quinolones first-line in a high-gonococcal setting. Ofloxacin/moxifloxacin are effective agents but should not be empirical first choice where gonococcal and quinolone-resistance risk is high — reserve moxifloxacin for confirmed or strongly suspected M. genitalium.
• Over-claiming a single regimen as universal. The right M. genitalium answer depends on local resistance: moxifloxacin where resistance is high, macrolide-based syndromic therapy where (as in SA) resistance remains low — stated without the local picture, either is half an answer.
• Treating PID without considering genital tuberculosis in a woman who is not improving on standard therapy in a high-TB-burden setting — a chronic, paucisymptomatic upper-tract infection that conventional PID antibiotics will not touch.

Evidence anchors

• PEACH trial — Ness et al., Am J Obstet Gynecol 2002 (DOI 10.1067/mob.2002.121625)
• Wiesenfeld et al. — ceftriaxone + doxycycline ± metronidazole for acute PID, Clin Infect Dis 2021 (DOI 10.1093/cid/ciaa101)
• Weström et al. — PID and fertility cohort, Sex Transm Dis 1992 (PMID 1411832)
• Chua et al. — macrolide and fluoroquinolone resistance in M. genitalium, Lancet Microbe 2025 (DOI 10.1016/j.lanmic.2024.101047)
• Peters et al. — lack of macrolide resistance in M. genitalium in SA pregnant women, Sex Transm Infect 2021 (DOI 10.1136/sextrans-2020-054583)
• Peters et al. — doxycycline post-exposure prophylaxis for STIs in South Africa, S Afr J HIV Med 2023 (DOI 10.4102/sajhivmed.v24i1.1510)
• CDC STI Treatment Guidelines 2021 — Pelvic Inflammatory Disease
• BASHH 2018 UK National Guideline for the Management of Pelvic Inflammatory Disease
• Southern African HIV Clinicians Society 2022 STI guideline — Peters et al., S Afr J HIV Med 2022 (DOI 10.4102/sajhivmed.v23i1.1450)
• South Africa NDoH Standard Treatment Guidelines & Essential Medicines List, Primary Healthcare (2020), Chapter 12 (Sexually Transmitted Infections) — syndromic "lower abdominal pain" algorithm: ceftriaxone + azithromycin + metronidazole, with partner treatment and HIV testing.