Incidence and outcomes of long QTc in acute medical admissions

Summary Aims Prolonged QT interval on electrocardiogram (ECG) increases the risk of ventricular arrhythmia. Patients admitted to acute medical units (AMU) may be at risk of QT prolongation from multiple, recognised risk factors. Few data exist regarding incidence or outcomes of QT prolongation in acute general medical admissions. The aims were to determine the incidence of Bazett's‐corrected QT (QTc) prolongation upon admission to AMU; the relationship between QTc and inpatient mortality, length of stay and readmission; proportion with prolonged QTc subsequently administered QT interval‐prolonging drugs. Methods Retrospective, observational study of 1000 consecutive patients admitted to an AMU in a large urban hospital. Exclusion criteria: age <18 years, ventricular pacing, poor quality/absent ECG . QTc determined manually from ECG obtained within 4‐hours of admission. QTc prolongation considered ≥470 milliseconds (males) and ≥480 milliseconds (females). In both genders, >500 milliseconds was considered severe. Study end‐points, (a) incidence of QTc prolongation at admission; (b) inpatient mortality, length of stay and readmission rates; (c) proportion with QTc prolongation subsequently administered QT interval‐prolonging drugs. Results Of 1000 patients, 288 patients were excluded, therefore final sample was n = 712. Patient age (mean ± SD) was 63.1 ± 19.4 years; females 49%. QTc prolongation was present in n = 50 (7%) at admission; 1.7% had QTc interval >500 ms. Of the 50 patients admitted with prolonged QTc, 6 (12%) were subsequently administered QT interval‐prolonging drugs. QTc prolongation was not associated with worse inpatient mortality or readmission rate. Length of stay was greater in those with prolonged QTc, 7.2 (IQR 2.4‐13.2) days vs 3.3 (IQR 1.3‐10.0; P = 0.004), however, in a regression model, presence of QTc did not independently affect length of stay. Conclusions QTc interval prolongation is frequent among patients admitted to AMU. QT interval‐prolonging drugs are commonly prescribed to patients presenting with prolonged QTc but whether this affects clinical outcomes is uncertain.

and so may be more vulnerable for drug-induced QTc prolongation and thence arrhythmia, as the QTc interval is known to increase with age. 2,3 In addition, elderly patients may have other risk factors for QTc interval prolongation, such as the presence of diabetes mellitus, polypharmacy or electrolyte disturbances. 4 QT prolongation has been shown to be present in up to 45% of patients after cerebrovascular accident (CVA) 5 and in 18% of patients on a coronary care unit (CCU) 6 but a much lower prevalence (0.9%) was recorded in a single centre study across all inpatients of a general hospital. 7 The incidence and prevalence of prolonged QTc in unselected medical admissions to an acute medical unit (AMU) is unknown and the frequency with which QT interval-prolonging drugs are prescribed to patients at admission and following admission, in this environment, are also unknown. We hypothesised that QT prolongation would be more frequently encountered in acute medical admissions than in a hospital-wide environment, 7 thereby posing a considerable clinical risk.
The objectives of this study were to (a) determine the incidence of long QT upon admission to a large urban teaching hospital with an acute medical illness; and (b) determine the proportion of patients receiving QT-prolonging drugs at the time of admission and the change in prescribing of QT-prolonging drugs after admission.

| ME THODS
A retrospective study of 1000 consecutive patients admitted to an AMU at a 950-bed, urban teaching hospital. At our institution, STelevation myocardial infarction and acute CVA are triaged directly to the cardiac and stroke services, respectively, and so are not referred to the AMU. Patients routinely have a 12-lead ECG within 12 hours of admission (Welch-Allyn ® ). The ECG trace was accessed using EDMonline ® software. The first satisfactory ECG obtained for each ED visit was included. If the ECG was noisy, obscuring the relevant onset or offsets, incomplete with some leads missing, or rapid with overlap of the terminal T wave and the next cycle P or QRS waves, then the next ECG obtained during the same ED visit was assessed.
If there was no satisfactory ECG, then the case was excluded. Other exclusions were patients aged less than 18 years of age, paced ECG, missing patient ECG or missing patients' identifiers.
The QT interval represents the time from the beginning of ventricular depolarisation to completion of repolarisation. However, because the QT interval encompasses ventricular depolarisation adjustment of QTc in those with left bundle branch block was made using the formula: measured QT interval minus 50% of LBBB duration. 8,9 The presence and severity of candidate risk factors (determined from the published literature) were recorded (obtained from the EPR system). These fell under three domains: basic demographic (age, gender 10,11 ), biochemistry (capillary glucose at the time of the ECG, 12 potassium, 13 calcium and magnesium concentrations 10,14 obtained closest temporally to ECG acquisition within 12 hours) and medications prescribed within 7-days of the index ECG. Community-issued medications are obtained by pharmacist reconciliation with primary care. All data were anonymised. A medication was classified as QTc-prolonging based upon listing at a pharmacy-reference website (www.crediblemeds.org). 15 In this listing, medications are sub-classified as 'known', probable' or 'conditional' for QTc prolongation, or whether they are at risk of inducing QTc prolongation in those with congenital long-QT syndrome.
The principal outcome measure was the incidence of long QTc using Bazett formula in acute medical admissions. Bazett formula is QTc = QT*(HR/60) 1/2 . 16 Secondary outcome measures were: proportion of patients with prolonged QTc who were subsequently administered drugs known to prolong QTc during their inpatient stay; length of admission; in-hospital mortality; rate of readmission at 1, 3 and 12 months; inter-observer agreement (kappa statistic) of QTc length. We also compared the incidence of QT based upon the automatic ECG output, which uses the Bazett formula by default and with the Fridericia formula (QTcF), for which the formula is QTc = QT*(HR/60) 1/3 . QTcF is thought to reflect a more accurate correction factor than Bazett's formula in subjects with faster heart rates. 17 QT intervals were determined manually by two investigators.
QT intervals were determined manually from lead II of 12-lead ECGs. QT intervals were measured from the earliest QRS deflection to the end of the T wave. The intersection of the down slop of the T wave with the baseline was taken as the end of the QT interval. With a sinus rhythm, QT and RR intervals were averaged over three consecutive complexes. During other rhythms, QT and RR intervals were averaged over all complexes on lead II.
QTc of ≥500 milliseconds was considered significantly prolonged in both genders; >470 milliseconds in men and >480 milliseconds in women was considered moderately prolonged 6,18 and represents the 99th percentile for QTc. 19 Cohen's kappa statistic,

What's known
There is a high prevalence and poor outcomes of prolonged QTc in cardiac and cerebrovascular disease. Bazett formula is considered the gold-standard means to control QT interval for heart rate.

What's new
The incidence of prolonged QTc interval (using Bazett formula) in unselected medical admissions is 7%. Automatic for inter-observer agreement, was made from a subgroup of 200 randomly chosen ECGs evaluated by a third observer. Ethical permission was sought but was considered not necessary by the hospital's Research and Innovation (R&I) department.

| Statistical analysis
Normality of distribution was assessed using the Kolmogorov-Smirnov test. Comparison of normally distributed data was made by student's unpaired t test assuming equal or unequal variance as necessary. Correlation of heart rate with Bazett formula and to
Admission potassium was available in n = 581 (82%), calcium in n = 703 (99%), magnesium in n = 603 (85%). The admission F I G U R E 1 Study flow diagram potassium and the proportion of patients using QT-prolonging drugs on admission were significantly different in those patients with prolonged QTc (Table 1).
Kappa inter-observer agreement for long QTc with manual Bazett was 0.65 indicating substantial agreement.

| In-hospital prescription of QTcprolonging drugs
At admission, 328 patients (46%) were receiving ≥1 QTc-prolonging drug of any kind. Of this group, 96 patients were using medication known to cause TdP, 70 were using drugs with a possible risk of TdP, 231 using medications with a conditional risk of TdP and 13 were using medications with a risk in congenital long-QT syndrome. The QTc-prolonging drugs observed in this study are shown in Table 2. The mostly frequently prescribed drug (with a known risk of TdP) was Citalopram, although clarithromycin was the most commonly instituted drug in this risk group. Furosemide was the most commonly prescribed drug with any risk of QTc prolongation.

| Patient outcomes
Length of stay was greater in those with prolonged QTc (median 7.2, IQR 2.4-13.2 days) than in those without (median 3.3, IQR 1.3-10.0 days; P = 0.004). However, in a linear regression model for length of stay, using predictor variables with P < 0.1 (from Table 1 (Table 3).

| D ISCUSS I ON
This study has shown that approximately 1 in 14 patients admitted to an acute unit had prolonged QT interval as judged by a manual Bazett method. Ventricular arrhythmias are most often associated with QTc values of 500 milliseconds or more 17,20 and in this group (severe long QTc) the incidence was 1.7%. This is the first major study reporting incidence in acute medical admissions.

TA B L E 1 Comparison of candidate factors reported to affect QTc
F I G U R E 2 Change in number of medications with potential to prolong QT interval, following admission, in the entire cohort By contrast, prevalence data have shown that prolonged QTc is frequently encountered in an acute stroke unit 5 or CCU. 6 These data likely reflect the case mix of the various cohorts, as transmural myocardial ischaemia 21 and intraventricular haemorrhage 22 are factors associated with long QT, but are usually seen within a high-dependency environment. However, long QT (particularly drug-induced long QT) has also been associated with increasing age, congestive cardiac failure, hypokalaemia 4 and anti-psychotic medication. Individuals with major psychiatric disorders, with cardiovascular diseases and elderly people comprise a sizeable proportion of the acute medical population. Acutely ill patients are likely to have multiple acquired risk factors for QT prolongation. 23 Our data suggest that low-normal plasma potassium has a role in prolonging the QT interval and is a concern given the large number of patients prescribed potassium wasting diuretics. Hypokalaemia Automatic QT interval measurement, integrated into the electrocardiographic device, is convenient for routine use but can be imprecise. 24 In our series, the automatic measure overestimated the QT interval compared to manual evaluation-even in patients without bundle branch block. Why this occurred is uncertain although discordance has been suggested to relate to difficulties in delineating the end of T wave either due to 'noise' or in cases where the T wave is flat, bifid, biphasic, or overlapping on a U wave. 25 Conversely, the Fridericia formula gave a far lower estimation of incidence of prolonged QT, with a mean difference of 23 milliseconds between Bazett and Fridericia formulae. This difference was larger than previously reported. 26 Fridericia formula may be more accurate in the context of tachycardia, 17 although Bazett's formula is thought to adequately correct for heart rates varying between 50 and 90 beats per minute 27,28 ; of note the mean heart rate in this study was 87/ min and 40% of patients had a pulse over 90/min. The correlation data suggested that QTc tended to shorten with tachycardia using Fridericia, whereas Bazett formula had a positive correlation with heart rate.
The prescribing patterns reflect the case mix seen on an AMU.
NICE guideline recommends consideration of dual antibiotic therapy with amoxicillin and a macrolide (such as clarithromycin) for patients with moderate-or high-severity community-acquired pneumonia. 29 Use of fluoroquinolones, which have been a cause for concern to their risk for TdP, 30 are not promoted as first line. 29 Safety issues have also been raised for use of QTc-prolonging anti-psychotic medications. 2,31 Such drugs are frequently used inhospital inpatients as delirium can affect a fifth of acute medical admissions. 32 NICE guidelines allow for the antipsychotics haloperidol and olanzapine to be used for up to 1-week as a therapy for delirium. 33 Whether QTc contributes to increased hospital mortality in those with delirium is unknown. 34 Prolonged QT interval is known to predict mortality in a variety of other conditions that are frequently encountered in the AMU such as: coronary artery disease, 35 heart failure 36 and diabetes mellitus. 37,38 Survival curves of those with/without prolonged QTc separated well within 50 days of admission, in a hospital-wide study. 7 Our data have shown in-hospital mortality of 3.8% in those with normal QTc and 6.0% of those with long QTc. The study was not powered for this outcome but such a difference would be clinically relevant; based upon these data, approximately 1500 patients would be needed to be studied to detect a difference with 80% power and alpha level of 5%. The more prolonged length of stay (in patients with long QTc) in the unadjusted analysis was unexpected but was no longer significant when correcting for patient age and presence of diabetes or electrolyte disturbance. In conclusion, we have shown that QT prolongation, judged by the Bazett formula, is relatively common in acute medical admissions and that low plasma potassium (albeit within the normal range) and QT-prolonging drugs contribute to this. The incidence of prolonged QTc varied considerably by the formula used for heart rate correction. Whether prolonged QTc interval affects clinical outcomes, in this group of patients, requires further study.  full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.