When Guidelines Collide: Drug-Drug Interactions in HIV and TB Care

Last updated: 27 September 2026

South Africa carries one of the world's heaviest dual burdens of HIV and tuberculosis. That is not only an epidemiological fact; it is a pharmacological one. A large number of patients are treated for both conditions at the same time, and the drugs used for each can work against the other.

The interactions between antiretroviral therapy (ART) and anti-tuberculosis treatment are among the best described in clinical pharmacology — and, in everyday practice, among the most frequently missed. The reason is rarely ignorance. It is structure. HIV and TB services are often separate, each with its own guidelines, its own stationery and its own prescriber. When two individually correct decisions are made in two different rooms, the interaction between them can still slip through the gap.

Why rifampicin dominates the problem

Rifampicin is the backbone of first-line TB treatment and the single most consequential drug interaction in this space. It is a potent inducer of drug-metabolising enzymes and transporters — CYP3A4 and related cytochromes, UGT1A1 and P-glycoprotein among them. Those are precisely the pathways that clear many antiretrovirals. Induce them and plasma concentrations fall, sometimes far enough to lose virological suppression and, with it, years of treatment gains.

Two further mechanisms matter, and they are worth separating because they are solved in different ways:

• Chelation. Polyvalent cations — aluminium, magnesium, calcium, iron, zinc — bind dolutegravir and other integrase inhibitors in the gut and reduce absorption. This is chemistry, not enzymology, and the fix is dose separation rather than a change of drug.
• Gastric pH. Absorption of atazanavir and rilpivirine depends on gastric acidity. Proton-pump inhibitors therefore sit badly with them, and this interaction is easy to miss because the acid suppression is often prescribed for an unrelated indication.

The interactions that shape real regimens

Dolutegravir with rifampicin. Rifampicin lowers dolutegravir exposure substantially. Current South African and WHO guidance is to increase dolutegravir to 50 mg twice daily for the duration of rifampicin treatment, returning to once daily afterwards. The twice-daily step is frequently forgotten at either end — neither started when TB treatment begins, nor stopped when it ends.

Efavirenz with rifampicin. Efavirenz can be used at standard dose alongside rifampicin without adjustment, which is one reason it remains useful in this setting despite no longer being first-line for most patients.

Boosted protease inhibitors with rifampicin. A ritonavir- or cobicistat-boosted protease inhibitor combined with rifampicin is not recommended: exposure falls far enough to risk virological failure and the emergence of resistance. Where a protease inhibitor must be used, rifabutin is the usual substitute — and rifabutin itself requires dose adjustment, in opposite directions depending on whether it is paired with a protease inhibitor or with efavirenz.

Integrase inhibitors with cation-containing medicines. Antacids, iron and calcium supplements, and some multivitamins reduce integrase inhibitor absorption. Separating administration, or taking calcium and iron with food, resolves most of it.

Opioid substitution therapy. Both rifampicin and efavirenz reduce methadone concentrations and can precipitate withdrawal. This is a case where the interaction is felt by the patient before it is seen by the prescriber, and where a planned dose increase prevents an avoidable relapse.

Rifampicin does not stop there. Warfarin, corticosteroids, hormonal contraceptives and calcineurin inhibitors are all affected, and the same principle applies across the board: the inducer does not respect specialty boundaries.

Why the screen gets skipped

Interaction checking fails for systemic reasons more often than individual ones.

• Siloed services. TB is managed in one clinic and HIV in another. Each reviews its own regimen against its own guideline.
• Single-disease decision tools. Many electronic systems check interactions within a therapeutic area, not across it.
• Regimen changes. Interactions appear and disappear at transitions — when TB treatment starts, when it ends, when a regimen is switched for tolerability. The steady state is safe; the transitions are not.
• Guideline lag. TB and HIV guidelines are revised on independent cycles, so the interaction guidance in each is written at different times.

What good looks like

The practical answer is not more knowledge — it is better placement of the knowledge that already exists.

1. An interaction screen at every regimen change, not only at initiation. Starting and stopping TB treatment should both trigger a review of the ART regimen, and the reverse.
2. Cross-disease checking built into clinical decision support, so the tool that knows about rifampicin also knows about dolutegravir, whatever clinic the patient is sitting in.
3. A named second pair of eyes — pharmacist, clinical pharmacologist or an interested clinician — for any patient on concurrent ART and ATT, at least at the first visit.
4. Authoritative, maintained references in the room. The University of Liverpool's HIV drug interaction resource and the National Department of Health's Standard Treatment Guidelines and Essential Medicines List are the working references; both are updated, and neither is useful if it is not consulted.

Interactions are one of the few areas of clinical pharmacology where the harm is almost entirely preventable. The information exists, the alternatives exist, and the adjustments are small. What is required is that the two conversations — the HIV one and the TB one — become a single conversation about one patient.

Where this meets technology

This is the kind of problem that software can carry a meaningful part of. Interaction data is structured, the trigger points are predictable, and the failure is a missed lookup rather than a missing fact. That makes it a good fit for decision support that works across disease areas rather than within one — which is a large part of why we build the tools we build.

This article is written for healthcare professionals as general education. It is not patient-specific advice and does not replace the current South African Standard Treatment Guidelines, the Essential Medicines List, or clinical judgement in an individual case. For detailed, updatable interaction data, consult the University of Liverpool HIV Drug Interactions resource (hiv-druginteractions.org) and the NDoH STG/EML.