Performance of the Medical Priority Dispatch System® in Identifying Patients Requiring Chest Compressions at Overdose Prevention Services: A Retrospective Cohort Study

Abstract

Background and Aims

The Medical Priority Dispatch System (MPDS)® is used to triage 9-1-1 calls according to acuity, with certain coding receiving telecommunicator cardiopulmonary resuscitation (T-CPR) for suspected out-of-hospital cardiac arrest (OHCA). However, this may be challenging for those with drug poisoning emergencies, who may resemble OHCA. We sought to examine the performance of the system to correctly identify cases requiring T-CPR, specifically at overdose prevention services (OPS).

Methods

This retrospective cohort study included patients attended by the provincial emergency medical system (EMS) (May 1, 2019–January 31, 2023). We calculated the diagnostic performance of MPDS® assessment of whether the case required T-CPR instructions against the gold standard of whether the patient was found pulseless on EMS clinician arrival. We compared performance among subgroups, specifically OPS vs other locations and drug poisoning-classified cases vs other case classifications.

Results

Comparing OPS to other locations, the sensitivity of MPDS^® was similar (66.7% vs 62.4%, p = 0.4), with lower specificity (87.3% vs 98.1%, p < 0.01) and positive predictive value (0.3% vs 35.7%, p < 0.01) and higher negative predictive value (99.9% vs 99.4%, p < 0.01). The negative likelihood ratio of MPDS^® was 0.381 at OPS locations, compared with 0.383 at other locations, while the positive likelihood ratio was 5.24, compared with 32.36. In patients with drug poisoning emergencies, compared with other 9-1-1 events, MPDS^® had higher sensitivity (83.6% vs 60.6%, p < 0.01) but lower specificity (77.6% vs 98.9%, p < 0.01) and positive predictive value (10.5% vs 48.5%, p < 0.01), and similar negative predictive value (99.33% vs 99.35%, p = 0.03). The negative likelihood ratio of MPDS® was 0.212 in drug poisoning emergencies compared with 0.398 for all other presentations, and the positive likelihood ratio was 3.73 compared with 57.88.

Discussion and Conclusions

The ability of MPDS^® to correctly identify patients needing telecommunicator cardiopulmonary resuscitation instructions differed between OPS settings and other locations, frequently recommending T-CPR for patients not suffering OHCA at an OPS. Different strategies developed in collaboration with people who use substances are required to better tailor dispatch instructions prior to EMS arrival to avoid delays in life-saving interventions.

Introduction

The unregulated toxic drug poisoning emergency continues to worsen across North America (1), and particularly in British Columbia, Canada (2, 3). British Columbia Emergency Health Services (BCEHS), the provincial emergency medical system (EMS) in BC, responded to a record number of 35,525 patient events involving drug poisoning in 2021 (4). Unfortunately, early data suggests that 2023 may match or even exceed this figure (5). Similar to many other jurisdictions, BCEHS utilizes the Medical Priority Dispatch System (MPDS)^® to triage 9-1-1 calls according to patient acuity. Unconscious patients who have an uncertain breathing status, which often pertains to those with drug poisoning emergencies (6), receive the highest acuity response. In accordance with American Heart Association and International Liaison Committee on Resuscitation recommendations, emergency medical call takers (EMCTs) instruct bystanders to perform cardiopulmonary resuscitation (T-CPR) to unresponsive patients who demonstrate absent or abnormal breathing (7, 8).

While this approach across the general population is associated with a favorable risk-benefit analysis in improving survival from out-of-hospital cardiac arrest (OHCA) with minimal impacts on false positive cases (7–10), the applicability of this approach to patients with known or suspected drug poisoning is unclear. People with opioid-related drug poisoning, a common presentation in some communities, will typically appear unresponsive with absent or abnormal breathing (11). However, these patients are most often not in cardiac arrest and thus are not best treated with T-CPR, but rather should receive assisted ventilations and naloxone (6, 11). This is of particular importance in the era of fentanyl and benzodiazepine contamination of the drug supply, with fentanyl shortening the time available for effective reversal of respiratory depression (12), and benzodiazepines and other sedating agents potentiating the effects of opioids and increasing the risk of substance-related mortality (13, 14).

Optimal strategies for responding to drug poisoning emergencies is particularly relevant with the continued expansion of overdose prevention services (OPS), or overdose prevention centers, where opioid-related drug poisonings occur with a high incidence (15). People experiencing drug poisonings in these facilities will exhibit unresponsiveness with abnormal breathing status, leading to 9-1-1 calls in which EMCTs in many systems will provide T-CPR instructions to trained staff in the facility. This is despite the fact staff at these facilities are specifically trained in the identification of drug poisoning and are capable of providing more appropriate treatments including oxygenation, ventilation, and naloxone (16).

We examined cases from a large EMS and sought to examine the diagnostic performance of MPDS^® in identifying patients requiring T-CPR instructions, in order to compare cases located at an OPS against other locations. We calculated the sensitivity and specificity of the 9-1-1 call assessment of whether the patient required T-CPR instructions (“the test”), against the gold standard indication of whether the patient was ultimately found to be pulseless when EMS clinicians arrived at the scene. Furthermore, we sought to compare results among patients with accidental drug poisoning events and those without, regardless of setting.

Methods

Study Design

We performed an observational retrospective analysis of cases responded to by BCEHS between May 1, 2019 and January 31, 2023. This study was reviewed by the University of British Columbia-affiliated Providence Health Care Research Ethics Board (H22-00430) and registered with Monash University Research Ethics Board (38532). A waiver of informed consent was granted due to the retrospective and de-identified nature of the data. This study is reported in accordance with guidelines for Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) (17).

Study Setting

In 2016 BC became the first province to declare a public health emergency in Canada’s history following a sharp increase in the rate of deaths attributable to substance use (18). Since this time, over 11,000 people have lost their lives to the unregulated drug poisoning emergency (19). In response to the public health emergency, the availability of OPS dramatically increased (15, 16). OPS in BC includes both overdose prevention sites and supervised consumption sites. Overdose prevention sites and supervised consumption sites are similar, in that they provide safe facilities for the use of pre-obtained non-prescription substances without the risk of incarceration (16). However, overdose prevention sites are temporary sites which have been permitted at a provincial level since the declaration of the public health emergency in 2016 and are typically staffed by trained by peer workers, while supervised consumption sites are federally authorized and have a nurse on-site and typically provide access to primary care (16).

BCEHS is the largest EMS in Canada, operates the provincial 9-1-1 medical emergency response dispatch center, and responds to over 500,000 emergency 9-1-1 calls every year (20). BCEHS provides an Anglo-American model of out-of-hospital care, utilizing paramedics and collaborating with municipal fire-rescue department first responders in the delivery of clinical care (20, 21). Following a 9-1-1 call, an EMCT will use a set of questions produced by MPDS^® to categorize and triage the incident, with certain MPDS^® coding triggering a recommendation to provide T-CPR which includes chest compressions and used of automated external defibrillators, if available (Appendix A). Following confirmation of the location of the 9-1-1 call, known entities such as OPS will be automatically populated in the computer aided dispatch (CAD) system, allowing for early identification of incidents at OPS.

Case Identification

Cases of all types were eligible for inclusion in this study if 9-1-1 was contacted and a BCEHS response was initiated between May 1, 2019 and January 31, 2023. Cases were excluded where the location of the incident could not be confirmed.

Data Collection and Variables of Interest

We collected data from 9-1-1 call records and EMS electronic patient care records (ePCRs), with data extracted into a standardized data-collection form. For events where multiple units responded, data from ePCRs was collated chronologically between records.

Variables of interest included patient location (separated into “OPS” versus “non-OPS”), age (categorized into 20-year age groups to protect patient identity), sex, initial vital signs (including Glasgow Coma Scale [GCS], heart rate, respiratory rate, and peripheral oxygen saturation), EMS clinician impression code, and time of day (divided into 6-h periods). We contacted the respective leads from each health authority in BC to identify the location and operational dates of OPS sites. As address fields in 9-1-1 call-logs are free-field and may contain many variations for the same address, with some OPS existing within housing units, addresses were matched initially based on street address and subsequently unit numbers or the presence of keywords such as “OPS,” “consumption site,” or the site name. Where it was unclear whether the emergency event occurred in an OPS, or non-OPS location, the case was excluded.

Statistical Analysis

Statistical Package for Social Sciences (SPSS) (Version 28.0, IBM Australia) was used for collation and analysis of results. Continuous variables are described using means with standardized deviations, with categorical data summarized as frequencies and proportions. We dichotomized cases based on location (OPS versus non-OPS), and compared differences in descriptive statistics using two-sided independent T tests for continuous variables and Chi-squared for categorical values with pairwise comparisons using multiple z-tests of two proportions with Bonferroni corrections where required for outcomes with multiple dependent categorical variables. Where assumptions of Chi-squared tests were violated, we employed Fishers exact tests. Missing data completely at random were managed with pairwise deletion.

Our primary analysis examined the ability of MPDS^® to correctly identify OHCAs requiring T-CPR instructions. Cases were classified as “T-CPR-recommended” (a “positive test”) and “non-T-CPR-recommended” (a “negative test”) groups, based on the assigned MPDS^® code. The “gold standard” measure was the on-scene EMS clinician assessment of OHCA (“positive disease”), shown by EMS clinician provision of chest compressions on arrival to scene. Where OHCA occurred after arrival on scene (EMS-witnessed OHCA), we considered this to be equivalent to “negative disease” and categorized it as such.

We calculated sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (LR+), negative likelihood ratio (LR-), and false positive rate (FPR), with 95% confidence intervals (CI). We calculated diagnostic performance for the overall population, as well as subgroups defined by location (OPS and non-OPS). In addition, we calculated diagnostic performance for patients with a EMS clinician impression code relating to a drug poisoning emergency, and those without. Differences in sensitivity, specificity, PPV, NPV, and FPR were calculated using one-sided z tests for equality of two percentages using independent samples.

Patient and Public Involvement

BC Center for Disease Control (BCCDC) developed the Peer Engagement and Evaluation Project (PEEP), subsequently, the Professionals for the Ethical Engagement of Peers, in response to an identified need from BC Harm Reduction Strategies and Services (BCHRSS) to collaborate with people with lived and living experience of substance use at all stages of the research process (22). The structure and function of PEEP has been previously described elsewhere (22). Peers were consulted during the development of the study protocol and provided input into meaningful outcomes for this research project, and were provided with an opportunity to review the draft manuscript prior to submission for publication. In recognition of the preferred terminology expressed by people who use substances, we use the terms “unregulated toxic drug poisoning emergency,” “unregulated drug supply” and “drug poisoning” in place of the historical terms “overdose crisis,” “opioid crisis,” “fentanyl crisis,” “illegal drug supply,” and “overdose” (23, 24). Peers were engaged in accordance with guidelines for peer engagement best practices and renumerated for their expertise (25).

Results

Patient Characteristics

Between May 1, 2019, and January 31, 2023, BCEHS attended to 1,652,845 9-1-1 events. We excluded 4,199 cases where we were unable to determine the location of the patient at the time of the case, leaving a sample of 1,648,646 for analysis (Figure 1).

Figure 1. Participant flow.

BCEHS = British Columbia Emergency Health Services; OPS = Overdose Prevention Services; OHCA = Out-of-Hospital Cardiac.

In comparison to non-OPS, patients presenting in OPS were more often between 21 and 60 years of age, and more likely to be male (79% male at OPS vs. 51% male at non-OPS, p < 0.01). Cases at an OPS occurred more commonly between 12:01 and 00:00 (p < 0.01) and were more likely to receive T-CPR instructions (13% at OPS vs. 2.9% at non-OPS, p < 0.01), despite having a lower rate of OHCA (0.1% at OPS vs. 1.7% at non-OPS, p < 0.01). We observed statistical differences in multiple initial vital signs between patients presenting at an OPS compared with those not at an OPS in patients not presenting with OHCA, although the clinical relevance of these differences was negligible (Table 1).

Table 1. Location-based patient characteristics.

Characteristic	Non-OPS (N = 1,643,735) n (%)	OPS (N = 4,911) n (%)	p-value
Age			<0.01^a
0–20	102,994 (6.3)	72 (1.5)
21 – 40	312,204 (19.0)	2,215 (45.1)
41 – 60	342,257 (20.9)	1,975 (40.3)
61 – 80	491,170 (29.8)	337 (6.7)
> 81	379,659 (23.1)	9 (0.2)
Unknown	15,451 (0.9)	312 (6.4)_
Sex			<0.01^a
Female	800,074 (48.7)	1,029 (21.0)
Male	840,675 (51.1)	3,858 (78.6)
Non-Binary	917 (0.1)	3 (<0.1)
Unknown	2,069 (0.1)	21 (0.1)
Time			<0.01^a
00:01 – 06:00	242,730 (17.2)	501 (10.2)
06:01 – 12:00	433,690 (24.0)	866 (17.6)
12:01 – 18:00	535,476 (32.5)	2,063 (42.0)
18:01–00:00	431,839 (26.3)	1,481 (30.2)
T-CPR Instructions			<0.01^a
Yes	48,431 (2.9)	626 (12.7)
No	1,595,304 (97.1)	4,285 (87.3)
OHCA			<0.01^b
Yes	27,727 (1.7)	3 (0.1)
No	1,616,008 (98.3)	4,908 (99.9)
Reason for OHCA^c			<0.01^b
Allergy	1 (<0.1)	0 (0)
Asphyxia	666 (<0.1)	0 (0)
Drowning	183 (<0.1)	0 (0)
Drug Poisoning	2,801 (10.1)	2 (66.6)
Electrocution	19 (<0.1)	0 (0)
Other Medical	1,516 (5.4)	0 (0)
Presumed Cardiac	13,141 (47.4)	1 (33.3)
Trauma	916 (3.3)	0 (0)
Not Specified	8,484 (30.6)	0 (0)
	Mean (SD)	Mean (SD)
Glasgow Coma Scale	14.4 (2.2)	12.9 (3.6)	<0.01^d
Respiratory Rate	19.2 (7.1)	17.0 (15.9)	0.64^d
SpO2	95.2 (8.2)	97.0 (4.5)	<0.01^d
Heart Rate	89.7 (23.2)	93.9 (19.6)	<0.01^d

Abbreviations: T-CPR = telephone cardiopulmonary resuscitation; OHCA = out-of-hospital cardiac arrest.

^a Chi-Squared.

^b Fishers exact test.

^c Recorded by BCEHS staff based on clinical impression at time of OHCA.

^d Independent t test.

Diagnostic Performance

Overall Cohort

In the overall sample; among cases with OHCA, 17,291/27,730 were correctly recommended for T-CPR (sensitivity = 62.4%, 95%CI 61.8–62.9%). Among cases without OHCA 1,589,150/1,620,916 were correctly not recommended for T-CPR (specificity = 98.04%, 95%CI 98.02–98.06%). Less than 2% (31,766/1,620,916) were incorrectly recommended for T-CPR (FPR = 1.95%, 95%CI 1.94–1.98). Among cases who were recommended for T-CPR, 17,291/49,057 actually had an OHCA (PPV = 35.3%, 95%CI 34.8–35.7%); and among cases who were not recommended for T-CPR 1,589,150/1,599,589 did not have an OHCA (NPV = 99.35%, 95%CI 99.33–99.36%). The overall LR + of MPDS® was 31.82 (95%CI 31.37–32.27), while the LR- was 0.383 (95%CI 0.379–0.389) (Table 2).

Table 2. 2 × 2 tables by characteristic.

Overall
	OHCA	No OHCA
T-CPR Instructions	17,291	31,766
No T-CPR Instructions	10,439	1,589,150
Non-OPS Location
	OHCA	No OHCA
T-CPR Instructions	17,289	31,142
No T-CPR Instructions	10,438	1,584,866
OPS Location
	OHCA	No OHCA
T-CPR Instructions	2	624
No T-CPR Instructions	1	4,284
Non-Drug Poisoning Etiology
	OHCA	No OHCA
T-CPR Instructions	15,482	16,414
No T-CPR Instructions	10,083	1,536,058
Drug Poisoning Etiology
	OHCA	No OHCA
T-CPR Instructions	1,809	15,352
No T-CPR Instructions	356	53,092

Abbreviations: OHCA = Out of Hospital Cardiac Arrest; T-CPR = Telecommunicator Cardiopulmonary Resuscitation; OPS = Overdose Prevention Service.

OPS versus Non-OPS Locations

The sensitivity of MPDS^® for the detection of patients requiring T-CPR was similar between OPS and non-OPS locations (66.7% vs. 62.4%, p = 0.4), while the specificity was significantly lower at OPS compared with non-OPS locations (87.3% vs. 98.1%, p < 0.01). The FPR was also significantly higher at OPS locations, compared with non-OPS locations (12.7% vs. 1.9% p < 0.01). The PPV of MPDS^® was also lower in an OPS setting compared with non-OPS settings (0.3% vs. 35.7%, p < 0.01), while the NPV was higher (99.9% vs. 99.4%, p < 0.01). The LR + of MPDS^® in an OPS setting was 5.24 (95% CI 2.34–11.71), compared with 32.36 (95%CI 31.89–32.37) in non-OPS settings, while the LR- was 0.381 (95%CI 0.077–1.892) compared with 0.383 (95%CI 0.379–0.389) (Table 3).

Table 3. Test performance of MPDS® by characteristic.

		Location-Based Groups		Impression Code Groups
Measure	Overall	Non-OPS	OPS	Non-Drug Poisoning	Drug Poisoning
Sensitivity (%)	62.35 (61.78–62.93)	62.35 (61.78–62.93)	66.67 (9.43–99.16)	60.56 (59.96–61.16)	83.56 (81.93–85.09)
Specificity (%)	98.04 (98.02–98.06)	98.07 (98.05–98.09)	87.29 (86.32–88.21)	98.94 (98.93–98.96)	77.57 (77.26–77.88)
PPV	35.25 (34.92–35.57)	35.70 (35.37–36.03)	0.32 (0.14–0.72)	48.54 (48.09–48.99)	10.54 (10.32–10.76)
NPV	99.35 (99.33–99.36)	99.35 (99.34–99.36)	99.98 (99.88–100)	99.35 (99.34–99.36)	99.33 (99.27–99.39)
LR+	31.82 (31.37–32.27)	32.36 (31.89–32.27)	5.24 (2.34–11.71)	57.88 (56.84–58.94)	3.73 (3.64–3.81)
LR-	0.383 (0.378–0.389)	0.383 (0.379–0.389)	0.38 (0.08–1.89)	0.398 (0.392–0.404)	0.21 (0.19–0.23)
FPR	1.96 (1.94–1.98)	1.93 (1.91–1.95)	12.71 (11.79–13.68)	1.06 (1.04–1.07)	22.43 (22.12–22.74)

Abbreviations: OPS = overdose prevention service; PPV = positive predictive value; NPV = negative predictive value; LR+ = positive likelihood ratio; LR- = negative likelihood ratio; FPR = false positive rate.

%, (95% Confidence Interval).

Drug Poisoning Emergencies versus Non Drug Poisoning Emergencies

When examining patients experiencing drug poisoning emergencies compared with those who were not experiencing drug poisoning emergencies, the sensitivity of MPDS^® was higher in patients experiencing drug poisoning emergencies (83.6% vs. 60.6%, p < 0.01), while the specificity was significantly lower (77.6% vs. 98.9%, p < 0.01). The FPR of MPDS^® was higher in the drug poisoning emergencies group (22.4% vs. 1.1%, p < 0.01). The PPV was also notably lower in the drug poisoning emergencies cohort (10.5% vs. 48.5%, p < 0.01), although the NPV was comparable between groups (99.33% vs. 99.35%, p = 0.03). The LR + was 3.73 (95%CI 3.64–3.81) in the drug poisoning emergencies cohort compared with 57.88 (95%CI 56.84–58.94) in the while the LR- was 0.211 (95%CI 0.193–0.233) compared with 0.398 (95%CI 0.392–0.404) (Table 3).

Discussion

We examined a cohort of over 1.6 million 9-1-1 responses in a Canadian province to examine the performance of a 9-1-1 call center in correctly classifying patients requiring T-CPR. Although one may hypothesize that OPS locations would have a high incidence of OHCA, 9-1-1 calls from these locations demonstrated an OHCA incidence 17 times lower than all other 9-1-1 calls from non-OPS locations, but had a 4-times higher rate of T-CPR instructions. We found that while 13% (n = 626) of 9-1-1 calls at OPS resulted in a recommendation for T-CPR, overall OHCA incidence at these sites was extremely uncommon (n = 3, 0.1%). This highlights a high incidence of inappropriate T-CPR recommendations and potential missed or delayed opportunities to provide other, likely more effective, therapies, such as ventilatory support and naloxone. Our data suggests that in settings with a high incidence of drug poisoning emergencies and trained professionals or para-professionals, the recommendation for T-CPR in individuals who are unresponsive and breathing abnormally may be delaying the provision of other, more appropriate, life-saving interventions.

Interestingly, despite the more frequent recommendation for chest compressions at OPS, the sensitivity of MPDS^® remained relatively low. In all analysis groups the sensitivity of MPDS^® for the detection of OHCA was poor, but not incongruent with previous research (26–29). The need for improvements in the detection of OHCA during emergency medical call-taking is well recognized and significant effort is underway to improve in this area (30, 31). However, our results complicate this work by identifying that patients at OPS and with drug poisoning emergencies appear to require an alternate pathway for emergency bystander interventions.

The mismatch between the pathology and EMCT-recommended interventions is likely more significant in a population historically marginalized from healthcare (32, 33). Although in the general population T-CPR is associated with low overall risk of complications, such as rib fracture, increasing the odds of these risks through unnecessary T-CPR may increase the risks of pneumonia, chronic pain, and psychological trauma, particularly if repeatedly exposed to unnecessary T-CPR (10, 32–35). Furthermore, trained attendants at OPS sites are asked to intervene in ways contrary to their drug-poisoning-specific training, which may cause distress and hesitation to engage with the emergency health system. While early identification of patients requiring T-CPR instructions due to OHCA is imperative, the importance of avoidance of over-triage and unnecessary T-CPR instructions in patients likely to benefit from more tailored responses deserves equal consideration.

Our findings raise similar concerns as those previously highlighted in the literature regarding the appropriateness of currently recommended protocols for supporting bystanders at the scene of a drug poisoning emergency (6). Specific to OPS settings our results show that the risk of OHCA is extremely low, and that chest compressions are infrequently required. This is similar to the findings of Rowe et al. (36), who found no patients suffering a drug poisoning event at an OPS ultimately required chest compressions. Given that tensions between emergency workers, peer workers, and OPS workers is often cited as a reason for hesitancy in contacting EMS following drug poisoning emergencies (37, 38), our findings support a reconsideration of the application of T-CPR protocols when responding to clearly identified drug poisoning emergencies within these facilities.

This work also has implications for community responses to witnessed drug poisoning emergencies in other settings. There has been a proliferation of healthcare and peer responders within communities with high incidence of drug poisoning emergencies (38, 39). These individuals, or teams, may be equipped with naloxone, oropharyngeal airways, bag-valve masks, and a supply of oxygen (39, 40). Given this, some EMS may find a higher volume of drug poisoning emergencies within the community receive a professional response prior to contact with emergency medical call takers. As drug poisoning emergencies are often misclassified into non-substance related codes under MPDS^® (6), professional and para-professional responders are likely to be advised against the administration of naloxone and active airway management and ventilation in patients who could potentially benefit from these interventions.

Limitations

Our study is inherently limited by the characteristic of retrospective observational research. Specifically, we are reliant on data entry by EMCTs and EMS clinicians that may be incomplete. It is beyond our ability to control for quality and may have impacted our results. A number of cases were excluded as we were unable to ascertain whether the 9-1-1 event occurred in an OPS or non-OPS setting. Additionally, there are also likely unofficial OPS unknown to the researchers that may be included in the non-OPS category. Finally, with only three OHCA occurring at an OPS during the study period we are limited in our analyses in this setting.

Conclusion

The ability of the MPDS^® to correctly identify those needing T-CPR differed between cases at high risk of drug poisoning and other cases, frequently recommending T-CPR for patients not suffering OHCA at an OPS, suggesting that differing strategies are required to determine which resuscitative techniques are required. Strategies to improve the triage and pre-arrival instructions for patients experiencing drug poisoning emergencies, particularly at OPS, should be urgently evaluated in collaboration with people who use substances to avoid delaying life-saving interventions.

Acknowledgments

The authors would like to acknowledge the BCEHS analysts who provided the data for analysis, Randy Slemko and Shirin Sabouri. We would like to acknowledge the efforts of peer advocates and responders, emergency medical call takers and dispatchers, and EMS clinicians in providing exceptional care in the face of the substantial challenges brought on by the COVID-19 pandemic and on-going unregulated toxic drug poisoning emergency. Brian Grunau acknowledges the support of Michael Smith Health Research BC.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Appendix A.

MPDS ® Codes Recommending Hands-Only Chest Compressions

2E1 – Allergic Reaction (Ineffective Breathing).

3D1 – Animal Attack (Arrest).

4D1 – Assault (Arrest).

7D1 – Burns (Person on Fire/Arrest).

8D1 – CO/Hazmat (Unconscious).

9D1 – Cardiac Arrest (Ineffective Breathing).

9D2 – Cardiac Arrest (Expected Death Questionable).

9E1 – Cardiac Arrest (Not Breathing at All).

9E2 – Cardiac Arrest (Uncertain Breathing).

9E3 – Cardiac Arrest (Hanging).

9E4 – Cardiac Arrest (Strangulation).

9E5 – Cardiac Arrest (Suffocation).

11E1 – Choking (Complete Obstruction/Ineffective Breathing).

12D1 – Seizures (Not Breathing).

14E1 – Drowning (Arrest – Out of Water).

15E1 – Electrocution (Not Breathing/Ineffective Breathing).

17D2 – Fall (Arrest).

19D5 – Heart Problems (Just Resuscitated and/or defibrillated).

21D1 – Hemorrhage (Arrest).

23D1 – Overdose/Poisoning (Unconscious).

23E1 – Overdose/Poisoning (Arrest).

25D1 – Psychiatric (Not Alert).

27D1 – Stab/Gunshot/Penetrating Trauma (Arrest).

29D6 – Traffic/Transportation Incidents (Arrest).

30D1 – Traumatic Injuries (Arrest).

31D1 – Unconscious/Fainting (Agonal/Ineffective Breathing).

31E1 – Unconscious/Fainting (Ineffective Breathing).

32D1 – Unknown Problem (Life Status Questionable).

33D1 – Transfer/Interfacility/Palliative Care (Suspected Cardiac or Respiratory Arrest).

33D2 – Transfer/Interfacility/Palliative Care (Just Resuscitated and/or defibrillated.

Appendix B.

2 × 2 Tables for Sensitivity, Specificity, Positive Predictive Value, Negative Predictive Value, Positive Likelihood Ratio and Negative Likelihood Ratio by Type of OPS

Table

Public (n = 1,643,731)
	OHCA	No OHCA
T-CPR Instructions	17,286	31,142
No T-CPR Instructions	10,437	1,584,866
Supervised Consumption Site (n = 1,556)
	OHCA	No OHCA
T-CPR Instructions	0	212
No T-CPR Instructions	1	1,343
Overdose Prevention Site (n = 3,359)
	OHCA	No OHCA
T-CPR Instructions	2	416
No T-CPR Instructions	0	2,941

OHCA = Out of Hospital Cardiac Arrest; T-CPR = Telecommunicator Cardiopulmonary Resuscitation; OPS = Overdose Prevention Service.