Critical Thresholds in Non-Pharmaceutical Interventions for Epidemic Control

TL;DR

The paper addresses how non-pharmaceutical interventions (NPIs) interact to control epidemics, focusing on the speed of contact tracing and the scale of social interactions. It introduces a probabilistic TTQ framework with variables $\\tau$ and $\\bar{k}_+$ and derives the universal threshold $R=1$ and a critical line in the $\\bar{k}_+-\\tau$ plane. Using Shenzhen's 2022 Omicron outbreak data (1,187 cases, 86,451 contacts), it validates the model and quantifies containment thresholds: tracing alone controls diseases with $R_0 < 2.12$, and with social distancing $R_0 < 7.82$. The work offers actionable guidance for designing budget-friendly NPIs and highlights requirements for data-driven adaptive policies and robust tracing infrastructure.

Abstract

Non-pharmaceutical interventions, such as contact tracing and social distancing, are critical for controlling epidemic outbreaks, yet their dynamic interactions remain underexplored. We introduce a probabilistic framework to analyze the synergy between contact tracing speed, quantified by the contact tracing period $τ$, and the average number of close contacts, $\bar{k}_+$, reflecting social distancing measures. We identify critical thresholds ($R=1$) that separate pandemic and contained phases in the $\bar{k}_{+}-τ$ plane, validated using high-resolution data from Shenzhen's 2022 Omicron outbreak (1,187 cases, 86,451 contacts). Our findings show that contact tracing alone can contain diseases with $R_0 < 2.12$ (95% CI 2.07-2.16), covering 43.33% of major infectious diseases, while combining with social distancing extends control to $R_0 < 7.82$ (95% CI 7.70-7.93), encompassing 86.67% of pathogens. These results, supported by empirical data, highlight the efficacy of rapid tracing and targeted social distancing as alternatives to mass PCR testing. Our framework offers actionable insights for optimizing NPI strategies, though challenges in scaling to regions with higher tracing miss rates or weaker infrastructure underscore the need for adaptive, data-driven policies.

Figures (3)

• Figure 1: Epidemic Control in Shenzhen, 2022. (a) Contact trajectories of positive cases. Nodes are geolocated to cases' residential addresses. Edges represent epidemiological links indicating close contact history between them. The color of the edge (olive green, warm peach, burnt orange) indicates three time periods divided equally according to the chronological order of close contact. (b) A real transmission chain (see Supplementary Sec. 3). Muted teal nodes represent individuals who have no symptoms and have not been tested positive or have not been tested; burnt orange nodes represent individuals who have symptoms and have been tested positive; warm peach nodes represent individuals who have symptoms and have not been tested. Warm beige circles around nodes indicate that individuals within each circle have been under quarantine till the corresponding date, and the directed edges between nodes represent the disease transmission path. The transmission chain started from a retiree, confirmed through the community's PCR testing on February 22nd, who was immediately isolated. A housekeeper, who lived in the same building as the retiree, was identified as a close contact and was quarantined on February 25th. Subsequently, three family members served by the housekeeper (nodes c, k, and d), the housekeeper's daughter (node e), and three other close contacts (nodes i, j, and l) were also quarantined. Until February 28th, with all 13 people on the transmission chain (of whom six tested positive) in quarantine, this transmission chain was cut off successfully. (c) The process of isolating positive cases and quarantining close contacts according to Shenzhen's Prevention and Control Manual (see Supplementary Sec. 4).
• Figure 2: Theoretical results versus empirical data analysis results. (a) Theoretical relationship between the critical reaction time $\tau_c$ (in hours) and the basic reproduction number $R_0$ for different variants of COVID-19 by Eq. S3. The dark dots mark the $R_0$ and the corresponding $\tau_c$ of four COVID-19 variants and the one spread in Shenzhen as listed in Table \ref{['table: R0 tauc']}. The olive green asterisks represent the tracing period set in the COVID-19 Prevention and Control Manual for Delta (17 hours) and Omicron (11 hours). (b) Theoretical relationship between $\tau_c$ and $R_0$ for three typical epidemics by Supplementary Eq. S3. The dots mark the $R_0$ and the corresponding $\tau_c$ of SARS, H1N1, H3N2 as listed in Table \ref{['table: R0 tauc']}. Distributions of disease transmission state durations are provided in Supplementary Sec. 2A. (c) Phase plot of the containment policy on the $\bar{k}_{+} - \tau$ plane and $\bar{k}_{+t}-\tau_t$ trajectory of Shenzhen's epidemic control from February 16th to April 2rd, 2022 where each dot is calculated by averaging over the positive cases confirmed from day $t-3$ to day $t + 3$. The error bars are 95% confidence intervals. The dash-dotted line marks the actual value of the average number of effective contacts ($\bar{k}_{+SZ}$) in Shenzhen. The dashed line marks the critical value ($\bar{k}_{+c}$ by Supplementary Eq. S4) with transmission rate $\beta_{SZ}$ in Shenzhen) in the absence of contact tracing or quarantine (i.e., $\tau\to\infty$). The solid curve represents the critical line obtained by $R=\bar{k}\beta G(\tau)$ and setting $R=1$. Note that while the solid curve represents the theoretical result, the points are based on real data. Please refer to Supplementary Sec. 6 for the computation of the time-varying $\tau_t$ and $\bar{k}_{+t}$, as well as $\bar{k}_{+SZ}$ and $\beta_{SZ}$. (d) The daily effective reproduction number $R_t$ and number of newly confirmed cases (see Supplementary Sec. 6E for more details).
• Figure 3: Policy with incomplete contact tracing. The olive green and burnt orange lines in (a)-(c) represent simulation results under the Shenzhen network without and with social distancing measures (w/o and w/ SDM), respectively. In each simulation, we miss a proportion $q$ of close contacts for each positive case during contact tracing. Based on Shenzhen's control data, we set $q=0.186$ as the proportion of positive cases identified by large-scale PCR testing instead of contact tracing. The contact tracing periods in all simulations are set to $\tau=11$ hours. (a)-(b) Stabilized effective reproduction number (a) and average stopping time (b) for diseases with different $R_0$. The average stopping time is the average period from the start of each simulation until all positive cases in the simulation are in the R (removed) or Q (quarantined) state. The basic reproduction numbers of H1N1 tan2013modeling, SARS riley2003transmission, Alpha liu2020reproductivecampbell2021increased, Delta liu2020reproductivecampbell2021increased and Omicron BA.1 wang2022reproduction are marked on the $x$-axis. (c) The relationship between the missing rate in contact tracing, versus the critical basic reproduction number $R^c_0$, i.e., the value of $R_0$ under which an infectious disease can be successfully controlled. Points with four different colors mark the $R_0^c$ corresponding to missing rates of Shenzhen, major regions in the UK keeling2020efficacy, major regions in the US lash2021covid, and Hong Kong yang2022universal. For details of the simulations, please refer to Supplementary Sec. 9 and 10. Note that these results are theoretical results based on Shenzhen's $\bar{k}_{+}$, generalization to other regions may differ in practice. (d) Histogram of the basic reproduction number ($R_{0}$) of 30 known major infectious diseases (see Supplementary Tab. S6 for details). 86.67% of the known major infectious diseases can be contained with the control policy, and 43.33% can be contained with contact tracing alone.