Open Access Research

A geographic analysis of population density thresholds in the influenza pandemic of 1918–19

Siddharth Chandra1*, Eva Kassens-Noor2, Goran Kuljanin3 and Joshua Vertalka4

Author Affiliations

1 Asian Studies Center, Michigan State University, 427 N Shaw Lane, Room 301, East Lansing, MI, 48824, USA

2 Urban and Transport Planning in the School of Planning, Design, and Construction and Global Urban Studies Program, 552 W Circle Drive, Room 201E, East Lansing, MI, 48824, USA

3 Department of Psychology, Psychology Building 316 Physics Room 262, East Lansing, MI, 48824, USA

4 Department of Geography, 673 Auditorium Road, Room 116, East Lansing, MI, 48824, USA

For all author emails, please log on.

International Journal of Health Geographics 2013, 12:9  doi:10.1186/1476-072X-12-9

Published: 20 February 2013



Geographic variables play an important role in the study of epidemics. The role of one such variable, population density, in the spread of influenza is controversial. Prior studies have tested for such a role using arbitrary thresholds for population density above or below which places are hypothesized to have higher or lower mortality. The results of such studies are mixed. The objective of this study is to estimate, rather than assume, a threshold level of population density that separates low-density regions from high-density regions on the basis of population loss during an influenza pandemic. We study the case of the influenza pandemic of 1918–19 in India, where over 15 million people died in the short span of less than one year.


Using data from six censuses for 199 districts of India (n=1194), the country with the largest number of deaths from the influenza of 1918–19, we use a sample-splitting method embedded within a population growth model that explicitly quantifies population loss from the pandemic to estimate a threshold level of population density that separates low-density districts from high-density districts.


The results demonstrate a threshold level of population density of 175 people per square mile. A concurrent finding is that districts on the low side of the threshold experienced rates of population loss (3.72%) that were lower than districts on the high side of the threshold (4.69%).


This paper introduces a useful analytic tool to the health geographic literature. It illustrates an application of the tool to demonstrate that it can be useful for pandemic awareness and preparedness efforts. Specifically, it estimates a level of population density above which policies to socially distance, redistribute or quarantine populations are likely to be more effective than they are for areas with population densities that lie below the threshold.

Influenza; Population loss; Population density; Thresholds; Spatial distribution; Population growth