Population characteristics
653 deaths of Salford residents occurred in weeks 12-19 of 2020, 106 died out of area. 522 formed the final review cohort (figure 1 shows all exclusions). The characteristics of these 522 patients are found in table 1: 51% female, 81% Caucasian, 9% of mixed or multiple ethnic origin, 10% for other ethnic minority backgrounds, age 79±9 years, 64% cardiovascular co-morbidities. 197 patients (38%) died in hospital, 190 (37%) in a care home, 120 (21%) at home, and the majority of the remainder in a hospice. Of hospital deaths, 13 died in the Emergency Department (ED) including 5 dead on arrival. 28 died in a critical care or respiratory high care area.
Table 1. Overview of characteristics of the cohort undergoing structured judgement review, divided according to place of death.
|
Place of death
|
|
Care home
|
Own home
|
Hospital
|
Other
|
Number of cases (% of total)
|
190 (36)
|
120* (23)
|
197 (38)
|
15# (3)
|
Co-morbid features:
|
Age (year, mean ± SD)
|
85 ± 8
|
74 ± 14
|
75 ± 12
|
69 ± 17
|
Male gender (%)
|
38
|
53
|
58
|
60
|
Diabetes (%)
|
22
|
26
|
25
|
13
|
Chronic kidney disease (%)
|
19
|
18
|
22
|
13
|
Respiratory (%)
|
20
|
40
|
39
|
32
|
Cardiovascular (%)
|
59
|
71
|
66
|
47
|
BMI (kg/m2, mean±SD)
|
22 ± 5
|
25 ± 8
|
28 ± 8
|
27 ± 9
|
uDNAR or EoL (%)
|
97
|
63
|
71
|
80
|
Ethnicity (%):
|
Caucasian
|
81
|
68
|
87
|
93
|
Asian
|
0
|
1
|
1
|
0
|
Black
|
1
|
2
|
3
|
0
|
Other BAME
|
18
|
29
|
10
|
7
|
Usual residence (%):
|
Own home
|
2
|
75
|
53
|
73
|
Nursing Home
|
72
|
1
|
9
|
0
|
Residential Home
|
20
|
0
|
5
|
0
|
Other
|
6
|
24
|
34
|
27
|
Medical contact prior to death (%):
|
Primary care
|
96
|
89
|
62
|
80
|
111
|
13
|
10
|
11
|
67
|
999
|
23
|
26
|
86
|
47
|
Emergency Department~
|
15
|
61
|
19
|
33
|
1a listed Cause of death (%):
|
COVID-19
|
54
|
59
|
6
|
13
|
Acute respiratory, other
|
23
|
29
|
18
|
7
|
Acute cardiovascular
|
2
|
10
|
15
|
0
|
Cancer
|
7
|
11
|
45
|
73
|
Other chronic disease
|
71
|
65
|
42
|
27
|
Other acute cause
|
2
|
12
|
5
|
7
|
Avoidability of death score (count)
|
1
|
-
|
-
|
-
|
-
|
2
|
-
|
-
|
2
|
-
|
3
|
2
|
2
|
10
|
-
|
4
|
4
|
7
|
16
|
-
|
5
|
51
|
19
|
56
|
1
|
6
|
133
|
92*
|
113
|
14
|
Key: SD = standard deviation; BMI = body mass index; uDNAR = unified do not resuscitation document; EoL = end of life. * this value includes 8 cases of sudden death where cause is unknown and so no score was able to be appropriately allocated. # 14 of 15 were deaths in a hospice. ~ indicates hospital visit separate to final admission.
Healthcare contact prior to death was most frequently with primary care (81% of patients). 46 patients (9%) had healthcare appointments cancelled (median 1 cancellation, range 1-9). Recurrent cancellations were most often anticoagulation and podiatry appointments.
Less than half of NHS 111 calls were answered during this period. 27 of these (46%) resulted in advice to seek further care from their GP or in the community. 9 (15%) resulted in a delay in receiving this care, 5 of whom were seen later in the day by other services and subsequently admitted to hospital.
23% of patients were tested for COVID-19 (n=203). 63% were positive (25% of the review population). The likelihood of a positive swab differed between location of death. (68% in care homes, 20% in hospital, 30% in own home). COVID-19 was overall the most commonly listed cause of death. 87 of these (43%) did not have a positive swab result. Of the 286 who did not have COVID-19 on the death certificate, 80 (28%) had symptoms consistent with the diagnosis.
Avoidability of death scoring
80% of cases had an avoidability of death score of 6 allocated by the panel (definitely not avoidable). 18% had a score consistent with some degree of avoidability. 15% had a score of 5 or 4 (slight or possible avoidability). 3% of cases a score of 3 or 2 (more than 50:50 likelihood of being avoidable). None scored 1 (definitely avoidable). 2%, although reviewed for themes, had no final score allocated as these were sudden deaths in a patient’s home with no further information available: the panel did not feel scoring was possible.
The inter-observer agreement of individual reviewer scores was 79%. The free-marginal kappa was 0.75 (95% confidence 0.69-0.80). The fixed marginal kappa was 0.27 (0.12-0.42).Within the context of place of death, the number of cases with at least some potential avoidability were 30% for care home, 23% for death in own home, and 43% for death in hospital. For the age groups <55 years, 55-75 years, >75 years the respective figures were 40% 34%, and 32%. In older patients, deaths in care homes scored <6 less frequently than death in other locations (29% versus 53%). For Caucasian patients and non-Caucasian patients the respective figures were 19% and 16%. For COVID-19 and non-COVID-19 deaths the figures were 49% and 23%. The highest frequency of potential avoidability was in patients with learning disabilities (LD), albeit in a small number of deaths (4 of 9 deaths, figure 3).
Of the 16 scores of 2 or 3, there were 38 individual examples of recurring themes identified. These were most frequently delay in accessing care (n=11), below optimal care (n=10), and language barriers (n=3). There was also potentially avoidable COVID-19 contact, delayed response, misdiagnosis, and refusal to attend ED (table 2).
There was a similar distribution of themes in scores of 4 or 5 (table 3). Themes more common in these scores were potentially avoidable COVID-19 contact (n=19 of 154 cases) and delayed response (n=15). Potentially avoidable COVID-19 contact was the most common theme in patients with a score of 6 (not avoidable).
Multiple themes were greater in cases where scores were more likely to reflect avoidability of death. For scores of 2 or 3, 44% of cases had more than 2 themes. For scores of 4 to 5, 35% of cases had more than 2 themes and for scores of 6, 17% of cases had more than 2 themes.
Table 2. Frequency of core themes occurring within the structured judgement reviews, divided according to the overall judgement of avoidability. Patients may have been exposed to more than one theme.
Theme
|
Avoidability score
|
1-3
(>50% avoidable)
|
4-5
(<50% avoidable)
|
6
(not avoidable)
|
TOTAL
|
38
|
151
|
461
|
Below optimal care
|
10
|
40
|
126
|
Delay in testing
|
8
|
32
|
99
|
Delay in Accessing care
|
11
|
29
|
20
|
Capacity Issues (community services)
|
2
|
3
|
3
|
Exposure to COVID-19
|
1
|
19
|
129
|
Delay in Response
|
1
|
15
|
16
|
COVID-19 on MCCD without diagnosis
|
1
|
2
|
47
|
Communication Barrier
|
1
|
2
|
|
Advised to Self-Isolate
|
1
|
1
|
2
|
Potential Misdiagnosis
|
1
|
|
5
|
Patient refusal to go to ED
|
1
|
|
|
Transfer to IMC no testing
|
|
5
|
3
|
Sudden Death
|
|
1
|
8
|
Remote consultation
|
|
1
|
|
Self-Discharge
|
|
1
|
|
Delay in COVID-19 Swab Results
|
|
|
1
|
Family Insistence on transfer to ED
|
|
|
1
|
Hospital Acquired Infection
|
|
|
1
|
Key: MCCD – medical certificate of cause of death; ED = emergency department; IMC = intermediate care.
Table 3. Frequency of core contributing factors by major point of care, for all patients and for patients with scores indicating higher likelihood of avoidability. A full breakdown of all factors in all settings is found in supplementary table 1s.
|
Care Home
|
111 & 999
|
GP
|
Patient
|
2ndary care
|
All patients
|
|
|
|
|
|
Below optimal care
|
10
|
2
|
9
|
2
|
32
|
Delay in testing
|
17
|
1
|
38
|
4
|
30
|
Delay in Access
|
1
|
1
|
0
|
37
|
2
|
Exposure to Covid-19
|
111
|
0
|
0
|
5
|
20
|
Patient (Voluntary Isolation)
|
0
|
0
|
0
|
10
|
0
|
Delay in Response
|
0
|
20
|
2
|
5
|
0
|
No Swabs but COVID-19 on MCCD
|
0
|
0
|
17
|
0
|
8
|
Transfer to IMC no testing
|
0
|
0
|
0
|
0
|
6
|
Score 2, 3 & 4 only (n=39)
|
|
|
|
|
|
Below optimal care
|
1
|
1
|
3
|
0
|
8
|
Delay in testing
|
2
|
1
|
3
|
1
|
6
|
Delay in Access
|
1
|
1
|
0
|
13
|
1
|
Exposure to Covid-19
|
1
|
0
|
0
|
2
|
0
|
Patient (Voluntary Isolation)
|
0
|
0
|
0
|
5
|
0
|
Delay in Response
|
0
|
2
|
0
|
1
|
0
|
Thematic review
An overview of the 53 patients (9%) had a delayed presentation to medical services. Within general Practice there were 9 cases of below optimal care, 2 delayed response by clinicians, and 4 examples of lack of service capacity.
For deaths in hospital, there were 255 occasions of themes identified, affecting 129 patients. These included below optimal care in 79 patients, delay in testing 37 patients. Themes most often occurred prior to admission (69% delays in testing, 67% below optimal care, 29 of 30 delays in access). There were 9 hospital deaths after delayed presentation due to patient self-isolation. 4 occasions of below optimal care in secondary care were found in patients who were allocated an avoidability score of 2 or 3. These were examples of absence of escalation to higher levels of care.
In care homes, residents were regularly exposed to coronavirus (Table 3). In cases of below optimal care, the majority were issues with Do Not Attempt CPR (uDNAR) directives and Advanced Care (ACP) or Gold Standard Framework (GSF) planning. This is within the broader context of generally excellent end of life planning (92% ACP, 97% uDNAR, and 69% GSF in place at the time of death). There were examples where ambulances were called despite all these being in place occurred.
294 patients had called 999 (56%). The main theme was delay in response (20 cases). Eleven of these were in patients whose death was considered definitely not avoidable. 59 were recorded as having called NHS 111 (11%). There were 9 delays in care as a result of 111 calls (15% of calls). Details of the outcomes of all 111 calls and associated themes are found in table 4. Significant delays in 999 response to Category 1 and 2 calls occurred contributing to avoidable factors in 9 deaths (3%) involving ambulance conveyance, significant delay being a response over the three times the 90th Centile (Category 2 90th centile 40 minutes). The ambulance response took over two hours in 44% of Category 2 (immediate and life threatening) cases in March 2020 (mean time 126 minutes, maximum wait 328 minutes). The national ambulance response protocol was amended from 2nd April 2020 to mitigate against the emerging pandemic associated delays. This led to a reduction in category 2 delays as detailed in Table 4. Alongside this, the category 1 numbers may contain a number of category 2 ‘upgrades’ i.e. where a response was delayed until the call became critical and then upgraded. This is likely to explain the long dispatch time noted in some category 1 responses. However, these upgraded calls were not able to be differentiated clearly within the available data. From the beginning of April 2020 no Category 2 patient waited longer than 96 minutes (mean 44 minutes, maximum wait 96 minutes, table 5).
For ambulance conveyance to hospital with time from call to arrival <2 hours, 78% of patients had a NEWS2 ≥5 on arrival of the paramedic crew to the patient (mean score 7.1±4.0). This compared with 68% for call to arrival 2-4 hours (6.3±3.3), and 48% for arrivals >4 hours (4.9±3.6).
An overall summary of the key thematic findings is found in box 1.
Table 4. Resolution for 111 calls.
|
N
|
Themes & comments
|
Advice only
|
3
|
Delays in care for all 3 patients (advised to self-isolate).
|
Ambulance sent and patient conveyed to hospital
|
11
|
|
Ambulance Dispatched and not taken to hospital
|
5
|
All dead on arrival or end of life.
|
HCP from 111 call back
|
1
|
Delay in care.
|
Clinical assessment service
|
3
|
|
End of life advice on last day of life
|
5
|
4 care home patients in last day of life.
|
No response from 111
|
1
|
Delay of 2 hours for 1 patient.
|
999 response from 111
|
6
|
Ambulance dispatch = 5 (4 conveyed to hospital, 3 from care home, 1 delay). No data for 1.
|
Refer to GP
|
20
|
15 care home patients (13 died same day). 3 others referred to hospital by GP. 2 expected deaths at home.
|
Referral to Pharmacy
|
1
|
|
Urgent Community Response
|
3
|
All 3 delays in care, 2 referred to ED same day by community team.
|
Total
|
59
|
|
Table 5. Ambulance response times divided by those before and after adaptations to the ambulance pandemic response strategy. All times in minutes.
|
Pre
|
Post
|
|
Med ian
|
IQR
|
90th cent ile
|
Med ian
|
IQR
|
90th cent ile
|
Category 1 (n=29 pre, 44 post)
|
|
|
|
|
|
|
Ambulance Response Call to Dispatch
|
4
|
11
|
51
|
3
|
5
|
34
|
Ambulance Response Call to Arrival at Patient
|
11
|
23
|
91
|
11
|
14
|
57
|
Ambulance Response Arrival at Patient to Arrival at Hospital
|
46
|
20
|
71
|
45
|
19
|
78
|
Ambulance Response Total time Call to Arrival at Hospital
|
70
|
46
|
115
|
89
|
54
|
220
|
Category 2 (n=25 pre, 12 post)
|
|
|
|
|
|
|
Ambulance Response Call to Dispatch
|
86
|
103
|
205
|
20
|
54
|
71
|
Ambulance Response Call to Arrival at Patient
|
99
|
122
|
218
|
35
|
50
|
86
|
Ambulance Response Arrival at Patient to Arrival at Hospital
|
62
|
33
|
85
|
64
|
23
|
78
|
Ambulance Response Total time Call to Arrival at Hospital
|
179
|
110
|
286
|
98
|
74
|
155
|
Category 3 (n=4 pre, 7 post)
|
|
|
|
|
|
|
Ambulance Response Call to Dispatch
|
16
|
61
|
1245
|
28
|
33
|
150
|
Ambulance Response Call to Arrival at Patient
|
26
|
33
|
140
|
49
|
73
|
183
|
Ambulance Response Arrival at Patient to Arrival at Hospital*
|
-
|
-
|
-
|
59
|
-
|
-
|
Ambulance Response Total time Call to Arrival at Hospital*
|
-
|
-
|
-
|
130
|
-
|
-
|
|
|
|
|
|
|
|
|
Key: IQR = interquartile range; * 1 conveyance to hospital only.