Comparing the efficiency of forward and backward contact tracing

TL;DR

This paper critically evaluates the effectiveness of forward and backward contact tracing strategies in infectious disease outbreaks, revealing that their success depends heavily on disease dynamics and simulation methods.

Contribution

It challenges prior claims by showing that the effectiveness of contact tracing strategies varies with disease context and emphasizes the need for parallel simulation of disease spread and mitigation.

Findings

Backward contact tracing is not universally more effective than forward tracing.

Disease dynamics critically influence the impact of tracing strategies.

Simulating disease spread and mitigation measures simultaneously is essential.

Abstract

Tracing potentially infected contacts of confirmed cases is important when fighting outbreaks of many infectious diseases. The COVID-19 pandemic has motivated researchers to examine how different contact tracing strategies compare in terms of effectiveness (ability to mitigate infections) and cost efficiency (number of prevented infections per isolation). Two important strategies are so-called forward contact tracing (tracing to whom disease spreads) and backward contact tracing (tracing from whom disease spreads). Recently, Kojaku and colleagues reported that backward contact tracing was ``profoundly more effective'' than forward contact tracing, that contact tracing effectiveness ``hinges on reaching the `source' of infection'', and that contact tracing outperformed case isolation in terms of cost efficiency. Here we show that these conclusions are not true in general. They were based in part on simulations that vastly overestimated the effectiveness and efficiency of contact tracing. Our results show that the efficiency of contact tracing strategies is highly contextual; faced with a disease outbreak, the disease dynamics determine whether tracing infection sources or new cases is more impactful. Our results also demonstrate the importance of simulating disease spread and mitigation measures in parallel rather than sequentially.