BMJ Open Is Committed to Open Peer Review. As Part of This Commitment We Make the Peer Review History of Every Article We Publish Publicly Available

Home , Stellenbosch

BMJ Open is committed to open peer review. As part of this commitment we make the peer review history of every article we publish publicly available.

When an article is published we post the peer reviewers’ comments and the authors’ responses online. We also post the versions of the paper that were used during peer review. These are the versions that the peer review comments apply to.

The versions of the paper that follow are the versions that were submitted during the peer review process. They are not the versions of record or the final published versions. They should not be cited or distributed as the published version of this manuscript.

BMJ Open is an open access journal and the full, final, typeset and author-corrected version of record of the manuscript is available on our site with no access controls, subscription charges or pay-per-view fees (http://bmjopen.bmj.com).

If you have any questions on BMJ Open’s open peer review process please email [email protected] BMJ Open

An evaluation of patient characteristics, management and outcomes for COVID-19 at district hospitals in the Western Cape, South Africa: Descriptive observational study ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2020-047016

Article Type: Original research

Date Submitted by the 16-Nov-2020 Author:

Complete List of Authors: Mash, Robert; Stellenbosch University, Family Medicine and Primary Care Presence-Vollenhoven, Mellisa; Stellenbosch University, Family Medicine and Primary Care; Western Cape Provincial Government, Metro Health Services Adeniji, Adeloye; Stellenbosch University, Family Medicine and Primary Care; Western Cape Provincial Government, Rural Health Services Christoffels, Renaldo; University of Cape Town, Family Medicine; Western Cape Provincial Government, Metro Health Services Doubell, Karlien; Stellenbosch University, Family Medicine and Primary Care; Western Cape Provincial Government, Rural Health Services Eksteen, Lawson; Stellenbosch University, Family Medicine and Primary Care; Western Cape Provincial Government, Rural Health Services Hendrikse, Amee; Western Cape Provincial Government, Metro Health Services Hutton, Lauren; Stellenbosch University, Family Medicine and Primary Care; Western Cape Provincial Government, Rural Health Services Jenkins, Louis; Stellenbosch University, Family Medicine and Primary Care; Western Cape Provincial Government, Rural Health Services Kapp, Paul; Stellenbosch University, Family Medicine and Primary Care; Western Cape Provincial Government, Rural Health Services Lombard, Annie; Stellenbosch University, Family Medicine and Primary Care; Western Cape Provincial Government, Metro Health Services Marais, Heleen; Western Cape Provincial Government, Metro Health Services Rossouw, Liezel; University of Cape Town, Family Medicine; Western Cape Provincial Government, Metro Health Services Stuve, Katrin; Stellenbosch University, Internal Medicine; Western Cape Provincial Government, Metro Health Services Ugoagwu, Abi; University of Cape Town, Family Medicine; Western Cape Provincial Government, Metro Health Services Williams, Beverley; Stellenbosch University, Family Medicine and Primary Care; Western Cape Provincial Government, Rural Health Services

COVID-19, ACCIDENT & EMERGENCY MEDICINE, Respiratory infections Keywords: < THORACIC MEDICINE

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 28 BMJ Open

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 28

1 2 3 4 5 6 7 8 9 I, the Submitting Author has the right to grant and does grant on behalf of all authors of the Work (as defined 10 in the below author licence), an exclusive licence and/or a non-exclusive licence for contributions from authors 11 who are: i) UK Crown employees; ii) where BMJ has agreed a CC-BY licence shall apply, and/or iii) in accordance 12 with the terms applicable for US Federal Government officers or employees acting as part of their official 13 duties; on a worldwide, perpetual, irrevocable, royalty-free basis to BMJ Publishing Group Ltd (“BMJ”) its 14 licensees and where the relevant Journal is co-owned by BMJ to the co-owners of the Journal, to publish the 15 Work in this journal and any other BMJ products and to exploit all rights, as set out in our licence. 16 17 The Submitting Author accepts and understands that any supply made under these terms is made by BMJ to 18 the Submitting Author Forunless you peer are acting as review an employee on behalf only of your employer or a postgraduate 19 student of an affiliated institution which is paying any applicable article publishing charge (“APC”) for Open 20 Access articles. Where the Submitting Author wishes to make the Work available on an Open Access basis (and 21 intends to pay the relevant APC), the terms of reuse of such Open Access shall be governed by a Creative 22 Commons licence – details of these licences and which Creative Commons licence will apply to this Work are set 23 out in our licence referred to above. 24 25 Other than as permitted in any relevant BMJ Author’s Self Archiving Policies, I confirm this Work has not been 26 accepted for publication elsewhere, is not being considered for publication elsewhere and does not duplicate 27 material already published. I confirm all authors consent to publication of this Work and authorise the granting 28 of this licence. 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 28 BMJ Open

1 2 An evaluation of patient characteristics, management and 3 4 outcomes for COVID-19 at district hospitals in the Western Cape, 5 6 South Africa: Descriptive observational study 7 8 Robert Mash, Division of Family Medicine and Primary Care, Stellenbosch University. 9 10 [email protected] [corresponding author] 11 12 Mellisa Presence- Vollenhoven, Division of Family Medicine and Primary Care, Stellenbosch 13 University and Metro Health Services, Western Cape Government. 14 15 [email protected] 16 17 Adeloye Adeniji, Division of Family Medicine and Primary Care, Stellenbosch University and 18 Rural Health Services,For Western peer Cape Government. review [email protected] only 19 20 Renaldo Christoffels, Division of Family Medicine, University of Cape Town and Metro Health 21 22 Services, Western Cape Government. [email protected] 23 24 Karlien Doubell, Division of Family Medicine and Primary Care, Stellenbosch University and 25 Rural Health Services, Western Cape Government. [email protected] 26 27 Lawson Eksteen, Division of Family Medicine and Primary Care, Stellenbosch University and 28 29 Rural Health Services, Western Cape Government. [email protected] 30 31 Amee Hendrikse, Department of Family Medicine, Metro Health Services, Western Cape 32 Government. [email protected] 33 34 Lauren Hutton, Division of Family Medicine and Primary Care, Stellenbosch University and 35 36 Rural Health Services, Western Cape Government. [email protected] 37 38 Louis Jenkins, Division of Family Medicine and Primary Care, Stellenbosch University and 39 Rural Health Services, Western Cape Government. [email protected] 40 41 Paul Kapp, Division of Family Medicine and Primary Care, Stellenbosch University and Rural 42 Health Services, Western Cape Government. [email protected] 43 44 45 Annie Lombard, Division of Family Medicine and Primary Care, Stellenbosch University and 46 Metro Health Services, Western Cape Government. [email protected] 47 48 Heleen Marais, Department of Family Medicine, Metro Health Services, Western Cape 49 Government. [email protected] 50 51 Liezel Rossouw, Division of Family Medicine, University of Cape Town and Metro Health 52 53 Services, Western Cape Government. [email protected] 54 55 Katrin Stuve, Department of Medicine, Stellenbosch University and Metro Health Services, 56 Western Cape Government. [email protected] 57 58 Abi Ugoagwu, Division of Family Medicine, University of Cape Town and Metro Health 59 60 Services, Western Cape Government. [email protected]

1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 28

1 2 Beverley Williams, Division of Family Medicine and Primary Care, Stellenbosch University 3 and Rural Health Services, Western Cape Government. [email protected] 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 28 BMJ Open

1 2 Keywords 3 4 COVID-19; coronavirus; signs and symptoms; disease management; risk factors; mortality; 5 6 South Africa; district hospitals 7 8 Abstract 9 10 Objectives: To describe the characteristics, clinical management and outcomes of patients 11 12 with COVID-19 at district hospitals. 13 14 Design: A descriptive observational cross-sectional study. 15 16 Setting: District hospitals (4 in Metro and 4 in Rural Health Services) in the Western Cape, 17 South Africa. District hospitals were small (<150 beds) and led by family physicians. 18 For peer review only 19 Participants: All patients who presented to the hospitals’ emergency centre and who tested 20 21 positive for COVID-19 between March-June 2020. 22 23 Primary and secondary outcome measures: Source of referral, presenting symptoms, 24 demographics, co-morbidities, clinical assessment and management, turn-around-time, 25 clinical outcomes, factors related to mortality, length of stay and location. 26 27 28 Results: 1376 patients (73.9% Metro, 26.1% Rural). Mean age 46.3 years (SD 16.3), 58.5% 29 females. The majority were self-referred (71%) and had co-morbidities (67%): hypertension 30 (41%), type 2 diabetes (25%), HIV (14%) and overweight/obesity (19%). Assessment of 31 32 COVID-19 was mild (49%), moderate (18%) and severe (24%). Test turnaround time (median 33 3.0 days (IQR 2.0-5.0 days)) was longer than length of stay (median 2.0 day (IQR 2.0-3.0)). 34 35 Commonest treatment was oxygen (41%) and only 0.8% were intubated and ventilated. 36 Overall mortality was 11%. Most were discharged home (60%) and only 9% transferred to 37 higher levels of care. Increasing age (OR 1.06 (CI 1.04-1.07)), male (OR 2.02 (CI 1.37-2.98)), 38 39 overweight/obesity (OR 1.58 (CI 1.02-2.46)), type 2 diabetes (OR 1.84 (CI 1.24-2.73)), HIV (OR 40 3.41 (CI 2.06-5.65)), chronic kidney disease (OR 5.16 (CI 2.82-9.43)) were significantly linked 41 42 with mortality (p<0.05). Pulmonary diseases (TB, asthma, chronic obstructive pulmonary 43 disease, post-TB structural lung disease) were not associated with increased mortality. 44 45 Conclusion: District hospitals supported primary care and shielded tertiary hospitals. 46 Patients had high levels of co-morbidities and similar clinical pictures to that reported 47 48 elsewhere. Most patients were treated as people under investigation. Mortality was 49 comparable to similar settings and risk factors identified. 50 51 52 53 54 55 56 57 58 59 60

3 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 28

1 2 Strengths and limitations of this study 3 4 5  The whole study population was included over a four-month period at the height of 6 the epidemic from eight district hospitals with few missing records. 7 8  A comprehensive dataset on presentation, assessment, management and outcomes 9 was captured. 10  The size of the study population enabled identification of risk factors associated with 11 12 mortality 13  Data was dependent on the accuracy of medical records and clinical skills 14 15  District hospitals without family physicians may have different quality of care, which 16 limits generalisability. 17 18 Word count 3637For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

4 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 28 BMJ Open

1 2 Introduction 3 4 Coronavirus disease 2019 (COVID-19) is a global pandemic that has affected all regions of the 5 6 world, although Africa has so far been less affected than predicted.(1) Given the number of 7 poor and developing countries, and relatively weak health systems, the pandemic is 8 9 expected to significantly impact African communities.(2) Within the African continent, South 10 Africa has had the most reported cases and the Western Cape has been one of the leading 11 hotspots.(1,3) 12 13 The beginning of March 2020 marked the zero point for COVID-19 infections in South Africa, 14 15 but since then numbers have grown exponentially and stands at 730,548 confirmed cases of 16 COVID-19 with 19,585 deaths in South Africa on 4 November 2020.(3) South Africa has the 17 th 18 13 largest numberFor of cases peerin the world review as of 4 November only 2020.(1) The Western Cape was 19 initially declared the epicentre of COVID-19 infections in South Africa, but was later 20 21 surpassed by Gauteng and KwaZulu-Natal.(3) As of 4 November 2020 the Western Cape had 22 118,009 cases and 4384 deaths.(3) 23 24 Clinical findings among confirmed COVID-19 cases in China showed the most common 25 complaints were fever (83%), cough (82%), difficulty breathing (31%), fatigue and myalgia 26 27 (11%).(4) Non-respiratory injury was identified by elevated levels of aspartate 28 aminotransferase (20%), creatinine (6%), and creatine kinase (15%) in laboratory tests.(5) 29 30 Findings from another study conducted in China showed that older patients and those with 31 32 comorbidity had poorer clinical outcomes.(5) Multi-morbidity was also correlated with 33 poorer clinical outcomes.(6) Co-morbidity with cardiopulmonary diseases was a particular 34 concern, such as diabetes, hypertension, asthma and chronic obstructive pulmonary disease 35 36 (COPD). In South Africa there was also concern with regard to the large numbers of people 37 with HIV, particularly those who were immunocompromised due to no or inadequate anti- 38 39 retroviral treatment. South Africa has 20% of all people living with HIV in the world and this 40 is accompanied by a high incidence of TB.(7,8) Patients with active TB or post-TB structural 41 lung damage were also a concern in our context. 42 43 44 Most clinical research has focused on evaluating patients admitted to tertiary hospitals with 45 high-care or intensive care units (ICU) and as the proposal was prepared we identified no 46 published studies from district hospitals. Little is known, therefore, about the types of 47 48 patients seen at district hospitals and their clinical course with the expertise and equipment 49 available at this level of the health system and in this context. In South Africa, district 50 51 hospitals usually have less than 150 beds and are run by generalists with male, female, 52 maternity and paediatrics wards as well as emergency centres. They are the first referral 53 point, particularly in remote and rural areas, for patients from primary care. 54 55 In Africa, primary or district hospitals do not provide intensive or critical care and have 56 57 limited capacity for prolonged ventilation. At the same time tertiary referral hospitals may 58 not be able to receive patients if their ICU facilities are full or they are very distant. 59 60 Elsewhere in Africa it has been suggested that district hospitals should focus more on the provision of oxygen therapy as a more valuable intervention than ventilation, as ventilation 5 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 28

1 2 requires adequate equipment and expertise, with a risk of harm to the patient and 3 transmission of the virus during intubation.(9) Guidelines have also been put in place to 4 5 determine which patients should be prioritised for ICU and which critically ill patients should 6 be managed with oxygen and if necessary palliative care.(10) In June 2020 Cape Town also 7 8 opened field hospitals, which will have the ability to manage patients with COVID-19 at a 9 level of care similar to a district hospital.(11) 10 11 This study will describe the type of patients referred to district hospitals run by family 12 13 physicians in the Western Cape and evaluate their presentation, clinical management and 14 outcomes. This data will help to provide a more complete picture of how COVID-19 is 15 affecting our population as the patient population seen at district hospitals is different to 16 17 that seen at tertiary hospitals and African populations may differ from those in Europe, Asia 18 or America. For peer review only 19 20 The aim of this study, therefore, was to describe the characteristics, clinical management 21 22 and outcomes of patients with COVID-19 presenting to district hospitals in the Western Cape 23 from March 2020 to June 2020. The specific objectives were to describe the original source 24 of referral, the presenting symptoms, the patients’ demographics, the presence of co- 25 26 morbidities, the clinical assessment and management, the turn-around-time for laboratory 27 results, the clinical outcomes and factors related to mortality, the length of stay and to 28 29 compare district hospitals in rural and metro health services. 30 31 Methods 32 33 Study design 34 35 This was a descriptive observational cross-sectional study by means of a retrospective audit 36 37 of medical records. 38 39 Setting 40 41 The Western Cape had 33 district hospitals: 28 small (< 150 beds), 3 medium (150 – 299 42 43 beds), and 2 large (300 – 600 beds). Small district hospitals in this study operated as 44 generalist environments with family physicians as the most senior clinicians (one district 45 hospital had an internal medicine physician running their in-patient COVID ward). A family 46 47 physician is a specialist in family medicine. In South Africa, family physicians are trained for 48 the district hospital setting as well as primary care. 49 50 In terms of the continuum of care for COVID-19, these district hospitals received patients 51 52 from the public sector primary care facilities in their catchment area. The private sector 53 could also refer patients, without insurance, to the district hospitals. Patients in primary 54 care, with more than mild symptoms, were referred for further management; although 55 56 those requiring critical or intensive care could be referred directly to regional or tertiary 57 hospitals. Therefore the profile of patients seen and treated at district hospitals will be 58 59 different to those referred to regional and tertiary hospitals as the capacity for critical care 60 and intensive care was much less or non-existent. These small district hospitals typically had

6 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 28 BMJ Open

1 2 emergency centres and re-organised their wards into ‘hot’ and ‘cold’ streams for COVID-19. 3 Patients could be intubated and ventilated in the emergency centre, prior to transfer, and 4 5 there was not usually access to high flow oxygen, which was only installed in June in some 6 hospitals. District hospitals could discharge patients to home, transfer them to a field 7 8 hospital or to higher-level care. 9 10 Study population and selection of participants 11 12 The study engaged the Stellenbosch University Family Physician Research Network 13 (SUFPREN) and the family physicians within that network who worked at small district 14 15 hospitals in the province. Eight district hospitals, four from the Metro Health Services (MHS) 16 in Cape Town and three in the Rural Health Services (RHS), opted to take part. The study 17 18 excluded medium-largeFor metropolitan peer district review hospitals that only were organised along specialist 19 departmental lines and which delivered a different package of care. George Regional 20 21 Hospital was also included in the RHS, as its Department of Family Medicine offered district 22 hospital services to the surrounding area and ran the emergency centre (EC). 23 24 From these eight facilities, all patients who presented to the EC and who tested positive for 25 COVID-19 were included between March-June 2020. There was no sampling or other 26 27 exclusion criteria. 28 29 Data collection 30 31 Patients were identified from the laboratory results and their folders drawn from the 32 33 records department. Data was extracted from the medical records using a standardised data 34 collection tool by the family physicians at each hospital. Data was collected electronically 35 using RedCap software on internet-connected devices available to the researchers. 36 37 The data collection tool was designed to collect data on the objectives listed above and the 38 39 tool was validated by all the family physicians involved in the study prior to data collection. 40 The tool was piloted by one district hospital prior to use. 41 42 43 Data analysis 44 45 Data was exported from Redcap to the Statistical Package for Social Sciences (SPSS) version 46 26. There was no missing data. Means with standard deviations were used to describe 47 continuous variables that were normally distributed and medians with interquartile ranges 48 49 to describe continuous data that were not normally distributed. Categorical data was 50 analysed using frequencies and percentages. 51 52 Categorical variables were compared by Pearson’s Chi-squared test. An independent T-test 53 54 was used to compare continuous variables with binary variables if data was normally 55 distributed and a Mann-Whitney U test if data was not normally distributed. Analysis of 56 57 variance (ANOVA) was used to compare nominal variables with normally distributed numeric 58 variables. 59 60

7 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 28

1 2 Univariate binary logistic regression was used to determine odds ratios for factors that might 3 be associated with mortality (age, sex, co-morbidities and location). Factors with a p value 4 5 <0.1 were then entered into a multiple variable forward stepwise binary logistic regression 6 to determine which factors remained significantly associated with mortality. 7 8 9 Results 10 11 Study sample 12 13 Overall 1376 patients were included in the study for this period and 1017 (73.9%) were from 14 the MHS and 359 (26.1%) from RHS. The mean age of patients was 46.3 years (SD 16.3 years) 15 16 and patients were significantly older in the MHS compared to RHS (MHS 48.1 years (SD 17 16.1), RHS 41.4 years (SD 15.7), mean difference 6.73 years (4.80-8.66), p<0.001). Overall, 18 For peer review only 19 there were 571 (41.5%) males and 805 (58.5%) females. There were significantly more 20 females in the MHS sample than RHS (60.9% vs 51.8% female, p=0.003) 21 22 Presentation and assessment 23 24 Table 1 shows the characteristics of patients on arrival at the EC. Only 10% of patients were 25 26 referred from public sector primary care facilities and the majority (70.6%) were self- 27 referred. Patients were significantly more likely to be referred from a primary care facility in 28 29 the RHS as shown in Table 1. 30 31 The commonest symptoms were cough, shortness of breath, fever, body pains/myalgia and 32 sore throat. The commonest abnormal clinical signs were a raised respiratory rate, 33 tachycardia, hyperglycaemia, decreased oxygen saturation and raised systolic blood 34 35 pressure. The impression was that MHS patients were sicker (more dyspnoea, lower oxygen 36 saturation, more confused) than those in the RHS at presentation and RHS patients had 37 38 more symptoms from an earlier phase of the disease (sore throat, myalgia, nasal symptoms, 39 cough). 40 41 Table 2 presents the co-morbidities of patients. A third of patients had no known co- 42 43 morbidities. The commonest co-morbidities were hypertension, type-2 diabetes, 44 overweight/obesity and HIV. Patients in the MHS had significantly more co-morbidities than 45 those in the RHS. On arrival the levels of prior control for many co-morbidities was unknown, 46 47 particularly in the rural areas. People with type 2 diabetes had the highest proportion that 48 were uncontrolled. 49 50 Table 3 presents the initial assessment and final disposition from the EC. There was some 51 52 mismatch between the initial triage and final assessment, with 53.1% of patients triaged as 53 yellow and 17.1% of patients triaged as orange (triage utilised the South African Triage 54 Scale(12)) being clinically assessed as mild COVID, and half (49.1%) of all patients were 55 56 assessed as mild. Patients from the RHS were significantly more likely to be assessed as mild 57 and less likely to be admitted. Overall 42.2% were discharged home from the EC, 47.4% 58 59 admitted and only 6.2% transferred immediately to a higher level of care. 60 Management in hospital 8 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 28 BMJ Open

1 2 Most patients were admitted as people under investigation and managed without a 3 definitive diagnosis, as the turnaround time (TOT) for the test result (median 3.0 days (IQR 4 5 2.0-5.0 days)) was longer than the length of stay (median 2.0 day (IQR 2.0-3.0)). There was 6 no difference in the TOT between MHS and RHS (MHS median 3.0 days (IQR 3.0-4.0) and RHS 7 8 2.0 days (IQR 2.0-3.0), p=0.113). 9 10 11 Half of all patients did not receive a chest radiograph in the EC or on admission (695 (50.5%)) 12 and this was significantly more likely in rural areas (MHS 41.9% vs RHS 74.7%, p<0.001). The 13 typical appearance was bilateral changes in the lower or mid-zone, a ground glass 14 15 appearance or consolidation. There were no major differences in the radiograph findings 16 between MHS and RHS. 17 18 For peer review only 19 Table 4 presents the treatments for COVID-19. Only 40.6% received any form of oxygen 20 therapy and very few were intubated (0.8%). Those admitted were usually treated with 21 22 oxygen, low molecular weight heparin (enoxaparin sodium), and antibiotics (ceftriaxone, 23 azithromycin or co-amoxiclav), and far fewer patients were treated with proning or steroids. 24 25 Proning was only used in 24% of those with severe or critical COVID-19. 26 27 Table 5 presents the final clinical outcomes. The overall mortality rate was 11.0% and 19.6% 28 29 for those admitted. Mortality rose to 57.3% for those that were critically ill, 21.8% for severe 30 cases, 7.5% for moderate cases and 1.8% for mild. Half of those admitted were discharged 31 32 home (49.7%) and only 8.7% were transferred to a higher level of care and this was 33 significantly more likely in the MHS (p<0.001). The MHS also made use of the field hospitals. 34 35 Risk factors for mortality 36 37 In the multiple variable analysis increasing age, male sex, overweight/obesity, type 2 38 39 diabetes, chronic kidney disease, cardiac failure, HIV and treatment for cancer were all 40 independently associated with a higher risk of death (Table 6). Chronic respiratory 41 42 conditions, such as asthma, COPD, post-TB lung damage and tobacco smoking were not 43 associated with increased risk of mortality. Hypertension and hypercholesterolemia were 44 also not retained as risk factors in the multiple variable analysis. There was no difference in 45 46 risk of mortality between the MHS and RHS. In addition there was a significant relationship 47 between poorer categories of diabetic control and mortality (OR 2.28 ((95%CI 1.25-4.16), 48 49 p=0.007), but not with poor control of HIV (unsuppressed viral load) (OR 1.96 (95% CI 0.62- 50 6.23), p=0.253). There was also no significant relationship with being on antiretroviral 51 treatment (OR 0.879 (95% CI 0.524-1.48), p=0.627) 52 53 54 Discussion 55 56 Contribution of district hospitals to the health system response 57 58 The majority of people with COVID-19 were self-referred and bypassed the gatekeeping role 59 expected of public sector primary care facilities. Primary care facilities are often closed on 60 weekends and afterhours making the district hospital EC the next available point of access. 9 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 28

1 2 Primary care facilities also de-escalated during the epidemic and people were turned away 3 or advised to stay away as much as possible,(13) while there was no restrictions on access to 4 5 district hospitals. In some areas primary care facilities temporarily closed when staff were 6 infected. Although primary care facilities could test, the district hospitals had larger and 7 8 more visible testing centres. This, in addition to fear of more severe disease and more trust 9 in hospital-based services, may have led people to present directly to the hospital. In rural 10 11 areas, where the hospital was more geographically distant, more people entered the system 12 via primary care facilities. District hospitals, therefore, played an important role in primary 13 care during the epidemic and this is reflected in the findings that 49% of patients seen had 14 15 mild COVID-19 and 42% were immediately discharged home from the EC. This also reflects 16 the need to improve access to primary care in South Africa.(14) 17 18 District hospitals reducedFor pressure peer on tertiary review hospitals with only ICU and critical care beds 19 20 particularly for those with moderate and severe COVID-19. Of those with moderate COVID- 21 19 only 9% were transferred to tertiary hospitals and mortality was 7.5%. For those with 22 severe disease 17% were transferred and mortality was 22%, which compares favourably 23 24 with reports from high income settings.(15) District hospitals were unable to manage 25 critically ill patients, as they did not have ICU or critical care facilities. 26 27 The introduction of field hospitals, such as at the Cape Town International Convention 28 29 Centre,(16) also took pressure off acute hospitals and 13.5% of patients were transferred for 30 ongoing care thus creating additional capacity at the district hospital. The first field hospital 31 32 only opened in June 2020, the last month of this study period. At the field hospital care 33 might include management of co-morbidities, palliative care or further recovery prior to 34 discharge home.(16) 35 36 37 Management of COVID-19 at district hospitals 38 39 Patients presented with the typical symptoms of COVID-19 that have also been reported 40 elsewhere in Africa.(17) Fever, however, was only found in 12% of patients and many 41 patients had hyperglycaemia and high blood pressure, reflecting underlying co-morbidities. 42 43 Family physicians reported that the procedures to use infrared thermometers to triage 44 patients may not have resulted in accurate measurements and therefore fever may be 45 46 under-observed.(18) Not all patients had a body mass index measured and 47 overweight/obesity was probably under-reported. 48 49 As the median TOT for COVID-19 tests was longer than the median length of admission, most 50 51 patients were managed as people under investigation, who were presumed to have COVID- 52 19. The overall length of stay was very short with a median duration of 1-day and 75th centile 53 of 3-days. The laboratory capacity and turnaround time for COVID-19 tests has been labelled 54 55 “the Achilles heel” of the local response to the pandemic.(19) The in-hospital turnaround 56 time improved once the criteria for community testing were changed, from all people with 57 58 relevant symptoms to only those over 55-years or with co-morbidities or with more than 59 mild disease.(20) 60

10 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 28 BMJ Open

1 2 Antibiotics were presumably given as the diagnosis was not confirmed in the majority of 3 patients and the patient were treated as a community acquired pneumonia with COVID-19 4 5 as part of the differential diagnosis. Steroids were only given in 28% of patients, but the 6 evidence of their effectiveness were only announced in June 2020,(21) which was the last 7 8 month of this study period. 9 10 Basic imaging with a chest radiograph was only performed in half the patients, which mirrors 11 the large number of mild cases who did not require imaging. All district hospitals had access 12 13 to radiography, although not always 24-hours a day and in some cases the clinician may have 14 judged that management of the patient would not be changed. In addition, radiography 15 might have been avoided in order to reduce exposure of people to COVID-19 as patients 16 17 would have to traverse the hospital to access the radiography unit if a mobile radiograph 18 was not available. NotFor all critically peer ill received review a radiograph only and this might be because they 19 20 died before a radiograph could be taken or were transferred. Patients were referred on the 21 basis of the clinical picture and a radiograph was not essential. 22 23 Clinical outcomes at district hospitals 24 25 Overall in-hospital mortality for COVID-19 was 11% and rose to 20% in those admitted. 26 27 Mortality was significantly higher in MHS facilities compared to RHS and this most likely 28 reflects the higher severity of patients, rather than differences in management of patients. 29 30 Mortality rates reported from a field hospital in Ethiopia were much lower (5.3%) despite a 31 similar profile of COVID-19 severity in those seen.(22) This may be explained by much lower 32 levels of co-morbidity. For example no HIV or TB was reported and diabetes was found in 33 34 14% as opposed to 25% of patients. 35 36 A local population cohort study also found an association between people living with HIV 37 (PLWH) and mortality from COVID-19 and no clear association with viraemia or 38 39 immunosuppression.(23) However, in district hospitals we found no association with 40 previous or current TB infection and mortality from COVID-19, while the population cohort 41 study found a twofold increase. The population study may have overestimated risks as 42 43 routine data did not include all co-morbidities and these patients were more likely to be 44 followed up and diagnosed with COVID-19. Other respiratory conditions such as asthma, 45 46 COPD and post-TB structural lung disease were also not associated with increased mortality. 47 This could suggest that immunological factors were maybe more important than pulmonary 48 49 factors in determining risk of death. Other conditions with impaired or altered immunity 50 were also associated with mortality, such as type 2 diabetes and people receiving treatment 51 for cancer. 52 53 The combination of increasing age, overweight/obesity, type-2 diabetes, chronic kidney 54 55 disease and cardiac failure as independent risk factors is important as 1 in 2 South Africans 56 over the age of 45-years have pre-diabetes or diabetes.(24) Increasing age, obesity and 57 58 chronic kidney disease were also identified as key risk factors in the Democratic Republic of 59 Congo (25) and diabetes in Ethiopia. The importance of non-communicable diseases, such as 60 diabetes, was brought into the public spotlight by COVID-19 as previously the focus of 11 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 28

1 2 attention was on HIV and TB. The level of control of diabetes was also directly related to risk 3 of mortality, which emphasized the need to improve self-management and treatment, 4 5 particularly as 57% had a history of poor control on admission. The absence of electronic 6 medical records and an integrated health information system explains why hospitals could 7 8 not obtain information on prior control of chronic diseases, particularly in the RHS. The 9 continuity of care for type-2 diabetes was also disrupted by the de-escalation of services and 10 11 most forms of patient education and counselling were stopped. There is a need to innovate 12 new ways of managing and empowering people with diabetes while reducing their risk of 13 exposure to COVID-19 at health facilities and support groups. 14 15 All results were dependent on the completeness and accuracy of medical records. For 16 17 example, clinicians may not have recorded all the symptoms experienced by the patients. 18 Family physicians didFor not report peer a problem review with missing medical only records. It would have been 19 20 helpful to record the anti-retroviral treatment regimen in patients with HIV and to 21 investigate any association between mortality and exposure to different combinations of 22 medications. 23 24 These hospitals all had family physicians heading their clinical teams and it is possible that 25 26 they had a higher quality of care than the hospitals that are still without FPs. The Western 27 Cape also has better infrastructure and a stronger health system than many other provinces 28 29 and the quality of care is likely to be lower in other parts of South Africa. 30 31 Conclusions 32 33 District hospitals provided an essential primary care service for many patients with mild 34 symptoms of COVID-19 during the epidemic. This also represented a deficiency in access to 35 36 and utilisation of primary care. District hospitals successfully treated a large number of 37 people with moderate-severe COVID-19 who did not need ventilation and took pressure off 38 39 higher-level facilities. Limited laboratory capacity meant that most patients were treated as 40 people under investigation without a definitive diagnosis. The clinical picture was similar to 41 that reported elsewhere. Mortality at this level of care was associated with increasing age, 42 43 male sex, HIV, type 2 diabetes, overweight/obesity, cardiac failure, chronic kidney disease 44 and treatment for cancer, but not with hypertension, TB, asthma, COPD or post-TB lung 45 46 damage. Patients in the MHS were more numerous, had more co-morbidity and more severe 47 COVID-19 disease than in the RHS. 48 49 50 51 52 Acknowledgements 53 54 No other acknowledgements. 55 56 Funding statement 57 58 This research received no specific grant from any funding agency in the public, commercial 59 60 or not-for-profit sectors.

12 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 28 BMJ Open

1 2 Competing interests statement 3 4 The authors declare no competing interests. 5 6 7 Author contributions 8 9 RM and MPV conceptualised the study. All authors except RM collected data. RM analysed 10 the data and drafted the manuscript. All authors commented on and edited the manuscript 11 12 as well as approved the final version. 13 14 Ethical considerations 15 16 The study was approved by the Health Research Ethics Committee at Stellenbosch University 17 (N20/06/032_COVID-19) who granted a waiver of informed consent for collection on 18 For peer review only 19 anonymised data from medical records. Study permission was obtained from the Provincial 20 Government of the Western Cape to conduct the study. 21 22 Data sharing statement 23 24 25 The dataset will be made available on reasonable request from the first author. 26 27 Patient and public involvement 28 29 It was not appropriate or possible to involve patients or the public in the design, or conduct, 30 or reporting, or dissemination plans of our research. 31 32 33 References 34 35 1. COVID-19 coronavirus pandemic [Internet]. 2020 [cited 2020 Jun 6]. Available from: 36 https://www.worldometers.info/coronavirus/ 37 38 2. Lone SA, Ahmad A. COVID-19 pandemic - an African perspective. Emerg Microbes 39 Infect [Internet]. Taylor & Francis; 2020 Dec [cited 2020 Nov 5];9(1):1300–8. Available 40 from: http://www.ncbi.nlm.nih.gov/pubmed/32458760 41 3. Coronavirus in South Africa [Internet]. 2020 [cited 2020 Jun 6]. Available from: 42 43 https://mediahack.co.za/datastories/coronavirus/dashboard/ 44 4. Chen N, Zhou M, Dong X, Qu J, Gong F, Han Y, et al. Epidemiological and clinical 45 characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a 46 descriptive study. Lancet [Internet]. 2020 Feb 15 [cited 2020 Mar 47 17];395(10223):507–13. Available from: 48 49 http://www.ncbi.nlm.nih.gov/pubmed/32007143 50 5. Zhao X-Y, Xu X-X, Yin H-S, Hu Q-M, Xiong T, Tang Y-Y, et al. Clinical characteristics of 51 patients with 2019 coronavirus disease in a non-Wuhan area of Hubei Province, China: 52 a retrospective study. BMC Infect Dis [Internet]. BioMed Central; 2020 Dec 29 [cited 53 2020 Jun 11];20(1):311. Available from: 54 55 https://bmcinfectdis.biomedcentral.com/articles/10.1186/s12879-020-05010-w 56 6. Guan W, Ni Z, Hu Y, Liang W, Ou C, He J, et al. Clinical Characteristics of Coronavirus 57 Disease 2019 in China. N Engl J Med [Internet]. 2020 Apr 30 [cited 2020 Jun 58 11];382(18):1708–20. Available from: 59 60 http://www.nejm.org/doi/10.1056/NEJMoa2002032 7. Allinder S. The World’s Largest HIV Epidemic in Crisis: HIV in South Africa | Center for 13 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 28

1 2 Strategic and International Studies [Internet]. Centre for Strategic and International 3 Studies. 2019 [cited 2020 Nov 5]. Available from: 4 5 https://www.csis.org/analysis/worlds-largest-hiv-epidemic-crisis-hiv-south-africa# 6 8. Massyn N, Barron P, Day C, Ndlovu N, Padarath A. District Health Barometer 2018/19 7 [Internet]. Cape Town; 2020 [cited 2020 Nov 5]. Available from: 8 https://www.hst.org.za/publications/Pages/DISTRICT-HEALTH-BAROMETER- 9 201819.aspx 10 11 9. Madzimbamuto F. Re-inventing the ventilator. African J Prim Heal Care Fam Med. 12 2020;In Press. 13 10. Western Cape Critical Care Triage Tool. Cape Town; 2020. 14 11. First patients admitted to special Covid-19 field hospital in Cape Town [Internet]. 15 16 [cited 2020 Jun 11]. Available from: https://www.timeslive.co.za/news/south- 17 africa/2020-06-08-first-patients-admitted-to-special-covid-19-field-hospital-in-cape- 18 town/ For peer review only 19 12. The South African Triage Scale (SATS) [Internet]. [cited 2020 Nov 5]. Available from: 20 https://emssa.org.za/special-interest-groups/the-south-african-triage-scale-sats/ 21 22 13. Mash R, Goliath C, Perez G. Re-organising primary health care to respond to the 23 Coronavirus epidemic in Cape Town, South Africa. African J Prim Heal Care Fam Med 24 [Internet]. 2020 Nov 5 [cited 2020 Nov 5];12(1):4. Available from: 25 https://phcfm.org/index.php/phcfm/article/view/2607/4263 26 27 14. Bresick G, von Pressentin KB, Mash R. Evaluating the performance of South African 28 primary care: a cross-sectional descriptive survey. South African Fam Pract [Internet]. 29 Taylor & Francis; 2019 May 4 [cited 2019 Jul 15];61(3):109–16. Available from: 30 https://www.tandfonline.com/doi/full/10.1080/20786190.2019.1596666 31 15. Santorelli G, Sheldon T, West J, Cartwright C, Wright J. COVID-19 in-patient hospital 32 33 mortality by ethnicity. Wellcome Open Res [Internet]. F1000 Research Limited; 2020 34 May 4 [cited 2020 Nov 5];5:86. Available from: 35 https://wellcomeopenresearch.org/articles/5-86/v1 36 16. Reid S, Ras T, Pressentin K von. The Cape Town International Convention Centre from 37 the inside: The family physicians’ view of the ‛Hospital of Hope’. African J Prim Heal 38 39 Care Fam Med [Internet]. 2020 Oct 26 [cited 2020 Nov 5];12(1):4. Available from: 40 https://phcfm.org/index.php/phcfm/article/view/2667/4331 41 17. Olumade TJ, Uzairue LI. Clinical characteristics of 4490 COVID-19 patients in Africa: A 42 meta-analysis. medRxiv [Internet]. Cold Spring Harbor Laboratory Press; 2020 Oct 21 43 44 [cited 2020 Nov 6];2020.10.20.20215905. Available from: 45 https://www.medrxiv.org/content/10.1101/2020.10.20.20215905v1 46 18. Chen H-Y, Chen A, Chen C. Investigation of the Impact of Infrared Sensors on Core 47 Body Temperature Monitoring by Comparing Measurement Sites. Sensors [Internet]. 48 Multidisciplinary Digital Publishing Institute (MDPI); 2020 May 19 [cited 2020 Nov 49 50 5];20(10). Available from: http://www.ncbi.nlm.nih.gov/pubmed/32438729 51 19. Porter JD, Mash R, Preiser W. Turnaround times – the Achilles’ heel of community 52 screening and testing in Cape Town, South Africa: A short report. African J Prim Heal 53 Care Fam Med [Internet]. 2020 Oct 2 [cited 2020 Nov 5];12(1):3. Available from: 54 55 https://phcfm.org/index.php/phcfm/article/view/2624/4257 56 20. David N, Mash R. Community-based screening and testing for Coronavirus in Cape 57 Town, South Africa: Short report. African J Prim Heal Care Fam Med [Internet]. 2020 58 Jun 3 [cited 2020 Jun 6];12(1):3. Available from: 59 https://phcfm.org/index.php/phcfm/article/view/2499 60 21. The RECOVERY Collaborative Group. Dexamethasone in Hospitalized Patients with 14 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 28 BMJ Open

1 2 Covid-19 — Preliminary Report. N Engl J Med [Internet]. Massachusetts Medical 3 Society; 2020 Jul 17 [cited 2020 Nov 5];NEJMoa2021436. Available from: 4 5 http://www.nejm.org/doi/10.1056/NEJMoa2021436 6 22. Leulseged TW, Hassen IS, Maru EH, Zewde WC, Chamiso NW, Bayisa AB, et al. 7 Characteristics and outcome profile of Hospitalized African COVID-19 patients: The 8 Ethiopian Context. medRxiv [Internet]. Cold Spring Harbor Laboratory Press; 2020 Oct 9 28 [cited 2020 Nov 6];2020.10.27.20220640. Available from: 10 11 https://www.medrxiv.org/content/10.1101/2020.10.27.20220640v1 12 23. Boulle A, Davies M-A, Hussey H, Ismail M, Morden E, Vundle Z, et al. Risk factors for 13 COVID-19 death in a population cohort study from the Western Cape Province, South 14 Africa. Clin Infect Dis [Internet]. Clin Infect Dis; 2020 Aug 29 [cited 2020 Nov 6]; 15 16 Available from: http://www.ncbi.nlm.nih.gov/pubmed/32860699 17 24. The Society for Endocrinology M and D of SAT 2 DGEC. SEMDSA Guidelines for the 18 ManagementFor of Type peer2 Diabetes |review Journal of Endocrinology, only Metabolism and Diabetes 19 of South Africa. JEMDSA [Internet]. 2017 [cited 2019 Mar 26];22(1 (Supplement)):S1– 20 196. Available from: http://jemdsa.co.za/index.php/JEMDSA/article/view/647 21 22 25. Nachega JB, Ishoso DK, Otokoye JO, Hermans MP, Machekano RN, Sam-Agudu NA, et 23 al. Clinical Characteristics and Outcomes of Patients Hospitalized for COVID-19 in 24 Africa: Early Insights from the Democratic Republic of the Congo. Am J Trop Med Hyg 25 [Internet]. Am J Trop Med Hyg; 2020 Oct 2 [cited 2020 Nov 5]; Available from: 26 27 http://www.ncbi.nlm.nih.gov/pubmed/33009770 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

15 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 28

1 2 Table 1: Characteristics of patients on arrival at EC. 3 4 All Metro Rural p value 5 6 N=1376 N=1017 N=359 7 n (%) n (%) n (%) 8 Source of referral 9 10 Primary care facility 137 (10.0) 69 (6.8) 68 (18.9) 11 Private general 147 (10.7) 105 (10.3) 42 (11.7) 12 13 practice <0.001 14 Self-referral 972 (70.6) 736 (72.4) 235 (65.5) 15 16 Higher hospital 5 (0.4) 5 (0.5) 0 (0.0) 17 Other 40 (2.9) 26 (2.6) 14 (3.9) 18 For peer review only 19 Not known 75 (5.5) 75 (7.4) 0 (0.0) 20 Symptoms recorded on arrival 21 Sore throat 332 (24.1) 186 (18.3) 146 (40.7) <0.001 22 23 Nasal symptoms 93 (6.8) 61 (6.0) 32 (8.9) 0.059 24 Body pains / myalgia 360 (26.2) 248 (24.4) 112 (31.2) 0.012 25 26 Fever 441 (32.0) 302 (29.7) 139 (38.7) 0.002 27 Cough 877 (63.7) 612 (60.2) 265 (73.8) <0.001 28 29 SOB 717 (52.1) 545 (53.6) 172 (47.9) 0.062 30 Fatigue 181 (13.2) 155 (15.3) 26 (7.2) <0.001 31 32 Loss smell 64 (4.7) 43 (4.2) 21 (5.8) 0.211 33 Loss taste 49 (3.6) 29 (2.9) 20 (5.6) 0.017 34 Diarrhoea 82 (6.0) 70 (6.9) 12 (3.3) 0.015 35 36 Headache 222 (16.1) 157 (15.5) 65 (18.1) 0.240 37 Other 298 (21.7) 250 (24.6) 47 (13.1) <0.001 38 39 Observations recorded on arrival 40 Temperature > 162 (12.4) 96 (10.0) 66 (19.0) <0.001 41 42 37.5C 43 Respiratory rate > 650 (50.5) 505 (52.7) 145 (43.9) 0.007 44 18/min 45 46 Pulse rate > 100/min 646 (47.9) 491 (49.3) 154 (43.8) 0.013 47 Systolic BP < 31 (2.6) 28 (3.2) 3 (1.0) 48 49 90mmHg 0.08 50 Systolic BP > 360 (30.0) 259 (29.3) 101 (32.3) 51 52 140mmHg 53 Diastolic BP < 101 (8.4) 87 (9.8) 14 (4.5) 54 60mmHg 0.008 55 56 Diastolic BP > 217 (18.1) 152 (17.2) 65 (20.7) 57 90mmHg 58 59 Oxygen saturation 557 (41.9) 439 (44.8) 118 (33.8) <0.001 60 <95%

16 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 28 BMJ Open

1 2 Random blood 22 (2.5) 15 (2.1) 7 (3.9) 3 4 glucose < 4.0mmol/l 0.228 5 Random blood 397 (44.3) 324 (45.4) 73 (40.3) 6 glucose >7.8mmol/l 7 8 Alert 1264 (91.9) 921 (90.6) 342 (95.3) 9 Confused but 82 (6.0) 69 (6.8) 13 (3.6) 10 11 responds to verbal 0.034 12 commands 13 14 Unconscious but 10 (0.7) 7 (0.7) 3 (0.8) 15 responds to pain 16 Unconscious with no 6 (0.4) 5 (0.5) 1 (0.3) 17 18 response to pain For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

17 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 28

1 2 Table 2: Co-morbidities of patients with COVID-19 (N=1376) 3 4 All Metro Rural p value 5 6 N=1376 N=1017 N=359 7 n (%) n (%) n (%) 8 Co-morbidity 9 10 Overweight/obesity 269 (19.5) 208 (20.5) 61 (17.0) 0.153 11 Type 1 diabetes 17 (1.2) 15 (1.5) 2 (0.6) 0.175 12 13 Type 2 diabetes 347 (25.2) 273 (26.9) 74 (20.6) 0.019 14 Hypercholesterolemia 83 (6.0) 77 (7.6) 6 (1.7) <0.001 15 16 Hypertension 564 (41.0) 454 (44.7) 109 (30.4) <0.001 17 Cardiac failure 58 (4.2) 51 (5.0) 7 (1.9) 0.013 18 For peer review only 19 Ischaemic heart disease 25 (1.8) 24 (2.4) 1 (0.3) 0.011 20 Asthma 67 (4.9) 52 (5.1) 15 (4.2) 0.477 21 COPD 50 (3.6) 38 (3.7) 12 (3.3) 0.729 22 23 Post-TB lung damage 12 (0.9) 8 (0.8) 4 (1.1) 0.567 24 HIV 195 (14.2) 156 (15.4) 39 (10.9) 0.036 25 26 Active TB on treatment 23 (1.7) 20 (2.0) 3 (0.8) 0.150 27 Previous TB 49 (3.6) 40 (3.9) 9 (2.5) 0.209 28 29 Cancer on treatment 10 (0.7) 9 (0.9) 1 (0.3) 0.244 30 Previous cancer 2 (0.1) 2 (0.2) 0 (0.0) 0.400 31 32 Chronic kidney disease 60 (4.4) 55 (5.4) 5 (1.4) 0.001 33 None 450 (32.7) 276 (27.2) 174 (48.5) <0.001 34 Tobacco smoker 95/621 (15.3) 80/490 (16.3) 15/131 (11.5) 0.168 35 36 Previous control of co-morbidity 37 Diabetes N=348 38 39 Normal HbA1c < 7% 32 (9.2) 30 (10.8) 2 (2.9) 40 Controlled HbA1c 7-8% 41 (11.8) 34 (12.2) 7 (10.1) 0.002 41 42 Uncontrolled HbA1c >8 68 (19.5) 56 (20.1) 12 (17.4) 43 and <10% 44 45 Very uncontrolled 131 (37.6) 110 (39.4) 21 (30.4) 46 >=10% 47 Unknown 76 (21.8) 49 (17.6) 27 (39.1) 48 49 Hypertension N=446 50 Well controlled 165 (37.0) 150 (41.6) 14 (16.7) 51 52 Uncontrolled 88 (19.7) 75 (20.8) 13 (15.5) <0.001 53 Not known 193 (43.3) 136 (37.7) 57 (67.9) 54 55 Asthma N=72 56 Well controlled 23 (31.9) 21 (51.2) 2 (6.5) 57 <0.001 58 Partly controlled 10 (13.9) 6 (14.6) 4 (12.9) 59 Uncontrolled 5 (6.9) 5 (12.2) 0 (0.0) 60 Unknown 34 (47.2) 9 (22.0) 25 (80.6) 18 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 28 BMJ Open

1 2 COPD N=56 3 4 Mild 7 (12.5) 5 (15.6) 2 (8.3) 5 Moderate 11 (19.6) 9 (28.1) 2 (8.3) <0.001 6 Severe 13 (23.2) 12 (37.5) 1 (4.2) 7 8 Unknown 25 (44.6) 6 (18.8) 19 (79.2) 9 HIV N=180 10 11 Well controlled 96 (53.3) 79 (57.7) 17 (39.5) 0.004 12 Uncontrolled 22 (12.2) 15 (10.9) 7 (16.3) 13 14 Unknown 43 (23.9) 25 (18.2) 18 (41.9) 15 No ART 16 (8.9) 16 (11.7) 0 (0.0) 16 17 New on ART 3 (1.7) 2 (1.5) 1 (2.3) 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

19 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 28

1 2 Table 3: Initial assessment and disposition from the emergency centre 3 4 All Metro Rural p value 5 6 N=1376 N=1017 N=359 7 n (%) n (%) n (%) 8 Initial assessment in emergency centre 9 10 Triage green 455 (33.1) 267 (26.4) 188 (53.1) 11 Triage yellow 292 (21.3) 200 (19.7) 91 (25.7) 12 13 Triage orange 516 (37.7) 453 (44.7) 63 (17.8) <0.001 14 Triage red 105 (7.7) 93 (9.2) 12 (3.4) 15 16 Mild COVID 676 (49.1) 443 (43.6) 232 (64.6) 17 Moderate COVID 252 (18.3) 191 (18.8) 61 (17.0) 18 For peer review only <0.001 19 Severe COVID 335 (24.3) 303 (29.8) 32 (8.9) 20 Critical COVID 82 (6.0) 72 (7.1) 10 (2.8) 21 Unknown COVID 31 (2.3) 7 (0.7) 24 (6.7) 22 23 Disposition from emergency centre 24 Discharged home 581 (42.2) 378 (37.2) 203 (56.5) 25 26 Transferred to higher 85 (6.2) 66 (6.5) 19 (5.3) <0.001 27 level 28 29 Transferred to assisted 12 (0.9) 2 (0.2) 10 (2.8) 30 isolation 31 32 Transferred to field 20 (1.5) 19 (1.9) 1 (0.3) 33 hospital 34 Admitted to district 652 (47.4) 531 (52.2) 121 (33.7) 35 36 hospital 37 Died 13 (0.9) 10 (1.0) 3 (0.8) 38 39 Other 13 (0.9) 11 (1.1) 2 (0.6) 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

20 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 28 BMJ Open

1 2 Table 4: Treatment received at the district hospital 3 4 All Admitted Metro Rural p value 5 N=1376 N=625 N=1017 N=359 6 n (%) n (%) n (%) n (%) 7 8 Any form of oxygen 558 (40.6) 461 (70.7) 452 106 <0.001 9 (44.5) (29.5) 10 11 Oxygen by nasal prongs 1- 265 (19.3) 212 (32.5) 226 39 <0.001 12 4L/min (22.2) (10.9) 13 Oxygen by facemask 6- 240 (17.4) 212 (32.5) 207 33 (9.2) <0.001 14 15 10L/min (20.4) 16 Oxygen with non-rebreather 177 (12.9) 130 (19.9) 136 41 0.339 17 18 reservoir bag 10-15L/minFor peer review only(13.4) (11.4) 19 Oxygen high flow >15L/min 12 (0.9) 8 (1.2) 6 (0.6) 6 (1.7) 0.058 20 21 Intubation and ventilation 11 (0.8) 3 (0.5) 5 (0.5) 6 (1.7) 0.031 22 Proning 142 (10.3) 105 (16.1) 135 7 (1.9) <0.001 23 (13.3) 24 25 Clexane any 583 (42.4) 503 (77.1) 504 79 <0.001 26 (49.6) (22.0) 27 28 Clexane 40mg/d 308 (22.4) 272 (41.7) 250 58 0.001 29 (24.6) (16.2) 30 31 Clexane 1mg/Kg daily 218 (15.8) 191 (29.3) 203 15 (4.2) <0.001 32 (20.0) 33 Clexane 1mg/kg BD 74 (5.4) 57 (8.7) 67 (6.6) 7 (1.9) 0.001 34 35 Ceftriaxone 556 (40.4) 460 (70.6) 501 55 <0.001 36 (49.3) (15.3) 37 38 Azithromycin 541 (39.3) 427 (65.5) 440 101 <0.001 39 (43.3) (28.1) 40 41 Co-amoxiclav 166 (12.1) 106 (16.3) 80 (7.9) 86 <0.001 42 (24.0) 43 Any steroid 253 (18.4) 183 (28.1) 196 57 0.151 44 45 (19.3) (15.9) 46 Dexamethasone 111 (8.1) 78 (12.0) 83 (8.2) 28 (7.8) 0.825 47 48 Hydrocortisone 15 (1.1) 12 (1.8) 13 (1.3) 2 (0.6) 0.257 49 Prednisone 131 (9.5) 97 (14.9) 102 29 (8.1) 0.277 50 51 (10.0) 52 53 54 55 56 57 58 59 60

21 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 24 of 28

1 2 3 Table 5: Clinical outcomes at the district hospitals 4 5 Clinical outcome Admitted All Metro Rural Mild Moderate Severe Critical 6 7 N=652 N=1376 N=1017 N=359 N=676 N=252 N=335 N=82 8 n (%) n (%) n (%) n (%) n (%) n (%) n (%) n (%) 9 10 Died 128 (19.6) 151 (11.0) 118 (11.6) 33 (9.2) 12 (1.8) 19 (7.5) 73 (21.8) 47 (57.3) 11 Discharged home 324 (49.7) 831 (60.4) 577 (56.8) 253 (70.5) 525 (77.7) 147 (58.3) 130 (38.8) 7 (8.5) 12 Transferred 57 (8.7) 118For (8.6) 115 peer (11.3) 3 (0.8) review20 (3.0) 23 (9.1)only55 (16.4) 20 (24.4) 13 14 tertiary hospital 15 Transferred 30 (4.6) 47 (3.4) 20 (2.0) 27 (7.5) 26 (3.8) 11 (4.4) 3 (0.9) 0 (0.0) 16 17 assisted isolation 18 Transferred field 88 (13.5) 105 (7.6) 104 (10.2) 1 (0.3) 1.3 (1.9) 28 (11.1) 55 (16.4) 7 (8.5) 19 20 hospital 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 22 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 25 of 28 BMJ Open

1 2 Table 6: Risk factors for death from COVID-19 3 4 Variable Unadjusted odds P value 5 6 ratio (95% CI) 7 Cancer on treatment 12.63 (3.5-42.3) <0.001 8 Chronic kidney disease 10.65 (6.21-18.28) <0.001 9 10 Cancer previous 8.16 (0.51-131.14) 0.138 11 Cardiac failure 5.63 (3.21-9.87) <0.001 12 13 Post-TB SLD 4.14 (1.23-13.9) 0.022 14 Hypertension 3.95 (2.73-5.70) <0.001 15 16 Type 2 diabetes 3.1 (2.2-4.3) <0.001 17 TB on treatment 2.94 (1.14-7.57) 0.026 18 For peer review only 19 Ischaemic heart disease 2.06 (0.76-5.58) 0.154 20 Hypercholesterolemia 1.88 (1.04-3.37) 0.035 21 22 TB previous 1.88 (0.89-3.95) 0.097 23 Overweight/obese 1.69 (1.15-2.49) 0.007 24 HIV 1.67 (1.09-2.56) 0.019 25 26 Male sex 1.63 (1.16-2.30) 0.005 27 COPD 1.34 (0.59-3.03) 0.487 28 29 Asthma 1.11 (0.52-2.36) 0.795 30 Type 1 diabetes 1.08 (0.24-4.78) 0.916 31 32 Increasing age 1.06 (1.05 – 1.07) <0.001 33 Tobacco smoking 1.01 (0.458-2.21) 0.986 34 35 Rural vs metro services 0.77 (0.51-1.16) 0.208 36 No comorbidities 0.17 (0.095-0.30) <0.001 37 Variable Adjusted P value 38 39 Odds ratio (95% CI) 40 Cancer on treatment 7.45 (1.87-29.89) 0.004 41 42 Chronic kidney disease 5.16 (2.82-9.43) <0.001 43 HIV 3.41 (2.06-5.65) <0.001 44 45 Cardiac failure 2.85 (1.52-5.35) 0.001 46 Male sex 2.02 (1.37-2.98) <0.001 47 Type 2 diabetes 1.84 (1.24-2.73) 0.002 48 49 Overweight/obese 1.58 (1.02-2.46) 0.04 50 Increasing age 1.06 (1.04-1.07) <0.001 51 52 53 54 55 56 57 58 59 60

23 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 26 of 28

1 2 3 4 Reporting checklist for cross sectional study. 5 6 7 Based on the STROBE cross sectional guidelines. 8 9 10 11 Instructions to authors 12 13 14 Complete this checklist by entering the page numbers from your manuscript where readers will find 15 16 each of the items listed below.For peer review only 17 18 19 Your article may not currently address all the items on the checklist. Please modify your text to 20 21 include the missing information. If you are certain that an item does not apply, please write "n/a" and 22 23 24 provide a short explanation. 25 26 27 Upload your completed checklist as an extra file when you submit to a journal. 28 29 30 In your methods section, say that you used the STROBE cross sectionalreporting guidelines, and cite 31 32 them as: 33 34 35 von Elm E, Altman DG, Egger M, Pocock SJ, Gotzsche PC, Vandenbroucke JP. The Strengthening 36 37 38 the Reporting of Observational Studies in Epidemiology (STROBE) Statement: guidelines for 39 40 reporting observational studies. 41 42 43 Page 44 45 Reporting Item Number 46 47 48 Title and abstract 49 50 51 Title #1a Indicate the study’s design with a commonly used term in the 1 52 53 54 title or the abstract 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 27 of 28 BMJ Open

1 2 Abstract #1b Provide in the abstract an informative and balanced summary 2 3 4 of what was done and what was found 5 6 7 Introduction 8 9 10 Background / #2 Explain the scientific background and rationale for the 4-5 11 12 rationale investigation being reported 13 14 15 Objectives #3 State specific objectives, including any prespecified 5 16 For peer review only 17 18 hypotheses 19 20 21 Methods 22 23 24 Study design #4 Present key elements of study design early in the paper 5 25 26 27 Setting #5 Describe the setting, locations, and relevant dates, including 5 28 29 periods of recruitment, exposure, follow-up, and data 30 31 collection 32 33 34 35 Eligibility criteria #6a Give the eligibility criteria, and the sources and methods of 6 36 37 selection of participants. 38 39 40 #7 Clearly define all outcomes, exposures, predictors, potential 6 41 42 confounders, and effect modifiers. Give diagnostic criteria, if 43 44 applicable 45 46 47 48 Data sources / #8 For each variable of interest give sources of data and details 6 49 50 measurement of methods of assessment (measurement). Describe 51 52 comparability of assessment methods if there is more than 53 54 one group. Give information separately for for exposed and 55 56 57 unexposed groups if applicable. 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 28 of 28

1 2 Bias #9 Describe any efforts to address potential sources of bias 6 3 4 5 Study size #10 Explain how the study size was arrived at 6 6 7 8 Quantitative #11 Explain how quantitative variables were handled in the 6 9 10 variables analyses. If applicable, describe which groupings were 11 12 chosen, and why 13 14 15 Statistical #12a Describe all statistical methods, including those used to 6-7 16 For peer review only 17 18 methods control for confounding 19 20 21 Statistical #12b Describe any methods used to examine subgroups and 6-7 22 23 methods interactions 24 25 26 Statistical #12c Explain how missing data were addressed 6 27 28 methods 29 30 31 Statistical #12d If applicable, describe analytical methods taking account of 6-7 32 33 34 methods sampling strategy 35 36 37 Statistical #12e Describe any sensitivity analyses NA 38 39 methods 40 41 42 Results 43 44 45 Participants #13a Report numbers of individuals at each stage of study—eg 7 46 47 48 numbers potentially eligible, examined for eligibility, 49 50 confirmed eligible, included in the study, completing follow- 51 52 up, and analysed. Give information separately for for 53 54 exposed and unexposed groups if applicable. 55 56 57 Participants #13b Give reasons for non-participation at each stage NA 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 29 of 28 BMJ Open

1 2 Participants #13c Consider use of a flow diagram NA 3 4 5 Descriptive data #14a Give characteristics of study participants (eg demographic, Table 1 6 7 clinical, social) and information on exposures and potential 8 9 confounders. Give information separately for exposed and 10 11 unexposed groups if applicable. 12 13 14 15 Descriptive data #14b Indicate number of participants with missing data for each NA 16 For peer review only 17 variable of interest 18 19 20 Outcome data #15 Report numbers of outcome events or summary measures. 7-8 21 22 Give information separately for exposed and unexposed 23 24 groups if applicable. 25 26 27 28 Main results #16a Give unadjusted estimates and, if applicable, confounder- 8 29 30 adjusted estimates and their precision (eg, 95% confidence 31 32 interval). Make clear which confounders were adjusted for 33 34 and why they were included 35 36 37 Main results #16b Report category boundaries when continuous variables were NA 38 39 40 categorized 41 42 43 Main results #16c If relevant, consider translating estimates of relative risk into NA 44 45 absolute risk for a meaningful time period 46 47 48 Other analyses #17 Report other analyses done—e.g., analyses of subgroups NA 49 50 and interactions, and sensitivity analyses 51 52 53 54 Discussion 55 56 57 Key results #18 Summarise key results with reference to study objectives 8-10 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 30 of 28

1 2 Limitations #19 Discuss limitations of the study, taking into account sources 11 3 4 of potential bias or imprecision. Discuss both direction and 5 6 magnitude of any potential bias. 7 8 9 Interpretation #20 Give a cautious overall interpretation considering objectives, 11 10 11 limitations, multiplicity of analyses, results from similar 12 13 14 studies, and other relevant evidence. 15 16 For peer review only 17 Generalisability #21 Discuss the generalisability (external validity) of the study 11 18 19 results 20 21 22 Other Information 23 24 25 Funding #22 Give the source of funding and the role of the funders for the 11 26 27 28 present study and, if applicable, for the original study on 29 30 which the present article is based 31 32 33 The STROBE checklist is distributed under the terms of the Creative Commons Attribution License 34 35 CC-BY. This checklist was completed on 16. November 2020 using https://www.goodreports.org/, a 36 37 tool made by the EQUATOR Network in collaboration with Penelope.ai 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open

Article Type: Original research

Date Submitted by the 29-Dec-2020 Author:

Complete List of Authors: Mash, Robert; University of Stellenbosch, Family Medicine and Primary Care Presence-Vollenhoven, Mellisa; University of Stellenbosch, Family Medicine and Primary Care; Western Cape Provincial Government, Metro Health Services Adeniji, Adeloye; University of Stellenbosch, Family Medicine and Primary Care; Western Cape Provincial Government, Rural Health Services Christoffels, Renaldo; University of Cape Town, Family Medicine; Western Cape Provincial Government, Metro Health Services Doubell, Karlien; University of Stellenbosch, Family Medicine and Primary Care; Western Cape Provincial Government, Rural Health Services Eksteen, Lawson; University of Stellenbosch, Family Medicine and Primary Care; Western Cape Provincial Government, Rural Health Services Hendrikse, Amee; Western Cape Provincial Government, Metro Health Services Hutton, Lauren; University of Stellenbosch, Family Medicine and Primary Care; Western Cape Provincial Government, Rural Health Services Jenkins, Louis; University of Stellenbosch, Family Medicine and Primary Care; Western Cape Provincial Government, Rural Health Services Kapp, Paul; University of Stellenbosch, Family Medicine and Primary Care; Western Cape Provincial Government, Rural Health Services Lombard, Annie; University of Stellenbosch, Family Medicine and Primary Care; Western Cape Provincial Government, Metro Health Services Marais, Heleen; Western Cape Provincial Government, Metro Health Services Rossouw, Liezel; University of Cape Town, Family Medicine; Western Cape Provincial Government, Metro Health Services Stuve, Katrin; University of Stellenbosch, Internal Medicine; Western Cape Provincial Government, Metro Health Services Ugoagwu, Abi; University of Cape Town, Family Medicine; Western Cape Provincial Government, Metro Health Services Williams, Beverley; University of Stellenbosch, Family Medicine and Primary Care; Western Cape Provincial Government, Rural Health Services

Primary Subject Infectious diseases Heading:

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 29 BMJ Open

1 2 3 General practice / Family practice, Global health, Health services 4 Secondary Subject Heading: research 5 6 COVID-19, ACCIDENT & EMERGENCY MEDICINE, Respiratory infections Keywords: 7 < THORACIC MEDICINE 8 9 10 11 12 13 14 15 16 For peer review only 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 29

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 29 BMJ Open

1 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 29

2 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 29 BMJ Open

1 2 Keywords 3 4 COVID-19; coronavirus; signs and symptoms; disease management; risk factors; mortality; 5 6 South Africa; district hospitals 7 8 Abstract 9 10 Objectives: To describe the characteristics, clinical management and outcomes of patients 11 12 with COVID-19 at district hospitals. 13 14 Design: A descriptive observational cross-sectional study. 15 16 Setting: District hospitals (4 in Metro and 4 in Rural Health Services) in the Western Cape, 17 South Africa. District hospitals were small (<150 beds) and led by family physicians. 18 For peer review only 19 Participants: All patients who presented to the hospitals’ emergency centre and who tested 20 21 positive for COVID-19 between March-June 2020. 22 23 Primary and secondary outcome measures: Source of referral, presenting symptoms, 24 demographics, co-morbidities, clinical assessment and management, laboratory turn- 25 around-time, clinical outcomes, factors related to mortality, length of stay and location. 26 27 28 Results: 1376 patients (73.9% Metro, 26.1% Rural). Mean age 46.3 years (SD 16.3), 58.5% 29 females. The majority were self-referred (71%) and had co-morbidities (67%): hypertension 30 (41%), type 2 diabetes (25%), HIV (14%) and overweight/obesity (19%). Assessment of 31 32 COVID-19 was mild (49%), moderate (18%) and severe (24%). Test turnaround time (median 33 3.0 days (IQR 2.0-5.0 days)) was longer than length of stay (median 2.0 day (IQR 2.0-3.0)). 34 35 Commonest treatment was oxygen (41%) and only 0.8% were intubated and ventilated. 36 Overall mortality was 11%. Most were discharged home (60%) and only 9% transferred to 37 higher levels of care. Increasing age (OR 1.06 (CI 1.04-1.07)), male (OR 2.02 (CI 1.37-2.98)), 38 39 overweight/obesity (OR 1.58 (CI 1.02-2.46)), type 2 diabetes (OR 1.84 (CI 1.24-2.73)), HIV (OR 40 3.41 (CI 2.06-5.65)), chronic kidney disease (OR 5.16 (CI 2.82-9.43)) were significantly linked 41 42 with mortality (p<0.05). Pulmonary diseases (TB, asthma, chronic obstructive pulmonary 43 disease, post-TB structural lung disease) were not associated with increased mortality. 44 45 Conclusion: District hospitals supported primary care and shielded tertiary hospitals. 46 Patients had high levels of co-morbidities and similar clinical pictures to that reported 47 48 elsewhere. Most patients were treated as people under investigation. Mortality was 49 comparable to similar settings and risk factors identified. 50 51 52 53 54 55 56 57 58 59 60

3 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 29

1 2 Strengths and limitations of this study 3 4 5  The whole study population was included over a four-month period at the height of 6 the epidemic from eight district hospitals with few missing records. 7 8  A comprehensive dataset on presentation, assessment, management and outcomes 9 was captured. 10  The size of the study population enabled identification of risk factors associated with 11 12 mortality 13  Data was dependent on the accuracy of medical records and clinical skills 14 15  District hospitals without family physicians may have different quality of care, which 16 limits generalisability. 17 18 Word count 3929For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

4 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 29 BMJ Open

1 2 Introduction 3 4 Coronavirus disease 2019 (COVID-19) is a global pandemic that has affected all regions of the 5 6 world, although Africa has so far been less affected than predicted.(1) Given the number of 7 poor and developing countries, and relatively weak health systems, the pandemic is 8 9 expected to significantly impact African communities.(2) Within the African continent, South 10 Africa has had the most reported cases and the Western Cape has been one of the leading 11 hotspots.(1,3) 12 13 Clinical findings among confirmed COVID-19 cases in China showed the most common 14 15 complaints were fever (83%), cough (82%), difficulty breathing (31%), fatigue and myalgia 16 (11%).(4) Non-respiratory injury was identified by elevated levels of aspartate 17 18 aminotransferase (20%),For creatinine peer (6%), review and creatine kinase only (15%).(5) 19 20 Findings from another Chinese study showed that older patients and those with comorbidity 21 had poorer clinical outcomes.(5) Multi-morbidity was also correlated with poorer clinical 22 23 outcomes.(6) Co-morbidity with cardiopulmonary diseases was a particular concern, such as 24 diabetes, hypertension, asthma and chronic obstructive pulmonary disease (COPD). In South 25 Africa there was also concern with regard to the large numbers of people with HIV, 26 27 particularly those who were immunocompromised, due to no or inadequate anti-retroviral 28 treatment. South Africa has 20% of all people living with HIV in the world and this is 29 30 accompanied by a high incidence of TB.(7,8) Patients with active TB or post-TB structural 31 lung damage were also a concern in our context. 32 33 Most clinical research has focused on tertiary hospitals with high-care or intensive care units 34 (ICU). Little is known, therefore, about the types of patients seen at district hospitals and 35 36 their clinical course with the expertise and equipment available at this level of the health 37 system and in this African context. In South Africa, district hospitals usually have less than 38 39 150 beds and are run by generalists with male, female, maternity and paediatrics wards as 40 well as emergency centres (EC). They are the first referral point, particularly in remote and 41 rural areas, for patients from primary care. 42 43 44 In Africa, primary or district hospitals do not provide intensive or critical care and have 45 limited capacity for prolonged ventilation. At the same time tertiary referral hospitals may 46 not be able to receive patients if their ICU facilities are full or they are very distant. 47 48 Elsewhere in Africa it has been suggested that district hospitals should focus more on the 49 provision of oxygen therapy as a more valuable intervention than ventilation, as ventilation 50 51 requires adequate equipment and expertise, with a risk of harm to the patient and 52 transmission of the virus during intubation.(9) Guidelines have also been put in place to 53 determine which patients should be prioritised for ICU and which critically ill patients should 54 55 be managed with oxygen and if necessary palliative care.(10) In June 2020 Cape Town also 56 opened field hospitals, which will have the ability to manage patients with COVID-19 at a 57 58 level of care similar to a district hospital.(11) 59 60 This study will describe the type of patients referred to district hospitals run by family physicians in the Western Cape and evaluate their presentation, clinical management and 5 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 29

1 2 outcomes. This data will help to provide a more complete picture of how COVID-19 is 3 affecting our population as the patient population seen at district hospitals is different to 4 5 that seen at tertiary hospitals and African populations may differ from those in Europe, Asia 6 or America. 7 8 The aim of this study, therefore, was to describe the characteristics, clinical management 9 10 and outcomes of patients with COVID-19 presenting to district hospitals in the Western Cape 11 from March 2020 to June 2020. The specific objectives were to describe the original source 12 13 of referral, the presenting symptoms, the patients’ demographics, the presence of co- 14 morbidities, the clinical assessment and management, the turn-around-time for laboratory 15 results, the clinical outcomes and factors related to mortality, the length of stay and to 16 17 compare district hospitals in rural and metro health services. 18 For peer review only 19 Methods 20 21 Study design 22 23 This was a descriptive observational cross-sectional study by means of a retrospective audit 24 25 of medical records. 26 27 Setting 28 29 The Western Cape had 33 district hospitals: 28 small (< 150 beds), 3 medium (150 – 299 30 31 beds), and 2 large (300 – 600 beds). Small district hospitals in this study operated as 32 generalist environments with family physicians as the most senior clinicians (one district 33 hospital had an internal medicine physician running their in-patient COVID ward). A family 34 35 physician is a specialist in family medicine. In South Africa, family physicians are trained for 36 the district hospital setting as well as primary care. 37 38 In terms of the continuum of care for COVID-19, these district hospitals received patients 39 40 from the public sector primary care facilities in their catchment area. The private sector 41 could also refer patients, without insurance, to the district hospitals. Patients in primary 42 43 care, with more than mild symptoms, were referred for further management; although 44 those requiring critical or intensive care could be referred directly to regional or tertiary 45 hospitals. Therefore the profile of patients seen and treated at district hospitals will be 46 47 different to those referred to regional and tertiary hospitals as the capacity for critical care 48 and intensive care was much less or non-existent. These small district hospitals typically had 49 50 ECs and re-organised their wards into ‘hot’ and ‘cold’ streams for COVID-19. Patients could 51 be intubated and ventilated in the EC, prior to transfer, and there was not usually access to 52 high flow oxygen, which was only installed in June in some hospitals. District hospitals could 53 54 discharge patients to home, transfer them to a field hospital or to higher-level care. 55 56 Study population and selection of participants 57 58 The study engaged the Stellenbosch University Family Physician Research Network 59 60 (SUFPREN) and the family physicians within that network who worked at small district hospitals in the province. Eight district hospitals, four from the Metro Health Services (MHS) 6 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 29 BMJ Open

1 2 in Cape Town and three in the Rural Health Services (RHS), opted to take part. The study 3 excluded medium-large metropolitan district hospitals that were organised along specialist 4 5 departmental lines and which delivered a different package of care. George Regional 6 Hospital was also included in the RHS, as its Department of Family Medicine offered district 7 8 hospital services to the surrounding area and ran the EC. 9 10 From these eight facilities, all patients who presented to the EC and who tested positive for 11 COVID-19 were included between March-June 2020. There was no sampling or other 12 13 exclusion criteria. 14 15 Data collection 16 17 Patients were identified from the laboratory results and their folders drawn from the 18 records department.For Data was peer extracted reviewfrom the medical only records using a standardised data 19 20 collection tool by the family physicians at each hospital. Data was collected electronically 21 using REDCap software on internet-connected devices available to the researchers. 22 2 23 Overweight and obesity were defined as a body mass index greater than 25 kg/m . The 24 South African Triage Scale was used by clinicians where red is for an immediate emergency, 25 orange is very urgent, yellow is urgent, and green is non-urgent.(12) COVID-19 was classified 26 27 according to clinical guidelines issued by the Western Cape Government: Health.(13) 28 29 The data collection tool was designed to collect data on the objectives listed above and the 30 tool was validated by all the family physicians involved in the study prior to data collection. 31 32 The tool was piloted by one district hospital prior to use. 33 34 Data analysis 35 36 Data was exported from REDCap to the Statistical Package for Social Sciences (SPSS) version 37 26. There was no missing data. Means with standard deviations were used to describe 38 39 continuous variables that were normally distributed and medians with interquartile ranges 40 to describe continuous data that were not normally distributed. Categorical data was 41 42 analysed using frequencies and percentages. 43 44 Categorical variables were compared by Pearson’s Chi-squared test. An independent T-test 45 was used to compare continuous variables with binary variables if data was normally 46 47 distributed and a Mann-Whitney U test if data was not normally distributed. Analysis of 48 variance (ANOVA) was used to compare nominal variables with normally distributed numeric 49 variables. 50 51 Univariate binary logistic regression was used to determine odds ratios for factors that might 52 53 be associated with mortality (age, sex, co-morbidities and location). Factors with a p value 54 <0.1 were then entered into a multiple variable forward stepwise binary logistic regression 55 56 to determine which factors remained significantly associated with mortality. 57 58 Results 59 60 Study sample

7 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 29

1 2 Overall 1376 patients were included in the study for this period and 1017 (73.9%) were from 3 the MHS and 359 (26.1%) from RHS. The mean age of patients was 46.3 years (SD 16.3 years) 4 5 and patients were significantly older in the MHS compared to RHS (MHS 48.1 years (SD 6 16.1), RHS 41.4 years (SD 15.7), mean difference 6.73 years (4.80-8.66), p<0.001). Overall, 7 8 there were 571 (41.5%) males and 805 (58.5%) females. There were significantly more 9 females in the MHS sample than RHS (60.9% vs 51.8% female, p=0.003) 10 11 Presentation and assessment 12 13 Table 1 shows the characteristics of patients on arrival at the EC. Only 10% of patients were 14 15 referred from public sector primary care facilities and the majority (70.6%) were self- 16 referred. Patients were significantly more likely to be referred from a primary care facility in 17 18 the RHS as shown inFor Table 1. peer review only 19 20 The commonest symptoms were cough, shortness of breath, fever, body pains/myalgia and 21 sore throat. The commonest abnormal clinical signs were a raised respiratory rate, 22 23 tachycardia, hyperglycaemia, decreased oxygen saturation and raised systolic blood 24 pressure. The impression was that MHS patients were sicker (more dyspnoea, lower oxygen 25 saturation, more confused) than those in the RHS at presentation and RHS patients had 26 27 more symptoms from an earlier phase of the disease (sore throat, myalgia, nasal symptoms, 28 cough). 29 30 Table 2 presents the co-morbidities of patients. A third of patients had no known co- 31 32 morbidities. The commonest co-morbidities were hypertension, type-2 diabetes, 33 overweight/obesity and HIV. Patients in the MHS had significantly more co-morbidities than 34 those in the RHS. On arrival the levels of prior control for many co-morbidities was unknown, 35 36 particularly in the rural areas. People with type 2 diabetes had the highest proportion that 37 were uncontrolled. 38 39 Table 3 presents the initial assessment and final disposition from the EC. There was some 40 41 mismatch between the initial triage and final assessment, with 53.1% of patients triaged as 42 yellow and 17.1% of patients triaged as orange, being clinically assessed as mild COVID, and 43 44 half (49.1%) of all patients were assessed as mild. Patients from the RHS were significantly 45 more likely to be assessed as mild and less likely to be admitted. Overall 42.2% were 46 discharged home from the EC, 47.4% admitted and only 6.2% transferred immediately to a 47 48 higher level of care. 49 50 Management in hospital 51 52 Most patients were admitted as people under investigation and managed without a 53 54 definitive diagnosis, as the turnaround time (TOT) for the test result (median 3.0 days (IQR 55 2.0-5.0 days)) was longer than the length of stay (median 2.0 day (IQR 2.0-3.0)). There was 56 no difference in the TOT between MHS and RHS (MHS median 3.0 days (IQR 3.0-4.0) and RHS 57 58 2.0 days (IQR 2.0-3.0), p=0.113). 59 60

8 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 29 BMJ Open

1 2 Half of all patients did not receive a chest radiograph in the EC or on admission (695 (50.5%)) 3 and this was significantly more likely in rural areas (MHS 41.9% vs RHS 74.7%, p<0.001). The 4 5 typical appearance was bilateral changes in the lower or mid-zone, a ground glass 6 appearance or consolidation. There were no major differences in the radiograph findings 7 8 between MHS and RHS. 9 10 11 Table 4 presents the treatments for COVID-19. Only 40.6% received any form of oxygen 12 therapy and very few were intubated (0.8%). Those admitted were usually treated with 13 oxygen, low molecular weight heparin (enoxaparin sodium), and antibiotics (ceftriaxone, 14 15 azithromycin or co-amoxiclav), and far fewer patients were treated with proning or steroids. 16 Proning was only used in 24% of those with severe or critical COVID-19. 17 18 For peer review only 19 Table 5 presents the final clinical outcomes. The overall mortality rate was 11.0% and 19.6% 20 for those admitted. Mortality rose to 57.3% for those that were critically ill, 21.8% for severe 21 22 cases, 7.5% for moderate cases and 1.8% for mild. Half of those admitted were discharged 23 home (49.7%) and only 8.7% were transferred to a higher level of care and this was 24 25 significantly more likely in the MHS (p<0.001). The MHS also made use of the field hospitals. 26 27 Risk factors for mortality 28 29 In the multiple variable analysis increasing age, male sex, overweight/obesity, type 2 30 diabetes, chronic kidney disease, cardiac failure, HIV and treatment for cancer were all 31 32 independently associated with a higher risk of death (Table 6). Chronic respiratory 33 conditions, such as asthma, COPD, post-TB lung damage and tobacco smoking were not 34 35 associated with increased risk of mortality. Hypertension and hypercholesterolemia were 36 also not retained as risk factors in the multiple variable analysis. There was no difference in 37 risk of mortality between the MHS and RHS. In addition there was a significant relationship 38 39 between poorer categories of diabetic control and mortality (OR 2.28 ((95%CI 1.25-4.16), 40 p=0.007), but not with poor control of HIV (unsuppressed viral load) (OR 1.96 (95% CI 0.62- 41 42 6.23), p=0.253). There was also no significant relationship with being on antiretroviral 43 treatment (OR 0.879 (95% CI 0.524-1.48), p=0.627) 44 45 Discussion 46 47 48 Contribution of district hospitals to the health system response 49 50 The majority of people with COVID-19 were self-referred and bypassed the gatekeeping role 51 expected of public sector primary care facilities. Primary care facilities are often closed on 52 weekends and afterhours making the district hospital EC the next available point of access. 53 54 Primary care facilities also de-escalated during the epidemic and people were turned away 55 or advised to stay away as much as possible,(14) while there was no restrictions on access to 56 57 district hospitals. In some areas primary care facilities temporarily closed when staff were 58 infected. Although primary care facilities could test, the district hospitals had larger and 59 more visible testing centres. This, in addition to fear of more severe disease and more trust 60 in hospital-based services, may have led people to present directly to the hospital. In rural 9 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 29

1 2 areas, where the hospital was more geographically distant, more people entered the system 3 via primary care facilities. District hospitals, therefore, played an important role in primary 4 5 care during the epidemic and this is reflected in the findings that 49% of patients seen had 6 mild COVID-19 and 42% were immediately discharged home from the EC. This also reflects 7 8 the need to improve access to primary care in South Africa.(15) 9 10 District hospitals reduced pressure on tertiary hospitals with ICU and critical care beds 11 particularly for those with moderate and severe COVID-19. Of those with moderate COVID- 12 13 19 only 9% were transferred to tertiary hospitals and mortality was 7.5%. For those with 14 severe disease 17% were transferred and mortality was 22%, which compares favourably 15 with reports from high income settings.(16) District hospitals were unable to manage 16 17 critically ill patients, as they did not have ICU or critical care facilities. 18 For peer review only 19 The introduction of field hospitals, such as at the Cape Town International Convention 20 Centre,(17) also took pressure off acute hospitals and 13.5% of patients were transferred for 21 22 ongoing care thus creating additional capacity at the district hospital. The first field hospital 23 only opened in June 2020, the last month of this study period. At the field hospital, care 24 might include management of co-morbidities, palliative care or further recovery prior to 25 26 discharge home.(17) 27 28 The differences between RHS and MHS appeared to represent differences in the type of 29 patients and geographic access rather than health services. RHS appeared to have younger 30 31 patients, more often referred by local primary care facilities, with fewer co-morbidities and 32 less severe disease, which translated into lower mortality and better outcomes. In the 33 34 Western Cape the rural district hospitals had good infrastructure, equipment and competent 35 clinical teams led by family physicians. 36 37 No other studies were found reporting on district hospitals from low and middle income 38 countries. The few studies from the UK reporting on district hospitals are not comparable as 39 40 these hospitals offer specialist services and serve a different type of population.(18) District 41 or primary hospitals in our setting are small, often rural or remote, and led by generalists or 42 43 family physicians. These types of hospitals are rare in high-income countries, but an 44 important part of African health systems. Although district hospitals in the Western Cape are 45 generally well resourced and led by family physicians, in other parts of Africa these hospitals 46 47 may have significant skills, equipment and infrastructure gaps.(19) African populations are 48 also younger and have a different profile of co-morbidities with communicable diseases 49 50 more prominent (such as HIV, TB and malaria). Poverty is also a major issue that impacts on 51 access to health care as well as food security and malnutrition.(19) It is therefore important 52 to evaluate how people with COVID-19 are managed and what their outcomes are at this 53 54 level of African health systems. 55 56 Management of COVID-19 at district hospitals 57 58 Patients presented with the typical symptoms of COVID-19 that have also been reported 59 60 elsewhere in Africa.(20) Fever, however, was only found in 12% of patients and many patients had hyperglycaemia and high blood pressure, reflecting underlying co-morbidities. 10 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 29 BMJ Open

1 2 Family physicians reported that the procedures to use infrared thermometers to triage 3 patients may not have resulted in accurate measurements and therefore fever may be 4 5 under-observed.(21) Not all patients had a body mass index measured and 6 overweight/obesity was probably under-reported. 7 8 As the median TOT for COVID-19 tests was longer than the median length of admission, most 9 10 patients were managed as people under investigation, who were presumed to have COVID- 11 19. The overall length of stay was very short with a median duration of 1-day and 75th centile 12 13 of 3-days. The laboratory capacity and turnaround time for COVID-19 tests has been labelled 14 “the Achilles heel” of the local response to the pandemic.(22) The in-hospital turnaround 15 time improved once the criteria for community testing were changed, from all people with 16 17 relevant symptoms to only those over 55-years or with co-morbidities or with more than 18 mild disease.(23) For peer review only 19 20 Antibiotics were presumably given as the diagnosis was not confirmed in the majority of 21 22 patients and the patient were treated as a community acquired pneumonia with COVID-19 23 as part of the differential diagnosis. Steroids were only given in 28% of patients, but the 24 evidence of their effectiveness were only announced in June 2020,(24) which was the last 25 26 month of this study period. 27 28 Basic imaging with a chest radiograph was only performed in half the patients, which mirrors 29 the large number of mild cases who did not require imaging. All district hospitals had access 30 31 to radiography, although not always 24-hours a day and in some cases the clinician may have 32 judged that management of the patient would not be changed. In addition, radiography 33 34 might have been avoided in order to reduce exposure of people to COVID-19 as patients 35 would have to traverse the hospital to access the radiography unit if a mobile radiograph 36 was not available. Not all critically ill received a radiograph and this might be because they 37 38 died before a radiograph could be taken or were transferred. Patients were referred on the 39 basis of the clinical picture and a radiograph was not essential. 40 41 Clinical outcomes at district hospitals 42 43 44 Overall in-hospital mortality for COVID-19 was 11% and rose to 20% in those admitted. 45 Mortality was significantly higher in MHS facilities compared to RHS and this most likely 46 reflects the higher severity of patients, rather than differences in management of patients. 47 48 Mortality rates reported from a field hospital in Ethiopia were much lower (5.3%) despite a 49 similar profile of COVID-19 severity in those seen.(25) This may be explained by much lower 50 51 levels of co-morbidity. For example no HIV or TB was reported and diabetes was found in 52 14% as opposed to 25% of patients. 53 54 A local population cohort study also found an association between people living with HIV 55 (PLWH) and mortality from COVID-19 and no clear association with viraemia or 56 57 immunosuppression.(26) However, in district hospitals we found no association with 58 previous or current TB infection and mortality from COVID-19, while the population cohort 59 60 study found a twofold increase. The population study may have overestimated risks as routine data did not include all co-morbidities and these patients were more likely to be 11 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 29

1 2 followed up and diagnosed with COVID-19. Other respiratory conditions such as asthma, 3 COPD and post-TB structural lung disease were also not associated with increased mortality. 4 5 This could suggest that immunological factors were maybe more important than pulmonary 6 factors in determining risk of death. Other conditions with impaired or altered immunity 7 8 were also associated with mortality, such as type 2 diabetes and people receiving treatment 9 for cancer. 10 11 The combination of increasing age, overweight/obesity, type-2 diabetes, chronic kidney 12 13 disease and cardiac failure as independent risk factors is important as 1 in 2 South Africans 14 over the age of 45-years have pre-diabetes or diabetes.(27) Increasing age, obesity and 15 chronic kidney disease were also identified as key risk factors in the Democratic Republic of 16 17 Congo (28) and diabetes in Ethiopia. The importance of non-communicable diseases, such as 18 diabetes, was broughtFor into thepeer public spotlight review by COVID-19 only as previously the focus of 19 20 attention was on HIV and TB. The level of control of diabetes was also directly related to risk 21 of mortality, which emphasized the need to improve self-management and treatment, 22 particularly as 57% had a history of poor control on admission. The absence of electronic 23 24 medical records and an integrated health information system explains why hospitals could 25 not obtain information on prior control of chronic diseases, particularly in the RHS. The 26 27 continuity of care for type-2 diabetes was also disrupted by the de-escalation of services and 28 most forms of patient education and counselling were stopped. There is a need to innovate 29 new ways of managing and empowering people with diabetes while reducing their risk of 30 31 exposure to COVID-19 at health facilities and support groups. 32 33 All results were dependent on the completeness and accuracy of medical records. For 34 example, clinicians may not have recorded all the symptoms experienced by the patients. 35 36 Family physicians did not report a problem with missing medical records. It would have been 37 helpful to record the anti-retroviral treatment regimen in patients with HIV and to 38 39 investigate any association between mortality and exposure to different combinations of 40 medications. No data was collected on laboratory results such as full blood count, urea or 41 creatinine or on the time between onset of symptoms and treatment, and this might have 42 43 been useful to investigate the relationship with clinical outcomes. It was not possible to 44 determine the outcomes of patients that were discharged from the EC with mild disease and 45 46 it is possible that some were re-admitted to other hospitals. None of the family physicians 47 collecting data reported that patients in the study were re-admissions. 48 49 These hospitals all had family physicians heading their clinical teams and it is possible that 50 they had a higher quality of care than the hospitals that are still without FPs. The Western 51 52 Cape also has better infrastructure and a stronger health system than many other provinces 53 and the quality of care is likely to be lower in other parts of South Africa. 54 55 Conclusions 56 57 58 District hospitals provided an essential primary care service for many patients with mild 59 symptoms of COVID-19 during the epidemic. This also represented a deficiency in access to 60 and utilisation of primary care. District hospitals successfully treated a large number of 12 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 29 BMJ Open

1 2 people with moderate-severe COVID-19 who did not need ventilation and took pressure off 3 higher-level facilities. Limited laboratory capacity meant that most patients were treated as 4 5 people under investigation without a definitive diagnosis. The clinical picture was similar to 6 that reported elsewhere. Mortality at this level of care was associated with increasing age, 7 8 male sex, HIV, type 2 diabetes, overweight/obesity, cardiac failure, chronic kidney disease 9 and treatment for cancer, but not with hypertension, TB, asthma, COPD or post-TB lung 10 11 damage. Patients in the MHS were more numerous, had more co-morbidity and more severe 12 COVID-19 disease than in the RHS. 13 14 Acknowledgements 15 16 No other acknowledgements. 17 18 For peer review only 19 Funding statement 20 21 This research received no specific grant from any funding agency in the public, commercial 22 or not-for-profit sectors. 23 24 25 Competing interests statement 26 27 The authors declare no competing interests. 28 29 Author contributions 30 31 32 RM and MPV conceptualised the study. All authors approved the study proposal. MPV, HA 33 and HM collected data at Eerste River Hospital; AL and KS collected data at Helderberg 34 Hospital; AU and RC collected data at Wesfleur Hospital; LR collected data at False Bay 35 Hospital; LJ collected data at George Hospital; LH and PK collected data at Knysna Hospital; 36 EL and BW collected data at Stellenbosch Hospital; AA and DK collected data at Ceres 37 38 Hospital. RM analysed the data. All authors gave feedback on their interpretation of the 39 results. RM and MPV wrote the draft manuscript. All authors gave feedback on the 40 manuscript and approved the final version. 41 42 43 Ethical considerations 44 45 The study was approved by the Health Research Ethics Committee at Stellenbosch University 46 (N20/06/032_COVID-19) who granted a waiver of informed consent for collection on 47 anonymised data from medical records. Study permission was obtained from the Provincial 48 Government of the Western Cape to conduct the study. 49 50 51 Data sharing statement 52 53 The dataset will be made available on reasonable request from the first author. 54 55 56 Patient and public involvement 57 58 It was not appropriate or possible to involve patients or the public in the design, or conduct, 59 or reporting, or dissemination plans of our research. 60

13 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 29

1 2 References 3 4 1. COVID-19 coronavirus pandemic [Internet]. 2020 [cited 2020 Jun 6]. Available from: 5 6 https://www.worldometers.info/coronavirus/ 7 2. Lone SA, Ahmad A. COVID-19 pandemic - an African perspective. Emerg Microbes 8 Infect [Internet]. Taylor & Francis; 2020 Dec [cited 2020 Nov 5];9(1):1300–8. Available 9 from: http://www.ncbi.nlm.nih.gov/pubmed/32458760 10 11 3. Coronavirus in South Africa [Internet]. 2020 [cited 2020 Jun 6]. Available from: 12 https://mediahack.co.za/datastories/coronavirus/dashboard/ 13 4. Chen N, Zhou M, Dong X, Qu J, Gong F, Han Y, et al. Epidemiological and clinical 14 characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a 15 descriptive study. Lancet [Internet]. 2020 Feb 15 [cited 2020 Mar 16 17 17];395(10223):507–13. Available from: 18 http://www.ncbi.nlm.nih.gov/pubmed/32007143For peer review only 19 5. Zhao X-Y, Xu X-X, Yin H-S, Hu Q-M, Xiong T, Tang Y-Y, et al. Clinical characteristics of 20 patients with 2019 coronavirus disease in a non-Wuhan area of Hubei Province, China: 21 a retrospective study. BMC Infect Dis [Internet]. BioMed Central; 2020 Dec 29 [cited 22 23 2020 Jun 11];20(1):311. Available from: 24 https://bmcinfectdis.biomedcentral.com/articles/10.1186/s12879-020-05010-w 25 6. Guan W, Ni Z, Hu Y, Liang W, Ou C, He J, et al. Clinical Characteristics of Coronavirus 26 Disease 2019 in China. N Engl J Med [Internet]. 2020 Apr 30 [cited 2020 Jun 27 28 11];382(18):1708–20. Available from: 29 http://www.nejm.org/doi/10.1056/NEJMoa2002032 30 7. Allinder S. The World’s Largest HIV Epidemic in Crisis: HIV in South Africa | Center for 31 Strategic and International Studies [Internet]. Centre for Strategic and International 32 Studies. 2019 [cited 2020 Nov 5]. Available from: 33 34 https://www.csis.org/analysis/worlds-largest-hiv-epidemic-crisis-hiv-south-africa# 35 8. Massyn N, Barron P, Day C, Ndlovu N, Padarath A. District Health Barometer 2018/19 36 [Internet]. Cape Town; 2020 [cited 2020 Nov 5]. Available from: 37 https://www.hst.org.za/publications/Pages/DISTRICT-HEALTH-BAROMETER- 38 39 201819.aspx 40 9. Madzimbamuto F. Re-inventing the ventilator. African J Prim Heal Care Fam Med. 41 2020;In Press. 42 10. Government of Western Cape: Health. Western Cape Critical Care Triage Tool. 43 [Internet] [cited 2020 Dec 18] . Available from: 44 45 https://www.westerncape.gov.za/assets/departments/health/COVID- 46 19/western_cape_critical_care_triage_tool_version_1.2_14th_may.pdf 47 11. First patients admitted to special Covid-19 field hospital in Cape Town [Internet]. 48 [cited 2020 Jun 11]. Available from: https://www.timeslive.co.za/news/south- 49 50 africa/2020-06-08-first-patients-admitted-to-special-covid-19-field-hospital-in-cape- 51 town/ 52 12. The South African Triage Scale (SATS) [Internet]. [cited 2020 Nov 5]. Available from: 53 https://emssa.org.za/special-interest-groups/the-south-african-triage-scale-sats/ 54 13. COVID-19 Health Worker Resources [Internet]. 2020. [cited 2020 Jun 6]. Available 55 56 from: https://knowledgetranslation.co.za/resources/covid-19-hw-resources/ 57 14. Mash R, Goliath C, Perez G. Re-organising primary health care to respond to the 58 Coronavirus epidemic in Cape Town, South Africa. African J Prim Heal Care Fam Med 59 [Internet]. 2020 Nov 5 [cited 2020 Nov 5];12(1):4. Available from: 60 https://phcfm.org/index.php/phcfm/article/view/2607/4263 14 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 29 BMJ Open

1 2 15. Bresick G, von Pressentin KB, Mash R. Evaluating the performance of South African 3 primary care: a cross-sectional descriptive survey. South African Fam Pract [Internet]. 4 5 Taylor & Francis; 2019 May 4 [cited 2019 Jul 15];61(3):109–16. Available from: 6 https://www.tandfonline.com/doi/full/10.1080/20786190.2019.1596666 7 16. Santorelli G, Sheldon T, West J, Cartwright C, Wright J. COVID-19 in-patient hospital 8 mortality by ethnicity. Wellcome Open Res [Internet]. F1000 Research Limited; 2020 9 May 4 [cited 2020 Nov 5];5:86. Available from: 10 11 https://wellcomeopenresearch.org/articles/5-86/v1 12 17. Reid S, Ras T, Pressentin K von. The Cape Town International Convention Centre from 13 the inside: The family physicians’ view of the ‛Hospital of Hope’. African J Prim Heal 14 Care Fam Med [Internet]. 2020 Oct 26 [cited 2020 Nov 5];12(1):4. Available from: 15 16 https://phcfm.org/index.php/phcfm/article/view/2667/4331 17 18. Knights H, Mayor N, Millar K, Cox M, Bunova E, Hughes M, et al. Characteristics and 18 outcomes of Forpatients withpeer COVID-19 review at a district general only hospital in Surrey, UK. Clin 19 Med [Internet]. Clin Med (Lond); 2020 [cited 2020 Dec 18];20(5). Available from: 20 https://pubmed.ncbi.nlm.nih.gov/32709637/ 21 22 19. Agyepong IA, Sewankambo N, Binagwaho A, Coll-Seck AM, Corrah T, Ezeh A, et al. The 23 path to longer and healthier lives for all Africans by 2030: the Lancet Commission on 24 the future of health in sub-Saharan Africa. Lancet [Internet]. 2017 Dec 23 [cited 2018 25 Oct 23];390(10114):2803–59. Available from: 26 27 http://www.ncbi.nlm.nih.gov/pubmed/28917958 28 20. Olumade TJ, Uzairue LI. Clinical characteristics of 4490 COVID-19 patients in Africa: A 29 meta-analysis. medRxiv [Internet]. Cold Spring Harbor Laboratory Press; 2020 Oct 21 30 [cited 2020 Nov 6];2020.10.20.20215905. Available from: 31 https://www.medrxiv.org/content/10.1101/2020.10.20.20215905v1 32 33 21. Chen H-Y, Chen A, Chen C. Investigation of the Impact of Infrared Sensors on Core 34 Body Temperature Monitoring by Comparing Measurement Sites. Sensors [Internet]. 35 Multidisciplinary Digital Publishing Institute (MDPI); 2020 May 19 [cited 2020 Nov 36 5];20(10). Available from: http://www.ncbi.nlm.nih.gov/pubmed/32438729 37 22. Porter JD, Mash R, Preiser W. Turnaround times – the Achilles’ heel of community 38 39 screening and testing in Cape Town, South Africa: A short report. African J Prim Heal 40 Care Fam Med [Internet]. 2020 Oct 2 [cited 2020 Nov 5];12(1):3. Available from: 41 https://phcfm.org/index.php/phcfm/article/view/2624/4257 42 23. David N, Mash R. Community-based screening and testing for Coronavirus in Cape 43 44 Town, South Africa: Short report. African J Prim Heal Care Fam Med [Internet]. 2020 45 Jun 3 [cited 2020 Jun 6];12(1):3. Available from: 46 https://phcfm.org/index.php/phcfm/article/view/2499 47 24. The RECOVERY Collaborative Group. Dexamethasone in Hospitalized Patients with 48 Covid-19 — Preliminary Report. N Engl J Med [Internet]. Massachusetts Medical 49 50 Society; 2020 Jul 17 [cited 2020 Nov 5];NEJMoa2021436. Available from: 51 http://www.nejm.org/doi/10.1056/NEJMoa2021436 52 25. Leulseged TW, Hassen IS, Maru EH, Zewde WC, Chamiso NW, Bayisa AB, et al. 53 Characteristics and outcome profile of Hospitalized African COVID-19 patients: The 54 55 Ethiopian Context. medRxiv [Internet]. Cold Spring Harbor Laboratory Press; 2020 Oct 56 28 [cited 2020 Nov 6];2020.10.27.20220640. Available from: 57 https://www.medrxiv.org/content/10.1101/2020.10.27.20220640v1 58 26. Boulle A, Davies M-A, Hussey H, Ismail M, Morden E, Vundle Z, et al. Risk factors for 59 COVID-19 death in a population cohort study from the Western Cape Province, South 60 Africa. Clin Infect Dis [Internet]. Clin Infect Dis; 2020 Aug 29 [cited 2020 Nov 6]; 15 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 29

1 2 Available from: http://www.ncbi.nlm.nih.gov/pubmed/32860699 3 27. The Society for Endocrinology M and D of SAT 2 DGEC. SEMDSA Guidelines for the 4 5 Management of Type 2 Diabetes | Journal of Endocrinology, Metabolism and Diabetes 6 of South Africa. JEMDSA [Internet]. 2017 [cited 2019 Mar 26];22(1 (Supplement)):S1– 7 196. Available from: http://jemdsa.co.za/index.php/JEMDSA/article/view/647 8 28. Nachega JB, Ishoso DK, Otokoye JO, Hermans MP, Machekano RN, Sam-Agudu NA, et 9 al. Clinical Characteristics and Outcomes of Patients Hospitalized for COVID-19 in 10 11 Africa: Early Insights from the Democratic Republic of the Congo. Am J Trop Med Hyg 12 [Internet]. Am J Trop Med Hyg; 2020 Oct 2 [cited 2020 Nov 5]; Available from: 13 http://www.ncbi.nlm.nih.gov/pubmed/33009770 14 15 16 17 18 For peer review only 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

16 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 29 BMJ Open

17 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 29

18 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 29 BMJ Open

1 2 Table 2: Co-morbidities of patients with COVID-19 (N=1376) 3 4 All Metro Rural p value 5 6 N=1376 N=1017 N=359 7 n (%) n (%) n (%) 8 Co-morbidity 9 10 Overweight/obesity 269 (19.5) 208 (20.5) 61 (17.0) 0.153 11 Type 1 diabetes 17 (1.2) 15 (1.5) 2 (0.6) 0.175 12 13 Type 2 diabetes 347 (25.2) 273 (26.9) 74 (20.6) 0.019 14 Hypercholesterolemia 83 (6.0) 77 (7.6) 6 (1.7) <0.001 15 16 Hypertension 564 (41.0) 454 (44.7) 109 (30.4) <0.001 17 Cardiac failure 58 (4.2) 51 (5.0) 7 (1.9) 0.013 18 For peer review only 19 Ischaemic heart disease 25 (1.8) 24 (2.4) 1 (0.3) 0.011 20 Asthma 67 (4.9) 52 (5.1) 15 (4.2) 0.477 21 COPD 50 (3.6) 38 (3.7) 12 (3.3) 0.729 22 23 Post-TB lung damage 12 (0.9) 8 (0.8) 4 (1.1) 0.567 24 HIV 195 (14.2) 156 (15.4) 39 (10.9) 0.036 25 26 Active TB on treatment 23 (1.7) 20 (2.0) 3 (0.8) 0.150 27 Previous TB 49 (3.6) 40 (3.9) 9 (2.5) 0.209 28 29 Cancer on treatment 10 (0.7) 9 (0.9) 1 (0.3) 0.244 30 Previous cancer 2 (0.1) 2 (0.2) 0 (0.0) 0.400 31 32 Chronic kidney disease 60 (4.4) 55 (5.4) 5 (1.4) 0.001 33 None 450 (32.7) 276 (27.2) 174 (48.5) <0.001 34 Tobacco smoker 95/621 (15.3) 80/490 (16.3) 15/131 (11.5) 0.168 35 36 Previous control of co-morbidity 37 Diabetes N=348 38 39 Normal HbA1c < 7% 32 (9.2) 30 (10.8) 2 (2.9) 40 Controlled HbA1c 7-8% 41 (11.8) 34 (12.2) 7 (10.1) 0.002 41 42 Uncontrolled HbA1c >8 68 (19.5) 56 (20.1) 12 (17.4) 43 and <10% 44 45 Very uncontrolled 131 (37.6) 110 (39.4) 21 (30.4) 46 >=10% 47 Unknown 76 (21.8) 49 (17.6) 27 (39.1) 48 49 Hypertension N=446 50 Well controlled 165 (37.0) 150 (41.6) 14 (16.7) 51 52 Uncontrolled 88 (19.7) 75 (20.8) 13 (15.5) <0.001 53 Not known 193 (43.3) 136 (37.7) 57 (67.9) 54 55 Asthma N=72 56 Well controlled 23 (31.9) 21 (51.2) 2 (6.5) 57 <0.001 58 Partly controlled 10 (13.9) 6 (14.6) 4 (12.9) 59 Uncontrolled 5 (6.9) 5 (12.2) 0 (0.0) 60 Unknown 34 (47.2) 9 (22.0) 25 (80.6) 19 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 29

20 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 29 BMJ Open

21 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 24 of 29

22 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 25 of 29 BMJ Open

1 2 3 Table 5: Clinical outcomes at the district hospitals 4 5 Clinical outcome Admitted All Metro Rural Mild Moderate Severe Critical 6 7 N=652 N=1376 N=1017 N=359 N=676 N=252 N=335 N=82 8 n (%) n (%) n (%) n (%) n (%) n (%) n (%) n (%) 9 10 Died 128 (19.6) 151 (11.0) 118 (11.6) 33 (9.2) 12 (1.8) 19 (7.5) 73 (21.8) 47 (57.3) 11 Discharged home 324 (49.7) 831 (60.4) 577 (56.8) 253 (70.5) 525 (77.7) 147 (58.3) 130 (38.8) 7 (8.5) 12 Transferred 57 (8.7) 118For (8.6) 115 peer (11.3) 3 (0.8) review20 (3.0) 23 (9.1)only55 (16.4) 20 (24.4) 13 14 tertiary hospital 15 Transferred 30 (4.6) 47 (3.4) 20 (2.0) 27 (7.5) 26 (3.8) 11 (4.4) 3 (0.9) 0 (0.0) 16 17 assisted isolation 18 Transferred field 88 (13.5) 105 (7.6) 104 (10.2) 1 (0.3) 1.3 (1.9) 28 (11.1) 55 (16.4) 7 (8.5) 19 20 hospital 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 23 43 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 26 of 29

1 2 Table 6: Risk factors for death from COVID-19 3 4 Variable Unadjusted odds P value 5 6 ratio (95% CI) 7 Cancer on treatment 12.63 (3.5-42.3) <0.001 8 Chronic kidney disease 10.65 (6.21-18.28) <0.001 9 10 Cancer previous 8.16 (0.51-131.14) 0.138 11 Cardiac failure 5.63 (3.21-9.87) <0.001 12 13 Post-TB SLD 4.14 (1.23-13.9) 0.022 14 Hypertension 3.95 (2.73-5.70) <0.001 15 16 Type 2 diabetes 3.1 (2.2-4.3) <0.001 17 TB on treatment 2.94 (1.14-7.57) 0.026 18 For peer review only 19 Ischaemic heart disease 2.06 (0.76-5.58) 0.154 20 Hypercholesterolemia 1.88 (1.04-3.37) 0.035 21 22 TB previous 1.88 (0.89-3.95) 0.097 23 Overweight/obese 1.69 (1.15-2.49) 0.007 24 HIV 1.67 (1.09-2.56) 0.019 25 26 Male sex 1.63 (1.16-2.30) 0.005 27 COPD 1.34 (0.59-3.03) 0.487 28 29 Asthma 1.11 (0.52-2.36) 0.795 30 Type 1 diabetes 1.08 (0.24-4.78) 0.916 31 32 Increasing age 1.06 (1.05 – 1.07) <0.001 33 Turnaround time 1.02 (0.99-1.06) 0.216 34 35 Tobacco smoking 1.01 (0.46-2.21) 0.986 36 Length of admission 0.97 (0.93-1.02) 0.219 37 Rural vs metro services 0.77 (0.51-1.16) 0.208 38 39 No comorbidities 0.17 (0.09-0.30) <0.001 40 Variable Adjusted P value 41 42 Odds ratio (95% CI) 43 Cancer on treatment 7.45 (1.87-29.89) 0.004 44 45 Chronic kidney disease 5.16 (2.82-9.43) <0.001 46 HIV 3.41 (2.06-5.65) <0.001 47 Cardiac failure 2.85 (1.52-5.35) 0.001 48 49 Male sex 2.02 (1.37-2.98) <0.001 50 Type 2 diabetes 1.84 (1.24-2.73) 0.002 51 52 Overweight/obese 1.58 (1.02-2.46) 0.04 53 Increasing age 1.06 (1.04-1.07) <0.001 54 55 56 57 58 59 60

24 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 27 of 29 BMJ Open