Inter-city movement pattern of notifiable infectious diseases in China: a social network analysis

Abstract

Background: Co-existence of efficient transportation networks and geographic imbalance of medical resources greatly facilitated inter-city migration of patients of infectious diseases in China.

Methods: To characterize the migration patterns of major notifiable infectious diseases (NIDs) during 2016-2020 in China, we collected migratory cases, who had illness onset in one city but were diagnosed and reported in another, from the National Notifiable Infectious Disease Reporting System, and conducted a nationwide network analysis of migratory cases of major NIDs at the city (prefecture) level.

Findings: In total, 2,674,892 migratory cases of NIDs were reported in China during 2016-2020. The top five diseases with the most migratory cases were hepatitis B, tuberculosis, hand, foot and mouth disease (HFMD), syphilis, and influenza, accounting for 79% of all migratory cases. The top five diseases with the highest proportions of migratory cases were all zoonotic or vector-borne (37.89%‒99.98%). The network analysis on 14 major diseases identified three distinct migration patterns, where provincial capitals acted as key node cities: short distance (e.g., pertussis), long distance (e.g., HIV/AIDS), and mixed (e.g., HFMD). Strong drivers for patient migration include population mobility and labor flow intensities between cities as well as the economic development level of the destination city.

Interpretation: Collaborative prevention and control strategies should target cities experiencing frequent patient migration and cater to unique migration patterns of each disease. Addressing disparity in healthcare accessibility can also help alleviate case migration and thereby reduce cross-regional transmission.

Funding: National Key Research and Development Program of China.

Keywords: Disease migration; Human mobility; Network analysis.

Fig. 1

Characteristics of migratory cases of notifiable infectious diseases during 2016–2020 in China. (A) Numbers (connected dots) and proportions (bars) of migratory cases by sex and age group; (B) Temporal distribution of proportions of migratory cases by day of the week; (C) Spatial distribution of proportion of inflow cases among all reported cases at the city level; and (D) Spatial distribution of proportion of outflow cases among all reported cases at the city level. Centers of provincial capital cities are marked by dots in (C) and (D).

Fig. 2

Network community structure for fourteen major infectious diseases. Cities in the same community are colored the same, representing a sub-network.

Fig. 3

Network backbones of migration networks for (A) pertussis, representing the short distance migration mode, (B) HIV/AIDS, representing the long distance migration mode, and (C) hand, foot and mouth disease, representing the mixed mode. The color of each directed edge changes from dark blue to bright blue to represent the direction of migration from the origin to the destination. The size of each node (city) represents the node strength of the city in the migration network. Each node is colored by the node betweenness in the network, with darker red indicating a larger value of node betweenness.

Fig. 4

The SHAP feature importance of predictors for infectious disease migration. The numbers in the grids represent the relative SHAP importance (measured as the mean absolute Shapley values) of the XGBoost models, scaled to sum up to 100% for each row. The grids are also colored in red (blue) for high (low) relative importance.