NHS Test and Trace works, new research finds, but the data glitch which led to over 15,000 late referrals caused more than 125,000 additional COVID-19 infections

NHS Test and Trace works, new research finds, but the data glitch which led to over 15,000 late referrals in October may have brought about more than 125,000 additional COVID-19 infections and over 1500 deaths.

On 3 October Public Health England revealed that, due to a technical error, at least 15,841 individuals who tested positive for COVID-19 between 20 September and 2 October had not been referred to the national contact tracing system. The technical error was due to an Excel spreadsheet reaching its maximum file size. New analysis shows that this data glitch, which by chance affected some regions more than others, led to a significant increase in new infections – and subsequently more deaths – in those areas.

In the new paper, ‘Does Contact Tracing Work? Quasi-Experimental Evidence from an Excel Error in England’, Dr Thiemo Fetzer of the CAGE Research Centre in the Economics Department at the University of Warwick and Dr Thomas Graeber of Harvard Business School consider the effectiveness of contact tracing by analysing the causal impact of the delays resulting from the data glitch. Key findings are:

· By chance, the loss of case information from was more severe for some areas of England than others.

· Areas that were more strongly affected by the lack of contact tracing subsequently experienced a sharp increase in COVID-19 infections and COVID-19 related deaths.

· The study’s most conservative estimates suggest that the failure to contact trace is associated with more than 125,000 additional infections and more than 1500 deaths.

· The delay in contact tracing was further associated with a subsequent worsening of the regional performance of contact tracing.

The research leverages daily data on the spread of COVID-19 at the Lower Tier Local Authority level, together with weekly data on NHS Test and Trace testing activity and contact tracing performance.

To demonstrate the causal connection between the delays in contact tracing and the spread of COVID-19, Dr Fetzer and Dr Graeber show that areas that had more delayed contact tracing cases saw a notable differential increase in infections ­– and subsequently in deaths – compared to other, otherwise comparable areas.

Dr Fetzer said, ‘Distinguishing correlation from causation in the analysis of non-pharmaceutical interventions is hard. This data glitch gives researchers a natural experiment to cast a light on whether contact tracing can make a difference as it resulted in a large number of cases that were not contact traced.’

The results strongly suggest that contact tracing is an important tool to contain the pandemic. Dr Graeber said, ‘The large size of the estimated effects implies that a well-functioning and robust contact tracing system may be even more important to fight COVID-19 than we previously thought’.

23 NOVEMBER 2020

NOTES TO EDITORS:

The contact tracing system in England is organised at the national level through NHS Test and Trace. Each day, laboratories report positive COVID-19 test results to Public Health England (PHE), which aggregates and forwards all case information to NHS Test and Trace. NHS Test and Trace contracts all elements of their tracing operations to private companies and has been reported to draw on a total £12 billion. The data transmission error occurred when PHE failed to transfer information on all positive cases to NHS Test and Trace, which then did not allow NHS Test and Trace to trace all positive cases.

PAPER AVAILABLE AT:

Does Contact Tracing Work? Quasi-Experimental Evidence from an Excel Error in England. Thiemo Fetzer (University of Warwick) and Thomas Graeber (Harvard Business School).

CAGE Working paper: https://bit.ly/394Ebuo

Learning about the effectiveness of contact tracing from when it failed – a natural experiment. Thiemo Fetzer (University of Warwick) and Thomas Graeber (Harvard Business School).

CAGE Policy briefing: https://warwick.ac.uk/fac/soc/economics/research/centres/cage/manage/publications/bn30.2020.pdf