Preventable deaths from SARS-CoV-2 in England and Wales: a systematic case series of coroners’ reports during the COVID-19 pandemic

34 35

To examine coroners’ Prevention of Future Deaths reports (PFDs) to identify deaths 36 involving SARS-CoV-2 that coroners deemed preventable. 37 38 Design Consecutive case series. 39 40 Setting England and Wales. 41 42 Participants Patients reported in 510 PFDs dated between 01 January 2020 and 28 June 2021, 43 collected from the UK’s Courts and Tribunals Judiciary website using web scraping to create an 44 openly available database, https://preventabledeathstracker.net/. 45 46 Main outcome measures Concerns reported by coroners. 47 48 Public and Patient Involvement Patients and members of the public were not involved in this 49 study. 50 51

SARS-CoV-2 was involved in 23 deaths reported by coroners in PFDs. Twelve deaths 52 were indirectly related to the COVID-19 pandemic, defined as those that were not medically 53 caused by SARS-CoV-2, but were associated with mitigation measures. In 11 cases the coroner 54 explicitly reported that COVID-19 had directly caused death. There was geographical variation 55 in the reporting of PFDs; most (39%) were written by coroners in the North-West of England. 56 The coroners raised 56 concerns, problems in communication being the most common (30%), 57 followed by failure to follow protocols (23%). Organizations in the National Health Service (NHS) 58 were sent the most PFDs (51%), followed by the Government (26%), but responses to PFDs by 59 these organizations were poor. 60 61

PFDs contain a rich source of information on preventable deaths that has 62 previously been difficult to examine systematically. Our openly available tool 63 (https://preventabledeathstracker.net/) streamlines this process and has identified many 64 concerns raised by coroners that should be addressed during the Government’s inquiry into the 65 handling of the COVID-19 pandemic, so that mistakes made are less likely to be repeated. 66 67 Study protocol pre-registration: https://osf.io/bfypc/ 68 69 70 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint 3 Summary box 71 72 What is already known about this subject? 73 • The UK Government has stated that there w ill be a public inquiry into the handling of the 74 COVID-19 pandemic, to learn lessons for future pandemics. 75 • Coroners in England and Wales have a duty to report and communicate information 76 about the deaths they investigate when the coroner believes that action should be taken 77 to prevent future deaths. 78 • These reports, called Prevention of Future Death reports (PFDs), had not yet been 79 systematically analysed to identify deaths that occurred during the COVID-19 pandemic. 80 81 What are the new findings? 82 • We created the Preventable Deaths Database ( https://preventabledeathstracker.net/) 83 using web scraping to systematically assess PFDs published on the Courts and Tribunal 84 Judiciary website. 85 • Between 01 January 2020 and 28 June 2021, one in 20 (4.5%, n=23) PFDs that were 86 published by coroners involved SARS-CoV-2. 87 • Coroners raised many concerns about the care of patients in hospitals, care homes, and 88 people in the community during the COVID-19 pandemic, which require action to prevent 89 future deaths. 90 91 How might it affect clinical practice in the foreseeable future? 92 • Preventable deaths that occurred during the COVID-19 pandemic should be referred to 93 the coroner so that an inquest can be performed and a PFD issued, highlightng actions 94 that could be avoided in improving the handling of future pandemics in both the UK and 95 elsewhere. 96 • Our tool, https://preventabledeathstracker.net/, can be used by others to examine 97 preventable deaths in England and Wales, and to identify signals for quality 98 improvement to reduce avoidable harms in clinical practice. 99 100 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint 4

101 102 Over five million deaths worldwide have been attributed to the severe acute respiratory 103 syndrome due to coronavirus 2 (SARS-CoV-2)1; some deaths may have been preventable. 104 105 In England and Wales, causes of deaths are investigated by coroners during an inquest, unless 106 the death is natural or referred to the criminal court. Under UK regulations, coroners have a duty 107 to report and communicate information about the deaths that they investigate when they believe 108 that actions should be taken to prevent similar deaths2 3,4. Such reports, previously called Rule 109 43 reports, are named Prevention of Future Deaths reports or PFDs. Despite these regulations, 110 there is no formal system in place for auditing or systematically analysing PFDs, so concerns 111 raised by coroners in such reports may go unrecognized and unreported, reducing the impact of 112 the PFD system. 113 114 The PFD system has three processes: 1) coroners write PFDs after the inquest and send 115 reports to those with the power to act; 2) addressees respond to coroners regarding the 116 concerns raised in PFDs within 56 days; and 3) actions to prevent such deaths are proposed 117 and ideally implemented. In December 2020, a series was launched in BMJ Evidence Based 118 Medicine to disseminate PFDs that highlight lessons for clinical practice and policy5. Case 119 reports in the series have identified deaths from ingesting alcohol-based hand sanitizer, 120 misdiagnosed adverse drug reactions, problems with repeat opioid prescriptions, and fire 121 hazards from emollient products6-9. Case series of PFDs have also been conducted to 122 investigate preventable deaths from medicines and misused drugs, suicides, cardiovascular 123 disease, and cycling10-13. 124 125 During the COVID-19 pandemic, healthcare professionals in England and Wales called for the 126 deaths of their colleagues to be reported to coroners and for PFDs to be issued14,15. However, 127 PFDs issued during the COVID-19 pandemic have not been analysed. We therefore aimed to 128 systematically analyse PFDs in which SARS-CoV-2 was directly or indirectly implicated in a 129 death. 130 131

132 133 We extracted a series of cases from the Courts and Tribunals Judiciary website and analysed 134 them according to a study protocol that was preregistered on an open repository16. 135 136 Data collection 137 PFDs are published on the Courts and Tribunals Judiciary website17. We used web scraping to 138 systematically collect all published PFDs and created a searchable database, the Preventable 139 Deaths Database18, which can be efficiently updated. The code for the web scraper is openly 140 available on GitHub and the methods have been previously described19,20. The Preventable 141 Deaths Database contains the case reference number; the date of the report; the name of the 142 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint 5 deceased; the coroner's name; the coroner’s jurisdiction; the category of death (as assigned by 143 the Chief Coroner’s office); to whom the report was sent; and the URL to the Judiciary website. 144 Eligibility of cases 145 We screened all 510 PFDs in the Preventable Deaths Database, dated from 01 January 2020 to 146 28 June 2021, for cases that mentioned COVID-19 as a direct or indirect cause of death. Direct 147 deaths were defined as those that the coroner explicitly attributed to COVID-19 as a cause of 148 death or were associated with a positive test for COVID-19 within 28 days of death. Indirect 149 deaths were defined as those that were not medically caused by COVID-19, but where coroners 150 mentioned that the associated mitigation measures during the COVID-19 pandemic (for 151 example, missed appointments due to lockdown), had contributed to the death. Cases that did 152 not mention SARS-CoV-2 or an effect of the COVID-19 pandemic were excluded. 153 Data extraction 154 For included cases, one study author (BS) manually extracted the following variables into a 155 predesigned Google Sheet, which was cross-examined by another study author (GCR): the 156 individuals or organizations to whom reports were sent and who responded; date of death; age; 157 sex; setting or location of death; medical cause(s) of death; the coroner’s conclusion(s) of the 158 inquest; relevant medical, mental health and social history; substance(s) implicated in the death 159 and the type of substance(s); and the coroner’s concerns. The data available for extraction were 160 limited by the information reported by coroners in the PFDs. 161 Data analysis 162 We used descriptive statistics to describe the numbers and types of cases that met the eligibility 163 criteria for inclusion. The numbers and types of individuals and organizations who received 164 PFDs were synthesized and response rates to coroners were calculated. A response rate of 165 100% meant that the individual or organization responded to all published PFDs issued by 166 coroners. Once all concerns were extracted, one study author (BS) read through all the 167 concerns and used content analysis to count and classify each concern inductively21. 168 Software 169 We used Tableau to present the coroners’ concerns visually and Data Wrapper to 170 geographically map the number of PFDs reported in each region of England and Wales. 171 Patient and public Involvement 172 Patients and the public were not involved in this study. 173 174

175 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint 6 Figure 1 shows the results of the extraction process. Of 510 PFDs reported on the Judiciary 176 website between 01 January 2020 and 28 June 2021, 23 (4.5%) were attributed to COVID-19 177 and deemed preventable by coroners (Table 1).178 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint 7 Table 1: Summary of 23 deaths involving SARS-CoV-2 as reported in Prevention of Future Death reports in England and Wales between 1 179 January 2020 and 28 June 2021, ordered by causes and date of death (created by the authors) 180 Dates Responses to PFDs Age Sex Death Inquest Report Causes of death Addressee(s) Date of reply± Directly-related to COVID-19 - M 13/04/2020 14/04/2020 01/12/2020 1) Community-acquired pneumonia 2) COVID-19 3) Dementia, chronic obstructive pulmonary disease, asbestos- related pulmonary fibrosis, pleural plaques, type 2 diabetes 1) CQC 2) Vicarage Residential Care Home 3) Public Health England 4) NHS England 5) Greater Manchester Health and Social Care Partnership 1) 04/02/2021 2) Undated 3) 26/01/2021 4) 02/03/2021 5) 19/02/2021 86 F 17/04/2020 03/07/2020 11/12/2020 Natural causes and COVID-19 Whipps Cross Hospital Received but not dated 74 M 21/04/2020 22/04/2020 01/12/2020 1) Hypovolemic shock 2) End-stage kidney disease 3) Polyneuropathy, frailty and COVID-19 1) Department of Health and Social Care 2) Royal London Hospital Not yet received^ - M 21/05/2020 26/05/2020 09/12/2020 1) COVID-19 pneumonia 2) Right sided neck of femur fracture, hypertension, atrial fibrillation 1) Public Health England 2) NHS England 1) 11/02/2021 2) 09/02/2021 18 M 31/07/2020 03/08/2020 30/03/2020 Drowning to which COVID-19 and asthma were contributory 1) Craven District Council 2) Yorkshire Dales National Park 3) Yorkshire Water Not yet received ^ - M 07/09/2020 07/09/2020 24/04/2021 1) Bronchopneumonia in combination with COVID-19 2) Falls with vertebral fractures, type 2 diabetes mellitus, pulmonary fibrosis, heart failure and epilepsy 1) Greater Manchester Health and Social Care Partnership 2) NHS England 1) 07/09/21 2) 02/07/21 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint 8 - M 15/11/2020 16/11/2020 11/06/2021 1) Covid 19 Pneumonitis 2) Chronic obstructive pulmonary disease, ischaemic heart disease, previous right upper lobe resection for lung adenocarcinoma, type 2 diabetes mellitus Tameside CCG 24/06/2021 90 M 28/01/2021 08/02/2021 23/04/2021 1) COVID-19 pneumonia 2) Dementia, heart failure, acute on chronic subdural haematoma, fall Medway Maritime Hospital 07/06/2021 - M 05/02/2021 08/02/2021 10/06/2021 1) COVID-19 on background of immunomodulatory treatment 2) Seborrheic eczema 3) Peripheral vascular disease 1) NHS England 2) Secretary of State of Health Not yet received ^ - M 03/04/2021 17/11/2020 14/06/2021 1) Aspiration pneumonia on a

of a choking incident, COVID-19 pneumonitis 2) Alzheimers dementia 1) MHRA 2) NHS Stockport CCG 1) 27/07/21 2) 06/08/21 - M - 30/10/2020 19/02/2021 COVID-19 pneumonitis 1) Brighton Sussex University NHS Hospital Trust 2) West Sussex NHS Hospital Trust 3) Medico-Legal 1) 19/03/2021 2) Not yet received ^ 3) Not yet received^ Indirectly-related to COVID-19 32 F 19/03/2020 30/03/2020 23/11/2020 1) Hanging 2) Bipolar affective disorder 1) Sussex Partnership Foundation NHS Trust 2) Brighton and Hove City Council 1) 10/02/2021 2) 10/02/2021 - M 17/04/2020 24/04/2020 07/12/2020 Methadone toxicity 1) Public Health England 2) Haverhill Pharmacy 1) 13/01/2021 2) Undated - M 24/04/2020 05/05/2020 19/11/2020 Suicide 1) Woolwich Station Medical Centre 2) Ministry of Defence 1) Not received^ 2) 16/02/2021 - F 20/06/2020 23/06/2020 11/02/2021 1) Bronchopneumonia 2) Frailty 3) Dementia 4) Hypertension 1) CQC 2) Department of Health and Social Care 1) 04/06/2021 2) 03/06/2021 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint 9 5) Fractured neck of femur 28 M 10/08/2020 14/08/2020 15/03/2021 Suicide Sussex Partnership NHS Foundation Trust Not yet received^ 76 F 28/09/2020 03/12/2020 16/12/2020 Atherosclerosis and complete blockage of one artery 1) NHS Pathways* 2) COVID-19 Pandemic Response Service 10/02/2021 - M 25/10/2020 26/10/2020 02/06/2021 Combined drug toxicology Stockport CCG 07/07/21 77 M - 08/10/2020 02/02/2021 1) Advanced dementia 2) Fractured neck of femur 3) Ischaemic heart disease 1) Adult Social Services, Norfolk County Council 2) Norfolk and Norwich University Hospital 1) 11/03/2021 2) 09/04/2021 88 M - 20/08/2020 05/02/2021 1) Bronchopneumonia 2) Heat stroke 3) Dehydration Care Outlook Ltd 18/04/2021 - M - 11/08/2020 07/05/2021 1) Small bowel obstruction and perforation 2) Ingestion of foreign body Norfolk and Norwich University Hospital NHS Foundation Trust 23/07/21 68 F - 20/09/2020 14/12/2020 1) Pneumothorax 2) Rib fractures 3) Fall 4) COPD and IHD West Midlands Ambulance Service 08/01/2021 87 M - 08/10/2020 07/05/2021 1) Urosepsis 2) Long term indwelling catheter not changed since October 2019 3) Alzheimer’s dementia, cerebrovascular accident, chronic kidney disease, bladder cancer, prostate cancer Lower Clapton Group Practice 24/06/21 *NHS Digital responded on behalf of NHS Pathways and the COVID-19 Pandemic Response Service. ±Recipients of PFDs have 56 days from the date of report to 181 respond to the coroner under Regulation 29 of The Coroners (Investigations) Regulations 2013. ^replies were still overdue on 18/10/21; however not all replies are 182 posted on the Judiciary website. 183 CCG: Clinical Commissioning Group; CQC: Care Quality Commission; COPD: Chronic Obstructive Pulmonary Disease; MHRA: Medicines and Healthcare 184 products Regulatory Agency; NHS: National Health Service. Deaths directly related were defined as those that the coroner explicitly reported COVID-19 as a 185 cause of death or a positive test for COVID-19 within 28 days of death. Indirectly related deaths were defined as those that were attributed to mitigation measures 186 during the COVID-19 pandemic. 187 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint 10 Most of the deaths (78%; n=18) occurred in men, and the median age at death was 76 years 188 (IQR: 50–87 years; range: 18–90 years; n=11). The deaths occurred between 13 April 2020 189 and 03 April 2021, but the date of death was unreported in six cases. 190 Causes of death 191 Eleven deaths (48%) were directly related to COVID-19, defined as those for which the 192 coroner explicitly reported COVID-19 as a cause of death or that were associated with a 193 positive test for COVID-19 within 28 days of death. Risk factors for death from COVID-19 194 included type 2 diabetes mellitus, kidney disease, hypertension, heart failure, and epilepsy 195 (Table 1). Pneumonitis co-occurred in 64% of deaths (n=7). One death was attributed to 196 drowning, to which COVID-19 and asthma contributed. In one-third (36%; n=4) of the direct 197 deaths, patients contracted COVID-19 as in-patients for other reasons, and another 198 contracted COVID-19 while a care-home resident. One man’s symptoms of COVID-19 were 199 undiagnosed during a telephone appointment and untreated, resulting in death. In four cases 200 (36%) it was unclear how the deceased had contracted COVID-19. 201 Twelve deaths (52%) were indirectly related to COVID-19, defined as those that were 202 attributed to mitigation measures during the pandemic. There were three suicides and two 203 cases of inappropriate prescription of medications during lockdown. Two deaths occurred 204 because the deceased refused to go to hospital or a care home, against medical advice, 205 owing to fears of COVID-19. A further two deaths occurred from complications with medical 206 devices when medical appointments were cancelled owing to lockdown and inadequate 207 follow-up; a catheter was not changed in one case, causing urosepsis, and a salivary bypass 208 tube was accidentally left inside the patient, where it perforated the small bowel. In two 209 cases, deaths were attributed to undiagnosed medical problems after incorrect diagnoses of 210 COVID-19 via remote telehealth appointments. One death was due to natural causes, 211 contributed to by several falls and a fractured neck of femur in a patient with advanced 212 dementia. 213 The 23 deaths were classified into nine groups by the Chief Coroner’s Office; hospital-214 related (28%; n=11), community healthcare (18%; n=7), care homes (13%; n=5), other 215 (13%; n=5), emergency services (10%; n=4), alcohol, drugs of abuse, and medications (8%; 216 n=3), mental health related (5%; n=2), suicide (3%; n=1), and service personnel related (3%; 217 n=1). 218 Geographical variation 219 Twelve coroners across 12 jurisdictions wrote PFDs relating to COVID-19. Most were written 220 by coroners in the North West of England (39%; n=9), followed by the South East (22%; 221 n=5) and London (17%; n=4) (Figure 2; Supplementary Table 1). Coroners in the South 222 West, North East, West Midlands, and Wales did not report any deaths deemed preventable 223 from COVID-19. 224 Coroners’ concerns 225 The coroners raised 56 individual concerns in the 23 PFDs. We categorized them into 28 226 groups and five higher-order categories (Figure 3; Supplementary Table 2). Poor 227 communication was reported in one-third of PFDs, followed by failure to follow protocols 228 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint 11 (23%), lack of education and training (19%), lack of resources (16%), and safety concerns 229 (12%). 230 1) Communication 231 232 Most (42%, n=8) concerns involved poor communication, followed by failure to keep 233 accurate medical records/care plans (n=5) and failure to consider how the COVID-19 234 pandemic would affect appointments (n=3). 235 Example 1: An elderly woman residing in a care home, at high risk of falls, became unwell. 236 Her GP remotely diagnosed suspected COVID-19, and she was isolated in her room, with 237 staff observation and sensor mats to ensure her wellbeing. She fell while unobserved and 238 was admitted to hospital with a fractured neck of femur, bronchopneumonia, and possible 239 COVID-19. A test for SARS-COV-2 was negative. She was unfit for surgery, deteriorated, 240 and later died in hospital from bronchopneumonia. The care home had a risk plan that 241 stipulated that she was to be observed during the day in communal areas. However, the 242 home was not staffed to provide one-to-one observation for residents in self-isolation. 243 The coroner believed that it was unclear how care homes were being advised to safely 244 manage residents at risk of falls when isolation was required, and the home knew of no 245 guidance they could follow to manage the risk. Furthermore, restrictions on family visitors 246 when in hospital presented significant communication problems, which reduced appropriate 247 support and timely clinical decision making. 248 2) Failure to follow protocols 249 The failure to regularly monitor/observe the patient was the most common form of failure 250 (36%; n=4), followed by a general failure to follow protocols, omission of necessary 251 treatment, and delayed discharge (Figure 3). 252 Example 2: A 28-year-old man with a lifelong history of low mood and depression with 253 suicidal ideation was socially isolated, which was exacerbated when he was furloughed from 254 his job. He had made five or six attempts to kill himself in 2020 alone and came to the 255 attention of an NHS Trust in May 2020. He registered with a GP at around the same time 256 and was assessed by a Mental Health practitioner, who referred him for treatment. The man 257 was aware of the referral and waited for treatment, but unfortunately the referral was lost. He 258 died by suicide at home 2 months later. 259 The coroner concluded that the Care Programme Approach (CPA), as set out in national 260 guidance “Refocusing the CPA – Policy and Positive Practical Guidance”, had not been 261 followed. It was clear at the inquest that staff members were not aware of how matters 262 should be dealt with and that this depended heavily on geographical location. 263 3) Education and training 264 The most common concern was poor clinical decision making (30%, n=3), followed by 265 inappropriate dosage of medication (20%, n=2). 266 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint 12 Example 3: A man with drug dependency had been receiving support from a Recovery 267 Network and was generally fit, well, and in good spirits. He received a regular prescription 268 for methadone in daily dosage bottles three times a week. During the COVID-19 pandemic 269 this regimen was changed to once every 14 days, giving him access to a much larger 270 quantity of methadone. The deceased was found at home with a very high blood 271 concentration of methadone. There was no evidence that he intended to take his own life. At 272 the start of the pandemic, Public Health England (PHE) guidance was issued that patients 273 taking short-term methadone should be moved to long-term treatment. The doctor who 274 changed the prescription stipulated that the drug must be supplied in single daily dosage 275 bottles, and explanatory letters had been sent to all the pharmacies that supplied opiate 276 replacement therapies to his patients. 277 The coroner saw evidence that the prescription was not in daily dose bottles and that there 278 was no measuring jug to enable accurate measurement of doses. The coroner believed that 279 it was probable that the man had guessed his first dose from the large methadone bottle. If 280 he had been given his daily dosage bottles, as prescribed, a measuring jug, and instructions 281 on how to use it, his death might have been prevented. 282 4) Resources 283 Lack of awareness of appropriate protocols and national guidelines was the most common 284 failure (40%, n=4), followed by understaffing (30%, n=3), a shortage or lack of availability of 285 medical equipment (20%, n=2), and unclear protocols (10%, n=1). 286 Example 4: A man was admitted to hospital after an accidental fall at home. After surgery 287 for a fractured hip, he developed a chest infection. When fit for discharge, he was moved to 288 several different wards and eventually put in a bay where patients had been exposed to 289 COVID-19. He subsequently tested positive for COVID-19, deteriorated rapidly, and died. 290 The coroner heard that the decision to move the deceased had been made on interpretation 291 of guidance from PHE. The Trust changed its policy, and such movements are reportedly no 292 longer taking place. However, when the PFD was written the guidance from PHE had not 293 been amended, and it was not known how other Trusts were choosing to interpret the 294 guidance, potentially putting other vulnerable in-patients at risk of COVID-19. Rapid and 295 national dissemination of coroner’s concerns might have prevented similar errors in other 296 hospitals. 297 5) Safety 298 Half of the concerns about safety related to facilities (50%, n=3), followed by poor systems 299 (33%, n=2) and inadequate discharge processes (17%, n=1). 300 301 Example 5: A resident in a care home, who had pulmonary fibrosis, had an unwitnessed fall. 302 He lay on the floor for over 4 hours awaiting an ambulance. Sepsis was diagnosed and it 303 was thought that he had symptoms consistent with COVID-19. He remained at the home 304 until 11 April, became unresponsive, deteriorated rapidly, was moved to palliative care, and 305 died on 13 April 2020. 306 307 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint 13 The coroner reported that the care had been of limited quality, notwithstanding the diagnosis 308 of COVID-19 and his vulnerability. The care home was unclear if staff had brought COVID-309 19 into the home or if admission of residents from the community who had not being tested 310 before admission had caused COVID-19 to enter the home. There was no risk assessment 311 in place relating to the admission of new residents, creating an unsafe environment for 312 vulnerable people. 313 314 315 Responses to PFDs 316 Forty-three unique individuals and organizations received PFDs from coroners (Table 2). 317 Most reports were sent to the NHS (51%; n=22), followed by the Government and related 318 bodies, other organizations, and professional bodies. Government organizations had the 319 highest response rates (64%; n=7), followed by professional bodies (50%; n=2), other 320 organizations (50%; n=3), and NHS organizations (45%; n=10). Of the 23 PFDs, nine had a 321 100% response rate and ten had no responses on the Judiciary website. 322 323 Table 2: Recipients of Prevention of Future Deaths reports involving COVID-19 in England and Wales 324 between 1 January 2020 and 28 June 2021 and their response rates (created by the authors) 325 Addressee No. of PFDs sent No. of responses± Response rate (%) NHS organizations 22 10 45% Trusts 5 2 40% NHS England 4 2 50% NHS Hospitals 4 3 75% CCGs 3 0 0% Health and Social Care Partnerships 2 1 50% NHS Pathways* 1 1 100% Ambulance services 1 1 100% GPs 2 0 0% Government 11 7 64% Public Health England 3 3 100% Department of Health and Social Care 2 1 50% Local authorities 3 2 67% COVID-19 pandemic response service* 1 1 100% Secretary of State of Health 1 0 0% Ministry of Defence 1 0 0% Professional bodies 4 2 50% CQC 2 2 100% General Pharmaceutical Council 1 0 0% MHRA 1 0 0% Other 6 3 50% Care homes/providers 2 2 100% . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint 14 Water board 1 0 0% National Park 1 0 0% Legal 1 0 0% Pharmacy 1 1 100% *NHS Digital responded on behalf of NHS Pathways and the COVID-19 Pandemic Response Service 326 ±Recipients of PFDs have 56 days from the date of the report to respond to the coroner under 327 Regulation 29 of The Coroners (Investigations) Regulations 2013 328 CCG: Clinical Commissioning Group; CQC: Care Quality Commission; GPs: General Practitioners; MHRA: 329 Medicines and Healthcare products Regulatory Agency; NHS: National Health Service 330 331

332 One in 20 PFDs published online involved COVID-19. Most of the deaths occurred in men 333 and older adults. There was wide geographical variation: no PFDs were reported by 334 coroners in Wales or in the North East, East Midlands, or South West of England. Coroners 335 raised several concerns, particularly regarding problems with communication and following 336 protocols. The largest numbers of PFDs were sent to NHS Trusts and the Government. 337 Eleven deaths were directly caused by SARS-CoV-2. In one-third of these cases, patients 338 acquired COVID-19 after admission to hospital for an unrelated reason, and another 339 acquired COVID-19 while in a care home. This suggests that measures to reduce the 340 transmission of SARS-COV-2 in healthcare settings were not adequate. Healthcare settings 341 should focus on evidence-based provisions, such as ventilation, personal protective 342 equipment (PPE), and regular testing to mitigate the risk to vulnerable patients22. 343 Twelve deaths were indirectly attributed to the COVID-19 pandemic. This finding highlights 344 the importance of considering the harms of measures and policies that were implemented to 345 reduce the transmission of SARS-COV-2 in the community. Reduced social interactions and 346 changed working conditions or loss of work and income have negatively affected adult 347 mental health in the UK23. Six million patients in the UK did not seek treatment in 2020 (so-348 called “missing patients”), owing to reprioritization of healthcare services24. In some cases 349 reduced access to care because of lockdown, despite telemedicine, led to deaths. 350 Healthcare professionals in England and Wales have called for the deaths of their 351 colleagues to be reported to coroners and for PFDs to be issued14,15. However, we did not 352 identify any PFDs that reported deaths of healthcare professionals. 353 The UK Government has stated that they will begin their public inquiry into the handling of 354 the COVID-19 pandemic to learn lessons for future pandemics25. We have identified several 355 areas that the Government should address during this inquiry, including poor communication 356 and gaps in education and training. PFDs should be examined by the Government and 357 healthcare providers to inform quality improvement and patient safety initiatives. 358 The Office of the Chief Coroner, who is responsible for uploading PFDs to the Courts and 359 Tribunals Judiciary website, categorized the COVID-19-related PFDs into nine groups, 360 including hospital-related, community healthcare, and care homes. A new category 361 specifically for the effects of pandemics should be added to the Judiciary website to assist 362 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint 15 the Government in examining these case reports so that policy measures can be 363 implemented for future pandemics. 364 Our study has several limitations. The 23 deaths do not represent all deaths during the 365 COVID-19 pandemic that could have been prevented in England and Wales. We are also 366 limited by the information reported by coroners in PFDs. In 52% of PFDs the age of the 367 deceased was not reported and 26% did not report the date of death. There were also 368 regions in England and Wales that did not report any PFDs. 369 Under-reporting of PFDs limits the capacity for actions to be taken to prevent future deaths. 370 However, it is likely that more PFDs relating to COVID-19 will be published, owing to the 371 backlog of inquests and the time it takes for inquests to conclude and PFDs to be written 372 and published. 373 There are no clear guidelines in England and Wales for referring deaths to coroners nor for 374 determining when a PFD should be issued and what information to include, hence the 375 missing data. There is also no auditing or quality control of PFDs and their responses. Thus, 376 whether action is taken to prevent such deaths, and the timeliness of such action, is 377 unknown and unmonitored. In the meantime, we encourage coroners across England and 378 Wales to continue writing PFDs when they believe that deaths could have been prevented. 379 PFDs contain a rich source of information that can be systematically analysed to share 380 information on preventing harms.The concerns identified in the 23 PFDs should be 381 considered during the UK Government’s inquiry, including how communication, protocols, 382 education, training, resources, and patient safety can be improved. Guidelines on how and 383 when to report deaths to coroners, thresholds for issuing PFDs, and the 384 necessarymandatory types of information to report in PFDs, including age and date of death, 385 are needed. Coroners in England and Wales should be encouraged to continue writing 386 PFDs during the COVID-19 pandemic, particularly when deaths involve frontline healthcare 387 professionals. 388 389 Declarations 390 391 Statement of ethical approval 392 We are using publicly available information, for which ethics committee approval is not 393 required. 394 395 Funding statement 396 No grant or research funding was obtained to undertake this study. 397 398 Competing interests statement 399 BS receives funding from Mustafa Bahceci (Bahceci Health Group, Istanbul, Turkey) for her 400 Doctor of Philosophy studies at the University of Oxford (2019-2022) and has received 401 financial renumeration for consultancy work in women’s health. CH is a National Institute for 402 Health Research (NIHR) Senior Investigator and has received expenses and fees for his 403 media work, received expenses from the WHO, FDA, and holds grant funding from the NIHR 404 School for Primary Care Research (SPCR) and the NIHR SPCR Evidence Synthesis 405 Working Group [Project 380], the NIHR BRC Oxford and the WHO. On occasion, CH 406 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint 16 receives expenses for teaching EBM and is also paid for his GP work in NHS out of hours 407 (contract with Oxford Health NHS Foundation Trust). JKA has published articles and edited 408 textbooks on adverse drug reactions and interactions and has often given medicolegal 409 advice, including appearances as an expert witness in coroners’ courts, often dealing with 410 the adverse effects of opioids and other medicines. DJH is the Director of Studies for 411 Sustainable Urban Development at the Department for Continuing Education, University of 412 Oxford. DJH has received financial remuneration for providing political and socioeconomic 413 country updates for Latin America and the Caribbean for IHS Global. GCR was financially 414 supported by the NIHR SPCR, the Naji Foundation, and the Rotary Foundation to study for a 415 Doctor of Philosophy (2017-2021), but no longer has any financial COIs. GCR is an 416 Associate Editor of BMJ Evidence Based Medicine and is developing 417 https://preventabledeathstracker.net/. The views expressed are those of the authors and not 418 necessarily those of the NHS, the NIHR, or the Department of Health and Social Care. 419 420 Contributorship statement 421 GCR developed the idea for this study, wrote the initial study protocol, ran the python code 422 to collect the most recent PFDs for screening, contributed to the first draft of the manuscript, 423 and provided supervisory support. BS contributed to the study protocol, screened the 510 424 PFDs for their eligibility, extracted the data from the 23 included cases, analysed the data, 425 and wrote the first draft of the manuscript. CH, JKA, and DJH contributed to the study 426 protocol, data analysis, and supervision of the project. All authors read, reviewed, and 427 approved the manuscript before submission. 428 Data sharing 429 The data, statistical code, and study materials are openly available via the Open Science 430 Framework (OSF) and GitHub16,19. 431 432

433 1. WHO Coronavirus (COVID-19) Dashboard. World Health Organization, 2021. 434 https://covid19.who.int 435 2. The Coroners Rules 1984. Rule 43. 436 http://www.legislation.gov.uk/uksi/1984/552/article/43/made. 437 3. Coroners and Justice Act 2009. 438 https://www.legislation.gov.uk/ukpga/2009/25/pdfs/ukpga_20090025_en.pdf. 439 4. The Coroners (Investigations) Regulations 2013. 440 http://www.legislation.gov.uk/uksi/2013/1629/pdfs/uksi_20131629_en.pdf. 441 5. Richards GC, Aronson JK, Heneghan C. Coroners’ concerns to prevent harms: a series 442 of coroners' case reports to serve patient safety and educate the public, clinicians and 443 policy-makers. BMJ Evid Based Med 2021; 26(2): 37-8. 444 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint 17 6. Richards GC. Alcohol-based hand sanitisers : a warning to mitigate future poisonings 445 and deaths. BMJ Evid Based Med 2021; 26(2): 65-8. 446 7. Cox AR, Ferner R. Tramadol: repeated prescriptions and repeated warnings. BMJ Evid 447 Based Med 2021 Mar 8; bmjebm-2020-111661. 448 8. Richards GC. Preventable opioid-related deaths in England and Wales. (2021) 449 doi:10.17605/OSF.IO/ECZ4R. 450 9. Bilip MK, Richards GC. Emollients and smok ing: a fire hazard that could be prevented 451 to reduce future deaths. BMJ Evid Based Med 2021; 26(3): 131-4. . 452 10. Ferner RE, Easton C, Cox AR. Deaths from medicines: a systematic analysis of 453 coroners’ Reports to Prevent Future Deaths. Drug Saf 2018; 41(1): 103-10. 454 11. Richards GC. et al. Preventable suicides: a systematic analysis of coroners’ prevent 455 future death reports in England and Wales. (2020) doi:10.17605/OSF.IO/AD4UP 456 12. Ains A, Heneghan C, Aronson JK, DeVito NJ, Richards GC. Deaths from cardiovascular 457 disease involving anticoagulants: a systematic synthesis of coroners’ case reports. 458 BJGP Open 2021; BJGPO.2021.0150. 459 13. Odgers L, Aronson JK, Richards GC. Preventable deaths involving cyclists in England 460 and Wales: solutions to improve their safety. Global Centre on Healthcare and 461 Urbanisation. https://www.gchu.org.uk/2021/09/gchu-report-preventable-deaths-462 involving-cyclists-in-england-and-wales (2021). 463 14. Kendrick D, Agius R, McKee M, Robertson J, Sewell H, Stewart M. COVID-19: Learning 464 lessons from the deaths of our colleagues. BJGP Life 2020; 465 https://bjgplife.com/2020/05/18/gp-deaths/. 466 15. Agius RM. Doctors’ deaths from COVID-19 should be reported to the coroner. BMJ 467 2020; 369 m1622. 468 16. Swift B, Heneghan C, Aronson JK, Howard D, Richards. Preventable deaths from 469 SARS-CoV-2 in England and Wales: a systematic analysis of coroners’ case reports. 470 OSF Registries 29 June 2021; https://osf.io/ej6ks 471 17. Judiciary. Prevention of Future Deaths. Courts and Tribunals Judiciary 2021; 472 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint 18 https://www.judiciary.uk/subject/prevention-of-future-deaths/. 473 18. Richards, G.C.Preventable Deaths Tracker. 2021. https://preventabledeathstracker.net 474 19. [dataset] Richards, G.C. &. DeVito, N.J. Data from: Preventable Deaths Database. 475 2021, November 12 2021. georgiarichards.github.io DOI: 10.5281/zenodo.567586620.476 477 20. DeVito NJ, Richards GC, Inglesby P. How we learnt to stop worrying and love web 478 scraping Nature 2020; 585(7826): 621-2. 479 21. Kleinheksel AJ, Rockich-Winston N, Tawfik H, Wyatt TR. Demystifying content analysis. 480 Am J Pharm Educ 2020; 84(1): 7113. 481 22. Richterman A, Meyerowitz EA, Cevik M. Hospital-acquired SARS CoV-2 infection 482 lessons for public health. JAMA 2020; 324(21): 2155-6. 483 23. Abbott A. COVID’s mental-health toll: how scientists are tracking a surge in depression. 484 Nature 2021; 590(7845): 194-5. 485 24. The Health Foundation. Unequal pandemic, fairer recovery: The COVID-19 impact 486 inquiry report. 2021; https://reader.health.org.uk/unequal-pandemic-fairer-recovery. 487 25. Booth R. Johnson says public inquiry into Covid will begin this parliament. The Guardian 488 11 May 2021. https://www.theguardian.com/world/2021/may/11/boris-johnson-public-489 inquiry-into-covid-begin-this-parliament 490 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 29, 2021. ; https://doi.org/10.1101/2021.07.15.21260589doi: medRxiv preprint