Characteristics and outcomes of clinically diagnosed RT-PCR swab negative COVID-19: a retrospective cohort study

Patients with strong clinical features of COVID-19 with negative real time polymerase chain reaction (RT-PCR) SARS-CoV-2 testing are not currently included in official statistics. The scale, characteristics and clinical relevance of this group are not well described. We performed a retrospective cohort study in two large London hospitals to characterize the demographic, clinical, and hospitalization outcome characteristics of swab-negative clinical COVID-19 patients. We found 1 in 5 patients with a negative swab and clinical suspicion of COVID-19 received a clinical diagnosis of COVID-19 within clinical documentation, discharge summary or death certificate. We compared this group to a similar swab positive cohort and found similar demographic composition, symptomology and laboratory findings. Swab-negative clinical COVID-19 patients had better outcomes, with shorter length of hospital stay, reduced need for > 60% supplementary oxygen and reduced mortality. Patients with strong clinical features of COVID-19 that are swab-negative are a common clinical challenge. Health systems must recognize and plan for the management of swab-negative patients in their COVID-19 clinical management, infection control policies and epidemiological assessments.


Scientific Reports
| (2021) 11:2455 | https://doi.org/10.1038/s41598-021-81930-0 www.nature.com/scientificreports/ little has been documented about the scale and clinical relevance of this subgroup of patients for whom there is no consensus on optimal management. Our study aims to assess the real-world prevalence and characteristics of clinically diagnosed swab-negative COVID-19, including factors associated with swab-negative disease, and whether their outcomes differ to swab-positive patients.

Methods
We retrospectively reviewed medical admissions from March 1 to April 12, 2020 at Imperial College Healthcare Trust (ICHNT) in London, UK from two admitting sites. We defined eligible cases as those who presented with clinical suspicion of COVID-19 or had symptoms compatible with COVID-19, were admitted to hospital and had a SARS-CoV-2 nasopharyngeal swab performed. We collected full demographic characteristics, time course of symptoms, time of presentation and testing, presenting symptoms, final diagnosis and outcome as well as radiological and laboratory findings for all patients with a negative swab from admission until discharge. Cases were evaluated using the Public Health England (PHE) testing criteria for possible COVID cases. We defined swab-negative clinical COVID-19 cases as follows: (a) clinical COVID-19 or high level of suspicion as defined by the treating medical team (as recorded in the medical notes, discharge documents or death certificate); and (b) RT-PCR swab-negative (on initial and any subsequent tests performed). Patients who tested positive on subsequent swabs while in hospital were not included in the reported cohort.
The swab-negative clinical COVID-19 group of patients was compared to a subgroup of a previously described swab-positive cohort who were similarly admitted via general medical admissions 12 . Chi-square, Fisher's and rank sum tests were used, as appropriate, to compare the cohorts' characteristics, and odds ratios (OR) calculated to assess differences in the outcomes of respiratory deterioration, defined as requiring greater than 60% oxygen, and death. Lastly, we assessed the cumulative risk of the competing outcomes of hospital discharge and death over time using the using the Nelson-Aelen estimator.
The study was approved by the ICHNT clinical governance team. As we report on routinely collected nonidentifiable clinical audit data, no individual informed consent was required under the UK policy framework for health and social care. All methods were conducted in accordance with relevant guidelines and regulations.

Results
We identified 1,119 emergency medical admissions with initial clinical suspicion of COVID-19 who had a SARS-CoV-2 swab performed (Fig. 1). Initial swab was negative in 456 (41%) and positive in 663 (59%) patients. 62% (281/456) of those who were swab-negative received an alternative diagnosis. 47 of the swab-negative cohort later tested positive for SARS-CoV-2 on repeat PCR testing and were not included within the clinical COVID-19 cohort. Patients who tested positive on subsequent testing had between 2 and 4 nasopharyngeal swabs in total with a median number of 2. The most common alternative diagnoses included community acquired pneumonia, lower respiratory tract infection and exacerbations of chronic lung disease such as chronic obstructive pulmonary disease.
Overall, 20% (94/456) of swab-negative cases were identified as swab-negative clinical COVID-19. 98% (92/94) of these cases fulfilled PHE criteria for patients eligible to be swabbed for SARS-CoV-2 compared to www.nature.com/scientificreports/ 84% of the entire swab-negative group. 41/94 had repeat swab testing performed and remained negative. Those who had repeat swabbing had between 2 and 3 swabs in total with a median number of 2.
The demographic profile and symptomatology of the swab-positive and swab-negative cohorts were similar, with high rates of influenza-like symptoms, cough and fever in both groups (Table 1). Typical features such as shortness of breath was higher in the swab-negative clinical COVID-19 cohort (75/94 (79.79%) vs 312/478 (66.67%) p = 0.017). The swab-negative cohort was more likely to have chest radiographs reporting typical COVID-19 appearances than the swab-positive cohort (53/77 (68.83%) vs 195/389 (50.13%); p = 0.004). Only 18 patients within the swab-negative clinical COVID-19 cohort had a CT chest performed during their admission. 15/18 CT chests were reported as being consistent with COVID-19 or had COVID-19 within the reported differential.
Overall haematological and biochemical findings that have been associated with COVID-19 including raised CRP, d-dimer and ferritin were similarly distributed between the two cohorts. A higher proportion of patients had moderate lymphopenia within the swab-negative group (65/94 (69.15%) vs 258/460 (56.09%); p = 0.026).
Patients in the swab-negative group presented slightly later in their clinical course after symptom onset, at a median of 7 days (IQR 4.3-13) for the RT-PCR-negatives compared to 6 days (IQR 3-10) within the RT-PCRpositives cohort (p < 0.001). Interestingly, RT-PCR-negative patients had better hospitalisation outcomes than RT-PCR-positives. The former had a 57% lower probability of requiring ≥ 60% supplementary oxygen during hospitalisation (OR 0.43, 95% CI 0.26-0.70, p < 0.001), a shorter length of hospital stay (median 5 days vs 6 days, p < 0.001) and 60% lower probability of death (OR 0.40, 95%CI 0.22-0.72, p < 0.001). The cumulative risk of being discharged alive was significantly higher in the swab-negative group than the positive group (p < 0.001). Furthermore, the swab-negative cohort have a lower cumulative risk of in-hospital mortality than positives (p < 0.001) (Fig. 2).

Discussion
We find one in five symptomatic patients admitted to the medical department who had a negative SARS-CoV-2 swab received a clinical diagnosis of COVID-19 despite negative testing. This represents a 13% increase in the total number of hospitalized COVID-19 patients over the study period compared to considering only swab positive patients. Overall the clinical, biochemical and radiological features of the swab-negative clinical COVID-19 cohort did not differ significantly to the swab-positive cohort. Typical features of COVID-19 such as shortness of breath, typical chest x-ray findings and lymphopenia are more common in the swab-negative cohort. The likely explanation is that swab-positive patients with a range of symptoms are classified as COVID-19 patients based on the positive swab test, whilst swab-negative patients diagnosed as having COVID-19 have their diagnosis based on typical symptoms. Swab-negative patients tended to have a longer delay between symptom onset and hospitalization, and had significantly better hospitalization outcomes, including mortality. Whether this potentially less aggressive phenotype is related to a lower viral load, variations in immune response or a different stage of illness presenting after period of peak viral replication is not yet known. Meta-analysis of studies reporting viral load and shedding of SARS-CoV-2 report that upper respiratory tract viral titres peak in the first week of illness 13 . Another analysis of published studies have reported the probability of false negative swab SARS-CoV-2 swab testing increases after day 3 of symptoms and upper respiratory tract SARS-CoV-2 viral load reduces over time 9,14 . Further research including immunological profiling would help further elucidate underlying disease mechanisms.
Our results provide evidence to validate swab-negative COVID-19 as a real clinical entity commonly encountered in hospital settings, findings which have important implications for current clinical practice and public health guidance 15 . This is supported by reported cohorts in the literature of patients with clinical features of COVID-19 with negative RT-PCR swab testing and positive serological testing [16][17][18] .
Our study highlights that clinicians caring for COVID-19 patients should maintain a high clinical suspicion, even in the presence of a negative swab result. Our results suggest that a combination of clinical, radiological and biochemical features in keeping with COVID-19 disease, rather than swab results alone, should guide clinical management.
Furthermore, managing swab-negative clinical COVID-19 within hospital poses disease control challenges. Cohorting this group with swab-positive patients presents potential risk of SARS-CoV-2 infection, if the clinical diagnosis is incorrect. Alternatively, isolating these patients from both swab-negative low-risk patients and swabpositive groups would place significant strain on hospital resources and may be unfeasible during an outbreak. Nasopharyngeal swabbing is vulnerable to inter-operator variation and poor sampling technique may explain why patients with high clinical suspicion of COVID-19 had negative swab testing. Repeated testing may be beneficial in some cases however in our cohort 41/94 swab negative clinical COVID-19 cases had a repeat swab which remained negative. Testing on sputum or bronchial lavage may provide greater sensitivity [6][7][8] . However, their use is limited by feasibility and potential risk to healthcare workers. Serological testing to confirm exposure to SARS-CoV-2 is likely the optimal second-line test as it is not operator-dependent, is low-risk to healthcare workers, and can feasibly be performed on a large number of patients. Serological proof of exposure could reduce concern regarding cohorting with swab-positive patients. Although serological testing has demonstrated good sensitivity and specificity performance can vary between serological kits 19 .
Lastly, but importantly, from a public health perspective, swab-negative clinically diagnosed COVID-19 patients may not be accurately captured in surveillance statistics, thereby underestimating healthcare demand. As the pandemic evolves countries with reducing incidence should consider this clinical group and how they should be addressed when planning interventions to manage second-wave outbreaks and contact tracing.
Our study is limited by its retrospective nature and likely under-estimates the proportion of patients with swab-negative clinical COVID-19, as it requires a high degree of clinical confidence without alternative diagnosis. www.nature.com/scientificreports/ Our relatively small cohort size is therefore a further limitation. The lack of positive confirmatory test means we cannot be sure of the true diagnosis in the clinically diagnosed swab negative group. However, this does not detract from the clinical problems this group presents to hospital management and epidemiological assessment.
In the absence of improved testing or routine second line testing for patients who have high clinical suspicion of COVID-19 but negative swab testing, this group is likely to continue to pose a clinical challenge. During the period of data collection changes in local testing policy meant that not all patients were retested for SARS-CoV-2 which could limit the generalizability of some of our results. However this is probably more reflective of the realworld context, particularly where resources are limited. Lastly, it is also possible that, given our relatively small cohort size, demographic and clinical differences between swab-positive and swab-negative patients were not evident. This finding warrants further investigation, as larger studies or meta-analyses are needed to allow the investigation of potential differences between such cohorts and even undertake multivariable regression analyses. Patients with strong clinical features of COVID-19 who have negative nasopharyngeal RT-PCR test results are a common but understudied clinical group. We found they do not differ significantly to similar swab-positive patients but seem to have better outcomes. Healthcare services should recognize and plan for the management of this group when making disease control interventions and epidemiological assessments.