Demographic of Case Controls
There were 249,390 people who had a positive SARS-CoV-2 test between 28/02/2020 and 26/08/2021. When these were propensity matched with controls this number reduced to 41,838 individuals, thus removing 83% of the data. The dataset was then further restricted by removing all matches for whom their test location was matched as missing.
Three matched cohorts are used in this study; COVID-19 test positive (case), COVID-19 test negative (control) and not tested (control). 23,015 and 69,566 individuals were identified to have had a positive and negative test respectively. Appendix C shows ‘Love Plots’ for the standardised mean distribution before and after the propensity matching had occurred. Censorship patterns were checked and were similar across the cohorts.
Death Following RT-PCR
The mortality rate was higher within the first four weeks of those who tested positive (n = 3019.77) compared with individuals who tested negative (n = 980.36) or who were untested (n = 128.23) (see figure 2). The untested population maintains a low mortality frequency for the entire follow up period (77.29 - 140.55). The mortality frequency in positive and negative controls converges at 5-8 weeks and declines towards the end of the follow-up period. The negative population had a marginally higher frequency during the 5–24-week period.
These findings are also observed in the cox proportional hazard models (see figures 5 - 8). Positive individuals were significantly more likely to die within 28 days of a positive test in both the community (HR: 4.10, 95%CI: 3.38 to 4.97, p<0.001) and hospital (HR: 2.83, 95%CI: 2.50 to 3.22, p<0.001) environments when compared controls. Untested individuals were significantly less likely to die (HR: 0.13, 95%CI: 0.11 to 0.16, p<0.001) compared with tested. After 5 weeks, positive individuals were significantly less likely to die if tested in the community (HR: 0.72, 95%CI: 0.58 to 0.91, p=0.005) more likely to die if tested in hospital (HR: 1.22, 95%CI: 1.01 to 1.47, p=0.037. Untested individuals were much less likely to die than those tested after 5 weeks (HR: 0.36, 95%CI: 0.32 to 0.41, p<0.001)
Risk ratios from the life table analysis support the observed trends and can be seen in table 2. They illustrate that individuals are more likely to die following a positive test in both the hospital (RR: 2.77, 95%CI: 2.45 to 3.14) and community (RR: 4.06, 95%CI: 3.35 to 4.92) environment in the first four weeks. After 5 weeks an individual was less likely to die if receiving a positive test in the community or if no test had been taken by this point.
Outcomes 1 - 4 - and 5 - 24-weeks Post Test
Combined Environments
The hazard ratios, confidence intervals and p-values for the occurrence of each outcome can be seen in figures 3-6. For both environments combined (figure 3), diagnosis of influenza (HR: 5.68, 5.08 to 6.43, p<0.001), fatigue (HR 1.77, 1.34 to 2.35, p<0.001), respiratory condition (HR: 1.53, 1.43 to 1.63, p<0.001), embolism (HR: 1.50, 1.15 to 1.97, p<0.003), requirement for a sick note (HR: 3.58, 95%CI:1.20 to 10.07, p = 0.022) and death (HR: 3.12, 95%CI:2.80 to 3.46, p<0.001) were higher in the first 4 weeks compared to their propensity matched controls. In the 5–24-week period, previously positive individuals had higher post-viral syndrome (HR: 4.57, 95%CI:1.77 to 11.08, p=0.002), embolism (HR: 1.51, 95%CI:1.13 to 2.02, p=0.005) and fatigue (HR: 1.47, 1.24 to 1.75, p<0.001). In addition, when compared to negative controls, survivors of COVID-19 were less likely to be diagnosed with anxiety, depression, or other mental or developmental condition codes.
Separate Test Environments
Hazard ratios for community and hospital tested individuals can be seen in figures 4 and 5. The use of the influenza codes were much higher in those diagnosed in the community (HR: 10.31, 95%CI: 8.53 to 12.50, p<0.001) compared to the hospital diagnosed (HR: 3.92, 95%CI: 3.40 to 4.51, p<0.001) in the first 4 weeks. Sick notes were more likely to be given following a positive test in hospital (HR: 3.04, 95%CI: 0.88 to 10.50, p=0.079) and community sites (HR: 6.43, 95%CI: 0.58 to 70.90, p=0.130). Fatigue occurred with a similar frequency 1-4 weeks following tests at either location Hospital (HR: 1.81, 95%CI:1.08 to 3.02, p = 0.024) and community (HR: 1.76, 95%CI: 1.26 to 2.46, <0.001). Embolism codes occurred more frequently within 1-4 weeks in the community (HR: 2.83, 95%CI: 1.91 to 4.21, p<0.001) vs Hospital (HR: 0.88, 95%CI: 0.55 to 1.27, p = 0.440).
Individuals were more at risk of post-viral syndrome following a test in the community (HR:4.65, 95%CI: 1.52 to 14.20, p<0.007) and hospital (HR: 4.40, 95%CI: 0.73 to 26.30, p<0.100). Regardless of the environment, individuals who had tested positive were more likely to have an embolism code 5-24 weeks following a test (HR: 1.50, 95%CI:1.01 to 2.21, p = 0.043) (community) and (HR:1.54, 95%CI 1.01 to 2.37, p = 0.047) (hospital). Survivors of COVID tested in the community appeared to be at a lower risk from anxiety, depression and self–harm (HR: 0.87, 95%CI: 0.77 to 1.00, p = 0.048), however those tested in the hospital were at an increased risk (HR: 1.20, 95%CI: 1.00 to 1.45, p = 0.053).
Untested Controls
Hazard ratios for the untested controls comparison to test negative individuals can be seen in figure 6. The people who did not have any record of a COVID test between 28/02/2020 and 26/08/2021 were at a lower risk of all outcomes.
Fatigue
Fatigue was higher for the entire follow-up period for those who had tested positive. Community rates were comparable to those in hospital. Those who had been in hospital (but tested negative) had higher levels of fatigue than people in the community who were untested (see figure 7).
Embolism
Across all groups, the overall volume of events is very low. The risk of embolism (figure 8) is much higher in those who were in hospital. The rates are equivalent in those testing either positive or negative. There appears to be a greater volume of events within the first 3 weeks for those who were hospital negative vs hospital positive. There seems to be an equivalent number of events up to around 140 days where there begin to be marginally more in the hospital negative group. Positive cases in the community also seem to have a greater volume of events compared to community negative. Beyond 3 weeks both community groups appear to have a similar frequency of events shown by a similar gradient.
Life Table Analysis
Table 2 shows individuals who tested positive in the community have an increased risk or increased trend of a first embolism and first fatigue code occurring through almost the entire follow up period. Community positive individuals are only less at risk of an embolism code 21-24 weeks following a test.