Descriptive data
As described in Figure 1, a total of 38 654 records were identified. Of these, 25 917 had a medical problem, with the remainder being either administrative assignments (e.g. stand-by or test records) or misdirected calls (e.g. police/fire calls or accidental calls). Of these, 16 464 were determined by a dispatch nurse to require an ambulance, with the remainder of patients referred to alternate forms of care (e.g. alternate forms of transport to definitive care, the nursing advice line, or to self-care pending a change in symptoms). Of these, 15 622 had both a documented personal identification number and a linked ambulance record. Finally, hospital records could be identified for 11 775 of these patients, with the remainder having either been treated and released on-scene (and did not visit a hospital-based care provider within 72 hours), or had a misregistered personal identification number resulting in a linkage failure. Thus, a total of 11 775 hospital visits with linked prehospital data were identified over the study period.
Of the patients included, 2 684 (23%) were tested for ongoing SARS-CoV-2 infection via PCR, of whom 356 (13%) were found to be positive at any point during the hospital episode. A total of 404 patients had a primary diagnosis code indicating Covid-19, of whom 125 were diagnosed with Covid-19 without a positive test (potentially due to the results of tests performed outside of the hospital, or via other means of diagnosis including radiology). An additional 77 patients were found to have a positive test result with no corresponding diagnosis, for a total of 481 “hospital-confirmed” cases of Covid-19.
Of the included records, 6 794 occurred on or after April 14 and were included in the analysis of prehospital predictive value. 2252 PCR tests were performed following this date resulting in 249 positive results, with an additional 99 Covid-19 patients identified via diagnosis code, for a total of 348 hospital confirmed cases during this time frame.
A total of 1381 patients had suspicion of Covid-19 documented by the dispatch center, while 1690 had suspicion documented in the ambulance journal. The volume of these patients over time is reported in figure 2 below.
Rolling averages for dispatch and ambulance suspicion are plotted from the dates they were implemented (March 19 and April 14, respectively)
Table 1 below presents descriptive statistics regarding the study cohort, divided into comparison groups.
Table 1 – Patient demographics and clinical characteristics
|
Pre-Covid
|
Assumed negative
|
Tested negative
|
Positive
|
N
|
2998
|
5983
|
2313
|
481
|
Percent Female
|
52.8 (51.0-54.6)
|
51.2 (50.1-52.5)
|
50.2 (48.2-52.3)
|
47.0 (42.2-51.1)
|
Age, median
|
70.0 (70.0-71.0)
|
65.0 (64.0-66.0)
|
77.0 (76.0-77.0)
|
73.0 (69.0-74.0)
|
High priority dispatch, %
|
48.1 (46.3-49.9)
|
46.7 (45.4-48.0)
|
40.9 (38.9-42.8)
|
30.1 (25.8-34.3)
|
Abnormal airway, %
|
2.9 (2.3-3.5)
|
1.9 (1.5-2.3)
|
3.7 (3.0-4.5)
|
1.3 (0.4-2.4)
|
Abnormal breathing, %
|
17.0 (15.7-18.4)
|
9.6 (8.8-10.3)
|
32.4 (30.5-34.4)
|
46.7 (42.3-51.2)
|
Abnormal circulation, %
|
16.8 (15.4-18.1)
|
12.6 (11.7-13.4)
|
25.3 (23.4-27.0)
|
26.0 (21.8-30.2)
|
Supplemental oxygen, %
|
15.3 (14.0-16.6)
|
7.2 (6.5-7.8)
|
29.5 (27.5-31.3)
|
36.4 (31.8-41.0)
|
Medication administration, %
|
38.6 (36.9-40.3)
|
32.3 (31.1-33.5)
|
38.3 (36.4-40.3)
|
28.5 (24.7-32.4)
|
High priority transport, %
|
15.4 (14.2-16.7)
|
11.0 (10.2-11.9)
|
22.0 (20.0-23.8)
|
14.1 (11.0-17.5)
|
Total mission duration (minutes), median
|
104 (102-105)
|
102 (101-103)
|
120 (118-122)
|
128 (123-132)
|
Admitted to hospital, %
|
49.3 (47.4-51.1)
|
37.6 (36.4-38.8)
|
88.6 (87.2-89.9)
|
83.4 (80.0-86.7)
|
Admitted to Intensive Care Unit, %
|
2.0 (1.5-2.6)
|
0.6 (0.4-0.8)
|
3.9 (3.1-4.7)
|
16.2 (12.9-19.5)
|
30-day mortality, %
|
5.9 (5.1-6.7)
|
3.4 (3.0-3.8)
|
11.3 (9.9-12.6)
|
23.7 (19.8-27.4)
|
All data presented with point estimates and bootstrapped 95% Confidence intervals. Grouped by patients cared for prior to the outbreak of Covid-19 from Jan 1-Feb 29 (Pre-Covid), those not tested for Covid-19 (Assumed negative), tested negative (Tested negative), and positive for Covid-19 per PCR-test and/or primary diagnosis code (Positive).
Covid-19 patients had a 30-day mortality rate of 23.7% (95% CI 19.8 – 27.4), and 16% (13 - 20) were admitted to an intensive care unit. 30.1% (25.8 – 34.3) of Covid-19 patients received a high priority ambulance from dispatch, while 14.1% (11 – 17.5) were transported by the ambulance with a high priority. 36.4% (31.8 – 41.0) received supplemental oxygen, while 28.5% (24.7 – 32.4) received other medications administered by ambulance crews. The median mission time (from receipt of call at the dispatch center to ambulance clearance from the hospital, including time spent sanitizing the ambulance) was 128 (123 - 132) minutes.
Table 2 presents the distribution of Covid-19 suspicion and test results across call types documented by the EMD center. A total of 1223 patients (18%, including 27 patients with an upper airway-related complaint) contacted the dispatch center with a primary complaint directly related to the Covid-19 suspicion guidelines, while 222 (46%) confirmed Covid-19 cases were among these callers.
Table 2 – Covid-19 suspicion and testing rates by primary complaint at dispatch with more than 100 occurrences from April 14 onwards, ordered by descending proportion of confirmed Covid-19 cases.
Primary Complaint
|
n
|
Suspicion at dispatch, %
|
Suspicion by ambulance, %
|
Tested at hospital, %
|
Hospital-confirmed * Covid-19, %
|
Infection
|
261
|
68.6 (62.8-74.3)
|
78.5 (73.9-83.5)
|
72.8 (67.4-78.2)
|
29.1 (23.8-34.9)
|
Fever
|
147
|
71.4 (64.6-79.6)
|
76.9 (70.1-83.7)
|
74.1 (67.3-81.0)
|
16.3 (10.9-22.4)
|
Difficulty Breathing
|
790
|
50.9 (47.5-54.3)
|
65.4 (61.8-68.5)
|
62.9 (59.9-66.1)
|
14.7 (12.3-17.1)
|
General Elderly†
|
344
|
17.2 (13.4-21.5)
|
29.1 (24.1-33.7)
|
54.7 (49.1-60.2)
|
6.4 (3.8-9.0)
|
General Adult†
|
139
|
28.8 (21.6-36.0)
|
29.5 (22.3-37.4)
|
38.1 (30.2-46.0)
|
5.8 (2.2-10.1)
|
Reduced consciousness
|
262
|
12.6 (8.8-16.4)
|
22.9 (18.3-27.9)
|
38.9 (33.2-45.0)
|
4.6 (2.3-7.3)
|
Fainting
|
117
|
12.0 (6.8-17.9)
|
19.7 (12.8-27.4)
|
22.2 (14.5-29.9)
|
3.4 (0.9-6.8)
|
Stroke
|
428
|
5.4 (3.5-7.5)
|
14.7 (11.4-18.0)
|
35.3 (30.8-40.2)
|
3.3 (1.9-5.1)
|
Seizure
|
207
|
4.3 (1.9-7.2)
|
10.6 (6.8-15.4)
|
20.3 (15.0-25.6)
|
2.4 (0.5-4.8)
|
Other
|
908
|
11.3 (9.1-13.4)
|
15.7 (13.4-18.1)
|
24.1 (21.3-27.1)
|
2.2 (1.3-3.2)
|
Dizziness
|
144
|
6.9 (3.5-11.8)
|
14.6 (9.0-20.8)
|
17.4 (11.1-23.6)
|
2.1 (0.0-4.2)
|
Major trauma
|
1203
|
2.8 (1.9-3.8)
|
8.1 (6.6-9.7)
|
26.4 (23.9-28.9)
|
1.8 (1.1-2.7)
|
Missing
|
118
|
0.0 (0.0-0.0)
|
11.0 (5.9-16.9)
|
23.7 (16.1-31.4)
|
1.7 (0.0-4.2)
|
Abdominal pain
|
548
|
8.4 (6.2-10.9)
|
17.2 (13.9-20.4)
|
17.9 (14.6-21.2)
|
1.5 (0.5-2.6)
|
Chest pain
|
813
|
9.5 (7.4-11.4)
|
17.7 (15.3-20.5)
|
19.7 (17.1-22.4)
|
1.2 (0.5-2.1)
|
Arrythmia
|
153
|
3.3 (0.7-6.5)
|
14.4 (9.2-20.3)
|
11.8 (6.5-17.0)
|
0.7 (0.0-2.6)
|
All data presented with point estimates and bootstrapped 95% Confidence intervals
* Confirmed based on either a positive PCR-test or by hospital diagnosis code
† These categories capture patients with non-specific complaints that cannot be further categorized
Figure 3 presents density functions for the initial vital signs of patients in each of the comparison groups. In comparing Covid-19-positive and patients tested negative, all vital signs except for pulse rates demonstrated statistically significant differences in central tendency per Wilcoxon rank sum tests. Body temperature demonstrated the most substantial effects, with 23% of patients with a negative Covid-19 test result presenting with a fever of >=38 degrees centigrade, and 51% of Covid-19 positive patients presenting with this degree of fever upon contact with the ambulance.
Prehospital assessments
Table 3 below presents the predictive values for dispatch and ambulance suspicion of Covid-19. Given the risk of misclassification-bias due to low testing rates, data are presented both for the full cohort of patients, and for only those with a documented PCR-test for SARS-CoV-2. The prevalence of SARS-CoV-2 in the full population was 5%, and 11% among tested patients. The apparent prevalence of Covid-19 (i.e., the percentage of patients with documented suspicion) in the full population was 17% at the dispatch, and 25% in the ambulance. Apparent prevalence among tested patients was 32% and 49% at the dispatch and ambulance, respectively. In the full population, the sensitivity of documented Covid-19 suspicion at dispatch was 75.9% (71.0 – 80.3) with a specificity of 86.4% (85.5 – 87.2) in the full cohort. The sensitivity of the ambulance suspicion was 82.2% (77.8 – 86.1), with a specificity of 78.2% (77.2 – 79.2). Levels of Specificity and negative predictive value were lower within the cohort of tested patients only.
Table 3 – Predictive value of dispatch and ambulance suspicion of Covid 19 among all patients, and tested patients only (April 14 – August 31)
|
All patients (n = 6 776)
|
Tested patients only (n = 2 252)
|
|
Dispatch
|
Ambulance
|
Dispatch
|
Ambulance
|
Apparent prevalence, %
|
16.8 (15.9 - 17.7)
|
24.9 (23.9 - 25.9)
|
31.9 (30.0 - 33.9)
|
49.3 (47.2 - 51.4)
|
True prevalence, %
|
5.1 (4.6 - 5.7)
|
5.1 (4.6 - 5.7)
|
11.1 (9.8 - 12.4)
|
11.1 (9.8 - 12.4)
|
Sensitivity, %
|
75.9 (71.0 - 80.3)
|
82.2 (77.8 - 86.1)
|
71.1 (65 - 76.6)
|
79.1 (73.5 – 84.0)
|
Specificity, %
|
86.4 (85.5 - 87.2)
|
78.2 (77.2 - 79.2)
|
73.0 (71.0 - 74.9)
|
54.4 (52.2 - 56.6)
|
Positive predictive value, %
|
23.1 (20.7 - 25.7)
|
16.9 (15.2 - 18.8)
|
24.7 (21.5 - 28)
|
17.7 (15.5 - 20.1)
|
Negative predictive value, %
|
98.5 (98.2 - 98.8)
|
98.8 (98.4 - 99.1)
|
95.3 (94.1 - 96.3)
|
95.4 (94.1 - 96.6)
|
All data presented with point estimates and exact binomial 95% Confidence intervals
Figure 3 presents this data per study month. Over time, both the apparent and true prevalence of Covid-19 declined (Note that structured documentation on the ambulance was implemented at the peak of the pandemic). Specificity and negative predictive value increased while positive predictive value declined substantially over time. Between April and May, dispatch sensitivity changed from 75% (65 - 83) to 82% (74 - 88), while ambulance sensitivity changed from 77% (67 - 84) to 89% (83 - 95), while in later months, the scarcity of true positive cases resulted in wide confidence intervals.