Corrected QT Interval and Torsades Risk Calculator

Evidence & Clinical Pearls

1. The Measurement Is Where Most of the Error Lives

Every formula on this page is arithmetic performed on a number somebody measured by eye. The arithmetic is exact and the measurement is not, so the corrected value can never be better than the QT that went into it. Measure from the first deflection of the QRS to the point where the T wave returns to the isoelectric line, in the lead where the T wave ends latest and is most clearly defined, conventionally lead II or V5. Where the downslope is gradual, draw a tangent to its steepest part and take the point at which that tangent crosses the baseline. Average over three consecutive beats where the rhythm is regular.

Two habits protect the number. Do not include a U wave: a low, late, rounded deflection after the T wave is a U wave and belongs to the next part of repolarisation, not to the QT. And treat the machine's automated reading as a screening figure only. Automated algorithms are reliable when the T wave is tall and clean, and unreliable exactly when it matters, which is when the T wave is flat, notched, biphasic or merged with a U wave. A drug should not be stopped on an automated reading that has not been confirmed by hand.

2. Why the Interval Has to Be Corrected at All

Ventricular repolarisation shortens as the heart rate rises, so the raw QT falls with tachycardia and lengthens with bradycardia in every normal heart. Comparing a QT of 400 ms at a rate of 50 with a QT of 400 ms at a rate of 120 therefore compares two entirely different states of repolarisation. Correction expresses the interval as what it would have been at a rate of 60 beats per minute, which is the only form in which two ECGs, or one ECG before and after a drug, can be set beside each other.

The correction is an empirical fit to population data, not a physiological law, which is why five formulas exist and none of them is right for everybody. They agree exactly at a rate of 60, where the RR interval is one second and every one of them reduces to the measured QT. They diverge progressively as the rate moves away from 60, and at the extremes they can differ from one another by more than 50 ms, which is the width of the whole decision.

3. The Five Formulas, and Which the Heart Rate Allows

This tool prints all five and leads with Fridericia, because a single displayed number invites the reader to treat it as the answer when the honest position is that the formulas disagree and the heart rate decides which disagreement matters. Where the rate is between 60 and 100 the five converge and the choice is immaterial. Where it is not, the gap between Bazett and Fridericia is the size of the error you would have carried if you had read the machine.

4. Where the Thresholds Come From, and Where They Disagree

This tool bands the corrected interval as normal at or below 450 ms in men and 460 ms in women, prolonged above that, and markedly prolonged above 500 ms. The 500 ms figure is the one that carries most of the clinical weight, because above it the risk of torsades de pointes rises steeply. The 2010 American Heart Association statement on preventing torsades in hospital treats either a QTc above 500 ms or a rise of more than 60 ms from the patient's own baseline as the point at which the drug list must be reviewed.

A disagreement worth knowing about, and not settled here. A second convention is widely attributed to the same 2009 AHA, ACCF and Heart Rhythm Society standardisation statement: normal to 430 ms in men and 450 ms in women, borderline from 431 to 450 and from 451 to 470, and prolonged only above 450 and 470 respectively. The programmatic guidance for drug-resistant tuberculosis uses a third variant, normal below 450 in men and below 470 in women. The primary statement could not be read directly during the building of this tool, so the two conventions are set out here rather than one being asserted. The practical consequence is confined to women with a QTc between 461 and 470, who are prolonged on the threshold this tool uses and borderline on the other. Above 500 ms, and for a rise of more than 60 ms from baseline, every convention agrees.

At the other end, a QTc below 350 ms sits below the usual normal range and is worth a second look rather than a diagnosis. The 2022 European Society of Cardiology guideline considers short QT syndrome at a QTc of 320 ms or less, and diagnoses it at 360 ms or less when a pathogenic variant, a family history or a survived arrest is also present. Hypercalcaemia, digoxin effect and hyperthermia shorten the interval far more often than the syndrome does.

5. The Wide QRS Problem

The QT interval contains both depolarisation and repolarisation. When the QRS is wide, part of the measured QT is simply the extra time the ventricles took to depolarise, and the interval is prolonged without repolarisation itself being abnormal. Treating that as drug-induced prolongation stops a drug the patient needed.

Bogossian and colleagues addressed this experimentally in left bundle branch block and derived a modified interval: subtract 48.5 per cent of the QRS duration from the measured QT, which at the bedside is close enough to subtracting half the QRS. The modified QT is then rate-corrected in the ordinary way. It was validated again in patients paced from the right ventricle. This tool applies it in left bundle branch block and in ventricular pacing, and declines to apply it in right bundle branch block or in a non-specific conduction delay, where the correction has not been established. In those two the honest answer is that the corrected interval cannot be read reliably, and the JT interval, which is the QT minus the QRS, or a cardiology opinion, is the way forward.

6. The Drug List, and the Indian Prescribing Reality

CredibleMeds, maintained by AZCERT, is the reference list and places each agent in one of four categories. Known risk means the drug prolongs the QT and has been shown to cause torsades even when taken as directed. Possible risk means it prolongs the QT but there is no established evidence of torsades on ordinary use. Conditional risk means torsades occurs only in defined circumstances, such as overdose, an interaction, hypokalaemia or an inherited long QT. Special risk covers agents that should be avoided by patients with congenital long QT syndrome for other reasons, largely the sympathomimetics.

The table below is a working subset, restricted to agents in common use in Indian practice and taken from the list revised on 12 December 2023. It is a teaching aid with a date on it, not a live register: crediblemeds.org is the authority and the categories move.

Three of these deserve naming separately, because they are prescribed here at a volume that has no Western parallel. Levosulpiride is a known-risk agent sold widely for functional dyspepsia and often continued for months. Domperidone, also known risk, is dispensed for nausea with very little scrutiny and was restricted in Europe for exactly this reason. Ondansetron is given almost reflexively for vomiting, including to patients who are vomiting because they are hypokalaemic, which stacks the two mechanisms. A patient can arrive on all three without any single prescriber having considered the QT.

Amiodarone is the recognised exception. It prolongs the QT in almost everyone who takes it, and yet torsades on amiodarone is distinctly uncommon, because it blocks several currents rather than the rapid potassium current alone and produces little transmural dispersion of repolarisation. A long QTc on amiodarone is expected and is not by itself a reason to stop the drug. It did not appear in any of the four categories in the source read for this table, so no category is assigned to it here.

7. The Electrolytes, and What to Do When Torsades Actually Starts

Hypokalaemia, hypomagnesaemia and hypocalcaemia each prolong repolarisation, and they travel together: diarrhoeal illness, diuretics, alcohol use and refeeding produce all three at once. Correcting potassium without correcting magnesium fails, because magnesium is required for the cell to retain potassium at all, which is why a resistant hypokalaemia is a magnesium problem until proved otherwise. Hypocalcaemia lengthens the ST segment specifically and is the one electrolyte cause that prolongs the QT without much changing the T wave itself.

If torsades de pointes occurs. Magnesium sulphate is given whether or not the serum magnesium is low, because its action here is on the membrane rather than on a deficiency. In Indian practice the ampoule is usually 50 per cent magnesium sulphate, in which 2 g is 4 mL, given intravenously over 10 to 15 minutes and repeated once if the rhythm returns. Stop every contributing drug, replace potassium to the upper half of the normal range, and treat the pause-dependence. Torsades is characteristically initiated after a pause. Raising the rate with isoprenaline or with temporary pacing to about 90 to 110 beats per minute suppresses it when magnesium alone has not. Defibrillate a sustained, haemodynamically unstable episode as ventricular fibrillation.

8. QT Monitoring Under the National Tuberculosis Elimination Programme

Drug-resistant tuberculosis regimens routinely stack QT-prolonging agents: bedaquiline and delamanid at possible risk, moxifloxacin at known risk, and clofazimine at conditional risk, given together for months. The programmatic guidance specifies Fridericia, not the Bazett value the ECG machine prints. That is the formula used in the phase II studies of both bedaquiline and delamanid, and in the STREAM trial. Where monitoring can be done often, the schedule is baseline, week 2, week 4, then every 4 weeks; where it cannot, baseline, week 4, week 8 and week 24, with additional ECGs whenever clinically indicated.

Levofloxacin is the useful practical point in that list. Its effect on the QT is small, and the guidance states that ECG monitoring is not critical for levofloxacin, whereas moxifloxacin's larger effect does warrant monitoring when it is combined with other QT-prolonging agents. Where a fluoroquinolone is needed in a regimen that already contains bedaquiline and clofazimine, that difference is a real choice.

Indian data. In a Mumbai cohort of 70 patients given bedaquiline and delamanid concomitantly under programmatic conditions, QTcF prolongation was recorded in about 7 per cent, and only two patients exceeded 500 ms, both within the first month and neither subsequently. The observed rate was lower than in comparable series. That is reassuring about the regimen and is also an argument for concentrating the scarce ECG capacity of a district programme in the first four to eight weeks, which is when the events actually occur.

9. The Poisoned Patient

In poisoning the QT is doing two jobs at once: it is a marker of the poison's effect on repolarisation, and it is a marker of severity. In organophosphorus poisoning, which is among the commonest reasons for an intensive care admission in rural southern India, a prolonged QTc at presentation has repeatedly been associated with respiratory failure, with the need for ventilation and with mortality. That signal arrives before any cholinesterase assay returns. It costs one ECG.

The correction formulas are least reliable in exactly this population, because poisoned patients are frequently tachycardic and Bazett over-corrects hardest there. The toxicological literature therefore prefers the Chan and Isbister QT nomogram, in which the uncorrected QT is plotted directly against the heart rate and a point above the boundary line is regarded as at risk. This tool does not implement that nomogram: the coordinates of the line could not be obtained from a primary source, and a boundary reconstructed from a picture would be a confident wrong number in the one setting where the decision is time-critical. Where the nomogram is what you want, plot it from the published figure. The Rautaharju value printed here is the closest of the five formulas to being usable at a high heart rate.

10. What This Tool Deliberately Does Not Do

It does not diagnose congenital long QT syndrome. That needs repeated ECGs, a family and syncope history, and frequently genetic testing, and the Schwartz score which formalises it is not reproduced here. It does not recommend stopping any specific drug, because the decision depends on what the drug is treating. It does not read an ECG image. And the Tisdale score it offers answers a different question from the corrected interval beside it: the interval says whether repolarisation is prolonged on this ECG, while the score says how likely this patient is to develop prolongation during the admission. A patient can score high with a normal QTc today, and that is the situation the score was built to find.