Utilising data for 27 million individuals in England, our study highlights marked variation in the impact of the COVID-19 pandemic on diagnosis rates across a range of diseases and sociodemographic characteristics. New diagnoses of all 19 long-term conditions fell sharply during the first year of the pandemic, but with a disproportionate impact on conditions including COPD, asthma, psoriasis and osteoporosis. Rebound increases in diagnoses were subsequently observed for several conditions – particularly CKD, for which reported cases have surged since 2022. In contrast, substantial deficits in diagnoses have persisted for other conditions nearly five years after the onset of the pandemic, most notably for depression. Analyses stratified by ethnicity and socioeconomic status highlighted compensatory increases in dementia diagnoses for individuals of White ethnicity and those from less deprived socioeconomic quintiles, but not for individuals of other ethnicities or more deprived socioeconomic backgrounds. Importantly, this study demonstrates how routinely collected health data can transform how we monitor disease epidemiology in close to real-time.

Data from early in the pandemic reported sharp decreases in incident diagnoses for several long-term conditions, including asthma, COPD and rheumatoid arthritis.^6,7,17,19,20 Little was known, however, about the extent of recovery in diagnosis rates as healthcare services emerged from the pandemic. Demonstrating this at a population level requires regularly updated data from primary and secondary care settings, which has only recently become possible with the introduction of the OpenSAFELY Trusted Research Environment in England.⁵ Our demonstration of a disproportionate impact of the early pandemic on respiratory conditions, such as COPD and asthma, supports the findings from a previous population-level study in Wales, which included data up to December 2021.¹⁷ Using contemporaneous data in OpenSAFELY, our study demonstrates that, while incidence rates for COPD and asthma have gradually recovered following these initial decreases, large deficits remained as of November 2024, suggesting many cases remain undiagnosed as a consequence of the pandemic. An important contributing factor is likely to be the sudden disruption to spirometry and lung function testing during the pandemic, resulting in large backlogs in testing.²¹ As such, equitable access to spirometry has been identified as a priority area for the NHS as it recovers from the pandemic.²¹

Of all conditions studied, the largest relative and absolute deficits in diagnoses were observed for depression. After an initial decrease in diagnoses during the early pandemic, the incidence of depression partly recovered by late 2021, but has declined markedly since 2022. While a disproportionate impact of the early pandemic on mental health diagnoses has been reported previously,^17,19 the recent fall in new diagnoses of depression warrants further investigation. Our findings are supported by decreases in new initiations of antidepressant medications and referrals to mental health services for depression in England since 2022, following large increases previously.^22,23 It is important to note, however, that the observed decline in new diagnoses may not necessarily equate to a decrease in the underlying incidence of depression. New claimants of disability-related benefits due to mental health conditions have increased by 190% in England between 2019/20 and 2023/24.²⁴ One possible explanation for this disconnect is that people with symptoms of depression are taking longer to be formally diagnosed due to increasing pressures on the NHS.^25,26 Primary care services have faced particular challenges with increased demand since the pandemic. Although the total number of face-to-face consultations has increased, the proportion of remote consultations has also grown, which together with pressures on time in consultations, may make it more challenging to identify non-verbal cues associated with depression.²⁷ It is also possible that individuals with symptoms of depression may be accessing services without having their diagnoses recorded in primary or secondary care records. Following a national drive to expand access to psychological therapies, referrals to NHS Talking Therapies services increased by nearly two-thirds between 2013/14 and 2023/24 (from 1.1 million to 1.8 million referrals, respectively), with self-referrals (i.e. without necessarily needing to consult a healthcare professional prior to referral) constituting 69% of all referrals.²⁸ Recommendations from the National Institute for Health and Care Excellence (NICE) – updated in June 2022 – emphasised the importance of therapy-based treatments over antidepressants for new cases of less severe depression.²⁹ This, in turn, may have contributed to a change in how clinicians are recording depression diagnoses. In our sensitivity analysis that included additional diagnostic codes for symptoms of depression (e.g. depressed mood), the relative deficit in depression diagnoses was attenuated; indicating a shift towards recording symptom-related codes over formal depression diagnoses. Previous studies reported similar findings after the introduction of the Quality and Outcomes Framework (QOF) incentivisation scheme in the NHS, which resulted in GPs switching from recording diagnostic codes for depression to symptom-related codes.³⁰

A large, persistent deficit in diagnoses was also observed for osteoporosis. Delays in diagnosis and the initiation of bone protection therapy can substantially impact prognosis for individuals with osteoporosis. Osteoporotic fractures are associated with a high burden of disability and functional impairment,³¹ with estimates of mortality in the first year after hip fracture ranging from 15% to 36%.³² Bisphosphonates reduce the risk of hip and vertebral fractures in post-menopausal women by 35% and 54%, respectively;³³ highlighting the importance of early diagnosis and treatment. Our findings therefore suggest that a national case-finding strategy may be warranted to identify individuals who have not yet been diagnosed with or treated for osteoporosis as a consequence of the pandemic.

In contrast, a large increase in newly-recorded diagnoses of CKD has been evident since 2022. While this may partially represent a catch-up in diagnoses following service disruption during the pandemic, the observed increase in CKD diagnoses was 33% higher than expected based upon pre-pandemic trends. Subgroup analyses highlighted that much of this increase occurred in individuals aged 80 years and over, and in those from less deprived socioeconomic quintiles. There are numerous factors that could have potentially contributed to this increase, including accelerated GFR decline following COVID-19 infection,^34,35 the inclusion of CKD as a risk factor in national cardiovascular prevention audits (e.g. CVDPREVENT),³⁶ initiatives to facilitate the use of novel therapies to prevent CKD progression (e.g. SGLT-2 inhibitors),³⁷ and remote albuminuria testing programmes.³⁸ In August 2021, updated guidance on CKD assessment and management was published by NICE, containing recommendations for the investigation of CKD, including screening of at-risk individuals, which may have led to enhanced case detection after its implementation.³⁹ Additionally, updated NICE guidelines recommended removal of ethnicity-related adjustment factors when estimating glomerular filtration rates (eGFR). Use of ethnicity-based eGFR was demonstrated to overestimate GFR and under-recognise CKD in individuals of Black ethnicity in the UK,^39,40 which may have contributed to the observed increase in individuals of Black ethnicity meeting criteria for CKD diagnosis following its removal. However, the comparable relative increase in CKD diagnoses among individuals of White ethnicity suggests that this factor alone is unlikely to explain the large overall increase in CKD diagnoses.

Importantly, the methodology utilised in this study provides a mechanism through which to highlight variation by age, sex, ethnicity, and socioeconomic status. Despite England having a universal, free at the point-of-access healthcare system, we observed compensatory post-pandemic increases in dementia diagnoses for individuals of White ethnicity and those from less deprived socioeconomic quintiles, whereas diagnosis rates remained lower than expected among individuals from other ethnic backgrounds and more deprived communities. By highlighting and exploring these disparities in more detail – for example, through more granular analyses by ethnicity and region – these data could inform the development and targeting of quality improvement interventions towards reducing inequity. Contemporaneous data sources in OpenSAFELY provide opportunities not only for surveillance and early detection of diagnostic trends and inequity, but also the ability to monitor the impact of quality improvement interventions, while minimising the burden of manual data collection and maximising the inclusion of populations who are often under-represented in clinical studies.^41-43 Although the primary focus of our study was to describe changes in the incidence and prevalence of long-term conditions, the methodology and analysis code can be readily adapted to assess care quality benchmarked against clinical guidelines,^6,7 and evaluate the impact of strategies aimed at improving healthcare delivery.⁴¹

Our study had several strengths, including the use of nationally representative data from over 27 million individuals to provide updated estimates of incidence and prevalence for 19 long-term conditions, spanning multiple disease areas. This approach can be expanded to assess the incidence, prevalence, and management of numerous other diseases at national, regional and local levels, including rare diseases with limited epidemiological data. Additionally, all analyses can updated regularly and made publicly available,⁴⁴ enabling near real-time monitoring of changing incidence rates – for example, in response to the implementation of clinical guidelines.

There are also limitations that must be acknowledged. As with other studies utilising coded health data, there is a potential for diagnostic misclassification. While these data can be used to highlight temporal trends in new diagnoses recorded in primary and secondary care, we were unable to infer whether the observed differences represented true changes in underlying disease incidence (e.g. following SARS-CoV-2 infection), changes in the recording of diagnoses (e.g. due to updated guidelines and/or incentivisation), changes in diagnostic testing (e.g. increased testing of B-type natriuretic peptide for heart failure), or changes in service delivery (e.g. care provision outside of primary and secondary care settings). Trends in diagnoses must be considered against the backdrop of abrupt changes in health care utilisation during the pandemic: the number of general practice appointments in England fell by one third between March and April 2020, before increasing above pre-pandemic levels since then.⁴⁵ To maximise case capture, we utilised data from both primary care and secondary care sources; however, it is important to acknowledge the differences between diagnoses made in primary care and hospital settings. We were unable to age and sex-standardise our subgroup analyses by socioeconomic status or ethnicity due to the computationally-intensive nature of these analyses; although crude and adjusted estimates for the overall cohort were comparable, suggesting this is unlikely to have meaningfully altered our findings. We reported incidence trends separated into five ethnicity categories for most, but not all, conditions, due to small numbers of incident cases and the potential for disclosure; this also precluded more granular reporting of differences within these broad ethnicity categories. Importantly, as these analyses were conducted using routinely collected data from the NHS of England, they may not necessarily be generalisable to other healthcare systems or settings.

In conclusion, our study utilised routinely collected health data for 27 million individuals to provide a comprehensive epidemiological analysis of chronic diseases in England. We highlighted large post-pandemic deficits in diagnoses for conditions including depression, COPD, asthma and osteoporosis, contrasting a surge in recorded CKD diagnoses since the pandemic. Importantly, this study demonstrates the potential for real-time health data to transform disease monitoring, identify inequities across under-served populations, and inform strategies to improve case detection and reduce diagnostic delays.