Association between mortality and phone-line waiting time for non-urgent medical care: a Danish registry-based cohort study

BACKGROUND AND IMPORTANCE

Telephone calls are often patients' first healthcare service contact, outcomes associated with waiting times are unknown.

OBJECTIVES

Examine the association between waiting time to answer for a medical helpline and 1- and 30-day mortality.

DESIGN, SETTING AND PARTICIPANTS

Registry-based cohort study using phone calls data (January 2014 to December 2018) to the Capital Region of Denmark's medical helpline. The service refers to hospital assessment/treatment, dispatches ambulances, or suggests self-care guidance.

EXPOSURE

Waiting time was grouped into the following time intervals in accordance with political service targets for waiting time in the Capital Region: <30 s, 0:30-2:59, 3-9:59, and ≥10 min.

OUTCOME MEASURES AND ANALYSIS

The association between time intervals and 1- and 30-day mortality per call was calculated using logistic regression with strata defined by age and sex.

MAIN RESULTS

In total, 1 244 252 callers were included, phoning 3 956 243 times, and 78% of calls waited <10 min. Among callers, 30-day mortality was 1% (16 560 deaths). For calls by females aged 85-110 30-day mortality increased with longer waiting time, particularly within the first minute: 9.6% for waiting time <30 s, 10.8% between 30 s and 1 minute and 9.1% between 1 and 2 minutes. For calls by males aged 85-110 30-day mortality was 11.1%, 12.9% and 11.1%, respectively. Additionally, among calls with a Charlson score of 2 or higher, longer waiting times were likewise associated with increased mortality. For calls by females aged 85-110 30-day mortality was 11.6% for waiting time <30 s, 12.9% between 30 s and 1 minute and 11.2% between 1 and 2 minutes. For calls by males aged 85-110 30-day mortality was 12.7%, 14.1% and 12.6%, respectively. Fewer ambulances were dispatched with longer waiting times (4%/2%) with waiting times <30 s and >10 min.

CONCLUSION

Longer waiting times for telephone contact to a medical helpline were associated with increased 1- and 30-day mortality within the first minute, especially among elderly or more comorbid callers.

A retrospective study on machine learning-assisted stroke recognition for medical helpline calls

Advanced stroke treatment is time-dependent and, therefore, relies on recognition by call-takers at prehospital telehealth services to ensure fast hospitalisation. This study aims to develop and assess the potential of machine learning in improving prehospital stroke recognition during medical helpline calls. We used calls from 1 January 2015 to 31 December 2020 in Copenhagen to develop a machine learning-based classification pipeline. Calls from 2021 are used for testing. Calls are first transcribed using an automatic speech recognition model and then categorised as stroke or non-stroke using a text classification model. Call-takers achieve a sensitivity of 52.7% (95% confidence interval 49.2-56.4%) with a positive predictive value (PPV) of 17.1% (15.5-18.6%). The machine learning framework performs significantly better (p < 0.0001) with a sensitivity of 63.0% (62.0-64.1%) and a PPV of 24.9% (24.3-25.5%). Thus, a machine learning framework for recognising stroke in prehospital medical helpline calls may become a supportive tool for call-takers, aiding in early and accurate stroke recognition.

Traumatic cardiac arrest - a nationwide Danish study

BACKGROUND

Cardiac arrest following trauma is a leading cause of death, mandating urgent treatment. This study aimed to investigate and compare the incidence, prognostic factors, and survival between patients suffering from traumatic cardiac arrest (TCA) and non-traumatic cardiac arrest (non-TCA).

METHODS

This cohort study included all patients suffering from out-of-hospital cardiac arrest in Denmark between 2016 and 2021. TCAs were identified in the prehospital medical record and linked to the out-of-hospital cardiac arrest registry. Descriptive and multivariable analyses were performed with 30-day survival as the primary outcome.

RESULTS

A total of 30,215 patients with out-of-hospital cardiac arrests were included. Among those, 984 (3.3%) were TCA. TCA patients were younger and predominantly male (77.5% vs 63.6%, p =  < 0.01) compared to non-TCA patients. Return of spontaneous circulation occurred in 27.3% of cases vs 32.3% in non-TCA patients, p < 0.01, and 30-day survival was 7.3% vs 14.2%, p < 0.01. An initial shockable rhythm was associated with increased survival (aOR = 11.45, 95% CI [6.24 - 21.24] in TCA patients. When comparing TCA with non-TCA other trauma and penetrating trauma were associated with lower survival (aOR: 0.2, 95% CI [0.02-0.54] and aOR: 0.1, 95% CI [0.03 - 0.31], respectively. Non-TCA was associated with an aOR: 3.47, 95% CI [2.53 - 4,91].

CONCLUSION

Survival from TCA is lower than in non-TCA. TCA has different predictors of outcome compared to non-TCA, illustrating the differences regarding the aetiologies of cardiac arrest. Presenting with an initial shockable cardiac rhythm might be associated with a favourable outcome in TCA.

Geographical Association Between Basic Life Support Courses and Bystander Cardiopulmonary Resuscitation and Survival from OHCA in Denmark

Introduction

Annually, approximately 4% of the entire adult population of Denmark participate in certified basic life support (BLS) courses. It is still unknown whether increases in BLS course participation in a geographical area increase bystander cardiopulmonary resuscitation (CPR) or survival from out-of-hospital cardiac arrest (OHCA). The aim of the study was to examine the geographical association between BLS course participation, bystander CPR, and 30-day survival from OHCA.

Methods

This nationwide register-based cohort study includes all OHCAs from the Danish Cardiac Arrest Register. Data concerning BLS course participation were supplied by the major Danish BLS course providers. A total of 704,234 individuals with BLS course certificates and 15,097 OHCA were included from the period 2016-2019. Associations were examined using logistic regression and Bayesian conditional autoregressive analyses conducted at municipality level.

Results

A 5% increase in BLS course certificates at municipality level was significantly associated with an increased likelihood of bystander CPR prior to ambulance arrival with an adjusted odds ratio (OR) of 1.34 (credible intervals: 1.02;1.76). The same trends were observed for OHCAs in out-of-office hours (4pm-08am) with a significant OR of 1.43 (credible intervals: 1.09;1.89). Local clusters with low rate of BLS course participation and bystander CPR were identified.

Conclusion

This study found a positive effect of mass education in BLS on bystander CPR rates. Even a 5% increase in BLS course participation at municipal level significantly increased the likelihood of bystander CPR. The effect was even more profound in out-of-office hours with an increase in bystander CPR rate at OHCA.

"I just haven't experienced anything like this before": A qualitative exploration of callers' interpretation of experienced conditions in telephone consultations preceding a myocardial infarction

OBJECTIVES

Callers with myocardial infarction presenting atypical symptoms in telephone consultations when calling out-of-hours medical services risk misrecognition. We investigated characteristics in callers' interpretation of experienced conditions through communication with call-takers.

METHODS

Recording of calls resulting in not having an ambulance dispatched for 21 callers who contacted a non-emergency medical helpline, Copenhagen (Denmark), up to one week before they were diagnosed with myocardial infarction. Qualitative content analysis was applied.

RESULTS

Awareness of illness, remedial actions and previous experiences contributed to callers' interpretation of the experienced condition. Unclear symptoms resulted in callers reacting to their interpretation by being unsure and worried. Negotiation of the interpretation was seen when callers tested the call-taker's interpretation of the condition and when either caller or call-taker suggested: "wait and see".

CONCLUSION

Callers sought to interpret the experienced conditions but faced challenges when the conditions appeared unclear and did not correspond to the health system's understanding of symptoms associated with myocardial infarction. It affected the communicative interaction with the call-taker and influenced the call-taker's choice of response.

PRACTICE IMPLICATIONS

Call-takers, as part of the decision-making process, could ask further questions about the caller's insecurity and worry. It might facilitate faster recognition of conditions warranting hospital referral.

Training in Basic Life Support and Bystander-Performed Cardiopulmonary Resuscitation and Survival in Out-of-Hospital Cardiac Arrests in Denmark, 2005 to 2019

IMPORTANCE

Strategies to improve survival from out-of-hospital cardiac arrest (OHCA) include mass education of laypersons with no official duty to respond to OHCA. In Denmark, basic life support (BLS) course attendance has been mandated by law in October 2006 for obtaining a driver's license for all vehicles and in vocational education programs.

OBJECTIVES

To examine the association between yearly BLS course participation rate and bystander cardiopulmonary resuscitation (CPR) and 30-day survival from OHCA and to examine if bystander CPR rate acted as a mediator on the association between mass education of laypersons in BLS and survival from OHCA.

DESIGN, SETTING, AND PARTICIPANTS

This cohort study included outcomes for all OHCA incidents from the Danish Cardiac Arrest Register between 2005 and 2019. Data concerning BLS course participation were supplied by the major Danish BLS course providers.

MAIN OUTCOMES AND MEASURES

The main outcome was 30-day survival of patients who experienced OHCA. Logistic regression analysis was used to examine the association between BLS training rate, bystander CPR rate, and survival, and a bayesian mediation analysis was conducted to examine mediation.

RESULTS

A total of 51 057 OHCA incidents and 2 717 933 course certificates were included. The study showed that the annual 30-day survival from OHCA increased by 14% (odds ratio [OR], 1.14; 95% CI, 1.10-1.18; P < .001) when BLS course participation rate increased by 5% in analysis adjusted for initial rhythm, automatic external defibrillator use, and mean age. An average mediated proportion of 0.39 (95% QBCI, 0.049-0.818; P = .01). In other words, the last result indicated that 39% of the association between mass educating laypersons in BLS and survival was mediated through an increased bystander CPR rate.

CONCLUSIONS AND RELEVANCE

In this cohort study of Danish BLS course participation and survival, a positive association was found between annual rate of mass education in BLS and 30-day survival from OHCA. The association of BLS course participation rate on 30-day survival was mediated by the bystander CPR rate; approximately 60% of the association of BLS course participation rate on 30-day survival was based on factors other than increased CPR rates.

When the machine is wrong. Characteristics of true and false predictions of Out-of-Hospital Cardiac arrests in emergency calls using a machine-learning model

BACKGROUND

A machine-learning model trained to recognize emergency calls regarding Out-of-Hospital Cardiac Arrest (OHCA) was tested in clinical practice at Copenhagen Emergency Medical Services (EMS) from September 2018 to December 2019. We aimed to investigate emergency call characteristics where the machine-learning model failed to recognize OHCA or misinterpreted a call as being OHCA.

METHODS

All emergency calls were linked to the dispatch database and verified OHCAs were identified by linkage to the Danish Cardiac Arrest Registry. Calls with either false negative or false positive predictions of OHCA were evaluated by trained auditors. Descriptive analyses were performed with absolute numbers and percentages reported.

RESULTS

The machine-learning model processed 169,236 calls to Copenhagen EMS and suspected 5,811 (3.4%) of the calls as OHCA, resulting in 84.5% sensitivity and 97.1% specificity. Among OHCAs not recognised by machine-learning model, a condition completely different from OHCA was presented by caller in 31% of the cases. In 28% of unrecognised calls, patient was reported breathing normally, and language barriers were identified in 23% of the cases. Among falsely suspected OHCA, the patient was reported unconscious in 28% of the cases, and in 13% of the false positive cases the machine-learning model interpreted calls regarding dead patients with irreversible signs of death as OHCA.

CONCLUSION

Continuous optimization of the language model is needed to improve the prediction of OHCA and thereby improve sensitivity and specificity of the machine-learning model on recognising OHCA in emergency telephone calls.

Patient characteristics and dispatch responses of urinary tract infections in a prehospital setting in Copenhagen, Denmark: a retrospective cohort study

BACKGROUND

Urinary tract infection (UTI) is particularly common in young women and the elderly. The Emergency Medical Services (EMS) in Copenhagen, Denmark can be reached by calling either of two dedicated telephone lines: 1-1-2 in case of an emergency and 1813 during general practitioner's (GP) out-of-office hours (OOH). This study investigated characteristics of patients with symptoms of UTI calling the Copenhagen EMS and the response they received.

METHODS

A retrospective observational cohort study was conducted in which 7.5 years of telephone data on UTI from the EMS in Copenhagen were analyzed. Descriptive statistics and multinomial logistic regression were used to analyze patient characteristics, the timing of the incident and response. Patients' age and gender were assessed and the use of urinary catheters, the timing of the incident, and the impact on the response were evaluated.

RESULTS

A total of 278.961 calls were included (78% female, mean age 47), with an average of 120 patients with UTI symptoms calling each day. Most people contacted the 1813-medical helpline (98%) and of those, the majority were referred to the emergency department (ED)(37%). Patients were more likely to be referred to the ED during the weekend compared to a weekday and less likely during OOH compared to in-office hours (IH). Patients with a urinary catheter were more likely to receive specialized care referred to as 'other'. For the smaller proportion of patients calling 1-1-2, most people got a B (urgent) response (1.5%). The most likely response to be given was an A (emergency) or F (non-emergency) response during OOH compared to IH and on weekends compared to weekdays. Patients with a urinary catheter were more likely to receive a D (unmonitored transport) response.

CONCLUSIONS

Since 2015, there was a decrease in 1813 antibiotic prescription rates and a subsequent increase in referral to the ED of UTI patients. Patients were referred less to the ED during OOH as they were likely to be sent to their GP the next day. During the weekend, patients were referred more to the ED for the likely reason that their GP is closed.

Characteristics of low acuity prehospital emergency patients with 48-h mortality, an observational cohort study

BACKGROUND

Every year an emergency medical technician or paramedic treats and transports up to several hundred patients. Only some patients are acutely seriously ill, and a few of these show only discrete signs and symptoms of their condition. This study aims to describe patients who died within 48 h of being admitted non-emergently to hospital by ambulance, examine the extent to which critically ill patients are recognized prehospitally, and identify clinical warning signs that might be frequently overlooked.

METHOD

Registry based follow-up study on patients receiving an ambulance from the Copenhagen EMS in 2018. Data was included regarding the dispatch of the ambulance from the emergency services disposition system, ICD-10 hospital admission diagnoses from the National Patient Register, 48-h mortality from the Central Person Register and assessment and treatment in the ambulance by reviewing the electronic pre-hospital patient record.

RESULTS

In 2018 2279 patients died within 48 h after contact with the EMS, 435 cases met inclusion criteria. The patients' median age was 83 years (IQR 75-90), and 374 (86.0%) had one or more underlying serious medical conditions. A triage category based on vitals and presentation was not assigned by the EMS in 286 (68.9%) cases, of which 38 (13.3%) would meet red and 126 (44.1%) orange criteria. For 409 (94.0%) patients, it was estimated that death within 48 h could not have been avoided prehospitally, and for 26 (6.0%) patients it was uncertain. We found 27 patients with acute aortic syndrome as admission diagnosis, of these nine (33.3%) had not been admitted urgently to a hospital with vascular surgery specialty.

CONCLUSIONS

It was estimated that death within 48 h could generally not be avoided prehospitally. The patients' median age was 83 years, and they often had serious comorbidity. Patients whose vital parameters met red or orange triage criteria were to a lesser degree triaged prehospitally, compared to patients in the yellow or green categories. Patients with acute aortic syndrome were not recognized by EMS 33.3% of the time.

Opioid administered prior to onset of myocardial infarction is associated with atypical symptoms of acute coronary ischemia in emergency calls

Background

Acute identification and treatment of myocardial infarction (MI) is crucial for survival. MI symptom presentation is typically characterized by chest pain. Regular use of opioids to relief pain of chronic diseases is common. However, it has not been investigated whether opioid administered prior to onset of MI is associated more with atypical presentation of ischemia compared to non-opioid users.

Purpose

To investigate if opioid therapy is associated with atypical symptoms of MI.

Methods

The study is registry-based. MI is defined as patients with the ICD-10 discharge diagnoses I20.0 and I21 as primary diagnosis &lt;24 hours after a call to the Capital Emergency Medical Services, including the Emergency Number (1–1-2) and Out-of-hours Service (OOHS). MI patients with opioid treatment prior to onset of MI are compared with a control group of MI patients without opioid treatment. The primary symptom of MI is registered in calls to the Capital Emergency Medical Services, from 2014 to 2018. Opioid treatment includes opioid prescribed and picked up &lt;30 days prior to onset of MI.

Results

In total, 866 MI patients with opioid treatment and 10,061 MI patients without opioid treatment were included. The median age was 72.0 vs 67.8 years, 50.5% vs 34.5% were women, 61.0% vs 54.7% called the Emergency Number (1–1-2), ischemic heart disease was present in 48.5% vs 34.3% and type 2 diabetes in 23.1% vs. 16.2%. Atypical symptoms of MI were associated with opioid treatment and most prevalent in calls to OOHS, in women and by higher age. Acute symptoms of MI are visualized in Figure 1 and 2.

Conclusion

Opioid treatment prior to onset of MI was associated with increased atypical symptoms of MI, especially dyspnea, and increased in calls to the OOHS, in women and by higher age. This might challenge the clinician in early diagnose of MI in patients with opioid treatment and thereby delay acute lifesaving treatment.

Funding Acknowledgement

Type of funding sources: Foundation. Main funding source(s): The Danish Heart Foundation

Impact of integrating out-of-hours services into Emergency Medical Services Copenhagen: a descriptive study of transformational years

Background

Many emergency medical services and out-of-hours systems are facing an increasing demand for primary, ambulance, and secondary care services caused by population aging and a higher prevalence of long-term and complex conditions. In order to ensure safety and efficiency for future demands, many systems are changing their dispersed healthcare services towards a more integrated care system. Therefore, an evaluation of the production and performance over time of such a unified system is desirable.

Methods

This retrospective quantitative study was performed with dispatch and financial accounting data of Copenhagen Emergency Medical Services for the period 2010–2019. Copenhagen Emergency Medical Services operates both an emergency number and a medical helpline for out-of-hours services. The number of calls to the emergency number, the centralized out-of-hours medical helpline, the number of dispatches, and the annual expenditure of the system are described for both the periods before and after the major reforms. Production of the emergency number and the centralized medical helpline were analyzed separately.

Results

The average number of dispatches increased from 328 per 10,000 inhabitants in 2010 to 361 per 10,000 inhabitants in 2019. The newly initiated medical helpline received 533 calls per 10,000 inhabitants in its first year and 5 years later 548 calls per 10,000 inhabitants. A cost increase of 10% was observed in the first year after the reforms, but it decreased again to 8% in the following year.

Conclusions

There is a population demand for a centralized telephone access point for (semi-)emergency medical services. A more integrated EMS system is promising for a sustainable healthcare provision for a growing population with complex healthcare demands and multi-morbidities.

The National Danish Cardiac Arrest Registry for Out-of-Hospital Cardiac Arrest - A Registry in Transformation

Aim of the Database

The aim of the Danish Cardiac Arrest Registry is to monitor the quality of prehospital cardiac arrest treatment, evaluate initiatives regarding prehospital treatment of cardiac arrest, and facilitate research.

Study Population

All patients with prehospital cardiac arrest in Denmark treated by the emergency medical services in whom resuscitation or defibrillation has been attempted.

Main Variables

The Danish Cardiac Arrest Register records descriptive and qualitative variables as outlined in the “Utstein” template for reporting out-of-hospital-cardiac arrest. Main variables include whether the case was witnessed, whether the cardiac arrest was electrocardiographically monitored, the timing of cardiopulmonary resuscitation, and the timing of the first analysis of the cardiac rhythm. The outcome measures are the status of the patient at handover to the hospital, return of spontaneous circulation, and 30-day survival after event.

Database Status

The Danish Cardiac Arrest Registry was established in June 2001, and all Danish emergency medical services are reporting to the database.

Conclusion

The Danish Cardiac Arrest Registry is among the oldest Danish national clinical registries, with a high quality of clinical data and coverage. This registry provides the prerequisite for all research on out-of-hospital cardiac arrest research in Denmark and is essential for monitoring and improving the quality of care for patients suffering from out-of-hospital cardiac arrest.

"Like a rainy weather inside of me": Qualitative content analysis of telephone consultations concerning back pain preceding out-of-hospital cardiac arrest

INTRODUCTION

Cardiac arrest patients presenting with back pain are at risk of not receiving the appropriate help when calling emergency medical services. In telephone consultations regarding patients with back pain preceding an out-of-hospital cardiac arrest, we investigated how communication between caller and call-taker influenced the call-taker's interpretation of back pain descriptions and decision-making about choice of response.

METHOD

The study was conducted using 20 recorded phone calls from 17 patients who contacted the Copenhagen Emergency Medical Services (Denmark) reporting back pain up to 24 hours before an out-of-hospital cardiac arrest. Qualitative content analysis was applied.

RESULTS

Two main categories emerged: (1) reasons, including subcategories: reported conditions, descriptions of conditions, patient's interpretation of condition and patient's own remedial actions; and (2) considerations, including subcategories: assessment of the severity, call-taker's interpretation of the condition, arguments for chosen response and conditions not facilitating further communication by the call-taker.

CONCLUSION

In telephone consultations regarding patients with back pain preceding an out-of-hospital cardiac arrest the communication was influenced by the communicative preconditions of the call-taker. Communication in consultations where ambulances were not dispatched was characterized by complex descriptions of symptoms not easily fitting into the health system's interpretations of conditions warranting an urgent response.

The “unclear problem” category: an analysis of its patient and dispatch characteristics and its trend over time

Background

An effective emergency medical dispatch process is vital to provide appropriate prehospital care to patients. It increases patient safety and ensures the sustainable use of medical resources. Although Copenhagen has a sophisticated emergency medical services (EMS) system with a significant focus on public welfare, more than 10% of emergency cases are still being categorized as an "unclear problem category" (UPC) and are thus not categorized as "symptom-specific". Therefore, the objective of this research is to gain a better understanding of the patient and dispatch characteristics of emergency cases categorized as "unclear".

Methods

This register-based study based on medical emergency cases data describes patient and dispatch characteristics of emergency cases categorized as “unclear” through the use of numbers and proportions. Moreover, these cases were compared to non UPC cases. Use of UPC was stratified by month to determine the impact of alerting medical dispatchers to reduce its use.

Results

From 296,398 included cases UPC accounted for 11.4% of the cases. The median age of those triaged with the UPC was 66 years vs 58 years for individuals triaged with other symptom-specific categories.

Moreover, after having been triaged with the UPC, 9,661 (34.7%) of the dispatched EMS vehicles ended up being cancelled. Sensitizing medical dispatchers about the use of the UPC likely contributed to the decreased use of the UPC over time.

Conclusion

The UPC has different dispatch characteristics than the symptom-specific categories, with potential negative effects on the medical dispatch process. Moreover, the median age of individuals triaged with the UPC is higher than those triaged with symptom-specific categories. Nonetheless, the use of the UPC decreased throughout the study period after the medical dispatchers were alerted about the implications of its use.

Inverse Association Between Bystander Use of Audiovisual Feedback From an Automated External Defibrillator and Return of Spontaneous Circulation

Background

Treatment with an automated external defibrillator (AED) improves outcome in out‐of‐hospital cardiac arrest (OHCA). Audiovisual feedback from an AED may assist bystanders achieve higher quality cardiopulmonary resuscitation. However, the association between audiovisual feedback and clinical outcomes is not well assessed in real‐life OHCA. The aim of this study was to assess the association between audiovisual feedback from an AED used in bystander resuscitation with rates of return of spontaneous circulation (ROSC) and 30‐day survival in a real‐life cohort of patients with OHCA.

Methods and Results

We included 325 patients treated with bystander AED use before arrival of emergency medical services during 2016 to 2019 from the Capital Region of Denmark. Patients were divided into a “feedback” and a “nonfeedback” group, depending on presence of audiovisual feedback from the AED. Audiovisual feedback was defined as voice prompts with continuous feedback to ongoing resuscitation. Rates of ROSC upon hospital admission and 30‐day survival were assessed, and univariate and multivariable models were applied to decide the association to audiovisual feedback. Multivariable models were adjusted for sex, age, primary heart rhythm, and location of OHCA. A total of 155 (48%) patients had a bystander AED applied with audiovisual feedback and 170 (52%) without audiovisual feedback. A lower rate of ROSC was found in the feedback group compared with the nonfeedback group (33% [n=51] versus 45% [n=76]; P =0.03). No association was observed between AV feedback and 30‐day survival (feedback=27% [n=42] and nonfeedback=31% [n=53]; P =0.49). In the unadjusted logistic regression model, audiovisual feedback was associated with a decreased chance of ROSC (odds ratio, 0.61; 95% CI, 0.38–0.95; P =0.03), which remained significant after adjusted analysis (odds ratio, 0.53; 95% CI, 0.29–0.97; P =0.04), whereas we found no significant association between audiovisual feedback and 30‐day survival in the unadjusted and adjusted analyses.

Conclusions

Audiovisual feedback from an AED used by bystanders was associated with a lower chance of ROSC at hospital admission, but we found no significant difference in 30‐day survival. Focus on early and correct bystander cardiopulmonary resuscitation and AED use remain key for OHCA survival.

Psychiatric Admissions, Referrals, and Suicidal Behavior Before and During the COVID-19 Pandemic in Denmark: A Time-Trend Study

OBJECTIVE

To assess the patterns in psychiatric admissions, referrals, and suicidal behavior before and during the COVID-19 pandemic.

METHODS

This study utilized health records from hospitals and Emergency Medical Services (EMS) covering 46% of the Danish population (n = 2,693,924). In a time-trend study, we compared the number of psychiatric in-patients, referrals to mental health services and suicidal behavior in years prior to the COVID-19 pandemic to levels during the first lockdown (March 11 - May 17, 2020), inter-lockdown period (May 18 - December 15, 2020), and second lockdown (December 16, 2020 - February 28, 2021).

RESULTS

During the pandemic, the rate of psychiatric in-patients declined compared to pre-pandemic levels (RR = 0.95, 95% CI = 0.94 - 0.96, p < 0.01), with the largest decrease of 19% observed three weeks into the first lockdown. Referrals to mental health services were not significantly different (RR = 1.01, 95% CI = 0.92 - 1.10, p = 0.91) during the pandemic; neither was suicidal behavior among hospital contacts (RR = 1.04, 95% CI = 0.94 - 1.14, p = 0.48) nor EMS contacts (RR = 1.08, 95% CI = 1.00 - 1.18, p = 0.06). Similar trends were observed across nearly all age groups, sexes, and types of mental disorders examined. In the age group <18, an increase in the rate of psychiatric in-patients (RR = 1.11, 95% CI = 1.07 - 1.15, p < 0.01) was observed during the pandemic; however, this did not exceed the pre-pandemic, upwards trend in psychiatric hospitalizations in the age group <18 (p = 0.78).

CONCLUSION

The COVID-19 pandemic has been associated with a decrease in psychiatric hospitalizations, while no significant change was observed in referrals to mental health services and suicidal behavior. Psychiatric hospitalizations among children and adolescents increased during the pandemic; however, this appears to be a continuation of a pre-pandemic trend.

Socio-demographic characteristics of basic life support course participants in Denmark

BACKGROUND

Bystander-initiated basic life support (BLS) plays an important role in improving survival after out-of-hospital cardiac arrest. In 2009, laws mandating BLS course participation when acquiring a driver's licence were implemented in Denmark. The aim of this study was to characterise Danish BLS course participants.

METHODS

This study is a Danish, registry-based, follow-up study that examined all Danish BLS course participants from 2016 to 2019. Data concerning BLS course participation were supplied by the major Danish BLS course providers. Socio-economic and healthcare data on all Danish inhabitants were assessed using national registers from Statistics Denmark.

RESULTS

Between January 1, 2016, and January 1, 2020, 3.6% of the entire adult population of Denmark attended certified BLS courses annually. Since the implementation of a law mandating BLS course participation when acquiring a driver licence in 2009, approximately 44% of the adult population has participated in a BLS course. BLS course participants were commonly younger and healthier than the general population (mean 31.3 years old vs. 51.3 years old, P < 0.001). Furthermore, law-mandated BLS course participants had a lower disposable income (adjusted OR: 0.23; 95% CI: 0.23-0.23; P < 0.001) and were more likely to live in rural areas (adjusted OR: 0.57; 95% CI: 0.57-0.58; P < 0.001).

CONCLUSION

In Denmark, 3.6% of the entire adult population attend certified courses annually. BLS participants are commonly male, younger, healthier, less likely to have small children in the household, and more likely to live in rural areas. Law-mandated BLS course participation prior to acquiring a driver's licence has been successful in reaching segments of the society that are known to have limited participation.

On Assessing Trustworthy AI in Healthcare. Machine Learning as a Supportive Tool to Recognize Cardiac Arrest in Emergency Calls

Artificial Intelligence (AI) has the potential to greatly improve the delivery of healthcare and other services that advance population health and wellbeing. However, the use of AI in healthcare also brings potential risks that may cause unintended harm. To guide future developments in AI, the High-Level Expert Group on AI set up by the European Commission (EC), recently published ethics guidelines for what it terms “trustworthy” AI. These guidelines are aimed at a variety of stakeholders, especially guiding practitioners toward more ethical and more robust applications of AI. In line with efforts of the EC, AI ethics scholarship focuses increasingly on converting abstract principles into actionable recommendations. However, the interpretation, relevance, and implementation of trustworthy AI depend on the domain and the context in which the AI system is used. The main contribution of this paper is to demonstrate how to use the general AI HLEG trustworthy AI guidelines in practice in the healthcare domain. To this end, we present a best practice of assessing the use of machine learning as a supportive tool to recognize cardiac arrest in emergency calls. The AI system under assessment is currently in use in the city of Copenhagen in Denmark. The assessment is accomplished by an independent team composed of philosophers, policy makers, social scientists, technical, legal, and medical experts. By leveraging an interdisciplinary team, we aim to expose the complex trade-offs and the necessity for such thorough human review when tackling socio-technical applications of AI in healthcare. For the assessment, we use a process to assess trustworthy AI, called 1Z-Inspection® to identify specific challenges and potential ethical trade-offs when we consider AI in practice.

Effect of Machine Learning on Dispatcher Recognition of Out-of-Hospital Cardiac Arrest During Calls to Emergency Medical Services: A Randomized Clinical Trial

IMPORTANCE

Emergency medical dispatchers fail to identify approximately 25% of cases of out-of-hospital cardiac arrest (OHCA), resulting in lost opportunities to save lives by initiating cardiopulmonary resuscitation.

OBJECTIVE

To examine how a machine learning model trained to identify OHCA and alert dispatchers during emergency calls affected OHCA recognition and response.

DESIGN, SETTING, AND PARTICIPANTS

This double-masked, 2-group, randomized clinical trial analyzed all calls to emergency number 112 (equivalent to 911) in Denmark. Calls were processed by a machine learning model using speech recognition software. The machine learning model assessed ongoing calls, and calls in which the model identified OHCA were randomized. The trial was performed at Copenhagen Emergency Medical Services, Denmark, between September 1, 2018, and December 31, 2019.

INTERVENTION

Dispatchers in the intervention group were alerted when the machine learning model identified out-of-hospital cardiac arrest, and those in the control group followed normal protocols without alert.

MAIN OUTCOMES AND MEASURES

The primary end point was the rate of dispatcher recognition of subsequently confirmed OHCA.

RESULTS

A total of 169 049 emergency calls were examined, of which the machine learning model identified 5242 as suspected OHCA. Calls were randomized to control (2661 [50.8%]) or intervention (2581 [49.2%]) groups. Of these, 336 (12.6%) and 318 (12.3%), respectively, had confirmed OHCA. The mean (SD) age among of these 654 patients was 70 (16.1) years, and 419 of 627 patients (67.8%) with known gender were men. Dispatchers in the intervention group recognized 296 confirmed OHCA cases (93.1%) with machine learning assistance compared with 304 confirmed OHCA cases (90.5%) using standard protocols without machine learning assistance (P = .15). Machine learning alerts alone had a significantly higher sensitivity than dispatchers without alerts for confirmed OHCA (85.0% vs 77.5%; P < .001) but lower specificity (97.4% vs 99.6%; P < .001) and positive predictive value (17.8% vs 55.8%; P < .001).

CONCLUSIONS AND RELEVANCE

This randomized clinical trial did not find any significant improvement in dispatchers' ability to recognize cardiac arrest when supported by machine learning even though artificial intelligence did surpass human recognition.

TRIAL REGISTRATION

ClinicalTrials.gov Identifier: NCT04219306.

Strategies to Handle Increased Demand in the COVID-19 Crisis: A Coronavirus EMS Support Track and a Web-Based Self-Triage System

BACKGROUND

In emergencies, such as the COVID-19 pandemic, there is an increased need for contact with emergency medical services (EMS), and call volume might surpass capacity. The Copenhagen EMS operates two telephone line the 1-1-2 emergency number and the 1813 medical helpline. A separate coronavirus support track was implemented on the 1813 medical helpline and a web-based self-triage (web triage) system was created to reduce non-emergency call volume. The aim of this paper is to present call volume and the two measures implemented to handle the increased call volume to the Copenhagen EMS.

METHODS

This is a cross sectional observational study. Call volume and queue time is presented in the first month of the COVID-19 pandemic (27th of February 2020 to 27^th of march) and compared to the equivalent month from the year before (2019). Descriptive statistics are conducted on call volumes and queue times and spearman's rank correlation test are performed to test correlation between web triage and call volume.

RESULTS

Total EMS call volume increase by 23.3% between 2019 and 2020 (92.670 vs. 114,250). The 1-1-2 emergency line total call volume increase by 4.4% (8,4942 vs. 8,870) and the 1813 medical helpline increased by 25.1% (84.176 vs. 105.380). The coronavirus support track handled 21,063 calls. The 1813 medical helpline queue time was a mean of 02 minutes and 23 seconds (CI: 2.22-2.25) in 2019 and 12 minutes and 2 seconds (CI 11:55-12:09) in 2020 (P < 0.001). The web triage was used 10,894 times. No correlation between call volume and web triage usage was seen.

CONCLUSIONS

In the first month of the ongoing COVID-19 pandemic a significant increase in call volume was observed in the 1813 medical helpline compared to 2019. A large number of calls were handled by the additional coronavirus track and diverted away from the regular tracks of the 1813 medical helpline. This likely contributed to mitigating increased call volumes and queue times. The web triage was widely used but no significant correlation was seen with 1813 medical helpline call volume. Other EMS organizations can learn from this to enhance capacity in a future epidemics.

Satisfaction of 30 402 callers to a medical helpline of the Emergency Medical Services Copenhagen: a retrospective cohort study

OBJECTIVES

To keep healthcare systems sustainable for future demands, many countries are developing a centralised telephone line for out-of-hours primary care services. To increase the quality of such services, more information is needed on factors that influence caller satisfaction. The aim of this study was to identify demographic and call-related characteristics that are associated with the patient satisfaction of callers to a medical helpline in Denmark.

DESIGN

Retrospective cohort study on patient registry data and questionnaire results.

SETTING

Non-emergency medical helpline in the Capital Region of Denmark.

PARTICIPANTS

A random sample of 30 402 callers to the medical helpline between May 2016 and May 2018.

PRIMARY AND SECONDARY OUTCOME MEASURES

Responses of a satisfaction questionnaire were linked to demographic and call-related dispatch data. Associations between the characteristics were analysed with multivariable logistic regression analysis with satisfaction as the dependent variable. A subgroup analysis was performed on callers for children aged between 0 and 4 years.

RESULTS

Of the 30 402 analysed callers, 73.0% were satisfied with the medical helpline. Satisfaction was associated with calling for a somatic injury (OR: 1.96, 95% CI: 1.72 to 2.23), receiving a face-to-face consultation (OR: 2.27, 95% CI: 2.04 to 2.50) and a waiting time less than 10 min (OR: 1.82, 95% CI: 1.56 to 2.08). Callers for a 0-year to 4-year-old patient were more likely to be satisfied when they called for a somatic illness or received a telephone consultation, compared with the rest of the population (p<0.0001).

CONCLUSION

Callers were in general satisfied with the medical helpline. Satisfaction was associated with reason for encounter, triage response and waiting time. People calling for 0-year to 4-year-old patients were, compared with the rest of the population, more frequently satisfied when they called for a somatic illness or received a telephone consultation.

Machine learning as a supportive tool to recognize cardiac arrest in emergency calls

BACKGROUND

Emergency medical dispatchers fail to identify approximately 25% of cases of out of hospital cardiac arrest, thus lose the opportunity to provide the caller instructions in cardiopulmonary resuscitation. We examined whether a machine learning framework could recognize out-of-hospital cardiac arrest from audio files of calls to the emergency medical dispatch center.

METHODS

For all incidents responded to by Emergency Medical Dispatch Center Copenhagen in 2014, the associated call was retrieved. A machine learning framework was trained to recognize cardiac arrest from the recorded calls. Sensitivity, specificity, and positive predictive value for recognizing out-of-hospital cardiac arrest were calculated. The performance of the machine learning framework was compared to the actual recognition and time-to-recognition of cardiac arrest by medical dispatchers.

RESULTS

We examined 108,607 emergency calls, of which 918 (0.8%) were out-of-hospital cardiac arrest calls eligible for analysis. Compared with medical dispatchers, the machine learning framework had a significantly higher sensitivity (72.5% vs. 84.1%, p < 0.001) with lower specificity (98.8% vs. 97.3%, p < 0.001). The machine learning framework had a lower positive predictive value than dispatchers (20.9% vs. 33.0%, p < 0.001). Time-to-recognition was significantly shorter for the machine learning framework compared to the dispatchers (median 44 seconds vs. 54 s, p < 0.001).

CONCLUSIONS

A machine learning framework performed better than emergency medical dispatchers for identifying out-of-hospital cardiac arrest in emergency phone calls. Machine learning may play an important role as a decision support tool for emergency medical dispatchers.