OpenSAFELY: Factors associated with COVID-19 related hospital death in the linked electronic health records of 17 million adult NHS patients, 28 April 2020
Published 5 June 2020
© Crown copyright 2020
This publication is licensed under the terms of the Open Government Licence v3.0 except where otherwise stated. To view this licence, visit nationalarchives.gov.uk/doc/open-government-licence/version/3 or write to the Information Policy Team, The National Archives, Kew, London TW9 4DU, or email: psi@nationalarchives.gov.uk.
Where we have identified any third party copyright information you will need to obtain permission from the copyright holders concerned.
This publication is available at https://www.gov.uk/government/publications/opensafely-factors-associated-with-covid-19-related-hospital-deaths-in-adult-nhs-patients-28-april-2020/opensafely-factors-associated-with-covid-19-related-hospital-death-in-the-linked-electronic-health-records-of-17-million-adult-nhs-patients-28-april
Abstract:
Background
To date there have been no large cohort studies with access to the full clinical history of all patients at risk of COVID-19. Establishing factors associated with a rapidly arising novel cause of death requires a new approach to epidemiological research. We therefore set out to deliver a secure analytics platform inside the data centre of major electronic health records vendors, running across the full live linked pseudonymised electronic health records of all NHS patients in England. The following results are preliminary.
Data sources
Primary care electronic health records managed by the electronic health record vendor TPP, pseudonymously linked to patient-level ECDS A&E presentation data, ICNARC ITU admissions data, CPNS hospital death data, and ONS death data, using the new OpenSAFELY platform.
Population
17,420,832 adults.
Time period
1 February 2020 to 16 April 2020.
Primary outcome
Death in hospital attributed to COVID-19 as recorded by the CPNS system.
Methods
Cohort study. Cox-regression to generate hazard ratios: crude, age and sex adjusted, and multiply adjusted for clinical co-variates selected prospectively on the basis of clinical interest and prior findings.
Results
Preliminary results are in tables 1 and 2 below. There were 3956 deaths attributed to COVID-19. In summary: death from COVID-19 was strongly associated with being male (hazard ratio 1.98, 95%CI 1.85 to 2.12); older age and deprivation (both with a strong gradient); diabetes (HR 1.98 95% CI 1.86 to 2.12); asthma (1.25 (1.14 to 1.38)); and various other prior medical conditions as per Table 2. Compared to people with ethnicity recorded as white, people of Asian origin were at higher risk of death, with only partial attenuation in hazard ratios from the fully adjusted model (age-sex adjusted HR 2.01 95% CI 1.76 to 2.30; fully adjusted HR 1.52 95% CI 1.32 to 1.74)); and similar findings for black people (age-sex adjusted HR 1.89 95% CI 1.30 to 2.75; fully adjusted HR 1.68 95% CI 1.15 to 2.45).
Conclusions
We have identified and quantified a range of clinical risk factors for death from COVID-19, in the largest cohort study conducted by any country to date. People from Asian and black groups are at markedly increased risk of death from COVID-19, only partially attributable to co-morbidity or other risk factors. Deprivation is also a major risk factor, with little of the excess risk explained by co-morbidity or other risk factors. We will update and extend these results regularly; our OpenSAFELY-EMIS platform is adding over 30 million additional current NHS patients’ records. The unprecedented statistical power offered by our approach means that associations with less common risk factors can be robustly assessed in more detail, at the earliest possible date, as the pandemic progresses.
Tables
Table 1: Numbers of patients with each outcome
| N (column %) | Number (% within stratum) experiencing outcomes | |||
|---|---|---|---|---|
| COVID-19 ITU admission | COVID-19 Hospital death (CPNS) | COVID-19 death (ONS) | ||
| Total | 17420832 (100.0) | 1098 (0.0) | 4009 (0.0) | 1152 (0.0) |
Age
| N (column %) | Number (% within stratum) experiencing outcomes | |||
|---|---|---|---|---|
| COVID-19 ITU admission | COVID-19 Hospital death (CPNS) | COVID-19 death (ONS) | ||
| 18 to 40 | 5974032 (34.3) | 88 (0.0) | 25 (0.0) | SMALL N |
| 40 to 50 | 2871930 (16.5) | 132 (0.0) | 66 (0.0) | 17 (0.0) |
| 50 to 60 | 3067912 (17.6) | 284 (0.0) | 243 (0.0) | 49 (0.0) |
| 60 to 70 | 2406215 (13.8) | 324 (0.0) | 488 (0.0) | 131 (0.0) |
| 70 to 80 | 1952015 (11.2) | 231 (0.0) | 1153 (0.1) | 354 (0.0) |
| 80+ | 1148728 (6.6) | 39 (0.0) | 2034 (0.2) | 596 (0.1) |
Sex
| N (column %) | Number (% within stratum) experiencing outcomes | |||
|---|---|---|---|---|
| COVID-19 ITU admission | COVID-19 Hospital death (CPNS) | COVID-19 death (ONS) | ||
| Male | 8726979 (50.1) | 305 (0.0) | 1447 (0.0) | 425 (0.0) |
| Female | 8693853 (49.9) | 793 (0.0) | 2562 (0.0) | 727 (0.0) |
Body Mass Index (BMI) kg/m2
| N (column %) | Number (% within stratum) experiencing outcomes | |||
|---|---|---|---|---|
| COVID-19 ITU admission | COVID-19 Hospital death (CPNS) | COVID-19 death (ONS) | ||
| less than 18.5 | 314037 (1.8) | 7 (0.0) | 108 (0.0) | 28 (0.0) |
| 18.5 to 24.9 | 4811487 (27.6) | 105 (0.0) | 1023 (0.0) | 288 (0.0) |
| 25 to 29.9 | 4727168 (27.1) | 319 (0.0) | 1168 (0.0) | 363 (0.0) |
| 30 to 34.9 | 2405830 (13.8) | 300 (0.0) | 827 (0.0) | 241 (0.0) |
| 35 to 39.9 | 930211 (5.3) | 130 (0.0) | 334 (0.0) | 84 (0.0) |
| more than 40 | 467093 (2.7) | 95 (0.0) | 208 (0.0) | 47 (0.0) |
| Missing | 3765006 (21.6) | 142 (0.0) | 341 (0.0) | 101 (0.0) |
Smoking
| N (column %) | Number (% within stratum) experiencing outcomes | |||
|---|---|---|---|---|
| COVID-19 ITU admission | COVID-19 Hospital death (CPNS) | COVID-19 death (ONS) | ||
| Never | 5686215 (32.6) | 322 (0.0) | 724 (0.0) | 220 (0.0) |
| Former | 7225197 (41.5) | 663 (0.0) | 2778 (0.0) | 791 (0.0) |
| Current | 3821922 (21.9) | 104 (0.0) | 492 (0.0) | 134 (0.0) |
| Missing | 687498 (3.9) | 9 (0.0) | 15 (0.0) | 7 (0.0) |
Ethnicity
| N (column %) | Number (% within stratum) experiencing outcomes | |||
|---|---|---|---|---|
| COVID-19 ITU admission | COVID-19 Hospital death (CPNS) | COVID-19 death (ONS) | ||
| White | 10961052 (62.9) | 613 (0.0) | 2493 (0.0) | 693 (0.0) |
| Black | 170729 (1.0) | 18 (0.0) | 30 (0.0) | 10 (0.0) |
| Asian | 1028813 (5.9) | 133 (0.0) | 277 (0.0) | 111 (0.0) |
| Mixed | 342239 (2.0) | 71 (0.0) | 115 (0.0) | 32 (0.0) |
| Other | 320915 (1.8) | 44 (0.0) | 47 (0.0) | 23 (0.0) |
| Missing | 4597084 (26.4) | 219 (0.0) | 1047 (0.0) | 283 (0.0) |
Indices of Multiple Deprivation (IMD) quintile
| N (column %) | Number (% within stratum) experiencing outcomes | |||
|---|---|---|---|---|
| COVID-19 ITU admission | COVID-19 Hospital death (CPNS) | COVID-19 death (ONS) | ||
| 1 (most deprived) | 3342763 (19.2) | 224 (0.0) | 927 (0.0) | 239 (0.0) |
| 2 | 3479242 (20.0) | 268 (0.0) | 854 (0.0 | 250 (0.0) |
| 3 | 3483284 (20.0) | 228 (0.0) | 792 (0.0) | 230 (0.0) |
| 4 | 3478604 (20.0) | 174 (0.0) | 722 (0.0) | 211 (0.0) |
| 5 (least deprived) | 3501396 (20.1) | 202 (0.0) | 697 (0.0) | 216 (0.0) |
| Missing | 135543 (0.8) | SMALL N | 17 (0.0 | 6 (0.0) |
Co-morbidities
| Number (% within stratum) experiencing outcomes | ||||
|---|---|---|---|---|
| N (column %) | COVID-19 ITU admission | COVID-19 Hospital death (CPNS) | COVID-19 death (ONS) | |
| Respiratory disease excluding asthma | ||||
| Present | 715539 (4.1) | 96 (0.0) | 948 (0.1) | 286 (0.0) |
| Absent | 16705293 (95.9) | 1002 (0.0) | 3061 (0.0) | 866 (0.0) |
| Asthma | ||||
| Present | 1481350 (8.5) | 171 (0.0) | 538 (0.0) | 172 (0.0) |
| Absent | 15939482 (91.5) | 927 (0.0) | 3471 (0.0) | 980 (0.0) |
| Chronic heart disease | ||||
| Present | 1180379 (6.8) | 148 (0.0) | 1467 (0.1) | 439 (0.0) |
| Absent | 16240453 (93.2) | 950 (0.0) | 2542 (0.0) | 713 (0.0) |
| Diabetes | ||||
| Present | 1732099 (9.9) | 350 (0.0) | 1722 (0.1) | 494 (0.0) |
| Absent | 15688733 (90.1) | 748 (0.0) | 2287 (0.0) | 658 (0.0) |
| Cancer excluding haematology | ||||
| Present | 82118 (0.5) | 7 (0.0) | 77 (0.1) | 31 (0.0) |
| Absent | 17338714 (99.5) | 1091 (0.0) | 3932 (0.0) | 1121 (0.0) |
| Cancer haematology / APL anaemia / bone marrow transplant | ||||
| Present | 67801 (0.4) | 17 (0.0) | 98 (0.1) | 25 (0.0) |
| Absent | 17353031 (99.6) | 1081 (0.0) | 3911 (0.0) | 1127 (0.0) |
| Liver disease | ||||
| Present | 114796 (0.7) | SMALL N | 81 (0.1) | 26 (0.0) |
| Absent | 17306036 (99.3) | 1093 (0.0) | 3928 (0.0) | 1126 (0.0) |
| Stroke/dementia | ||||
| Present | 377601 (2.2) | 41 (0.0) | 702 (0.2) | 172 (0.0) |
| Absent | 17043231 (97.8) | 1057 (0.0) | 3307 (0.0) | 980 (0.0) |
| Other neurological | ||||
| Present | 172934 (1.0) | 13 (0.0) | 213 (0.1) | 59 (0.0) |
| Absent | 17247898 (99.0) | 1085 (0.0) | 3796 (0.0) | 1093 (0.0) |
| Organ transplant | ||||
| Present | 20210 (0.1) | 10 (0.0) | 34 (0.2) | SMALL N |
| Absent | 17400622 (99.9) | 1088 (0.0) | 3975 (0.0) | 1148 (0.0) |
| Spleen diseases | ||||
| Present | 28261 (0.2) | SMALL N | 19 (0.1) | 8 (0.0) |
| Absent | 17392571 (99.8) | 1094 (0.0) | 3990 (0.0) | 1144 (0.0) |
| Rheumatoid, Lupus and Psoriasis | ||||
| Present | 886338 (5.1) | 87 (0.0) | 377 (0.0) | 123 (0.0) |
| Absent | 16534494 (94.9) | 1011 (0.0) | 3632 (0.0) | 1029 (0.0) |
Table 2: Age and sex adjusted, and fully adjusted hazard ratios from stratified cox regression
(Ethnicity does not feature in the main fully adjusted model due to missing data. Hazard ratios are derived from a separate complete case fully adjusted model.)
Age
| CPNS Death HR (95% CI) | ||
|---|---|---|
| Age-sex adjusted | Fully adjusted | |
| 18 to 40 | 0.05 (0.03 to 0.08) | 0.06 (0.04 to 0.09) |
| 40 to 50 | 0.28 (0.21 to 0.36) | 0.31 (0.23 to 0.41) |
| 50 to 60 | 1.00 (ref) | 1.00 (ref) |
| 60 to 70 | 2.76 (2.36 to 3.23) | 2.32 (1.99 to 2.72) |
| 70 to 80 | 8.36 (7.27 to 9.63) | 6.05 (5.25 to 6.99) |
| 80+ | 27.06 (23.63 to 30.98) | 17.01 (14.77 to 19.59) |
Sex
| CPNS Death HR (95% CI) | ||
|---|---|---|
| Age-sex adjusted | Fully adjusted | |
| Female | 1.00 (ref) | 1.00 (ref) |
| Male | 2.31 (2.16 to 2.47) | 1.98 (1.85 to 2.12) |
Body Mass Index (BMI) kg/m2
| CPNS Death HR (95% CI) | ||
|---|---|---|
| Age-sex adjusted | Fully adjusted | |
| less than 40 | 1.00 (ref) | 1.00 (ref) |
| more than 40 | 3.33 (2.88 to 3.84) | 2.41 (2.08 to 2.79) |
Smoking[footnote 1]
| CPNS Death HR (95% CI) | ||
|---|---|---|
| Age-sex adjusted | Fully adjusted | |
| Non-smoker | 1.00 (ref) | 1.00 (ref) |
| Current | 0.90 (0.82 to 0.99) | 0.80 (0.72 to 0.88) |
Ethnicity[footnote 2]
| CPNS Death HR (95% CI) | ||
|---|---|---|
| Age-sex adjusted | Fully adjusted | |
| White | 1.00 (ref) | 1.00 (ref) |
| Black | 1.89 (1.30 to 2.75) | 1.68 (1.15 to 2.45) |
| Asian | 2.01 (1.76 to 2.30) | 1.52 (1.32 to 1.74) |
| Mixed | 2.18 (1.79 to 2.65) | 1.78 (1.45 to 2.17) |
| Other | 1.45 (1.07 to 1.95) | 1.35 (1.00 to 1.82) |
Indices of multiple deprivation (IMD) quintile
| CPNS Death HR (95% CI) | ||
|---|---|---|
| Age-sex adjusted | Fully adjusted | |
| 1 (most deprived) | 1.00 (ref) | 1.00 (ref) |
| 2 | 0.78 (0.70 to 0.85) | 0.83 (0.75 to 0.91) |
| 3 | 0.64 (0.58 to 0.71) | 0.72 (0.65 to 0.79) |
| 4 | 0.55 (0.50 to 0.61) | 0.64 (0.58 to 0.71) |
| 5 (least deprived) | 0.48 (0.44 to 0.54) | 0.58 (0.53 to 0.65) |
Blood pressure[footnote 1]
| CPNS Death HR (95% CI) | ||
|---|---|---|
| Age-sex adjusted | Fully adjusted | |
| Normal | 1.00 (ref) | 1.00 (ref) |
| High | 0.73 (0.68 to 0.77) | 0.79 (0.74 to 0.84) |
Co-morbidities
| CPNS Death HR (95% CI) | ||
|---|---|---|
| Age-sex adjusted | Fully adjusted | |
| Respiratory disease ex asthma | 2.35 (2.18 to 2.53) | 1.87 (1.72 to 2.02) |
| Asthma | 1.63 (1.49 to 1.79) | 1.25 (1.14 to 1.38) |
| Chronic heart disease | 1.86 (1.74 to 2.00) | 1.40 (1.30 to 1.50) |
| Diabetes | 2.39 (2.24 to 2.55) | 1.98 (1.86 to 2.12) |
| Cancer excluding haematology | 1.64 (1.31 to 2.05) | 1.54 (1.23 to 1.93) |
| Cancer haematology / APL anaemia / bone marrow transplant | 2.64 (2.16 to 3.23) | 1.48 (1.20 to 1.82) |
| Liver disease | 2.33 (1.87 to 2.91) | 1.68 (1.34 to 2.10) |
| Stroke and dementia | 2.24 (2.06 to 2.43) | 1.80 (1.65 to 1.96) |
| Other neurological | 2.65 (2.31 to 3.05) | 2.22 (1.93 to 2.56) |
| Organ transplant | 7.09 (5.05 to 9.94) | 5.19 (3.68 to 7.32) |
| Spleen diseases | 2.04 (1.30 to 3.21) | 1.73 (1.10 to 2.71) |
| Rheumatoid, Lupus and Psoriasis | 1.40 (1.26 to 1.56) | 1.26 (1.13 to 1.40) |
-
Smoking and hypertension: we suspect these surprising findings may be explained by the inclusion of ex-smokers in the non-smoker group; and our coding blood pressure as most recent reading rather than a diagnosis of hypertension (as the latter combines pathology with health-seeking behaviour); we are now conducting further additional rapid analyses to assess these issues. ↩ ↩2
-
Based on the 73.6% of participants with ethnicity recorded. ↩