Skip to main content

Publication

Coronavirus as Recorded in Primary Care - EXPERIMENTAL STATISTICS

Page contents

5. Caveats

There are several data limitations that should be considered when interpreting the results of this analysis. First the analysis is restricted only to patients residing in England with a valid NHS number and registered at an English general practice. Further, the results rely on events captured by primary care. These will either be by patients engaging with the general practitioner face-to-face, remotely, or test results sent to the practice. For instance, there is no mandate for lateral flow tests to be sent to general practice; some of them will be captured but some will be missed. Therefore, interpreting primary care data in isolation might underestimate the total number of COVID-19 infections in the population.

Diagnoses are captured only using SNOMED codes and no free text or local codes have been considered. The SNOMED codes were classified manually and reviewed by clinical advisors; however, they can be open to interpretation. This analysis does not consider the severity of infection.

In this analysis we only consider the first diagnosis of a patient. However, if a patient was recorded with suspected COVID-19 in the early stages of the pandemic but is subsequently reinfected with a test confirmation, they will only be included at their second infection.

In this analysis, the primary care record is the sole source of patient demographics, except for ethnicity which can also include hospital data. Geographic location and deprivation are estimated for each patient using the LSOA recorded in their latest GP record within the study window. Age is calculated from date of birth on event date. Ethnicity is built using multiple records from primary and secondary care. There is some variance in how ethnicity is captured in medical records. For some, ethnicity is self-reported whereas for others it might be ascribed to patient.

Reported infection rates are determined using denominators calculated from a snapshot of the full GDPPR data set on 19th October 2020.  Patients were counted only if they have a valid age (between 0 – 130), sex, or an LSOA in England. They were grouped according to their most recent demographic characteristics recorded on or before 19 October 2020.  For patients with two conflicting entries on the same day, an entry was selected at random.

The findings in this report are descriptive only.  Any differences presented can reflect either actual differences in infection rates or behaviour differences in how the different groups tested or approached their GP. Patient characteristics were examined in isolation and no confounding has been considered. For instance, difference in infection rates across ethnic groups could reflect deviations due to regional variability. 



Last edited: 19 May 2021 1:57 pm