Atypical presentations of COVID‐19 in care home residents presenting to secondary care: A UK single centre study

Abstract Background Atypical presentations of COVID‐19 pose difficulties for early isolation and treatment, particularly in institutional care settings. We aimed to characterize the presenting symptoms and associated mortality of COVID‐19 in older adults, focusing on care home residents admitted to secondary care. Methods A retrospective cohort study of 134 consecutive inpatients over 80 years old hospitalized with PCR confirmed COVID‐19 in the United Kingdom. Symptoms at presentation and frailty were analysed. Differences between community dwelling and care home residents, and associations with mortality, were assessed using between‐group comparisons and logistic regression. Results Care home residents were less likely to experience cough (46.9% vs 72.9%, P = .002) but more likely to present with delirium (51.6% vs 31.4%, P = .018), particularly hypoactive delirium (40.6% vs 24.3%, P = .043). Mortality was more likely with increasing frailty (OR 1.25, 95% CI 1.00, 1.58, P = .049) and those presenting with anorexia (OR 3.20, 95% CI 1.21, 10.09, P = .028). There were no differences in mortality or length of stay based on residential status. Conclusion COVID‐19 in older adults often presents with atypical symptoms, particularly in those admitted from institutional care. These individuals have a reduced incidence of cough and increased hypoactive delirium. Individuals presenting atypically, especially with anorexia, have higher mortality.

present without classic symptoms. 5 Data are emerging that presentation with COVID-19 in this cohort may also be atypical, potentially with geriatric syndromes including falls, delirium and anorexia. [6][7][8][9][10][11] Atypical presentation is of particular concern for residential and nursing home residents. Evidence exists that transmission within nursing homes is rapid due to difficulties in identifying new COVID- 19 infections. 12 Isolation of suspected cases is critical to protect staff and other residents, 13 but requires understanding of the characteristics of COVID-19 in this population.
Barnet is the most populous of London's boroughs with an estimated population of 402 700 residents and approximately 16 800 people over the age of 80. There are more than 100 care homes in the borough. 14 Barnet Hospital is a general suburban hospital with 440 beds and was affected by a high volume of COVID-19 cases early in the course of the UK pandemic. 15 Using admission data from this period, we set out to (a) characterize the presenting symptoms in older adults admitted with COVID-19, (b) determine whether care home residents exhibited different presentations to the general older population, and (c) examine the mortality associations with typical and atypical presentations. We hypothesized that atypical presentation would be more common in care home residents and associated with increased mortality.

| Study design, setting and population
Patients aged 80 or over admitted to Barnet Hospital with COVID-19 confirmed on consecutive polymerase chain reaction (PCR) testing for SARS-CoV2 were included for analysis. Individuals who had continuously been an inpatient for 14 days beforehand were excluded on the basis of having acquired COVID-19 in hospital and therefore unrelated symptoms on admission. Patients with a clinical diagnosis of COVID-19 without PCR confirmation were not included. Subgroups were defined within this population based on their care needs prior to admission, categorized as "community dwelling" (living in their own home, including those receiving carers at home) and care home residence (individuals residing in either nursing homes or residential care homes). Data were collected retrospectively from the electronic patient record. Further information is available here. 15

| Ascertainment of key variables
Standardized data were collected on demographic features, ethnicity, length of stay and the presence of comorbidities (prior diagnosis of cardiac disease, hypertension, diabetes, respiratory disease, immunosuppression and dementia). Frailty was determined by the Clinical Frailty Scale (CFS), 16

| Statistical analyses
Demographic and clinical characteristics of the cohort were described using measures of central tendency and variability to explore differences between residential status (community dwelling versus care home residents). Pearson's chi-square and Fisher's exact tests were used to test for relationships between categorical outcome variables and residential category. For relationships between continuous variables and residential category, the Wilcoxon signed-rank test was used. CFS was treated as a continuous variable.
We used univariable logistic regression to test the association between mortality and presenting symptoms. A series of logistic regressions were run to assess independent associations between the covariates and mortality. Any that were identified as statistically or clinically significant confounders (P < .1), through regression analyses, were sequentially adjusted for in the final multivariable regression model for this association, with backwards elimination to remove variables one at a time until only statistically significant (P < .05) variables remained in the model. All analyses were performed using R on a complete-case basis, including only participants with full data on delirium status and medication. All analysis was performed using R Statistics version 3.6.3. Code and data are available on reasonable request.

| Participant eligibility
One hundred and fifty individuals over the age of 80 were admitted to Barnet Hospital with SARS-CoV2 confirmed on PCR between March 10, 2020 and April 8, 2020, and all were included in initial analyses. Of these individuals, 16 were then excluded on the basis of having hospital acquired COVID-19. The remaining 134 individuals had complete data for all analysed variables (Figure 1).

| Cohort and clinical characteristics
The median age was 86 years, and most participants were white (76.1%, n = 102). There was a slight male predominance (54.5%, n = 73) and a nearly equal split between residential status (52.2% community dwelling participants, n = 70). All CFS categories were represented in the complete cohort with mild and moderate frailty (CFS 5 & 6) most prevalent. Most participants had one or more comorbidities, with hypertension (55.2%, n = 74), cardiac disease (56.7%, n = 76) and diabetes mellitus (29.1%, n = 39) most common.
The overall mortality was high (64.9%, n = 87) with a median length of stay of 11 days for those that survived.
When compared to community dwelling older adults, care home residents admitted with COVID-19 tended to be older (median age 88.5 vs 85, P < .001) and have higher frailty as indicated by the CFS (P < .001). Both groups had a similar gender and ethnicity distribution, and a similar prevalence of comorbidities, although dementia was more prevalent in care home residents (37.5% vs 15.7%, P = .004). Care home residents were less likely to present with cough (46.9% vs 72.9%, P = .002) and more likely to present with delirium (51.6% vs 31.4%, P = .018) particularly hypoactive delirium (40.6% vs 24.3%, P = .043). There was a suggestion that fully atypical presentation was more common in care home residents, though total numbers were small, limiting significance (18.8% vs 8.6%, P = .084).
There was no notable difference in either mortality or length of stay between these two groups ( Figure 2).
Full sample characteristics are provided in Table 1.

| Associations with mortality
Individuals were noted to be more likely to die when presenting with a higher CFS score (Odds Ratio (OR) 1. 25 Table 2.
For care home residents, these associations persisted when combined in a multivariable model. Our multivariable model for care home residents is presented in Table 3.

| D ISCUSS I ON
This single-centre study found that inpatients over the age of 80 with PCR positive COVID-19 presented not only with cough, fever and dyspnoea, but also with a high proportion of hypoactive delirium, falls and anorexia. These trends were more pronounced in individuals presenting from residential or nursing care, who were less likely to present with cough, and more likely to present with delirium.
Atypical presentations that featured none of the cardinal features of cough, dyspnoea and fever were common in patients presenting from institutional care. Individuals presenting from care homes were no more likely to die overall than those presenting from the community, though presentation with higher levels of frailty or anorexia was associated with increased mortality for both groups. Taken together, our findings suggest a high prevalence of atypical symptoms in older adults, particularly institutionalized populations, and a risk of increased mortality for those with less classic presentations.
Our study has several strengths. Firstly, the data were collected contemporaneously as part of routine clinical work, allowing for easy characterization of presenting complaints and patient histories. To accurately categorize delirium, delirium subtype, CFS and presentation with falls or anorexia, geriatric syndromes were independently corroborated by a full review of medical records by two consultant geriatricians. Concordance between CFS ratings was high, as seen in prior CFS validation studies. 18,19 Our data were collected from an area of the UK affected by high COVID-19 caseload early in the course of the UK pandemic. These data were also not without limitations. The comparisons drawn between institutional care and community care in our data do not account for differences between nursing home and residential home care, which cannot always be determined from address and admission history. Key differences may still exist between these subgroups. Likewise, community dwelling older adults have differing care needs, and homogenizing this group risks ignoring those with high levels of social care including that provided by the private sector. However, the higher trend in CFS displayed in those in institutional care suggests this was not a major issue in our sample. Using an endpoint of discharge versus death limits mortality analysis, as some of those discharged may subsequently die in the community; however this number is likely lower than would be expected in an older population, as few palliative discharges were made due to limitations on carer visits and return to residential care imposed by the infective nature of COVID-19. Likewise, collected biomarker data from patients were sporadic in this population, limiting its use in models of mortality without extensive imputation.
Finally, our smaller sample size, particularly in relation to rarer presenting symptoms, may have masked some associations between mortality and symptoms. Fully atypical presentations, diarrhoea and delirium suggested a trend towards increased mortality, and further study in larger populations would be worthwhile.
Although the spectrum of symptoms seen here with COVID-19 presentation is in keeping with that seen in the general UK population, 4 the high proportion of falls, delirium and anorexia suggests prior case reports 6-10 are indicative of a more widespread pattern.
The suggestion that COVID-19 presents atypically more frequently in older adults, and particularly in care home residents, supports a prior study in a United States nursing home, where atypical symptoms alone were noted in just under 10% of PCR positive patients. 12 In particular, the predominance of hypoactive delirium in this population is in keeping with response to critical illness in patients with pre-existing cognitive impairment. 7 Our findings are also consistent with previously reported associations between high mortality and older age. 15,20-22 While we did not find associations between mortality and dyspnoea, 23  with lower levels of frailty. 28,29 Notably none of our cohort received mechanical ventilatory support. CFS score should not be the sole determinant of treatment suitability. 30 Of key interest are novel associations seen here between anorexia, cough and mortality, particularly in institutionalized populations.
Associations between increased mortality and anorexia might be explained by late presentation of those with COVID-19 to secondary medical services. Anorexia is a subtle sign, becoming apparent over a few days, and so might not be detected until further into the disease course. Likewise, those with cough might have been referred earlier given the prominence of cough in Public Health England (PHE) and WHO criteria for COVID-19 diagnosis, 31  Jaffer for assistance with planning the database.

CO N FLI C T S O F I NTE R E S T
Nothing to disclose.

AUTH O R CO NTR I B UTI O N
MJR was responsible for data analysis with additional input from SEB. MJR and DLB devised the research question and wrote the first draft of the manuscript. SEB conceived and initiated the data collection project. All authors contributed to interpreting the data and writing the final paper.

E TH I C A L A PPROVA L
The data presented here were collected during routine clinical practice and formal Research Ethics Committee review was not required. Note: Odds ratios rounded to 2 decimal places. P values indicate significance of association between outcome and (level of) covariate.