Article Text
PDF
Abstract
Objectives To determine patient-reported symptoms and clinical factors associated with mimics and differences in health-seeking behaviour versus stroke.
Design This is a post-hoc analysis of a cross-sectional survey of interviews on patient-reported factors in patients admitted with suspected stroke. Patients were categorised as genuine stroke or mimic. The surveys were conducted from February 2018 to January 2019.
Setting Two non-comprehensive stroke centres in Denmark.
Participants Patients≥18 years (no upper age limit) admitted with symptoms of stroke to one of the non-comprehensive stroke centres or transferred from a comprehensive- or primary stroke centre were eligible for inclusion. 592 patients were included.
Outcome measures Symptoms or clinical factors associated with stroke mimics. Logistic regression analysis was performed to identify factors associated with mimics. Secondarily, the number of strokes versus mimics presenting at a healthcare facility within 3 hours contacted the emergency medical service (EMS) and arrived by ambulance.
Results Of 592 suspected patients with stroke, 113 (19.1%) were mimics; most frequently peripheral vertigo (24.7%) and migraine (11.5%). Factors associated with a higher likelihood of mimics were female sex (OR 1.79, 95% CI 1.14 to 2.79), high Scandinavian Stroke Scale scores (OR 1.05, 95% CI 1.02 to 1.09, per point increase), and vertigo (OR 1.86, 95% CI 1.18 to 2.95). Factors associated with a lower likelihood of mimics were increasing age (OR 0.96, 95% CI 0.95 to 0.98 per year increase), reported limb weakness (OR 0.52, 95% CI 0.30 to 0.89) and difficulty steering (OR 0.51, 95% CI 0.28 to 0.93).
There was no difference between groups in the proportion of patients for whom time from symptom onset to healthcare services contact exceeded 3 hours (52.2% vs 53.7%, p=0.78). Fewer mimics contacted the EMS first, were accepted at a primary stroke centre and arrived by ambulance (p<0.05 for all variables).
Conclusion Patient-reported vertigo and migraine are common stroke mimics. Increasing age and unilateral limb symptoms increase the likelihood of a stroke. Although symptoms are similar, prehospital pathways differ between mimics and genuine patients with stroke.
- Stroke medicine
- Stroke
- Emergency Service, Hospital
Data availability statement
Data are available upon reasonable request. The data that support the findings of this study are available from the corresponding author upon reasonable request and in adherence to Danish legislation.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
STRENGTHS AND LIMITATIONS OF THIS STUDY
Data are based on patient reported symptoms from a non-comprehensive stroke unit and may not be applicable to general stroke populations.
Observational nature of the study precludes assessment of cause-effect relationships.
Sample size may be too small as this is a post-hoc analysis.
Confirmed stroke population may also comprise patients clinically classified with transient ischaemic attack (TIA). Mimics may also have been misclassified as a TIA (or stroke).
Patients transferred from a comprehensive stroke centre may already have a diagnosis on interview, introducing a possibility of recall bias.
Introduction
With the increasing demand for rapid decisions on stroke diagnosis required for acute treatment with tissue plasminogen activator (tPA), there is a risk for unintended treatment of patients without stroke with tPA,1 resulting in unwanted and unnecessary side effects. In a general population admitted with suspected stroke, 5–30% may present with conditions that are often non-vascular and clinically mimic a stroke, also known as stroke mimics.2 3 Potential aetiologies of stroke mimics include conditions such as migraine, seizures, central nervous system tumours, infections, subdural haemorrhages, psychiatric disorders and neuropathy.3–7 Distinguishing stroke mimics from genuine stroke is a challenge when trying to optimise stroke diagnostics and reduce unnecessary treatment, and healthcare costs.
Screening tools that incorporate clinical findings on admission and vascular risk factors to predict a non-stroke diagnosis such as Facial droop, Atrial fibrillation and Age, Blood pressure, Seizures and Sensory symptoms score (FABS)8 and the Recognition of Stroke in the Emergency Room (ROSIER)9 scale have been made. However, these tools do not take patient-reported symptoms into account. Treatment with tPA is increasingly used in patients with minor stroke (ie, National Institute of Health Stroke Scale (NIHSS) ≤ 2)10 who present with partial remission of symptoms on evaluation where healthcare workers often must rely on patient-reported symptoms. In this study, we tried to identify patient-reported factors and help-seeking behaviour specific to stroke mimics using interviews with 592 patients admitted with suspected stroke. These results may present useful information on possible differences in patient-reported factors and help-seeking behaviour reported by stroke mimics compared with patients with genuine stroke. As such, our results may aid the healthcare worker to correctly identify stroke mimics.
Methods and materials
Design
We performed a post-hoc analysis using data from a cross-sectional survey on patients admitted with symptoms of stroke or transient ischaemic attack (TIA).11 The survey was conducted at two non-comprehensive stroke centres in the Capital Region of Denmark (Herlev Gentofte Hospital and North Zealand Hospital) from February 2018 to January 2019. Patients were included and interviewed on admission and prior to confirmation of diagnosis by neuroimaging or clinical examination to reduce recall bias. Patients referred from a comprehensive stroke centre may however already have had knowledge of their diagnosis. This procedure caused patients without stroke to be included in interviews (online supplemental figure 1). In the present analysis, interviews from patients with a non-stroke diagnosis were compared with those from patients with confirmed stroke (International Classification of Diseases codes (I61): non-traumatic intracerebral haemorrhage, (I63): Ischaemic stroke and (G45): TIA). A diagnosis of stroke or TIA was based on a clinical examination by a neurologist and routine acute neuroimaging (CT or MRI scans). The study design and cohort have previously been described in detail.11 Since this was a post-hoc analysis of observational data with no previous data available on patient-reported symptoms and admission pathways relating to mimics versus confirmed stroke, no prior power analysis was done.
Supplemental material
Patient and public involvement
There was no patient and public involvement in the design, conduct, reporting or dissemination plans of our research.
Setting
In Denmark, all citizens are provided tax-financed universal health coverage including access to emergency medical service (EMS),12 general practitioners (GP) and out-of-hours primary care (OOH-PC). Patients can present with symptoms either at a general practitioner, OOH-PC or contact EMSs directly. If the symptoms raise a suspicion of stroke, the GP or OOH-PC may consult the neurologist on call or call 1-1-2 directly for immediate transfer to thrombolysis. If the patient or bystanders contact the EMSs an ambulance may be dispatched, and the arriving paramedics can also consult the neurologist on call after their initial assessment of the patient. All individuals who present with symptoms of stroke within the time window (4.5 hours or 24 hours, thrombolysis and thrombectomy candidates, respectively) for revascularisation therapy are referred to primary stroke centres for intravenous thrombolysis (IVT) or comprehensive stroke centres for endovascular treatment (EVT), followed by transfer to local non-comprehensive stoke centres within 24–48 hours from arrival. Individuals who contact the healthcare system after the time window or who are not candidates for revascularisation therapy (eg, patients with remitted symptoms or due to contraindications) are admitted directly to emergency departments at local hospitals with non-comprehensive stroke centres for diagnostics, treatment, and rehabilitation.12 The described pathways are the model pathways, however in a minor part of cases there may be misinterpretation of symptoms, and an ambulance is not dispatched, or the patients choose to arrive by means of own transport and hence do not arrive by ambulance.13
Patients
Patients≥18 years (no upper age limit) admitted with symptoms of stroke to one of the non-comprehensive stroke centres described or transferred from a comprehensive or primary stroke centre were eligible for inclusion.
Data collection
Research assistants trained in administering a study-specific structured questionnaire performed the interviews, and data were stored using a REDCap database (REDCap consortium, Vanderbilt University, USA, V.9.1.0 hosted at Region Hovedstaden).14 15 Details on the questionnaire have been presented previously.11
Variables
The study cohort was categorised into stroke mimics or confirmed stroke according to the discharge diagnosis from patients’ medical files. In the case of an unspecific discharge diagnosis for stroke mimics (eg, ICD-10 (International Classification of Diseases, 10th revision) code ZO3.03; Observation for suspected nervous system disorder), we identified the most prominent clinical symptom(s) described in medical records to assign a descriptive diagnosis.
We retrieved data on age, sex, stroke symptom severity (assessed with the Scandinavian Stroke Scale (SSS)), prior stroke or TIA, comorbidities, medical treatment and accepting healthcare unit from the patients’ medical records. The SSS score (mandatory for the Danish National Stroke Registry database)16 was categorised into severe (0–25 points), moderate (26–42 points) and mild (43–58 points) stroke symptoms.17
Prehospital medication for stroke risk factors was defined as receiving≥1 of the following: antihypertensives, lipid-lowering agents, antidiabetics, anticoagulants or apnoea treatment.
Data on patient-related factors, including level of education, personal experience with stroke, knowledge of stroke symptoms and treatment, patient-reported stroke symptoms and their perceived symptom severity, were retrieved from the patient interviews. Patient-perceived symptom severity was rated on a numeric scale from 0 to 100 (100 being most severe).18 We defined values above the median for the whole cohort (30) as a high level of perceived symptom severity. Prior knowledge of stroke symptoms was defined as: knowledge of more than two, one or none of the signs from B(alance), E(yes), F(ace), A(rms), S(peech), T(ime), BE-FAST.19 Prior knowledge of revascularisation therapy with either thrombolysis or thrombectomy was noted.
Information on help-seeking behaviour was gathered from patient interviews and exact times were retrieved from medical or prehospital records.
We defined early arrival as hospital arrival within 3 hours of symptom onset to allow time for transfer to a comprehensive stroke unit in accordance with stroke management guidelines.20
Statistical analysis
Baseline characteristics for patients with stroke and stroke mimics are reported as frequencies with percentages or medians with 25th and 75th quartiles. Categorical variables were analysed using χ2 or Fisher’s exact test, and continuous variables were analysed using the Wilcoxon rank-sum test. A univariate analysis was performed to identify significant (p<0.05) clinically relevant demographic factors, risk factors and patient-reported symptoms. These variables were then included in a multivariable logistic regression analysis to identify factors independently associated with a diagnosis of stroke mimics. The model was a stepwise logistic regression. The following covariates were included: Age, sex, SSS, a history of hypertension, vertigo, nausea and/or vomiting, limb weakness, difficulty steering, visual symptoms (visual field loss, diplopia and other), facial drooping, speech difficulty and headache. Only variables with six or more observations were used. The significance level for covariates to enter and stay in the model was set to 0.05. The OR are reported with 95% CIs.
Analyses were performed using SAS statistical software V.9.4 (SAS Institute, Cary, North Carolina, USA).
Results
Baseline and clinical characteristics
We included 592 patients, of whom 479 (80.9%) had a discharge diagnosis of stroke. The stroke subcategories constituted each of ischaemic stroke (64.5%), TIA (27.3%) and non-traumatic intracerebral haemorrhage (8.2%). Of the 113 patients, who were not diagnosed with a stroke, the most common diagnoses were peripheral vertigo (24.7%) and migraine (11.5%).
Baseline and clinical characteristics are shown in table 1. Compared with patients with a confirmed stroke diagnosis, patients diagnosed with stroke mimics were younger (67 vs 74 years, p<0.001), more often female (55.8% vs 40.1%, p=0.003) and had less severe symptoms (mean SSS score 55.1 vs 50.9, p<0.001). In stroke mimics, a lower proportion compared with confirmed strokes had a history of hypertension (40.7% vs 55.1%, p=0.006). There was no difference between the groups in other vascular risk factors or prehospital medication. Of patients with confirmed stroke, 57 patients (11.9%) received intravenous tPA treatment and 12 patients (2.5%) underwent thrombectomy. In patients with stroke mimics, 2 (1.8%) patients received intravenous tPA and none had a thrombectomy.
Baseline characteristics, stroke severity, vascular risk factors and prehospital medication according to diagnosis
Patient-reported symptoms
For the entire population, the most frequently reported symptoms were arm and/or leg weakness (43.4%), difficulty steering the extremity (32.1%) and vertigo (28.9%). Patients with stroke mimics more frequently than confirmed strokes reported symptoms of vertigo, nausea and/or vomiting, inattention to one side, visual field loss, double vision or other visual symptoms and headache (p<0.05 for all symptoms). Less often than confirmed strokes, patients with stroke mimics reported symptoms of arm and/or leg weakness, difficulty steering, facial palsy or drooping and speech difficulty (p<0.05 for all symptoms) (table 2).
Presenting concern, subdivided into stroke and mimics
Patients with stroke mimics less often reported two or more BE-FAST symptoms (36.3% vs 58.9%, p<0.001). The proportion of patients reporting a sudden onset of symptoms was not significantly different between groups (51.3% vs 59.3%, p=0.12).
Stroke mimics more often had a perceived symptom severity above the overall median (61.3% vs 45.5%, p = 0.0029).
Patient knowledge
A numerically higher proportion of stroke mimics reported knowledge of≥2 BE-FAST symptoms (53.1% vs 41.1%, p=0.05). Prior knowledge of stroke therapy and experience with stroke from friends or family was not different between groups (60.2% vs 55.5%, p=0.83, and 50.4% vs 48.4%, p=0.70). The proportion of patients ever having heard of stroke was also similar between groups (84.1% vs 83.1%, mimics and confirmed strokes, respectively, p=0.80).
Prehospital stroke pathway
There was no difference between groups in the proportion of patients for whom time from symptom onset to healthcare services contact exceeded 3 hours (52.2% vs 53.7%, p=0.78). However, fewer mimics than confirmed strokes chose the EMS as first contact (15.9% vs 30.7%, p=0.0018), fewer mimics were admitted to a comprehensive or primary stroke centre (9.73% vs 25.6%, p<0.001) and fewer mimics arrived by ambulance compared with patients with stroke (56.6% vs 71.8%, p=0.002) (table 3).
Help-seeking behaviour and way through healthcare system
Mimics had a higher median time from symptom onset to healthcare contact (557 vs 240 min, p=0.058), but with a large range. The median time from symptom onset to hospital arrival was significantly higher for mimics (980 vs 320 min, p=0.0011).
Causes of stroke mimics
There were numerous causes of stroke mimics in this study. Signs of peripheral vertigo and migraine were the most common disease entities constituting over one-third of stroke mimics combined (table 4).
Causes of stroke mimics as per ICD-10 (International Classification of Diseases, 10th revision) coding (n=113)
Logistic regression analysis
Logistic regression analysis on stroke mimics versus confirmed stroke is shown in online supplemental figure 2. Factors associated with a higher likelihood of stroke mimics were female sex (OR 1.79, 95% CI 1.14 to 2.79), increasing SSS (OR 1.05, 95% CI 1.02 to 1.09) and vertigo (OR 1.86, 95% CI 1.18 to 2.95). Factors associated with a lower likelihood of stroke mimics were increasing age (OR 0.96, 95% CI 0.95 to 0.98 per year increase), reporting a hemiparesis (OR 0.52, 95% CI 0.30 to 0.89) or difficulty in steering the extremities (OR 0.51, 95% CI 0.28 to 0.93).
Supplemental material
Discussion
In this study, we addressed the significance of patient-reported symptoms on stroke to differentiate between stroke and stroke mimics. In patients admitted to the stroke unit with symptoms suspected to be signs of acute stroke, we found 19.1% of patients to be stroke mimics. Stroke mimics compared with patients with confirmed strokes, showed different patient-reported symptoms, with a preponderance of atypical symptoms such as vertigo, headache and visual symptoms. Furthermore, mimics showed a different pattern in choice of prehospital contact, as they less frequently called EMS. Despite this, there was no difference in prehospital delay between the two groups.
A 10-year cohort study on 8178 consecutive patients with suspected stroke from the National Institute of Health Stroke Programme reported a mimic proportion of 30%.2 In contrast, the proportion of stroke mimics in 1691 patients selected for possible tPA treatment, the proportion of mimics was as low as 8.8%.6 The reported proportion of stroke mimics may thus vary significantly between the hospital setting and the assessed population, for example, all patients with suspected stroke including TIA versus patients evaluated for tPA treatment. Our study included all patients with suspected stroke and TIA admitted to a general stroke unit, which could account for the relatively high number of mimics in our cohort.
The most common causes of stroke mimics in our study were signs of peripheral vertigo followed by signs of migraine, which constituted more than one-third of the mimic population combined (table 4). These particular conditions share symptomatology with posterior circulation strokes, known to present with amorphous symptoms such as dizziness, vertigo, ophthalmoplegia and visual symptoms.21 Other underlying diagnoses of mimicking symptoms such as hypo/paraesthesia and speech or visual disturbances could be anxiety-induced hyperventilation, diabetes, dehydration, peripheral nerve entrapment or conversion disorder. Epileptic seizures have been reported as a frequent aetiology to stroke mimics,3 6 with proportions as high as 47% in a stroke population eligible for tPA treatment.22 A ‘history of seizures’ is also included as factors associated with a non-stroke diagnosis in FABS and ROSIER, both of which are tools for screening and stratifying stroke mimics from acute cerebral ischaemia.8 9 In our study, only 1.8% of stroke mimics were caused by seizures of any kind. This discrepancy in the proportion of seizures as stroke mimics between ours and other studies may be attributed to a difference in referral procedures, as patients with suspected seizures were referred directly to our general neurology unit. The low number of seizures in our stroke mimic population could also be explained by a difference in the study population. The highest proportion of epileptic seizures among stroke mimics was found in tPA-treated populations, and not in non-thrombolysing stroke units such as ours.22 23 These tPA-treated populations may be biased towards patients with a higher symptom severity where postictal confusion and palsy could be misinterpreted as a large vessel occlusion stroke.
In this study, the odds of a mimic were 1.05 for every point higher scored on the SSS (95% CI 1.02 to 1.09), where high numbers indicate less symptom severity. This corresponds to previous studies on clinical symptoms of stroke mimics using the NIHSS.3 6 7 22 24 NIHSS scores can be converted to the inverse proportional SSS scores.25 Within recent years, a trend towards increased use of tPA therapy in patients with minor stroke has emerged,10 despite inconclusive evidence on the benefits of tPA in minor stroke.26 27 This change in treatment practice may pose a significant challenge in acute stroke treatment, as it could increase the number of stroke mimics receiving tPA treatment, since mimics seem to present with low symptom severity. It should be noted that only a low rate of symptomatic intracerebral haemorrhage is reported in mimics treated with tPA, which indicates a low-risk profile when tPA treatment is given to mimics.23 28 29 In spite of this, unintended tPA treatment of stroke mimics is undesirable also, as it increases the risk of delayed treatment of the underlying condition, as well as patient worries, and healthcare costs. In stroke research, a high prevalence of mimics in clinical trials on stroke treatment may weaken possible treatment effects, which can induce falsely neutral or negative results.30 31 Conversely, a high prevalence of mimics in stroke outcome studies may lead to overestimation of positive outcomes.32
Patient-reported vertigo was significantly associated with stroke mimics in the present study. In the case of acute vestibular syndrome the presence of a normal head impulse test, direction-changing nystagmus in eccentric gaze or skew deviation is 100% sensitive and 96% specific for stroke.33 This three-step oculomotor examination (head impulse, nystagmus and test of skew (HINTS test)) also appears to be more sensitive for stroke than early MRI (<48 hours of symptom onset).33 34 As such, clinicians have a well-developed bedside tool to distinguish peripheral causes of vertigo from central causes. The large number of mimics with peripheral vertigo in this study may suggest that more widespread training in and use of HINTS in the emergency department is warranted.
Of note, we found that female sex was associated with stroke mimics. Previous studies find that female sex is more likely to report atypical symptoms, which may resemble other conditions.35 Conversely increasing age, and patient-reported limb weakness and difficulty steering on admission were associated with less likelihood of a mimic diagnosis. Previous studies report similar results on age2 7 8 24 36 and sex.7 Other studies found no facial drooping on admission,7 8 22 36 no hypertension,2 22 24 no diabetes2 24 and a lack of atrial fibrillation2 8 24 to be associated with mimics. We found no significant difference in the frequency of atrial fibrillation between mimics and stroke on admission ECG. In our analysis, both facial drooping and history of hypertension entered our logistic model but did not remain in the final model. The fact that facial drooping did not remain in the model is surprising, as the absence of facial drooping alone was 94% sensitive and 71% specific for the diagnosis of mimics in FABS which was tested on a cohort of 784 patients (41% mimics).8
Our study suggests clinical factors sex, age and stroke symptom severity, and patient-reported symptoms vertigo, hemiparesis and difficulty in steering the extremities, may be helpful in distinguishing strokes from mimics. Any of these factors or symptoms alone should not deter clinicians from administering tPA, but they may be helpful in identifying which patients may benefit from a more comprehensive diagnostic approach. This is especially true in cases of suspected TIA where a partial or total remission of symptoms is present, forcing clinicians to rely on patient-reported symptoms. Symptoms are formed in an interplay between biological, psychological and cultural factors of which the clinician must be mindful in their assessment.37 In cases where clinical findings are lacking, an increased emphasis on vascular risk profile may be placed. Interestingly, apart from hypertension being more prevalent in the stroke group, the vascular risk profile was similar between mimic and patients with stroke in this study (table 1). This may suggest a bias towards referring patients with vascular risk factors to our stroke unit. Similarly, presymptom Modified Rankin Scale scores would have been an interesting parameter to apply for assessment of presymptom disability between groups; however, this was not a part of standard evaluation on admission.
In our data on health-seeking behaviour, we found that mimics arrived by ambulance less frequently than confirmed strokes. Another study reported a significantly higher mimic rate among walk-in patients.2 Interestingly, we also found that significantly fewer mimics chose to contact the EMS first. This does not seem to be explained by a lack of understanding or knowledge of stroke symptoms; in our study stroke mimics had a higher perception of symptom severity than their stroke counterparts, and there was no significant difference in knowledge of stroke symptoms. The overall perception of symptom severity, however, was surprisingly low (median=30 on a scale from 1 to 100). The low overall perception of symptom severity may represent an untapped potential for stroke awareness campaigns to increase the acknowledgement of stroke symptoms as a healthcare emergency requiring immediate medical evaluation. Also, the continued need for efforts in linking the identification of stroke symptoms to the appropriate behaviour towards receiving timely treatment is highlighted by the fact that only 30.7% of patients with stroke and 15.9% of mimics chose the EMS as their first contact. The fact that a higher number of patients with stroke called the EMS first, may be explained by a higher symptom severity, as the median SSS was lower in the stroke group. Such an association between increasing symptom severity and the use of the EMS has been reported among patients with stroke.38 39 The low proportion of mimics calling the EMS first compared with confirmed strokes may also indicate that some symptoms are more likely to evoke an EMS call. As an example of this, limb weakness, which was associated with a lower likelihood of a mimic diagnosis in this study, has previously been shown to increase the odds of patients with stroke calling the EMS.40
Limitations
Some limitations may apply to this study. First, the observational nature of this study precludes the assessment of cause-effect relationship. Second, no prior sample calculation was done as it was a post-hoc analysis. Hence minor differences in admission pathways or symptoms between patients presenting stroke mimics and those with confirmed diagnosis may have been overlooked. Third, the symptoms are patient-reported only and not verified objectively by medical personnel. Fourth, our confirmed stroke population also comprised clinically classified patients with TIA. Final diagnoses were not confirmed by an adjudication committee and as such there is a possibility that some mimics may have been misclassified as a TIA (or stroke). This, however, reflects clinical practice. Fifth, patients were interviewed only on admission to non-comprehensive stroke units. As such, patients referred from a comprehensive stroke unit may already have knowledge of their diagnosis, introducing a possibility of recall bias with regard to their symptoms.
Conclusion
Stroke mimics comprised a large entity of different diagnoses with peripheral vertigo and migraine being the most common. Clinical factors, including female sex and stroke symptom severity evaluated by SSS score and patient-reported vertigo, were associated with a higher likelihood of stroke mimics. Increasing age and patient-reported limb weakness and difficulty steering were associated with a lower likelihood of stroke mimics. These factors may help clinicians in distinguishing strokes from mimics; however, larger studies with a primary endpoint of investigating mimic-associated factors are needed. Low rates of contact directed at the EMS in patients with stroke-like symptoms warrant continued efforts to increase awareness of stroke as a medical emergency.
Data availability statement
Data are available upon reasonable request. The data that support the findings of this study are available from the corresponding author upon reasonable request and in adherence to Danish legislation.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was approved by the original study was approved by the Capital Region Ethics Committee (no. 2012-58-004) and the Danish Data Protection Agency (no. 2012-58-0004; internal reference: HGH-2017-110, I-Suite no. 06014). Patients (or bystanders) provided written informed consent before study enrollment.
References
Footnotes
Funding This work was supported by TrygFonden (application ID 128669). Christina Kruuse was supported by Novo Nordisk Foundation Borregaard stipend, grant number NNF18OC0031840.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.