BMJ Open Is Committed to Open Peer Review. As Part of This Commitment We Make the Peer Review History of Every Article We Publish Publicly Available

BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from

BMJ Open is committed to open peer review. As part of this commitment we make the peer review history of every article we publish publicly available.

When an article is published we post the peer reviewers’ comments and the authors’ responses online. We also post the versions of the paper that were used during peer review. These are the versions that the peer review comments apply to.

The versions of the paper that follow are the versions that were submitted during the peer review process. They are not the versions of record or the final published versions. They should not be cited or distributed as the published version of this manuscript.

BMJ Open is an open access journal and the full, final, typeset and author-corrected version of record of the manuscript is available on our site with no access controls, subscription charges or pay-per-view fees (http://bmjopen.bmj.com).

If you have any questions on BMJ Open’s open peer review process please email [email protected] http://bmjopen.bmj.com/ on September 24, 2021 by guest. Protected copyright. BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from

QTc prolongation and prognosis in patients with suspected poisoning in the emergency department – a transnational propensity score matched cohort

ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2017-020036

Article Type: Research

Date Submitted by the Author: 21-Oct-2017

Complete List of Authors: Schade Hansen, Camilla; Department of Emergency medicine Pottegard, Anton; University of Southern Denmark, Clinical Pharmacology, Institute of Public Health Ekelund, U; Lunds Universitet, Kildegaard Jensen, Helene; Odense Universitetshospital, Department of Emergency Medicine Lundager Forberg, Jakob; Helsingborgs lasarett, Department of Emergency Medicine and Prehospital Care Brabrand, Mikkel; Odense University Hospital, Department of Emergency Medicine Lassen, Annmarie; Odense University Hospital, Department of Emergency Medicine http://bmjopen.bmj.com/ Primary Subject Emergency medicine Heading:

Secondary Subject Heading: Epidemiology, Cardiovascular medicine

Keywords: EPIDEMIOLOGY, CARDIOLOGY, TOXICOLOGY

on September 24, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 QTc prolongation and prognosis in patients with suspected 5 6 poisoning in the emergency department – a transnational 7 8 propensity score matched cohort 9

10 Camilla Schade Hansen MD1*, Anton Pottegård MScPharm PhD2, Ulf Ekelund MD, PhD3, Helene Kildegaard Jensen 11 1 4 1,5 1 12 MD , Jakob Lundager Forberg MD PhD , Mikkel Brabrand MD, PhD , Annmarie Touborg Lassen MD, DMSci 13 14 1. Department of Emergency Medicine, Odense University Hospital, Odense, Denmark 15 2. Clinical Pharmacology and Pharmacy, Department of Public Health, University of Southern Denmark, Odense, 16 Denmark For peer review only 17 3. Department of Emergency Medicine, Skåne University Hospital, Lund, Sweden 18 4. Department of Emergency Medicine, Helsingborg Hospital, Helsingborg, Sweden 19 20 5. Department of Emergency Medicine, Hospital of South West Jutland, Esbjerg, Denmark 21 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 24, 2021 by guest. Protected copyright. 41

42 43 44 45 46 Word count text: 2806 47 48 49 *Correspondence to: 50 [email protected] 51 Telephone: + 45 28155105 52 Sdr. Boulevard 29 53 5000 Odense C 54 Denmark 55 56 57 58 1 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 ABSTRACT 5 6 7 OBJECTIVES: Poisoning is a frequent cause of admission to the emergency department (ED), and 8 9 may involve drugs known to prolong the QT interval. The aims of this study were to describe the 10 prevalence of QTc prolongation among ED patients with suspected poisoning and to calculate the 11 12 absolute and relative risk of mortality or cardiac arrest associated with a prolonged QTc interval. 13 14 15 METHODS: We performed a register-based cohort study, including all adult first time contacts 16 For peer review only 17 with suspected poisoning to the ED of two Swedish hospitals (January 2010 to December 2014) 18 19 and two Danish hospitals (March 2013 to April 2014). We used propensity score matching to 20 21 calculate hazard ratios (HR) for all-cause mortality or cardiac arrest (combined endpoint) within 30 22 days after contact comparing patients with a prolonged QTc interval (≥450 ms men, ≥460 ms 23 24 women) with patients with a QTc interval of <440 ms. 25 26 27 RESULTS: Among all first-time contacts with suspected poisoning that had an ECG recorded within 28 29 four hours after arrival (n = 3869), QTc prolongation occurred in 6.5%. The overall mortality after a 30 31 30-day follow-up period was 0.8% (95% CI, 0.6-1.2), with an absolute risk of mortality or cardiac

32 http://bmjopen.bmj.com/ 33 arrest in patients with QTc prolongation of 3.2% (95% CI, 1.4-6.1). A prolonged QTc interval on 34 arrival was associated with a HR of 3.6 (95% CI, 1.0-12.2). 35 36 37 38 CONCLUSION: In the ED, a prolonged QTc interval in patients arriving with suspected poisoning 39

40 seems to be associated with a three-fold increased risk of 30-day all-cause mortality or cardiac on September 24, 2021 by guest. Protected copyright. 41 arrest. 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 2 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 Strengths and limitations of the study 5 6 Patients were included from four different hospitals – two Swedish and two Danish. 7 Propensity score matching was used to adjust for several confounders. 8 9 Subgroups analysis was not possible due to a small number of events. 10 11 The included ECGs were all automatic readouts and the length of the QT interval was not 12 confirmed manually. 13 14 15 16 INTRODUCTION For peer review only 17 1,2 18 Poisoning is a frequent cause of admission to the emergency department (ED), and involves a 19 variety of different drugs and substances. A wide range of drugs have been linked to QTc 20 21 prolongation,3 which has been associated with all-cause-mortality, cardiovascular death, and 22 4-9 23 sudden cardiac death. As an increased risk of mortality has been documented in patients treated 24 10-13 25 with potential QTc prolonging drugs, one may hypothesize that the risk is even higher among 26 poisoned patients. Therefore, cardiac monitoring is recommended in patients poisoned by 27 28 potentially proarrhythmic agents and drugs that can lead to torsades de pointes.14 29 30 Only few studies have investigated the relationship between QTc prolongation and adverse 31 outcomes in a population of undifferentiated poisoned patients.15,16 The absolute and relative risk 32 http://bmjopen.bmj.com/ 33 of mortality and cardiac arrest associated to QTc prolongation in poisoned patients remains 34 35 unknown. Therefore, we aimed to: (1) describe the prevalence of QTc prolongation found among 36 37 patients with suspected poisoning in the emergency department; (2) to investigate if QTc 38 prolongation is associated with an increased risk of mortality or cardiac arrest within 30 days after 39

40 arrival to the emergency department. on September 24, 2021 by guest. Protected copyright. 41 42 43 44 MATERIALS AND METHODS 45 Study design and setting 46 47 This is a register-based cohort study. The study is based on ED data from January 1 2010 to 48 49 December 31 2014 from two Swedish hospitals (Skåne University Hospital, Lund and Helsingborg 50 51 Hospital) and from two Danish hospitals (Odense University Hospital and the Hospital of South 52 West Jutland, Esbjerg) from March 1 2013 to April 30 2014. In both Denmark and Sweden, the 53 54 healthcare systems are tax-funded and all residents have free access to healthcare. The University 55 56 57 58 3 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 Hospital Skåne has a contingency population of approximately 310,000, whereas Odense 5 6 University Hospital covers a population of 290,000 people. The two regional hospitals have a 7 contingency population of 250,000 people (Helsingborg), and 220,000 people (Esbjerg). 8 9 10 11 Selection of participants 12 We identified all adults (≥18 years), who arrived to the EDs with suspected poisoning. The contacts 13 14 were eligible for the main analysis if they had a 12-lead ECG recorded within 4 hours after arrival. 15 16 A missing QTc intervalFor on the peerrecorded ECG, review or a QRS duration onlyof ≥120 ms were both reasons for 17 18 exclusion. Patients with multiple contacts were included only at their first contact with suspected 19 poisoning within the study period. Information regarding identification of patients with suspected 20 21 poisoning is outlined in Appendix A. 22 23 24 25 Data sources 26 In both Denmark and Sweden, all residents have a unique personal civil registration number, 27 28 which allows cross-linkage at personal level between databases. We extracted data from several 29 17 30 registries: The logistic systems in the ED at the Region of Southern Denmark and Region of 31 Skåne, the electronic central ECG databases at Region of Southern Denmark and Region of Skåne, 32 http://bmjopen.bmj.com/ 33 the Danish National Patient Registry18 and Region of Skåne Health care databases, the Danish 34 19 20 35 National Prescription Registry, the Swedish Pharmacy Registry, and finally The Danish Civil 36 21 22 37 Registration System and the Swedish Population Register. Further information regarding the 38 data sources is provided in Appendix A. 39

40 on September 24, 2021 by guest. Protected copyright. 41 ECG measurements and definitions 42 43 The QT interval was measured at the first ECG recorded after contact to the ED. All the QT 44 45 intervals were calculated automatically as a median value and stored in either MUSE Cardiology 46 Information System (GE Healthcare) or Philips Diagnostic ECG. The GE Marquette 12SL ECG 47 48 Analysis Program provided QTc intervals for ECGs recoded in MUSE.23 ECGs recorded by Phillips 49 24 50 were analyzed by the DXL-algorithm. Only QT intervals corrected for heart rate (QTc) was used in 51 52 our analysis. For correction, we chose the Framingham Formula (QTcFramingham =QT + 0.154 (1- 53 RR)).25 Additional details about ECG measurements are outlined in Appendix B. 54 55 56 57 58 4 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 Exposure and outcome 5 6 Our primary outcome was a combined endpoint of all-cause mortality or cardiac arrest (defined in 7 Appendix C.1) within 30 days from the day of arrival to the ED. Patients who died in relation to 8 9 cardiac arrest were classified as dead rather than cardiac arrests. The primary exposure was QTc 10 26 11 prolongation, defined as a QTc of ≥450 ms for men and ≥460 ms for women. Patients with a 12 normal QTc length were defined as having a QTc interval of <450 ms (men) or <460 ms (women). 13 14 15 16 Analysis For peer review only 17 18 The prevalence of QTc prolongation overall, and in relation to specific groups of poisoning was 19 described in a cross-sectional description. In this description, we identified all patients with a 20 21 discharge diagnosis of poisoning (International Classification of Diseases (ICD-10) codes T36*- 22 23 T65*, F100*, F110*, F120*, F130*, F140*, F150*, F160*, F170*, F180* or F190* as a primary or 24 25 secondary diagnosis). All patients, who had a discharge diagnosis of poisoning, were subdivided 26 into five poisoning groups: 1. Analgesics and drugs of abuse, 2. Psychotropic drugs including drugs 27 28 affecting the central nervous system, 3. Organic and chemical substances, non-medical, 4. Others, 29 30 and 5. Multidrug (see Appendix C.2). 31 The association between QTc prolongation and all-cause mortality and cardiac arrest was 32 http://bmjopen.bmj.com/ 33 evaluated using propensity score matching.27,28 We calculated a propensity score for all included 34 35 patients by use of logistic regression with QTc ≥450 ms (men) or ≥460 ms (women) as the outcome 36 37 (binary outcome). Patients with a QTc interval between 440-449 ms (men) and 440-459 ms 38 (women) were excluded in the model to avoid near-overlapping ranges. The following possible 39

40 confounders were included in the propensity score model: sex, age, comorbidity (measured as on September 24, 2021 by guest. Protected copyright. 41 29,30 42 Charlson Comorbidity Index ), history of myocardial infarction or congestive heart failure 43 (Appendix C3. and C.4), prescription of QT prolonging drugs within 90 days (defined in Appendix 44 45 C.5), heart rate, and study center. We performed a 1:2 parallel balanced nearest neighbor 46 47 matching without replacement and with a caliper of 0.05.31 In the matched cohorts, 30-day 48 49 mortality was modelled using Cox regression. 50 51 52 Statistics 53 54 55 56 57 58 5 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 The absolute risk of event in patients with suspected poisoning was calculated overall, for those 5 6 with QTc prolongation and for those without QTc prolongation. In the propensity score matched 7 cohort, the risk associated with QTc prolongation was estimated as hazard ratios (HR). We 8 9 estimated 95% confidence intervals based on a binominal distribution. To illustrate the impact of 10 11 QTc prolongation on 30-days all-cause mortality or cardiac arrest we generated a Kaplan Meier 12 failure curve. 13 14 In a sensitivity analysis, we restricted the material to individuals who were both suspected of 15 16 being poisoned on arrivalFor and peer received a discharge review diagnose of only poisoning. The prevalence of QTc 17 18 prolongation and the propensity score analyses were repeated using the Bazett formula for QT 19 correction.32 20 21 Statistical analyses were performed using STATA version 14 (StataCorp LP, College Station, Texas). 22 23 The study was approved by the Danish Data Protection Agency (No. 2008-58-0035, Journal nr. 24 25 15/21632) and The Danish Health Authority (No. 3-3013-1031). In consistency with Swedish law 26 the study was approved by the Regional Ethics Committee in Lund and by Region Skåne. 27 28 29 30 RESULTS 31

32 Characteristics of the study cohort http://bmjopen.bmj.com/ 33 At the four hospitals, we identified a total of 6838 ED contacts with suspected poisoning. After 34 35 exclusion of those aged <18 years (n=22), an ECG not recorded in an acceptable time-interval 36 37 (n=1411), multiple contacts within the study period (n=1412), a missing QT interval (n=1), or QRS 38 39 duration ≥120 ms (n=123) the final cohort comprised 3869 patients with suspected poisoning 40 (48.0% men, median age 38) (Figure 1). Of these, 69.2% (n=2676) had a discharge diagnose of on September 24, 2021 by guest. Protected copyright. 41 42 poisoning. 43 44 45 Patients with a prolonged QTc interval were older, had more comorbidity, and more commonly 46 47 had a history of heart disease than those without QTc prolongation (Table 1). 48 49 Table 1: Baseline characteristics of the study population 50 51 All* Before propensity score matching After propensity score matching 52 QTc Prolonged QTc QTc Prolonged QTc 53 <440 ms (men and ≥450 ms (men) <440 ms (men and ≥450 ms (men) 54 women) ≥460 ms (women) women) ≥460 ms (women) 55 56 57 58 6 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 N 3869 3296 253 496 248 Sex 5 1859 (48.0) 1634 (49.6) 121 (47.8) 229 (46.2) 119 (48.0) Male (%) 6 Age (median, IQR) 38 (25-53) 36 (24-51) 52 (36-68) 53 (37-69) 51 (35-66) 7 18-50 – n (%) 2747 (71.0) 2444 (74.2) 119 (47.0) 236 (48.0) 119 (48.0) 8 51-69 – n (%) 788 (20.4) 611 (18.5) 77 (30.4) 140 (28.5) 77 (31.0) 9 ≥70 – n (%) 334 (8.6) 241 (7.3) 57 (22.5) 116 (23.6) 52 (21.0) Charlson Comorbidity Index – n (%) 10 CCI = 0 2747 (71.0) 2395 (72.7) 140 (55.3) 263 (53.0) 140 (56.5) 11 CCI = 1 718 (18.6) 587 (17.8) 60 (23.7) 133 (26.8) 58 (23.4) 12 CCI ≥ 2 404 (10.4) 314 (9.5) 53 (20.9) 100 (20.2) 50 (20.2) Myocardial infarction or congestive 185 (4.8) 136 (4.1) 32 (12.6) 55 (11.1) 29 (11.7) 13 heart failure – n (%) 14 QT-prolonging drugs – n (%) 1518 (39.2) 1248 (37.9) 110 (43.5) 213 (42.9) 109 (44.0) 15 ECG measurements 16 Heart rate (median, IQR) For peer85 (73-99) 87review (74-101) 76 (65-84) only 76 (65-87) 76 (65-85) 17 QTc ≥500 ms - n (%) 27 (0.7) - 27 (10.7) - 27 (10.9) 18 Any diagnose of poisoning – n (%) 2676 (69.2) 2282 (69.2) 153 (60.5) 310 (62.5) 151 (60.9) Group of poisoning – n (%) 19 1.Analgesics and drugs of abuse 397 (14.8) 333 (14.6) 21 (13.7) 41 (13.2) 21 (13.9) 20 2.Psychotropic drugs and drugs 805 (30.1) 695 (30.5) 49 (32.0) 103 (33.2) 49 (32.5) 21 affecting the central nervous system 3.Organic and chemical substances, 502 (18.8) 437 (19.1) 24 (15.7) 50 (16.1) 24 (15.9) 22 non-medical 23 4.Others 470 (17.6) 392 (17.2) 30 (19.6) 54 (17.4) 29 (19.2) 24 5.Multidrug 502 (18.8) 425 (18.6) 29 (19.0) 62 (20.0) 28 (18.5) Clinics – n (%) 25 The University Hospital Skåne, Lund 1794 (46.4) 1539 (46.7) 125 (49.4) 247 (49.8) 124 (50.0) 26 Odense University Hospital 501 (12.9) 419 (12.7) 28 (11.1) 50 (10.1) 28 (11.3) 27 Helsingborg Hospital 1372 (35.5) 1176 (35.7) 81 (32.0) 170 (34.3) 79 (31.9) 28 Hospital of South West Jutland 202 (5.2) 162 (4.9) 19 (7.5) 29 (5.8) 17 (6.9) Abbreviations: CCI, Charlson Comorbidity Index; IQR, interquartile range. 29 *In the total cohort patients with a near-overlapping QTc interval (440-449 ms men, 440-459 ms women) are included (n =320). 30 31

32 In addition, prescription of QT prolonging drugs was more frequent in the group with a prolonged http://bmjopen.bmj.com/ 33 34 QTc interval. Among the included patients, 6.5% (95% CI, 5.9-7.4) had QTc prolongation, while the 35 prevalence of severe QTc prolongation (≥500 ms) was 0.7% (95% CI, 0.5-1.0). The prevalence of 36 37 QTc prolongation in relation to specific groups of poisoning varied within the range 4.8-6.2%, with 38 39 the highest prevalence in the group categorized as “others” (6.2%; 95% CI, 4.8-8.7) (Table 2).

40 on September 24, 2021 by guest. Protected copyright. 41 Table 2: QTc prolongation in relation to poisoning groups

42 Psychotropic Chemical and 43 Analgesics and drugs including biological drugs of abuse drugs affecting 44 substances, non- Others Multidrug the central medical 45 nervous system

47 T39-T40, F110, ≥2 of the 48 T42-T44, F130, T51-T65, F100, T36-T38, T41, ICD-10 codes or definition F120, F140, F150, described 49 F190 F170, F180 T45-T50 F160 poisoning groups 50 51 N 397 805 502 470 502 52 QTc prolongation – n (%, CI 95%) 53 ≥450 ms (men) 21 (5.3%; 3.3-8.0) 49 (6.1%; 4.5-8.0) 24 (4.8%; 3.1-7.0) 29 (6.2%; 4.2-8.7) 28 (5.6%; 3.7-8.0) 54 ≥460 ms (women)

55 Abbreviation: CI, confidence interval. 56 57 58 7 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 5 6 Prognosis 7 Overall, the 30-day risk of all-cause mortality or cardiac arrest was 0.8% (95% CI, 0.6-1.2, n=32). 8 9 Among individuals with QTc prolongation (n= 253), death within 30 days after contact to the ED 10 11 occurred in 7 patients, whereas one patient suffered from cardiac arrest. Among those with a 12 normal QTc interval (n=3616), we found 24 events during the follow-up period. The absolute risk 13 14 of event within 30 days was 3.2% (95% CI, 1.4-6.1) and 0.7% (95% CI, 0.5-1.0) for patients with and 15 16 without QTc prolongation,For respectively. peer review only 17 18 The propensity score analysis included 248 patients with a QTc of ≥450 ms (men) or ≥460 ms 19 (women) matched with 496 patients with a QTc interval <440 ms. Acceptable balance of baseline 20 21 variables was achieved (Table 1). QTc prolongation was associated with a HR of 3.6 (95% CI, 1.0- 22 23 12.2) for 30-day all-cause mortality or cardiac arrest (Table 3 and Figure 2). 24 Table 3: Risk assessment in the study population 25 Propensity score matched cohort 26 27 n Events (No.) HR** (95% CI) 28 29 Suspected poisoning 30 Normal QTc interval 496 n<5 1.0 (ref) <440 ms 31 32 QTc prolongation 248 8 3.6 (1.0-12.2) http://bmjopen.bmj.com/ 33 ≥450 ms, men 34 ≥460 ms, women 35 Diagnose of poisoning* 36 Normal QTc interval 310 n<5 1.0 (ref) 37 <440 ms 38 39 QTc prolongation 151 6 10.5 (1.2-90.0) ≥450 ms, men on September 24, 2021 by guest. Protected copyright. 40 ≥460 ms, women 41 42 Abbreviations: CI, confidence interval; HR, hazard ratio. If the number of events in the analysis was less than 5 (marked by n<5), the number of 43 patients in the strata is not shown. 44 *Patient who arrived with suspected poisoning and had a discharge diagnose of poisoning. **Cox regression calculated after 1:2 propensity score matching comparing patients with QTc prolongation to patients without QTc prolongation. In 45 this population, patients with near-overlapping ranges of the QTc interval were excluded (QTc 440-449 ms, men and 440-459 ms, women). 46 47 48 Subgroups and sensitivity analyses 49 50 Our results from the subgroup analysis are outlined in Appendix D. When restricting to those who 51 52 also received a discharge diagnose of poisoning, we found an overall 30-day risk of 0.7% (95% CI, 53 0.4-1.1) and QTc prolongation yielded an overall HR of 10.5 (95% CI, 1.2-90.0). When we corrected 54 55 56 57 58 8 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 the QT interval with the Bazett formula a total of 1112 patients had QTc prolongation (28.7%), 5 6 which was associated with a HR of 1.0 (95% CI, 0.2-5.5). 7 8 9 DISCUSSION 10 11 In this transnational cohort of patients with suspected poisoning arriving to the ED, QTc 12 13 prolongation was common (6.5%). A prolonged QTc interval was associated with a three-fold 14 increased risk of 30-day all-cause mortality or cardiac arrest and an absolute risk of 3.2%. 15 16 For peer review only 17 18 This study has several strengths. First, this was a multicenter cohort study with data from two 19 20 Swedish and two Danish EDs which ensured a broad representability. Use of personal 21 identification numbers in all contacts to the hospital system in Sweden and Denmark provide the 22 23 possibility to follow individual patients in and out of hospital and loss of follow-up or unmeasured 24 17-22 25 registration of death did not occur. In addition, we implemented several confounders in our 26 27 propensity score model, and thus managed to control for these despite a low event-rate. We 28 included patients who were suspected for being poisoned on arrival to the ED. These patients do – 29 30 in contrast to patients identified by their discharge diagnosis – represent the clinical situation at 31

32 the door in the ED. At this point, the doctors have to decide whether or not to observe the http://bmjopen.bmj.com/ 33 patients using telemetry. 34 35 36 37 This study also has several limitations. First of all, the design was an observational design. The ECG 38 39 measures were all automatic readouts, and we did not manually validate the length of the QT 40 intervals. However, this method has been validated in a previous Danish study using the same on September 24, 2021 by guest. Protected copyright. 41 42 technique, which showed a good overall agreement between manual QTc interval and the digital 43 8 44 record of the QTc interval with a mean difference of 1.3 ms. Further, we did not exclude ECGs 45 with diagnoses complicating QTc measuring, e.g. atrial fibrillation. We did not have information 46 47 regarding previous ECGs, and we do not know if some patients had a previous ECG with QTc 48 49 prolongation before arrival with suspected poisoning. The dose of drug or substance was 50 51 unknown, and we were ignorant of the timing of the ECG recording in relation to peak drug 52 concentration. The poisonings were not confirmed by blood samples or by urine tests, but were 53 54 extracted from predefined ICD-10 codes. 55 56 57 58 9 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 The small number of events was a limitation in its own and did not allow for meaningful subgroup 5 6 analysis. As cardiac arrest was identified based on hospital registration an eventually event of 7 unregistered cardiac arrest, where the patient survived, is not included as an event. The number 8 9 of these events is believed to be small as registration of cardiac arrest is mandatory in both the 10 11 Swedish and Danish health care system. With a small number of events any miscounting of events 12 would lead to considerable change in risk estimates. If we have overlooked one event of cardiac 13 14 arrest who survived in the group of patients with QTc prolongation it would increase the absolute 15 16 risk from 3.2% to 3.6%,For while peerthe risk of event review in the entire study only population would increase from 17 18 0.8% to 0.9%. 19 20 21 The event-rate in our cohort (0.8%) is in accordance with previous studies of poisoned patients 22 16,33,34 23 (0.5-1.2%). In contrast, the prevalence of QTc prolongation is substantially lower (6.5%) than 24 35 25 in a previous study of unselected ED patients (35%). This is probably due to the choice of QT 26 correcting formula. If the Bazett formula had been chosen for main analysis, the prevalence of QTc 27 28 prolongation in our study population would have been 28,7%. It is of broad consensus that the 29 30 more widely used Bazett formula tends to overcorrect at heart rates at 80-90 beats per minute 31 and above resulting in a higher prevalence of QTc prolongation.32,36 As a high percentage of acute 32 http://bmjopen.bmj.com/ 33 patients have tachycardia at arrival, this probably explains most of the difference between the 34 35 occurrence of QTc prolongation in our study and in the study of unselected ED patients. The 36 37 Framingham formula used in our study is considered superior compared with the more widely 38 used Bazett formula.36 39

40 on September 24, 2021 by guest. Protected copyright. 41 42 The clinical impact of our findings is the difference in risk of all-cause mortality and cardiac arrest 43 within 30 days in respect to QTc prolongation. We found an absolute risk of 0.7% in patients with 44 45 suspected poisoning without QTc prolongation, whereas patients with a prolonged QTc interval 46 47 have an absolute risk of 3.2%, which translates into a HR of 3.6 (95% CI, 1.0-12.2). In the general 48 49 population, a meta-analysis reported a pooled relative risk of 1.35 (95% CI, 1.24-1.46) for long- 50 term mortality in patients with QTc prolongation.7 A recent study including all patients who had an 51 52 ECG recorded at the hospital for any reason reported QTc prolongation to be associated with a HR 53 36 54 of 7.3 (95% CI, 4.10-13.05) for 30-day mortality. Combined, these studies support the hypothesis 55 56 57 58 10 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 that patients with a prolonged QTc are at increased risk. Whether or not this is directly linked to 5 6 the increased QTc interval or due to other risk factors associated with a prolonged QTc remains 7 unknown. 8 9 As demonstrated in our cohort the prevalence of QTc prolongation is strongly associated to the 10 11 correction formula. Further, the difference between the HR calculated in the main analysis using 12 Framingham (HR 3.6; CI 95%, 1.0-12.2) versus the sensitivity analysis using Bazett (HR 1.0; 95% CI, 13 14 0.2-5.5) is remarkable. We suspect that using the Bazett formula dilutes the association by 15 16 including more patientsFor at low peer risk as a result review of overcorrection. only 17 18 19 Despite the use of a propensity score model adjusting for several covariates, we cannot exclude 20 21 residual confounding. From a clinical point of view, this means that the patients with a prolonged 22 23 QTc probably need special care and attention. However, the needed care is not necessarily limited 24 25 to telemetry and increased cardiac awareness. Of note, a ventricular arrhythmia with fatal 26 outcome caused by drug-induced QTc prolongation, would be expected to happen within a 27 28 relatively short time-interval after exposure. This was not the case in our study with the first event 29 30 occurring three days after contact (see Figure 2). In addition, a QTc interval threshold for 31 identification of patients in need of cardiac telemetry is not well-established. Unfortunately, our 32 http://bmjopen.bmj.com/ 33 cohort was too small to do further subdivisions of the QTc interval. 34 35 Ventricular arrhythmias, especially torsades de pointes, are feared consequences of QT 36 37,38 37 prolongation and may be the cause of death in some poisonings. However, as torsades de 38 pointes is a rare condition,37,38 it is unlikely to have influenced our results. 39

40 In this cohort of patients with suspected poisoning 69.2% received a discharge diagnose of on September 24, 2021 by guest. Protected copyright. 41 42 poisoning. This is in contrast to results from a previous Danish study, which found an agreement of 43 79% for suspected poisoning on arrival and a discharge diagnose of poisoning.17 In our cohort, only 44 45 those who had an ECG recorded were included, and several common poisonings, e.g. alcohol 46 47 intoxication, are usually not followed by ECG recording. 48 49 QTc prolongation was most frequent in the group of poisoning labeled “others” (Table 2). In this 50 group, the ICD-10 code T50.9 for unspecific poisonings was given to the majority of the patients. 51 52 These patients might have been too sick to tell about their poisoning or perhaps denied to do so. 53 54 This reflects a common clinical problem in the ED, and indicates that a specific poisoning diagnosis 55 56 57 58 11 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 can be difficult to establish. Further, lack of precision in coding procedure may contribute to 5 6 unspecific diagnoses. 7 8 9 In conclusion, we found QTc prolongation in a mixed population of patients with suspected 10 11 poisoning in the emergency department of two Swedish and two Danish hospitals to be associated 12 with a three-fold risk of 30-day all-cause mortality or cardiac arrest and an absolute risk of 3.2%. 13 14 15 Founding 16 For peer review only 17 This study was funded by an independent grant from The Research Foundation of Odense 18 19 University Hospital. The funding sources had no role in the design of the study, data analysis or 20 21 interpretation of the results. 22 23 24 Competing interests 25 26 ATL was supported by an unrestricted grant to the University of Southern Denmark from 27 28 TrygFoundation. All other authors declare no conflicts of interest. 29 30 31 Author contributions

32 http://bmjopen.bmj.com/ 33 CSH designed the study, interpreted the results and drafted the paper. AP analyzed the data. CSH, 34 AP, ATL, HKJ, UE, MB, and JLF conceived the study. ATL, AP and HKJ provided statistical advice and 35 36 advice on the study design. AP, ATL, HKJ, UE, MB, and JLF critically reviewed the paper, assisted 37 38 with interpretation of the results, and have approved the final edition. CSH takes responsibility for 39

40 the final paper. on September 24, 2021 by guest. Protected copyright. 41 42 43 Data sharing statement 44 45 Due to Danish law regarding personal data we are not allowed to share data in public dataset. 46 47 However, we welcome every researcher who wants to repeat the analysis or do new analysis in 48 the dataset. Please contact professor Annmarie Lassen ([email protected]) and she will 49 50 help the researcher to get access to the data. 51 52 53 54 55 56 57 58 12 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 24, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 13 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 REFERENCES 5 6 7 1. Carter-Storch R, Olsen UF, Mogensen CB. Admissions to emergency department may be classified 8 into specific complaint categories. Dan Med J. 2014;61(3):A4802. 9 2. Safwenberg U, Terént A, Lind L. Differences in Long-term Mortality for Different Emergency 10 Department Presenting Complaints. Acad Emerg Med. 2008;15(1):9-16. 11 3. Woosley RL, Romero KA. QTdrugs List. AZCERT, Inc. https://www.crediblemeds.org/new-drug-list/. 12 Updated August 10, 2016. Accessed December 17, 2015. 13 4. de Bruyne MC, Hoes AW, Kors JA, et al. Prolonged QT interval predicts cardiac and all-cause 14 mortality in the elderly. The Rotterdam Study. Eur Heart J. 1999;20(4):278-284. 15 5. Straus SM, Kors JA, De Bruin ML, et al. Prolonged QTc interval and risk of sudden cardiac death in a 16 population ofFor older adults. peer J Am Coll Cardiol. review 2006;47(2):362-367. only 17 6. Robbins J, Nelson JC, Rautaharju PM, et al. The association between the length of the QT interval 18 and mortality in the Cardiovascular Health Study. Am J Med. 2003;115(9):689-694. 19 20 7. Zhang Y, Post WS, Blasco-Colmenares E, et al. Electrocardiographic QT interval and mortality: a 21 meta-analysis. Epidemiology. 2011;22(5):660-670. 22 8. Nielsen JB, Graff C, Rasmussen PV, et al. Risk prediction of cardiovascular death based on the QTc 23 interval: evaluating age and gender differences in a large primary care population. Eur Heart J. 24 2014;35(20):1335-1344. 25 9. Panoulas VF, Toms TE, Douglas KM, et al. Prolonged QTc interval predicts all-cause mortality in 26 patients with rheumatoid arthritis: an association driven by high inflammatory burden. 27 Rheumatology (Oxford). 2014;53(1):131-137. 28 10. Straus SM, Sturkenboom MC, Bleumink GS, et al. Non-cardiac QTc-prolonging drugs and the risk of 29 sudden cardiac death. Eur Heart J. 2005;26(19):2007-2012. 30 11. Ray WA, Chung CP, Murray KT, et al. Atypical antipsychotic drugs and the risk of sudden cardiac 31 death. N Engl J Med. 2009;360(3):225-235. 32 12. Ray WA, Meredith S, Thapa PB, et al. Cyclic antidepressants and the risk of sudden cardiac death. http://bmjopen.bmj.com/ 33 Clin Pharmacol Ther. 2004;75(3):234-241. 34 13. Ray WA, Murray KT, Hall K, et al. Azithromycin and the risk of cardiovascular death. N Engl J Med. 35 2012;366(20):1881-1890. 36 14. Drew BJ, Califf RM, Funk M, et al. Practice standards for electrocardiographic monitoring in hospital 37 settings: an American Heart Association scientific statement from the Councils on Cardiovascular 38 Nursing, Clinical Cardiology, and Cardiovascular Disease in the Young: endorsed by the International 39

40 Society of Computerized Electrocardiology and the American Association of Critical-Care Nurses. on September 24, 2021 by guest. Protected copyright. 41 Circulation. 2004;110(17):2721-2746. 42 15. Manini AF, Nelson LS, Skolnick AH, et al. Electrocardiographic predictors of adverse cardiovascular 43 events in suspected poisoning. J Med Toxicol. 2010;6(2):106-115. 44 16. Manini AF, Hoffman RS, Stimmel B, et al. Clinical risk factors for in-hospital adverse cardiovascular 45 events after acute drug overdose. Acad Emerg Med. 2015;22(5):499-507. 46 17. Norgaard B, Mogensen CB, Teglbjaerg LS, et al. Diagnostic packages can be assigned accurately in 47 emergency departments. A multi-centre cohort study. Dan Med J. 2016;63(6):A5240. 48 18. Schmidt M, Schmidt SA, Sandegaard JL, et al. The Danish National Patient Registry: a review of 49 content, data quality, and research potential. Clin Epidemiol. 2015;7:449-490. 50 19. Pottegard A, Schmidt SA, Wallach-Kildemoes H, et al. Data Resource Profile: The Danish National 51 Prescription Registry. Int J Epidemiol. 2016. 52 20. Astrand B, Hovstadius B, Antonov K, et al. The Swedish National Pharmacy Register. Stud Health 53 Technol Inform. 2007;129(Pt 1):345-349. 54 21. Schmidt M, Pedersen L, Sorensen HT. The Danish Civil Registration System as a tool in 55 epidemiology. Eur J Epidemiol. 2014;29(8):541-549. 56 57 58 14 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 22. Ludvigsson JF, Otterblad-Olausson P, Pettersson BU, et al. The Swedish personal identity number: 5 possibilities and pitfalls in healthcare and medical research. Eur J Epidemiol. 2009;24(11):659-667. 6 23. MarquetteTM 12SLTM ECG Analysis Program. Physician's Guide. 2056246-002 Revision B. Milwaukee, 7 WI: GE Healthcare; 13 July 2015. 8 24. Läkarhandbok för Phillips DXL-algoritm för EKG. Publikationsnummer 453564149891. Andover, MA: 9 Phillips Medical Systems; 1 June 2009. 10 25. Sagie A, Larson MG, Goldberg RJ, et al. An improved method for adjusting the QT interval for heart 11 rate (the Framingham Heart Study). Am J Cardiol. 1992;70(7):797-801. 12 26. Rautaharju PM, Surawicz B, Gettes LS, et al. AHA/ACCF/HRS recommendations for the 13 standardization and interpretation of the electrocardiogram: part IV: the ST segment, T and U 14 waves, and the QT interval: a scientific statement from the American Heart Association 15 Electrocardiography and Arrhythmias Committee, Council on Clinical Cardiology; the American 16 College of CardiologyFor Foundation; peer and thereview Heart Rhythm Society: only endorsed by the International 17 Society for Computerized Electrocardiology. Circulation. 2009;119(10):e241-250. 18 27. Sturmer T, Joshi M, Glynn RJ, et al. A review of the application of propensity score methods yielded 19 20 increasing use, advantages in specific settings, but not substantially different estimates compared 21 with conventional multivariable methods. J Clin Epidemiol. 2006;59(5):437-447. 22 28. Rassen JA, Shelat AA, Myers J, et al. One-to-many propensity score matching in cohort studies. 23 Pharmacoepidemiol Drug Saf. 2012;21 Suppl 2:69-80. 24 29. Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognostic comorbidity in 25 longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373-383. 26 30. Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9-CM 27 and ICD-10 administrative data. Med Care. 2005;43(11):1130-1139. 28 31. Austin PC. Optimal caliper widths for propensity-score matching when estimating differences in 29 means and differences in proportions in observational studies. Pharm Stat. 2011;10(2):150-161. 30 32. C. BH. An analysis of time-relations of electrocardiograms. Heart. 1920;7:353-370. 31 33. Manini AF, Nelson LS, Stimmel B, et al. Incidence of adverse cardiovascular events in adults 32 following drug overdose. Acad Emerg Med. 2012;19(7):843-849. http://bmjopen.bmj.com/ 33 34. Gunnell D, Ho D, Murray V. Medical management of deliberate drug overdose: a neglected area for 34 suicide prevention? Emerg Med J. 2004;21(1):35-38. 35 35. Seftchick MW, Adler PH, Hsieh M, et al. The prevalence and factors associated with QTc 36 prolongation among emergency department patients. Ann Emerg Med. 2009;54(6):763-768. 37 36. Vandenberk B, Vandael E, Robyns T, et al. Which QT Correction Formulae to Use for QT 38 Monitoring? Journal of the American Heart Association. 2016;5(6):e003264. 39

40 37. Al-Abri SA, Woodburn C, Olson KR, et al. Ventricular dysrhythmias associated with poisoning and on September 24, 2021 by guest. Protected copyright. 41 drug overdose: a 10-year review of statewide poison control center data from California. Am J 42 Cardiovasc Drugs. 2015;15(1):43-50. 43 38. Yap YG, Camm AJ. Drug induced QT prolongation and torsades de pointes. Heart. 44 2003;89(11):1363-1372. 45

47 48 49 50 51 52 53 54 55 56 57 58 15 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 Legends for figures 5 6 7 Figure 1: Flowchart of the study population 8 Figure 2: Kaplan Meier failure estimate 9 10 Abbreviations: QTclong = 0, patients without QTc prolongation; QTclong = 1, patients with QTc prolongation. 11 12 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 24, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 16 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 24, 2021 by guest. Protected copyright. 41 42 Figure 1: Flowchart of the study population 43 203x220mm (300 x 300 DPI) 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Figure 2: Kaplan Meier failure estimate 31 Abbreviations: QTclong = 0, patients without QTc prolongation; QTclong = 1, patients with QTc

32 prolongation. http://bmjopen.bmj.com/ 33

34 203x144mm (300 x 300 DPI) 35 36 37 38 39

40 on September 24, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 APPENDIX 5 6 Appendix A – Data sources 7 8 In the logistic system in the EDs of the Region of Southern Denmark and the Region of Skåne 9 information regarding triage and presumed diagnoses are shared among the health care personal. 10 11 The system also provides information on time of arrival to the ED as well as time of discharge. At 12 13 arrival all patients are registered by main reason for contact. Lund and Helsingborg have 43 14 somatic contact possibilities, Odense and Esbjerg have 40 contact possibilities. We used these 15 16 systems to identify Forpatients withpeer suspected review poisoning. only 17 18 From the electronic central ECG databases at the Region of Southern Denmark and the Region of 19 20 Skåne all ECGs measures were extracted. These databases contain information of all ECGs 21 recorded in any hospital in the respective region. 22 23 24 25 The Danish National Patient Register was established in 1977. Since 1994 diagnostic information 26 27 has been recorded in accordance with ICD-10. The content includes information regarding 28 discharge diagnoses from which comorbidities were derived. In Sweden, all health care 29 30 consultations are recorded in the respective region databases. From the Skåne Healthcare 31

32 Register, we retrieved information corresponding to information from the Danish National Patient http://bmjopen.bmj.com/ 33 Register. In these regional databases diagnoses from both the primary and secondary health care 34 35 system are available. Charlson Comorbidity Index was calculated based on data extracted for a 10 36 37 year-period ending on the day of contact in the Danish data. From the Swedish database, Charlson 38 39 Comorbidity Index was calculated for a 2-year period ending on the day of contact to the ED. 40 on September 24, 2021 by guest. Protected copyright. 41 42 The Swedish National Pharmacy Register contains data on all prescriptions dispensed at 43 44 pharmacies since 2005. Besides date of dispensing, name, amount, and dose of the redeemed 45 medication are accessible from this register. The Danish National Prescription Registry provides 46 47 similar information of individual-level prescription of medication since 1995. Information of 48 49 redeemed prescription of QT prolonging drugs (see Appendix c.5) within 90 days before contact 50 51 was obtained through these registers. 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 The Danish Civil Registration System contains data regarding birth, emigration, and vital status for 5 6 the entire populations. Information regarding vital status in Sweden was retrieved from the 7 Swedish population registry. The Danish Civil Registration System was established in 1968, and by 8 9 law all Danish residents have a unique civil registration number of ten digits. Since 1967, all 10 11 Swedish residences are assigned a ten-digit personal identity number. These unique numbers 12 allow complete and accurate linkage between registries on an individual level. 13 14 15 Appendix B – Supplementary methods 16 For peer review only 17 ECG measurements 18 19 All ECGs were recorded and stored either in MUSE® Cardiology Information System (GE 20 21 Healthcare), or by Philips Diagnostic ECG (Philips). All ECGs in our analysis were 12 lead ECGs. 22 ECGs recorded in MUSE were later processed using the Marquette 12SL algorithm, which calculate 23 24 the QTc interval using the Bazett Formula, the Fridericia Formula, and the Framingham Formula. In 25 26 MUSE, the QT interval is measured as a median value from the 12 leads. ECG recorded by Philips 27 were analyzed by the DXL-algorithm. In this algorithm, the QT interval is measured as a median 28 29 value from reliable leads. A lead is defined as reliable if little variation in beat-to-beat variation. 30 31 Philips provided QTc intervals corrected by the Bazett Formula and the Fridericia Formula.

32 http://bmjopen.bmj.com/ 33 To correct for heart rate, we chose the Framingham Formula (QTcFramingham =QT + 0.154 (1-RR)). 34 Because Philips DXL-algorithm does not routinely correct the QT interval using the Framingham 35 36 Formula, we calculated the QTcFramingham ourselves. As we did not include RR intervals in our 37 38 analysis, the formula was used in another edition than the original formula. Therefore, we used 39

40 following formula for correction in ECGs recorded by Philips: QTcFramingham = QT + 154 (1-60/heart on September 24, 2021 by guest. Protected copyright. 41 rate). The QRS duration was measured as a median value in both algorithms. 42 43 44 45 The Marquette 12SL algorithm defines onsets as the earliest deflection in any lead, and offsets as 46 the latest deflection in any lead. The QT interval is measured from the earliest detection of 47 48 depolarization in any lead to the latest detection of repolarization in any lead. Similarly, the QRS 49 50 duration was measured from the earliest onset in any lead to the latest deflection in any lead. 51 52 The Philips DXL-algorithm first identifies waveform component and measures every beat in each 53 lead individually. After the approximate waveform locations are known, onsets and offsets are 54 55 defined. Once the onsets and offsets are known, duration of intervals are calculated. 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 5 6 In both Denmark and Sweden ECG recording is performed by well-educated heath care personals 7 instructed only to accept ECGs of satisfying quality. Otherwise it is considered a routine to record 8 9 another ECG. If multiple ECGs were recorded in a single individual, the first ECG recorded within 4 10 11 hours after arrival was included in the analysis. 12

13 14 15

16 For peer review only 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

14 SKS codes* Administrative codes 15 AVAA07: Sudden cardiac arrest 16 For AVAA06:peer Sudden review cardiac arrest only 17 18 Procedure codes 19 ZZ0401: Standby for cardiac arrest 20 21 Treatment codes 22 BFFA6: Chest compressions 23 BFFA60: External chest compressions 24 BFFA60A: External chest compressions by use of mechanical chest compressions 25 26 KVÅ-codes** DF012: Chest compressions 27 DF017: Mechanical chest compressions 28 DF028: Cardiopulmonary resuscitation 29 30 *SKS = Sundhedsvæsenets klassifikations system, available on http://medinfo.dk/sks/brows.php, Danish codes. 31 **KVÅ= Klassifikation av vårdåtgärder, available on

32 http://www.socialstyrelsen.se/klassificeringochkoder/atgardskoderkva, Swedish codes. http://bmjopen.bmj.com/ 33 34 C.2 35 36 Poisoning divided into groups by use of ICD-10 codes 37 38 Analgesics and drugs of abuse T39*-T40*, F110*, F120*, F140*, F150*, F160* 39 Psychotropic drugs and drugs affecting the T42*-T44*, F130*, F190* 40 central nervous system on September 24, 2021 by guest. Protected copyright. 41 Organic and chemical substances, non- T51*-T65*, F100*, F170*, F180* 42 medical substances 43 Others T36*-T38*, T41*, T45*-T50* 44 45 Multidrug ≥2 of the above mentioned poisoning groups 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 C.3 5 Charlson Comorbidity Index* 6 7 Condition Assigned weight ICD-10 codes 8 Peripheral vascular disease 1 I70.x, I71x, I73.1, I73.8, I73.9, I77.1, I79.0, I79.2, K55.1, K55.8, K55.9, Z95.8, Z95.9 9 Cerebrovascular disease 1 G45.x, G46.x, H34.0, I60.x-I69.x 10 Dementia 1 F00.x-F03.x, F05.1, G30.x, G31.1 11 Chronic pulmonary disease 1 I27.8, I27.9, J40.x-J47.x, J60.x-J67.x, J68.4, J70.1, J70.3 12 Rheumatic disease 1 M05.x, M06.x, M31.5, M32.x-M34.x, M35.1, M35.3, M36.0 13 Peptic ulcer disease 1 K25.x-K28.x 14 Mild liver disease 1 B18.x, K70.0-K70.3, K70.9, K71.3-K71.5, K71.7, K73.x, K74.x, 15 K76.0, K76.2-K76.4, K76.8, K76.9, Z94.4 16 Diabetes without chronicFor peer1 review E10.0, E10.1, E10.6, only E10.8, E10.9, E11.0, E11.1, E11.6, 17 complications E11.8, E11.9, E12.0, E12.1, E12.6, E12.8, E12.9, E13.0, 18 E13.1, E13.6, E13.8, E13.9, E14.0, E14.1, E14.6, E14.8, E14.9 19 Hemiplegia or paraplegia 2 G04.1, G11.4, G80.1, G80.2, G81.x, G82.x, G83.0-G83.4, 20 G83.9 21 Renal disease 2 I12.0, I13.1, N03.2-N03.7, N05.2- N05.7, N18.x, N19.x, 22 N25.0, Z49.0- Z49.2, Z94.0, Z99.2 23 Diabetes with chronic 2 E10.2-E10.5, E10.7, E11.2-E11.5, E11.7, E12.2-E12.5, E12.7, 24 complications E13.2- E13.5, E13.7, E14.2-E14.5, E14.7 25 Cancer 2 C00.x-C26.x, C30.x-C34.x, C37.x- C41.x, C43.x, C45.x-C58.x, 26 C60.x- C76.x, C81.x-C85.x, C88.x, C90.x-C97.x 27 Metastatic cancer 3 C77.x-C80.x 28 Moderate or severe liver disease 3 I85.0, I85.9, I86.4, I98.2, K70.4, K71.1, K72.1, K72.9, K76.5, 29 K76.6, K76.7 30 AIDS/HIV** 6 B20.x-B22.x, B24.x 31 *Diagnostic codes for myocardial infarction and heart failure are not included in the index, but are included in this 32 study as covariates. http://bmjopen.bmj.com/ 33 ** AIDS = acquired immunodeficiency syndrome, HIV = human immunodeficiency virus 34 35 36 C.4 37 Variables from Charlson Comorbidity Index, included separately in the analysis 38 Myocardial infarction I21.x, I22.x, I25.2 39 Congestive heart failure I09.9, I11.0, I13.0, I13.2, I25.5, I42.0, I42.5-I42.9, I43.x,

40 I50.x, P29.0 on September 24, 2021 by guest. Protected copyright. 41 42 43 C.5 44 * 45 List of drugs associated with QT prolongation and risk of TdP 46 Drug category ATC-codes** 47 Alimentary tract and metabolism Domperidone (A03FA03), Granisetron (A04AA02), Metoclopramide (A03FA01), 48 Ondansetron (A04AA01), Pantoprazole (A02BC02) 49 Cardiovascular system Amiodarone (C01BD01), Dronedarone (C01BD07), Flecainide (C01BC04), Furosemide 50 (C03CA01, C03EB01), Hydrochlorothiazide (C03EA01, C09DA01, C09DA06, C09DA04, 51 C09BA02), Indapamide (C03BA11), Isradipine (C08CA03), Ivabradine (C01EB17), Sotalol 52 (C07AA07) 53 Genito urinary system and sex Alfuzosin (G04CA01), Mifepristone (G03XB01), Mirabegron (G04BD12), Solifenacin 54 hormones (G04BD08, G04CA53), Tolterodine (G04BD07), Vardenafil (G04BE09) 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 24 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 Systemic hormonal preparations, 4 Oxytocin (H01BB02), Pasireotide (H01CB05) 5 excl. sex hormones and insulins Anti-infectives for systemic use 6 Atazanavir (J05AR15), Azithromycin (J01FA10), Bedaquiline (J04AK05), Ciprofloxacin 7 (J01MA02), Clarithromycin (J01FA09), Erythromycin (J01FA01), Fluconazole (J02AC01), 8 Foscarnet (J05AD), Itraconazole (J02AC02), Ketoconazole (J02AB02), Metronidazole (J01XD01), Moxifloxacin (J01MA14), Posaconazole (J02AC04), Rilpivirine (J05AG05), 9 Ritonavir (J05AE03), Roxithromycin (J01FA06), Saquinavir (J05AE01), Voriconazole 10 (J02AC03) 11 Antineoplastic and Anagrelide (L01XX35), Bortezomib (L01XX32), Bosutinib (L01XE14), Ceritinib (L01XE28), 12 immunomodulatory agents 13 Crizotinib (L01XE16), Dabrafenib (L01XE23), Dasatinib (L01XE06), Degarelix (L02BX02), 14 Eribulin mesylate (L01XX41), Fingolimod (L04AA27), Lapatinib (L01XE07), Leuprolide (L02AE02), Nilotinib (L01XE08), Oxaliplatin (L01XA03), Panobinostat (L01XX42), 15 Pazopanib (L01XE11), Sorafenib (L01XE05), Sunitinib (L01XE04), Tacrolimus (L04AD02), 16 For peerTamoxifen (L02BA01), review Vandetanib (L01XE12), only Vemurafenib (L01XE15) 17 Musculo-skeletal system 18 Tizanidine (M03BX02) 19 Nervous system 20 Amantadine (N04BB01), Amisulpride (N05AL05), Amitriptyline (N06AA09), 21 Apomorphine (N04BC07), Aripiprazole (N05AX12), Asenapine (N05AH05), Atomoxetine (N06BA09), Citalopram (N06AB04), Clomipramine (N06AA04), Clozapine (N05AH02), 22 Dexmedetomidin (N05CM18), Doxepin (N06AA12), Droperidol (N05AD08), Escitalopram 23 (N06AB10), Fluoxetine (N06AB03), Galantamine (N06DA04), Haloperidol (N05AD01), 24 Hydroxyzine (N05BB01), Imipramine (N06AA02), Levomepromazine (N05AA02), Lithium 25 (N05AN01), Methadone (N07BC02), Mirtazapine (N06AX11), Nortriptyline (N06AA10), 26 Olanzapine (N05AH03), Paliperidone (N05AX13), Paroxetine (N06AB05), Pimozide 27 (N05AG02), Pipamperone (N05AD05), Propofol (N01AX10), Quetiapine (N05H04), Risperidone (N05AX08), Sertindole (N05AE03), Sertraline (N06AB06), Sevoflurane 28 (N01AB08), Sulpiride (N05AL01), Tetrabenazine (N07XX06), Venlafaxine (N06AX16), 29 Ziprasidone (N05AE04) 30 Antiparasitic products, insecticides 31 Chloroquine (P01BC02), Hydroxychloroquine (P01BA02), Metronidazole (P01AB01),

and repellents http://bmjopen.bmj.com/ 32 Pentamidine (P01CX01), Quinine sulfate (P01BC01) 33 Respiratory system 34 Diphenhydramine (R06AA02), Promethazine (R06AD02) 35 Various Perflutren lipid microspheres (V08DA01) 36 37 *From QTDrug list, https://crediblemeds.org/, version December 17, 2015. Only drugs available in Denmark are 38 included at our list. The list includes drugs with known risk of TdP, possible risk of TdP, and conditional risk of TdP. 39 **ATC = Anatomical Therapeutic Chemical

18 Male n<5 n<5 2.7 (0.5-16.3) 19 4.4 (0.8-24.3) 20 Female n<5 n<5 21 Age ≥50 years 6 (131) n<5 2.7 (0.7-9.9) 22 Odense or Esbjerg n<5 n<5 5.8 (0.6-57.4) 23 Lund n<5 n<5 1.7 (0.2-12.2) 24 4.4 (0.4-49.2) 25 Helsingborg n<5 n<5 26 Confirmed poisoning 27 Total 6 (151) n<5 10.5 (1.2-90.0) 28 Psychotropic drugs n<5 n<5 2.0 (0.1-32.5) 29 Abbreviations: HR; hazard ratio, CI; confidence interval. 30 *Hazard ratio from Cox regression. 31 **If the number of events in the analysis was less than 5 (marked by n<5), the number of patients in the strata is not shown. 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 24, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 26 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 STROBE Statement—checklist of items that should be included in reports of observational studies 3 4 Item 5 No Recommendation 6 Title and abstract 1 (a) Indicate the study’s design with a commonly used term in the title or the abstract 7 8 Page 1 and 2 9 (b) Provide in the abstract an informative and balanced summary of what was done 10 and what was found 11 Page 2 12 13 Introduction 14 Background/rationale 2 Explain the scientific background and rationale for the investigation being reported 15 Page 3 16 Objectives For3 peer State specific objectives, review including any prespecifiedonly hypotheses 17 Page 3 18 19 Methods 20 Study design 4 Present key elements of study design early in the paper 21 Page 3 22 Setting 5 Describe the setting, locations, and relevant dates, including periods of recruitment, 23 24 exposure, follow-up, and data collection 25 Page 3-5 26 Participants 6 (a) Cohort study—Give the eligibility criteria, and the sources and methods of 27 selection of participants. Describe methods of follow-up 28 Page 4 and 5 29 Case-control study—Give the eligibility criteria, and the sources and methods of 30 31 case ascertainment and control selection. Give the rationale for the choice of cases

32 and controls http://bmjopen.bmj.com/ 33 Cross-sectional study—Give the eligibility criteria, and the sources and methods of 34 selection of participants 35 (b) Cohort study—For matched studies, give matching criteria and number of 36 37 exposed and unexposed 38 Page 5 (As we did a propensity score analysis the number of exposed and 39 unexposed are outlined in the results because it cannot be provided before the 40 analysis) on September 24, 2021 by guest. Protected copyright. 41 Case-control study—For matched studies, give matching criteria and the number of 42 controls per case 43 44 Variables 7 Clearly define all outcomes, exposures, predictors, potential confounders, and effect 45 modifiers. Give diagnostic criteria, if applicable 46 Page 4, 5, and Appendix B and C 47 Data sources/ 8* For each variable of interest, give sources of data and details of methods of 48 measurement assessment (measurement). Describe comparability of assessment methods if there 49 50 is more than one group 51 Page 4 and Appendix A, B, and C 52 Bias 9 Describe any efforts to address potential sources of bias 53 Page 6 54 Study size 10 Explain how the study size was arrived at 55 56 Page 4 and figure 1 57 Quantitative variables 11 Explain how quantitative variables were handled in the analyses. If applicable, 58 59 1 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 27 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 describe which groupings were chosen and why 3 Page 5 4 Statistical methods 12 (a) Describe all statistical methods, including those used to control for confounding 5 Page 5 and 6 6 7 (b) Describe any methods used to examine subgroups and interactions 8 Page 5 and 6 9 (c) Explain how missing data were addressed 10 No missing data. 11 (d) Cohort study—If applicable, explain how loss to follow-up was addressed 12 No loss to follow-up. 13 14 Case-control study—If applicable, explain how matching of cases and controls was 15 addressed 16 For peerCross-sectional studyreview—If applicable, describe only analytical methods taking account of 17 sampling strategy 18 (e) Describe any sensitivity analyses 19 20 Page 6 21 Continued on next page 22 23 24 25 26 27 28 29 30 31

32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 24, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 2 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 28 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 Results 4 Participants 13* (a) Report numbers of individuals at each stage of study—eg numbers potentially eligible, 5 examined for eligibility, confirmed eligible, included in the study, completing follow-up, and 6 7 analysed 8 Page 6 9 (b) Give reasons for non-participation at each stage 10 Page 6 11 (c) Consider use of a flow diagram 12 13 Figure 1 14 Descriptive 14* (a) Give characteristics of study participants (eg demographic, clinical, social) and information 15 data on exposures and potential confounders 16 ForPage 6 and peer 7, including table review 1 only 17 (b) Indicate number of participants with missing data for each variable of interest 18 19 None missing 20 (c) Cohort study—Summarise follow-up time (eg, average and total amount) 21 Page 8 22 Outcome data 15* Cohort study—Report numbers of outcome events or summary measures over time 23 Page 8 and table 3 24 25 Case-control study—Report numbers in each exposure category, or summary measures of 26 exposure 27 Cross-sectional study—Report numbers of outcome events or summary measures 28 Main results 16 (a) Give unadjusted estimates and, if applicable, confounder-adjusted estimates and their 29 precision (eg, 95% confidence interval). Make clear which confounders were adjusted for and 30 31 why they were included

32 Page 7 and 8 including table 3 http://bmjopen.bmj.com/ 33 (b) Report category boundaries when continuous variables were categorized 34 Page 5 35 (c) If relevant, consider translating estimates of relative risk into absolute risk for a meaningful 36 37 time period 38 Page 8 39 Other analyses 17 Report other analyses done—eg analyses of subgroups and interactions, and sensitivity 40 analyses on September 24, 2021 by guest. Protected copyright. 41 The prevalence of QTc prolongation in relation to specific subgroups: page 7 and table 2. 42 Appendix D includes a stratified analysis. 43 44 Sensitivity analysis: page 8 and 9. 45 Discussion 46 Key results 18 Summarise key results with reference to study objectives 47 Page 9 48 49 Limitations 19 Discuss limitations of the study, taking into account sources of potential bias or imprecision. 50 Discuss both direction and magnitude of any potential bias 51 Page 9 and 10. 52 Interpretation 20 Give a cautious overall interpretation of results considering objectives, limitations, multiplicity 53 of analyses, results from similar studies, and other relevant evidence 54 55 Page 10 and 11. 56 Generalisability 21 Discuss the generalisability (external validity) of the study results 57 Page 9 and 10. 58 59 3 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 29 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 Other information 3 Funding 22 Give the source of funding and the role of the funders for the present study and, if applicable, 4 for the original study on which the present article is based 5 Page 12 6 7 8 *Give information separately for cases and controls in case-control studies and, if applicable, for exposed and 9 unexposed groups in cohort and cross-sectional studies. 10 11 Note: An Explanation and Elaboration article discusses each checklist item and gives methodological background and 12 13 published examples of transparent reporting. The STROBE checklist is best used in conjunction with this article (freely 14 available on the Web sites of PLoS Medicine at http://www.plosmedicine.org/, Annals of Internal Medicine at 15 http://www.annals.org/, and Epidemiology at http://www.epidem.com/). Information on the STROBE Initiative is 16 available at www.strobe-statement.org.For peer review only 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 24, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 4 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from

The association between QTc prolongation and mortality in patients with suspected poisoning in the emergency department – a transnational propensity score matched cohort study

ForJournal: peerBMJ Open review only

Manuscript ID bmjopen-2017-020036.R1

Article Type: Research

Date Submitted by the Author: 17-Mar-2018

Complete List of Authors: Schade Hansen, Camilla; Odense Universitetshospital, Department of Emergency Medicine Pottegard, Anton; University of Southern Denmark, Clinical Pharmacology, Institute of Public Health Ekelund, U; Skåne University Hospital, Department of Emergency Medicine Kildegaard Jensen, Helene; Odense Universitetshospital, Department of Emergency Medicine Lundager Forberg, Jakob; Helsingborg Hospital, Department of Emergency Medicine Brabrand, Mikkel; Odense University Hospital, Department of Emergency Medicine http://bmjopen.bmj.com/ Lassen, Annmarie; Odense University Hospital, Department of Emergency Medicine

Primary Subject Emergency medicine Heading:

Secondary Subject Heading: Epidemiology, Cardiovascular medicine

Keywords: EPIDEMIOLOGY, CARDIOLOGY, TOXICOLOGY on September 24, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 Ç vÇ 5 6 7 8 9 10 11 12 / { I a5 ! t a{ t t 5 Ü 9 a5 t 5! I Y W 13 a5 W $ [ C$ a5 t 5' a .$ a5 t 5) ! Ç $ [ a5 5a{ 14 15 16 5 + For 9 , peer a h review Ü. , I+ h only 5 17 / t , t , 5 + t $ I Ü. , { 5 h 18 5 19 5 + 9 , a { Ü. , I+ [ {/ 20 5 + 9 , a I $ I+ I $ {/ 21 5 + 9 , a I+ { í W 9$1 5 22 23 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 24, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 í 23 44 48 49 / + 3 50 5 5 6, 5 51 Ç + 3 7 ') 4))8) 52 { 5 . . 9 53 )888 h / 54 5 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 !.{Çw!/Ç 5 6 7 h.W9/ÇLë9{( t : , + ;95< 8 9 , .. / + vÇ .5 Ç , / $ 10 + . vÇ + 95 + / + + 11 12 $ . , / + vÇ .5 13 14 15 a9ÇIh5{( í + >$ , 16 For peer review only 17 / + + 95 / {/ + ;W , 88 5 $ 8'< 18 19 / 5 + ;a 8! !+ 8'<5 í ++ , 20 21 ? ;Iw< > , ; $ +< / !8 22 , + + / + vÇ . ;A')8 A'B8 23 24 / < / + / vÇ . C''8 5 25 26 27 w9{Ü[Ç{( ! > / + + 9/D / 28 29 . ; E !4B9< vÇ + B5)F5 Ç . , 30 31 !8> , /> + + / 854F ;9)F /L 85B>5< / $ ,

32 http://bmjopen.bmj.com/ 33 + / vÇ + !5F ;9)F /L 5'>B5<5 ! + vÇ . 34 . / / Iw !5B ;9)F /L 58>5<5 35 36 37 38 /hb/[Ü{Lhb( L 95 + vÇ . + . / + + 39

40 $ / > !8> , > , on September 24, 2021 by guest. Protected copyright. 41 5 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 { / 5 6 t / + H / {/ / 5 5 7 t+ , / 1 . 5 8 9 { $ + , / +$ $ . 5 10 11 Ç 9/D / vÇ . / 12 ,5 13 14 15 16 LbÇwh5Ü/ÇLhb For peer review only 17 18 t : , + ;95< .. 19 . , $ 5 ! / . $ vÇ 20 21 +! / $ / > >, . 22 '>9 23 5 ! , $ + 24 8>! 25 / + vÇ + , ,+ ? . 26 + + 5 Ç + + $, 27 28 + , + , + 5' 29 30 h, / . . + $ / vÇ + . 31 ++ + + 5)B Ç $ . 32 http://bmjopen.bmj.com/ 33 , vÇ + + + 34 35 /5 Ç / 3 ;< $ + . vÇ + 36 37 + / + + , + I ;< . vÇ 38 + / , / !8 , 39

40 . , + 5 on September 24, 2021 by guest. Protected copyright. 41 42 43 44 a!Ç9wL![{ !b5 a9ÇIh5{ 45 { 46 47 Ç >$ ,5 Ç , $ 95 W , 88 48 49 5 $ ! 8' / {/ + ;{ Ü. , I+ [ I $ 50 51 I+< / 5 + ;h Ü. , I+ I+ { 52 í W 9$1 < a 8! !+ !8 8'5 L $ 5 {/ 53 54 , 2> . 5 Ç Ü. , 55 56 57 58 ! 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 I+ { , ++ ++2 , !8888 / h 5 6 Ü. , I+ . ++ 98888 + + 5 Ç / + . 7 , ++ )8888 + + ;I $< 8888 + + ;9$1 <5 8 9 10 11 { / 12 í ;A4 , < / . 95 / + +5 Ç 13 14 / $ , , > 9/D / ' .5 15 16 ! vÇ .For peer 9/D review vw{ only A8 / $ 17 18 2 5 t / + / , / + 19 + / , + 5 L + / + 20 21 + !++ 2 !5 22 23 24 25 5 26 L $ 5 {/ . : + . $ 27 28 / / > + . $ / $ 5 í 2 . 29 J 30 3 Ç , 95 w { 5 w 31 { 9/D $ w { 5 w { 32 http://bmjopen.bmj.com/ 33 5 b t w ,4 w { I $ 5 34 9 8 35 b t + w , {/ t , w , , Ç 5 /. 36 37 w {, {/ t+ w 5 C 38 +. !++ 2 !5 39

40 on September 24, 2021 by guest. Protected copyright. 41 9/D / 42 43 Ç vÇ . / 9/D 955 ! vÇ 44 45 . / , . aÜ{9 / , 46 L {, ;D9 I < t + 5 9/D5 Ç D9 a: {[ 9/D 47 48 !, t +. vÇ . 9/D aÜ{95! 9/D $, t + 49 ' 50 / ,? $, 5ó[> 5 h, vÇ . ;vÇ < / 51 52 ,5 C / C C ;vÇ C EvÇ 7 85)' ;> 53 ww<<5) ! $ 9/D !++ 2 .5 54 55 56 57 58 ' 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 91 5 6 h +, / $ + > , ; 7 !++ 2 /5< / !8 , , . 955 ! + / / !8 8 9 , + 5 t / 10 11 / 5 Ç +, 2+ / vÇ 12 + vÇ A')8 A'B8 / 5B t / 13 14 vÇ / . vÇ . C')8 ; < C'B8 ;/ <5 15 16 For peer review only 17 18 ! 19 Ç + . vÇ + . + + + / 20 21 $ > +5 L + / + / 22 23 + ;L / 5 ;L/5>8< Ç!B> 24 25 ÇB) C88 C8 C8 C!8 C'8 C)8 CB8 CJ8 C48 C98 +, 26 , <5 ! + / + / $ . 27 28 . + +3 5 ! $ 5 t, + 29 30 . , !5 h $ > '5 h 31 )5 a ; !++ 2 /5<5 32 http://bmjopen.bmj.com/ 33 Ç $ / vÇ + > , / 34 J4 35 . ++ , 5 í ++ , 36 37 + $, / vÇ A')8 ; < A'B8 ;/ < 38 ;$, <5 t / vÇ . $ / ''8>''9 ; < ''8>')9 39

40 ;/ < / 2 . >. ++ 5 Ç / +$ on September 24, 2021 by guest. Protected copyright. 41 42 / ++ , 3 2 $ , ; 43 / /$ , L 29!8< , , . 44 45 ;!++ 2 /!5 /5'< + + vÇ + / 98 , ; !++ 2 46 47 /5) , 50 , / /2 5 51 52 53 54 { 55 56 57 58 ) 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 Ç $ . + / + + / . 5 6 / vÇ + / vÇ +5 L ++ , 7 / vÇ + / ? ;Iw<5 í 8 9 9)F . $ $ $ 5 Ç + 10 11 vÇ + !8> , > , / Y+ a 12 . 5 13 14 L ., , / . / / $ + 15 16 $ + .For peer . review only +5 Ç + . vÇ 17 18 + ++ , , / + .? vÇ 19 5!! 20 21 { , / + {Ç!Ç! . ' ;{/+ [t / { Ç 2<5 22 23 Ç , / ++. $, 5 5 t ! , ;b5 884>)4>88!) W 5 24 25 )MB!< Ç 5 I ! , ;b5 !>!8!>8!<5 L , / {/ / 26 , / ++. $, w 9 / [ $, w { 5 27 28 29 30 t 33 31 Ç / , / + 5 ! / $ 5 32 http://bmjopen.bmj.com/ 33 34 35 w9{Ü[Ç{ 36 37 / / 38 39 ! + / B4!4 95 / + +5 ! 40 2 C4 , ;E< 9/D +$ > . on September 24, 2021 by guest. Protected copyright. 41 42 ;E'< + / , + ;E'< vÇ . ;E< vw{ 43 44 A8 ;E!< + !4B9 + / + + 45 ;'458F !4< ;C <5 h B95F ;EBJB< 46 47 +5 48 49 50 51 t / + vÇ . / $ , , 52 , / vÇ + ;Ç$ <5 53 54 Ç ( . / 55 56 57 58 B 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 !4 ./ !/ 5 6 vÇ t vÇ vÇ t vÇ C''8 ; A')8 ; < C''8 ; A')8 ; < 7 / < A'B8 ;/ < / < A'B8 ;/ < 8 9 b 567 76 76 5 {1 10 4)9 ;'458< B!' ;'95B< ;'J54< 9 ;'B5< 9 ;'458< a ;F< 11 ! ; Lvw< !4 ;)>)!< !B ;'>)< ) ;!B>B4< )! ;!J>B9< ) ;!)>BB< 12 4>)8 H ;F< J'J ;J58< ''' ;J'5< 9 ;'J58< !B ;'458< 9 ;'458< 13 )>B9 H ;F< J44 ;85'< B ;45)< JJ ;!85'< '8 ;45)< JJ ;!58< AJ8 H ;F< !!' ;45B< ' ;J5!< )J ;5)< B ;!5B< ) ;58< 14 / / L 1 H ;F< 15 //L E 8 J'J ;J58< !9) ;J5J< '8 ;))5!< B! ;)!58< '8 ;)B5)< 16 //L E For peerJ4 ;45B< )4Jreview ;J54< B8 ;!5J< only !! ;B54< )4 ;!5'< 17 //L A '8' ;85'< !' ;95)< )! ;859< 88 ;85< )8 ;85< a / 3 4) ;'54< !B ;'5< ! ;5B< )) ;5< 9 ;5J< 18 / H ;F< 19 vÇ8 H ;F< )4 ;!95< '4 ;!J59< 8 ;'!5)< ! ;'59< 89 ;''58< 20 9/D 21 I ; Lvw< 4) ;J!>99< 4J ;J'>8< JB ;B)>4'< JB ;B)>4J< JB ;B)>4)< 22 vÇ A)88 > ;F< J ;85J< > J ;85J< > J ;859< ! / H ;F< BJB ;B95< 4 ;B95< )! ;B85)< !8 ;B5)< ) ;B859< 23 D / H ;F< 24 5! $ !9J ;'54< !!! ;'5B< ;!5J< ' ;!5< ;!59< 25 5t, + 48) ;!85< B9) ;!85)< '9 ;!58< 8! ;!!5< '9 ;!5)< 26 . , !5h $ )8 ;454< '!J ;95< ' ;)5J< )8 ;B5< ' ;)59< 27 > 28 '5h 'J8 ;J5B< !9 ;J5< !8 ;95B< )' ;J5'< 9 ;95< 29 )5a )8 ;454< ') ;45B< 9 ;958< B ;858< 4 ;45)< / H ;F< 30 Ç Ü. , I+ { [ J9' ;'B5'< )!9 ;'B5J< ) ;'95'< 'J ;'954< ' ;)858< 31 h Ü. , I+ )8 ;59< '9 ;5J< 4 ;5< )8 ;85< 4 ;5!<

32 I $ I+ !J ;!)5)< JB ;!)5J< 4 ;!58< J8 ;!'5!< J9 ;!59< http://bmjopen.bmj.com/ I+ { í W 8 ;)5< B ;'59< 9 ;J5)< 9 ;)54< J ;B59< 33 !$$ .3 //L / /$ , L 2I Lvw : 5 34 L + / >. ++ vÇ . ;''8>''9 ''8>')9 / < ; E!8<5 35 36 37 L + + vÇ + / : + / + 38 39 vÇ .5 ! + / + + vÇ + J5)F 40 + vÇ . / 454F / vÇ + on September 24, 2021 by guest. Protected copyright. 41 42 + vÇ .5 ! + B5)F ;9)F /L )59>J5'< vÇ 43 44 + / + . . vÇ + ;A)88 < / 85JF ;9)F /L 85)> 45 58<5 Ç + . vÇ + + + + . / 46 47 '54>B5F / + . + ? N O ;B5FI 9)F 48 49 /L '54>45J< ;Ç$ <5 50 51 Ç ( vÇ 52 53 54 55 56 57 58 J 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3

4 t, + / ! 5 $ $ $ > h a 6 7 . ,

8 9 Ç!9>Ç'8 C8 A Ç'>Ç'' C!8 Ç)>ÇB) C88 Ç!B>Ç!4 Ç' 10 L/5>8 C8 C'8 C)8 $ C98 CJ8 C48 Ç')>Ç)8 11 CB8 + + 12 13 b !9J 48) )8 'J8 )8 14 vÇ H ;F /L 9)F< 15 A')8 ; < ;)5!FI !5!>458< '9 ;B5FI '5)>458< ' ;'54FI !5>J58< 9 ;B5FI '5>45J< 4 ;)5BFI !5J>458< A'B8 ;/ < 16 For peer review only 17 !$$ .3 /L .5 18 19 20 t 21 22 h. !8> , > , / 854F ;9)F /L 85B>5 E!<5 23 ! . / vÇ + ;E )!< / !8 , 95 24 25 J + / + 5 ! / 26 27 vÇ . ;E!BB< / ' . /> + + 5 Ç $ 28 29 . / !8 , / !5F ;9)F /L 5'>B5< 85JF ;9)F /L 85)>58< + / 30 / vÇ + + . ,5 31

32 Ç ++ , , '4 + / vÇ A')8 ; < A'B8 http://bmjopen.bmj.com/ 33 34 ;/ < / '9B + / vÇ . C''8 5 ! +$ $ $ 35 .$ / . ;Ç$ <5 vÇ + / / Iw !5B ;9)F /L 58> 36 37 5< !8> , > , ;Ç$ ! C <5 38 39 Ç ( w 9

40 t on September 24, 2021 by guest. Protected copyright. 41 9. ;b5< Iw ;9)F /L< 42 { 43 b vÇ . 58 ; < '9B C) 44 C''8

46 vÇ + !5B ;58>5< '4 4 A')8 47 A'B8 / 48

49 5 /

50 b vÇ . 58 ; < !8 C) C''8 51

52 vÇ + 85) ;5>9858< 53 ) B A')8

54 A'B8 / 55 56 57 58 4 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 !$$ .3 /L For .I Iw ? peer 5 L $ review . , only / ) ; $, C)< $ 17 + /5 18 t / . / + + +5 /2 3 ++ , + + / vÇ + + / vÇ +5 L 19 ++ + / >. ++ vÇ . / 2 ;vÇ ''8>''9 ''8>')9 / <5 20 21 22 { 3 23 24 h $ + , !++ 2 55 í / 25 26 . + / . !8> , 85JF ;9)F /L 27 85'>5< vÇ + , . Iw 85) ;9)F /L 5>9858<5 í / 28 29 vÇ . / .? + vÇ + ;45JF< 30 31 / / / Iw 58 ;9)F /L 85>)5)<5

32 http://bmjopen.bmj.com/ 33 34 5L{/Ü{{Lhb 35 36 L + / + + . 95 vÇ 37 38 + / ;B5)F<5 ! + vÇ . / / > 39

40 !8> , > , $ !5F5 on September 24, 2021 by guest. Protected copyright. 41 42 43 Ç , . 5 C / , / / 44 45 {/ / 5 95 / $ + $,5 Ü + 46 47 $ + , {/ 5 +. 48 +$, / . + + /> + 49 50 5 J> L / + . 51 52 ++ , + / . > 5 í 53 + / / + $ + . 955 Ç + H 54 55 + $, H + 56 57 58 9 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 955 ! + . / $ . 5 6 + ,5 7 8 9 Ç , . 5 C / $ . 5 Ç 9/D 10 11 / / , . vÇ 12 .5 I/ . $ . + . 5 , 13 14 : / / . $ / vÇ . 15 16 vÇ .For / peer review 5! 54 C only / 2 9/D 17 18 / + vÇ 55 $5 í . 19 + . 9/D / / + + . 9/D / vÇ 20 21 + $ . / + +5 Ç $ / 22 23 / / / 9/D + 24 25 5 Ç + / $, $ + $, $ / 26 2 + L/5>8 5 L +$ 27 28 , $ / /5 29 30 Ç $ . / / / $ + 31 ,5 ! / $ + . , . 32 http://bmjopen.bmj.com/ 33 / + .. . 5 Ç $ 34 35 . $ . $ , $ 36 37 {/ 5 , 5 í $ . , . 38 / $ 5 L / . . . 39

40 / .. + + / vÇ + / $ on September 24, 2021 by guest. Protected copyright. 41 42 !5F !5BF / . , ++ / 43 854F 859F5 44 45 46 47 Ç . > ;854F< / + . + + 48 B!'!) 49 ;85)>5F<5 L + . vÇ + $, / ;B5)F< 50 + . , 95 + ;!)F<5!B Ç +$$, vÇ 51 52 5 L .? $ , + . vÇ 53 54 + , ++ / . $ 4JF5 L $ 55 56 57 58 8 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 / , .? . 48>98 $ + 5 !!!J 6 $. + . vÇ +5 ! + 7 + . , . +$$, 2+ $ / 8 9 vÇ + , , 95 + 5 Ç 10 11 C , + + / / , 12 .? 5!J 13 14 15 16 Ç + For peer review > only , 17 18 / !8 , + vÇ +5 í $ 85JF + / 19 + + / vÇ + / + / + vÇ . 20 21 . $ !5F / Iw !5B ;9)F /L 58>5<5 L 22 23 ++ >, + + . 5!) ;9)F /L 5'>5'B< > 24 J 25 , + / vÇ +5 ! , + / 26 9/D + , + vÇ + $ / Iw 27 28 J5! ;9)F /L '58>!58)< !8> , ,5!J /$ ++ ,+ 29 30 + / + vÇ 5 í , 31 vÇ . / + vÇ 32 http://bmjopen.bmj.com/ 33 /5 34 35 ! + . vÇ + , 36 37 5 C $ / Iw , 38 C ;Iw !5BI /L 9)F 58>5< . ., , .? ;Iw 58I 9)F /L 39

40 85>)5)< $ 5 í + .? $, on September 24, 2021 by guest. Protected copyright. 41 42 + / . 5 43

44 45 5 + ++ , 1 . . / 2 46 47 5 C + . / + / + 48 49 vÇ +$$, + 5 I/ . , 50 , / 5 h . , / 51 52 $, > vÇ + / $ 2+ ++ / 53 54 . , > . 2+ 5 Ç / , / . 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 , ; C <5 L vÇ . 5 6 + , / > $ 5 Ü , 7 / $ . vÇ .5 8 9 ë , + , + : vÇ 10 !4!9 11 + , $ +5 I/ . 12 + !4!9 , . 5 13 14 L + / + + B95F . 15 16 +5 Ç For peer review + . 5 only , / 17 J 18 J9F + + . +5 L , 19 / 9/D / . + 55 20 21 2 , / $, 9/D 5 22 23 vÇ + / : + + $ N O ;Ç$ <5 L 24 25 + L/5>8 Ç)859 + + / . 1, + 5 26 Ç + . $ $ + + + 5 27 28 Ç +$ 95 + + 29 30 $ $ 5 C + + , $ 31 + 5 32 http://bmjopen.bmj.com/ 33 34 35 L / vÇ + 2 ++ + / + 36 37 + , + / {/ / 5 + $ 38 / > !8> , > , $ !5F5 39

40 on September 24, 2021 by guest. Protected copyright. 41 C 42 43 Ç , / $, + Ç w C h 44 45 Ü. , I+5 Ç , , 46 47 + 5 48 49 50 / 51 52 !Ç[ / ++ $, Ü. , { 5 53 Ç,C 5 ! 5 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 ! 5 6 /{I , + ++ 5 !t ,? 5 /{I 7 !t !Ç[ IYW Ü9 a. W[C . ,5 !Ç[ !t IYW +. . 8 9 . , 5 !t !Ç[ IYW Ü9 a. W[C , . / ++ 10 11 / + . ++. 5 /{I +$, 12 ++ 5 13 14 15 16 5 For peer review only 17 18 5 5 / + / / + $ 5 19 I/ . / / . , / / + , / , 20 21 5 t + ! [ ;! 5[ 6, 5 < / 22 23 + 5 24 25 26 27 28 29 30 31 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 24, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 ! 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 w9C9w9b/9{ 5 6 7 5 / >{ w h ÜC a /.5 ! , + , $ 8 + + 5 5 a W 8'IB;!<3!'485 9 5 {/ $ Ü Ç Q ! [ [5 5 [> a, 5 9 , 10 5 + t /+5 ! 9 a 884I);<39>B5 11 !5 í , w[ w Y!5 vÇ [5 !ù/9wÇ L 5 +3MM///5 $ 5M /> >M5 12 Ü+ ! 8 8B5 ! 5 $ J 8)5 13 '5 . , a/ I !í Y W! 5 t vÇ . + > 14 , ,5 Ç w { ,5 9 I W 999I8;'<3J4>4'5 15 )5 { {a Y W! 5 . a[ 5 t vÇ . 16 ++ For 5peer W ! / / review 88BI'J;<3!B>!BJ5 only 17 B5 w$$ W b W/ w 1 ta 5 Ç $ / vÇ . 18 , / . I { ,5 ! W a 88!I);9<3B49>B9'5 19 20 J5 ù ò t í{ . >/ 9 5 9 + vÇ . ,3 21 >,5 9 8I;)<3BB8>BJ85 22 45 b W. D / w të 5 w + . $ vÇ 23 .3 . +, ++ 5 9 I W 24 8'I!);8<3!!)>!''5 25 95 t ëC Ç Ç9 5 Ya 5 t vÇ . + > , 26 + / 3 . $, , $ 5 27 w h 8'I)!;<3!>!J5 28 85 { {a { $ a/ . D{ 5 b> vÇ >+ 29 5 9 I W 88)IB;9<388J>85 30 5 w, í! / /t a , YÇ 5 !,+ +, 31 5 b 9 W a 889I!B8;!<3)>!)5 32 5 w, í! a { Ç + t. 5 /, + 5 http://bmjopen.bmj.com/ 33 / t Ç 88'IJ);!<3!'>'5 34 !5 w, í! a , YÇ I Y 5 !? , . 5 b 9 W a 35 8I!BB;8<344>4985 36 '5 5 / .W / wa C a 5 t + + 37 3 ! I ! / / . 38 b / / , / . 5 ò 3 $, L 39

40 { , /+ ? 9 , ! ! / >/ b 5 on September 24, 2021 by guest. Protected copyright. 41 / 88'I8;J<3J>J'B5 42 )5 a !C b [{ { !I 5 9 + + . . 43 . + +5 W a Ç 88IB;<38B>)5 44 B5 a !C I w{ { . 5 / > + . . 45 . . 5 ! 9 a 8)I;)<3'99>)8J5 46 J5 b . a /. Ç $1 [{ 5 5 + $ , 47 , + 5 ! > ,5 5 a W 8BIB!;B<3!)'85 48 45 { a { {! { W[ 5 Ç 5 b t w ,3 . / 49 : , + 5 / 9 8)IJ3''9>'985 50 95 t ! { {! í >Y I 5 5 w t 3 Ç 5 b 51 t + w ,5 L W 9 8B5 52 85 ! . I. . !. Y 5 Ç {/ b t , w 5 { I 53 Ç L 88JI9;t <3!')>!'95 54 5 { a t [ { IÇ5 Ç 5 /. w {, 55 + ,5 9 W 9 8'I9;4<3)'>)'95 56 57 58 ' 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 5 [ . WC h $ >h t t .Ü 5 Ç {/ + , $ 3 5 +$ + 5 9 W 9 889I';<3B)9>BBJ5 6 !5 a%Ça &'{[Ça 9/D !** t t* +* D ',-.'/.0,,' w1* . a/ 7 íL3 D9 I I ! W , 8)5 8 '5 [3454 6 t* 5ó[0 6 9YD t54* /-9-./&/:;:& ! . a!3 9 t + a {, I W 8895 10 )5 { ! [ aD D $ wW 5 ! +. 1 vÇ . 11 ; C I { ,<5 ! W / 99IJ8;J<3J9J>485 12 B5 w 1 ta { / ? . D [{ 5 !I!M!//CMIw{ 13 ? + 3 + Lë3 {Ç Ç Ü 14 /. vÇ .3 ! I ! 15 9 + , ! , / / / / ,I ! 16 / / ,For C peer I review I w , { ,3 only $, L 17 { , /+ ? 9 ,5 / 889I9;8<3 '>)85 18 J5 { Ç W a D, wW 5 ! . / ++ ++ , , 19 20 . + $ $, + 21 / . .$ 5 W / 9 88BI)9;)<3'!J>''J5 22 45 w W! { !! a, W 5 h >>, ++ , 5 23 t 5 { 8I { ++ 3B9>485 24 95 / a9 t+ t ! Y[ 5 ! / , + $ , 25 3 . + . 5 W / 5* 94JI'8;)<3!J!>!4!5 26 !85 v I { 1 ë I t 5 / $ L/5>9>/a 27 L/5>8 . 5 a / 88)I'!;<3!8>!95 28 !5 t ?? 9 w 9 5 $ L 5 5 >L ! , 3 . D+ . / 29 t + , Y/ / t 5 5 { 8JI'83'B>'B'5 30 !5 ! t/5 h+ + / ++ ,> / 31 ++ $ . 5 t { 8I8;<3)8>B5 32 !!5 /5 .I5 ! , > 5 I 98IJ3!)!>!J85 http://bmjopen.bmj.com/ 33 !'5 a !C b [{ { . 5 L . . . 34 / . 5 ! 9 a 8I9;J<34'!>4'95 35 !)5 D 5 I 5 a , ë5 a $ . 3 36 + . T 9 a W 88'I;<3!)>!45 37 !B5 { aí ! tI I a 5 Ç + . / vÇ 38 + , + + 5 ! 9 a 889I)';B<3JB!>JB45 39

40 !J5 ë $ . ë 9 w$, Ç 5 í vÇ / C Ü vÇ on September 24, 2021 by guest. Protected copyright. 41 aT W ! I !** 8BI);B<3 88!B'5 42 !45 !>!$ {! í $ / h Yw 5 ë , , / + 43 . 3 8>, . / / + /5 ! W 44 /1* 5* 8)I);<3'!>)85 45 !95 ò+ òD / !W5 5 vÇ + + 5 I 46 88!I49;<3!B!>!J5 47 48

49 50 51 52 53 54 55 56 57 58 ) 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 [ / / 5 6 7 C ( C / 8 C ( Y a / 9 10 !$$ .3 vÇ E 8 + / vÇ +I vÇ E + / vÇ +5 11 12 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 24, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 B 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 5

6 This image cannot currently be displayed. 7 8 9 10 11 12 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

32 prolongation. http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

40 on September 24, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 !tt9b5Ló 5 6 !ppendix ! t 5ata sources 7 8 Ln the logistic system in the 95s of the wegion of {outhern 5enmark and the wegion of {kåne 9 information regarding triage and presumed diagnoses are shared among the health care personal. 10 11 Çhe system also provides information on time of arrival to the 95 as well as time of discharge. !t 12 13 arrival all patients are registered by main reason for contact. [und and Ielsingborg have 43 14 15 somatic contact possibilities, hdense and 9sbjerg have 40 contact possibilities. íe used these 16 systems to identify Forpatients withpeer suspected review poisoning. only 17 18 Crom the electronic central 9/D databases at the wegion of {outhern 5enmark and the wegion of 19 20 {kåne all 9/Ds measures were extracted. Çhese databases contain information of all 9/Ds 21 recorded in any hospital in the respective region. 22 23 24 25 Çhe 5anish bational tatient wegister was established in 1977. {ince 1994 diagnostic information 26 27 has been recorded in accordance with L/5-10. Çhe content includes information regarding 28 discharge diagnoses from which comorbidities were derived. Ln {weden, all health care 29 30 consultations are recorded in the respective region databases. Crom the {kåne Iealthcare 31

32 wegister, we retrieved information corresponding to information from the 5anish bational tatient http://bmjopen.bmj.com/ 33 wegister. Ln these regional databases diagnoses from both the primary and secondary health care 34 35 system are available. /harlson /omorbidity Lndex was calculated based on data extracted for a 10 36 37 year-period ending on the day of contact in the 5anish data. Crom the {wedish database, /harlson 38 39 /omorbidity Lndex was calculated for a 2-year period ending on the day of contact to the 95. 40 on September 24, 2021 by guest. Protected copyright. 41 42 Çhe {wedish bational tharmacy wegister contains data on all prescriptions dispensed at 43 44 pharmacies since 2005. .esides date of dispensing, name, amount, and dose of the redeemed 45 46 medication are accessible from this register. Çhe 5anish bational trescription wegistry provides 47 similar information of individual-level prescription of medication since 1995. Lnformation of 48 49 redeemed prescription of vÇ prolonging drugs (see !ppendix c.5) within 90 days before contact 50 51 was obtained through these registers. 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 Çhe 5anish /ivil wegistration {ystem contains data regarding birth, emigration, and vital status for 5 6 the entire populations. Lnformation regarding vital status in {weden was retrieved from the 7 {wedish population registry. Çhe 5anish /ivil wegistration {ystem was established in 1968, and by 8 9 law all 5anish residents have a unique civil registration number of ten digits. {ince 1967, all 10 11 {wedish residences are assigned a ten-digit personal identity number. Çhese unique numbers 12 allow complete and accurate linkage between registries on an individual level. 13 14 15 !ppendix . t {upplementary methods 16 For peer review only 17 9/D measurements 18 19 !ll 9/Ds were recorded and stored either in aÜ{9® /ardiology Lnformation {ystem (D9 20 21 Iealthcare), or by thilips 5iagnostic 9/D (thilips). !ll 9/Ds in our analysis were 12 lead 9/Ds. 22 9/Ds recorded in aÜ{9 were later processed using the aarquette 12{[ algorithm, which calculate 23 24 the vÇc interval using the .azett Cormula, the Cridericia Cormula, and the Cramingham Cormula. Ln 25 26 aÜ{9, the vÇ interval is measured as a median value from the 12 leads. 9/D recorded by thilips 27 28 were analyzed by the 5ó[-algorithm. Ln this algorithm, the vÇ interval is measured as a median 29 value from reliable leads. ! lead is defined as reliable if little variation in beat-to-beat variation. 30 31 thilips provided vÇc intervals corrected by the .azett Cormula and the Cridericia Cormula.

32 http://bmjopen.bmj.com/ 33 Ço correct for heart rate, we chose the Cramingham Cormula (vÇcCramingham =vÇ + 0.154 (1-ww)). 34 .ecause thilips 5ó[-algorithm does not routinely correct the vÇ interval using the Cramingham 35 36 Cormula, we calculated the vÇcCramingham ourselves. !s we did not include ww intervals in our 37 38 analysis, the formula was used in another edition than the original formula. Çherefore, we used 39

40 following formula for correction in 9/Ds recorded by thilips: vÇcCramingham = vÇ + 154 (1-60/heart on September 24, 2021 by guest. Protected copyright. 41 rate). Çhe vw{ duration was measured as a median value in both algorithms. 42 43 44 45 Çhe aarquette 12{[ algorithm defines onsets as the earliest deflection in any lead, and offsets as 46 the latest deflection in any lead. Çhe vÇ interval is measured from the earliest detection of 47 48 depolarization in any lead to the latest detection of repolarization in any lead. {imilarly, the vw{ 49 50 duration was measured from the earliest onset in any lead to the latest deflection in any lead. 51 52 Çhe thilips 5ó[-algorithm first identifies waveform component and measures every beat in each 53 lead individually. !fter the approximate waveform locations are known, onsets and offsets are 54 55 defined. hnce the onsets and offsets are known, duration of intervals are calculated. 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 5 6 Ln both 5enmark and {weden 9/D recording is performed by well-educated heath care personals 7 instructed only to accept 9/Ds of satisfying quality. htherwise it is considered a routine to record 8 9 another 9/D. Lf multiple 9/Ds were recorded in a single individual, the first 9/D recorded within 4 10 11 hours after arrival was included in the analysis. 12

13 14 15

16 For peer review only 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

14 {Y{ codes* !dministrative codes 15 !ë!!07: {udden cardiac arrest 16 For peer!ë!!06: {udden review cardiac arrest only 17 18 trocedure codes 19 ùù0401: {tandby for cardiac arrest 20 21 Çreatment codes 22 .CC!6: /hest compressions 23 .CC!60: 9xternal chest compressions 24 .CC!60!: 9xternal chest compressions by use of mechanical chest compressions 25 26 Yë)-codes** 5C012: /hest compressions 27 5C017: aechanical chest compressions 28 5C028: /ardiopulmonary resuscitation 29 30 *{Y{ = {undhedsvæsenets klassifikations system, available on http://medinfo.dk/sks/brows.php, 5anish codes. 31 **Yë)= Ylassifikation av vårdåtgärder, available on

32 http://www.socialstyrelsen.se/klassificeringochkoder/atgardskoderkva, {wedish codes. http://bmjopen.bmj.com/ 33 34 /.2 35 36 toisoning divided into groups by use of L/5-10 codes 37 38 !nalgesics and drugs of abuse Ç39*-Ç40*, C110*, C120*, C140*, C150*, C160* 39 tsychotropic drugs and drugs affecting the Ç42*-Ç44*, C130*, C190* 40 central nervous system on September 24, 2021 by guest. Protected copyright. 41 hrganic and chemical substances, non- Ç51*-Ç65*, C100*, C170*, C180* 42 medical substances 43 hthers Ç36*-Ç38*, Ç41*, Ç45*-Ç50* 44 45 aultidrug Hî }( šZ }À uvš]}v ‰}]•}v]vP PŒ}µ‰• 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 /.3 5 /harlson /omorbidity Lndex* 6 7 /ondition !ssigned weight L/5-10 codes 8 teripheral vascular disease 1 L70.x, L71x, L73.1, L73.8, L73.9, L77.1, L79.0, L79.2, Y55.1, 9 Y55.8, Y55.9, ù95.8, ù95.9 /erebrovascular disease 1 D45.x, D46.x, I34.0, L60.x-L69.x 10 5ementia 1 C00.x-C03.x, C05.1, D30.x, D31.1 11 /hronic pulmonary disease 1 L27.8, L27.9, W40.x-W47.x, W60.x-W67.x, W68.4, W70.1, W70.3 12 wheumatic disease 1 a05.x, a06.x, a31.5, a32.x-a34.x, a35.1, a35.3, a36.0 13 teptic ulcer disease 1 Y25.x-Y28.x 14 aild liver disease 1 .18.x, Y70.0-Y70.3, Y70.9, Y71.3-Y71.5, Y71.7, Y73.x, Y74.x, 15 Y76.0, Y76.2-Y76.4, Y76.8, Y76.9, ù94.4 16 5iabetes without chronicFor peer1 review910.0, 910.1, 910.6, only 910.8, 910.9, 911.0, 911.1, 911.6, 17 complications 911.8, 911.9, 912.0, 912.1, 912.6, 912.8, 912.9, 913.0, 18 913.1, 913.6, 913.8, 913.9, 914.0, 914.1, 914.6, 914.8, 914.9 19 Iemiplegia or paraplegia 2 D04.1, D11.4, D80.1, D80.2, D81.x, D82.x, D83.0-D83.4, 20 D83.9 21 wenal disease 2 L12.0, L13.1, b03.2-b03.7, b05.2- b05.7, b18.x, b19.x, 22 b25.0, ù49.0- ù49.2, ù94.0, ù99.2 23 5iabetes with chronic 2 910.2-910.5, 910.7, 911.2-911.5, 911.7, 912.2-912.5, 912.7, 24 complications 913.2- 913.5, 913.7, 914.2-914.5, 914.7 25 /ancer 2 /00.x-/26.x, /30.x-/34.x, /37.x- /41.x, /43.x, /45.x-/58.x, 26 /60.x- /76.x, /81.x-/85.x, /88.x, /90.x-/97.x 27 aetastatic cancer 3 /77.x-/80.x 28 aoderate or severe liver disease 3 L85.0, L85.9, L86.4, L98.2, Y70.4, Y71.1, Y72.1, Y72.9, Y76.5, 29 Y76.6, Y76.7 30 !L5{/ILë** 6 .20.x-.22.x, .24.x 31 *5iagnostic codes for myocardial infarction and heart failure are not included in the index, but are included in this 32 study as covariates. http://bmjopen.bmj.com/ 33 ** !L5{ = acquired immunodeficiency syndrome, ILë = human immunodeficiency virus 34 35 36 /.4 37 ëariables from /harlson /omorbidity Lndex, included separately in the analysis 38 ayocardial infarction L21.x, L22.x, L25.2 39 /ongestive heart failure L09.9, L11.0, L13.0, L13.2, L25.5, L42.0, L42.5-L42.9, L43.x,

40 L50.x, t29.0 on September 24, 2021 by guest. Protected copyright. 41 42 43 /.5 44 45 [ist of drugs associated with vÇ prolongation and risk of Çdt* 46 5rug category !Ç/-codes** 47 !limentary tract and metabolism 5omperidone (!03C!03), Dranisetron (!04!!02), aetoclopramide (!03C!01), 48 hndansetron (!04!!01), tantoprazole (!02./02) 49 /ardiovascular system !miodarone (/01.501), 5ronedarone (/01.507), Clecainide (/01./04), Curosemide 50 (/03/!01, /039.01), Iydrochlorothiazide (/039!01, /095!01, /095!06, /095!04, 51 /09.!02), Lndapamide (/03.!11), Lsradipine (/08/!03), Lvabradine (/019.17), {otalol 52 (/07!!07) 53 Denito urinary system and sex !lfuzosin (D04/!01), aifepristone (D03ó.01), airabegron (D04.512), {olifenacin 54 hormones (D04.508, D04/!53), Çolterodine (D04.507), ëardenafil (D04.909) 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 24 of 29 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 {ystemic hormonal preparations, 4 hxytocin (I01..02), tasireotide (I01/.05) 5 excl. sex hormones and insulins !nti-infectives for systemic use 6 !tazanavir (W05!w15), !zithromycin (W01C!10), .edaquiline (W04!Y05), /iprofloxacin 7 (W01a!02), /larithromycin (W01C!09), 9rythromycin (W01C!01), Cluconazole (W02!/01), 8 Coscarnet (W05!5), Ltraconazole (W02!/02), Yetoconazole (W02!.02), aetronidazole (W01ó501), aoxifloxacin (W01a!14), tosaconazole (W02!/04), wilpivirine (W05!D05), 9 witonavir (W05!903), woxithromycin (W01C!06), {aquinavir (W05!901), ëoriconazole 10 (W02!/03) 11 !ntineoplastic and !nagrelide ([01óó35), .ortezomib ([01óó32), .osutinib ([01ó914), /eritinib ([01ó928), 12 immunomodulatory agents 13 /rizotinib ([01ó916), 5abrafenib ([01ó923), 5asatinib ([01ó906), 5egarelix ([02.ó02), 14 9ribulin mesylate ([01óó41), Cingolimod ([04!!27), [apatinib ([01ó907), [euprolide ([02!902), bilotinib ([01ó908), hxaliplatin ([01ó!03), tanobinostat ([01óó42), 15 tazopanib ([01ó911), {orafenib ([01ó905), {unitinib ([01ó904), Çacrolimus ([04!502), 16 For peerÇamoxifen ([02.!01), review ëandetanib ([01ó912), only ëemurafenib ([01ó915) 17 ausculo-skeletal system 18 Çizanidine (a03.ó02) 19 bervous system 20 !mantadine (b04..01), !misulpride (b05![05), !mitriptyline (b06!!09), 21 !pomorphine (b04./07), !ripiprazole (b05!ó12), !senapine (b05!I05), !tomoxetine (b06.!09), /italopram (b06!.04), /lomipramine (b06!!04), /lozapine (b05!I02), 22 5exmedetomidin (b05/a18), 5oxepin (b06!!12), 5roperidol (b05!508), 9scitalopram 23 (b06!.10), Cluoxetine (b06!.03), Dalantamine (b065!04), Ialoperidol (b05!501), 24 Iydroxyzine (b05..01), Lmipramine (b06!!02), [evomepromazine (b05!!02), [ithium 25 (b05!b01), aethadone (b07./02), airtazapine (b06!ó11), bortriptyline (b06!!10), 26 hlanzapine (b05!I03), taliperidone (b05!ó13), taroxetine (b06!.05), timozide 27 (b05!D02), tipamperone (b05!505), tropofol (b01!ó10), vuetiapine (b05I04), 28 wisperidone (b05!ó08), {ertindole (b05!903), {ertraline (b06!.06), {evoflurane (b01!.08), {ulpiride (b05![01), Çetrabenazine (b07óó06), ëenlafaxine (b06!ó16), 29 ùiprasidone (b05!904) 30 !ntiparasitic products, insecticides 31 /hloroquine (t01./02), Iydroxychloroquine (t01.!02), aetronidazole (t01!.01),

and repellents http://bmjopen.bmj.com/ 32 tentamidine (t01/ó01), vuinine sulfate (t01./01) 33 wespiratory system 34 5iphenhydramine (w06!!02), tromethazine (w06!502) 35 ëarious 36 terflutren lipid microspheres (ë085!01) 37 *Crom vÇ5rug list, https://crediblemeds.org/, version 5ecember 17, 2015. hnly drugs available in 5enmark are 38 included at our list. Çhe list includes drugs with known risk of Çdt, possible risk of Çdt, and conditional risk of Çdt. 39 **!Ç/ = !natomical Çherapeutic /hemical 40 on September 24, 2021 by guest. Protected copyright. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 25 of 29 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 1 2 3 4 !ppendix 5 t {upplementary results 5 6 wesults from stratified analysis. !lthough further strata than shown in this table were preplanned 7 (e.g. stratification on all poisoning groups and age), we only stratified when possible due to a small 8 9 number of events. 10 11 vÇc prolongation bormal vÇc interval Iw* 12 vÇ Hðñì u•U uv <440 ms, men and women (95% /L) 13 vÇ Hðòì u•U Á}uv 14 15 9vents (n)** 9vents (n)** 16 {uspected poisoning For peer review only 17 Çotal 8 (248) n<5 (496) 3.6 (1.0-12.2)

18 aale n<5 n<5 2.7 (0.5-16.3) 19 4.4 (0.8-24.3) 20 Cemale n<5 n<5 21 !ge H50 years 6 (131) n<5 2.7 (0.7-9.9) 22 hdense or 9sbjerg n<5 n<5 5.8 (0.6-57.4) 23 [und n<5 n<5 1.7 (0.2-12.2) 24 4.4 (0.4-49.2) 25 Ielsingborg n<5 n<5 26 /onfirmed poisoning 27 Çotal 6 (151) n<5 10.5 (1.2-90.0) 28 tsychotropic drugs n<5 n<5 2.0 (0.1-32.5) 29 !bbreviations: Iw; hazard ratio, /L; confidence interval. 30 *Iazard ratio from /ox regression. 31 **Lf the number of events in the analysis was less than 5 (marked by n<5), the number of patients in the strata is not shown. 32 http://bmjopen.bmj.com/ 33 34 35 36 37 38 39

1 2 STROBE Statement—checklist of items that should be included in reports of observational studies 3

4 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 5 Item 6 No Recommendation 7 Title and abstract 1 (a) Indicate the study’s design with a commonly used term in the title or the abstract 8 Page 1 and 2 9 10 (b) Provide in the abstract an informative and balanced summary of what was done 11 and what was found 12 Page 2 13 14 Introduction 15 Background/rationale 2 Explain the scientific background and rationale for the investigation being reported 16 Page 3 17 Objectives 3 State specific objectives, including any prespecified hypotheses 18 For peer review only 19 Page 3 20 Methods 21 Study design 4 Present key elements of study design early in the paper 22 23 Page 3 24 Setting 5 Describe the setting, locations, and relevant dates, including periods of recruitment, 25 exposure, follow-up, and data collection 26 Page 3-5 27 28 Participants 6 (a) Cohort study—Give the eligibility criteria, and the sources and methods of 29 selection of participants. Describe methods of follow-up 30 Page 4 and 5 31 32 Case-control study—Give the eligibility criteria, and the sources and methods of 33 case ascertainment and control selection. Give the rationale for the choice of cases 34 and controls 35 Cross-sectional study—Give the eligibility criteria, and the sources and methods of

36 http://bmjopen.bmj.com/ 37 selection of participants 38 (b) Cohort study—For matched studies, give matching criteria and number of 39 exposed and unexposed 40 Page 5 (As we did a propensity score analysis the number of exposed and 41 42 unexposed are outlined in the results because it cannot be provided before the 43 analysis)

44 Case-control study—For matched studies, give matching criteria and the number of on September 24, 2021 by guest. Protected copyright. 45 controls per case 46 47 Variables 7 Clearly define all outcomes, exposures, predictors, potential confounders, and effect 48 modifiers. Give diagnostic criteria, if applicable 49 Page 4, 5, and Appendix B and C 50 Data sources/ 8* For each variable of interest, give sources of data and details of methods of 51 52 measurement assessment (measurement). Describe comparability of assessment methods if there 53 is more than one group 54 Page 4 and Appendix A, B, and C 55 Bias 9 Describe any efforts to address potential sources of bias 56 57 Page 6 58 Study size 10 Explain how the study size was arrived at 59 Page 4 and figure 1 60 Quantitative variables 11 Explain how quantitative variables were handled in the analyses. If applicable,

1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 27 of 29 BMJ Open

1 2 describe which groupings were chosen and why 3 Page 5

4 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 5 Statistical methods 12 (a) Describe all statistical methods, including those used to control for confounding 6 Page 5 and 6 7 (b) Describe any methods used to examine subgroups and interactions 8 Page 5 and 6 9 10 (c) Explain how missing data were addressed 11 No missing data. 12 (d) Cohort study—If applicable, explain how loss to follow-up was addressed 13 No loss to follow-up. 14 15 Case-control study—If applicable, explain how matching of cases and controls was 16 addressed 17 Cross-sectional study—If applicable, describe analytical methods taking account of 18 For peer review only sampling strategy 19 20 (e) Describe any sensitivity analyses 21 Page 6 22 Continued on next page 23 24 25 26 27 28 29 30 31 32 33 34 35

36 http://bmjopen.bmj.com/ 37 38 39 40 41 42 43

44 on September 24, 2021 by guest. Protected copyright. 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 28 of 29

1 2 3 Results

4 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 5 Participants 13* (a) Report numbers of individuals at each stage of study—eg numbers potentially eligible, 6 examined for eligibility, confirmed eligible, included in the study, completing follow-up, and 7 analysed 8 9 Page 6 10 (b) Give reasons for non-participation at each stage 11 Page 6 12 (c) Consider use of a flow diagram 13 14 Figure 1 15 Descriptive 14* (a) Give characteristics of study participants (eg demographic, clinical, social) and information 16 data on exposures and potential confounders 17 Page 6 and 7, including table 1 18 For peer review only 19 (b) Indicate number of participants with missing data for each variable of interest 20 None missing 21 (c) Cohort study—Summarise follow-up time (eg, average and total amount) 22 23 Page 8 24 Outcome data 15* Cohort study—Report numbers of outcome events or summary measures over time 25 Page 8 and table 3 26 Case-control study—Report numbers in each exposure category, or summary measures of 27 28 exposure 29 Cross-sectional study—Report numbers of outcome events or summary measures 30 Main results 16 (a) Give unadjusted estimates and, if applicable, confounder-adjusted estimates and their 31 precision (eg, 95% confidence interval). Make clear which confounders were adjusted for and 32 33 why they were included 34 Page 7 and 8 including table 3 35 (b) Report category boundaries when continuous variables were categorized

36 http://bmjopen.bmj.com/ Page 5 37 38 (c) If relevant, consider translating estimates of relative risk into absolute risk for a meaningful 39 time period 40 Page 8 41 42 Other analyses 17 Report other analyses done—eg analyses of subgroups and interactions, and sensitivity 43 analyses

44 The prevalence of QTc prolongation in relation to specific subgroups: page 7 and table 2. on September 24, 2021 by guest. Protected copyright. 45 Appendix D includes a stratified analysis. 46 47 Sensitivity analysis: page 8 and 9. 48 Discussion 49 Key results 18 Summarise key results with reference to study objectives 50 51 Page 9 52 Limitations 19 Discuss limitations of the study, taking into account sources of potential bias or imprecision. 53 Discuss both direction and magnitude of any potential bias 54 Page 9 and 10. 55 56 Interpretation 20 Give a cautious overall interpretation of results considering objectives, limitations, multiplicity 57 of analyses, results from similar studies, and other relevant evidence 58 Page 10 and 11. 59 Generalisability 21 Discuss the generalisability (external validity) of the study results 60 Page 9 and 10.

3 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 29 of 29 BMJ Open

1 2 Other information 3 Funding 22 Give the source of funding and the role of the funders for the present study and, if applicable,

4 BMJ Open: first published as 10.1136/bmjopen-2017-020036 on 7 July 2018. Downloaded from 5 for the original study on which the present article is based 6 Page 12 7 8 9 *Give information separately for cases and controls in case-control studies and, if applicable, for exposed and 10 unexposed groups in cohort and cross-sectional studies. 11 12 Note: An Explanation and Elaboration article discusses each checklist item and gives methodological background and 13 14 published examples of transparent reporting. The STROBE checklist is best used in conjunction with this article (freely 15 available on the Web sites of PLoS Medicine at http://www.plosmedicine.org/, Annals of Internal Medicine at 16 http://www.annals.org/, and Epidemiology at http://www.epidem.com/). Information on the STROBE Initiative is 17 available at www.strobe-statement.org. 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35

36 http://bmjopen.bmj.com/ 37 38 39 40 41 42 43

44 on September 24, 2021 by guest. Protected copyright. 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

4 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml