	Table of contents

	Current issue

	Search journal

	Archived issues

	NZMJ Obituaries

	Classifieds

	Hotline (free ads)

	How to subscribe

	How to contribute

	How to advertise

	Contact Us

	Copyright

	Other journals

Journal of the New Zealand Medical Association, 15-June-2007, Vol 120 No 1256

Was rurality protective in the 1918 influenza pandemic in New Zealand?

Kirsten McSweeny, Atalie Colman, Nick Fancourt, Melinda Parnell, Sara Stantiall, Geoffrey Rice, Michael Baker, Nick Wilson

Abstract

Aims This study aimed to examine the impact of rurality on mortality rates from pandemic influenza in New Zealand in 1918.

Methods Mortality data was obtained from death certificates (in a published source) and denominator population data from the 1916 census (for the European population only). Analyses were conducted on cities (n = 4), towns (n = 111), counties (n = 97).

Results The influenza mortality rate for the towns and cities was more than twice that of the counties that represented rural settings (rate ratio (RR) = 2.13, 95% CI = 2.00–2.27). However, larger towns (population >2000 people) had a significantly lower mortality rate than smaller towns (RR = 0.81, 95%CI = 0.74–0.88). Similarly, cities had a lower mortality rate than larger towns (RR = 0.89, 95%CI = 0.83–0.95).

Conclusions These results are suggestive that rurality may have provided some protection from mortality during this influenza pandemic. This may have been due to a mix of remoteness and greater social distancing among rural residents. However, the differences in mortality rates between towns and cities may have reflected other factors such as the more organised provision of community care in the larger towns and cities, when compared to smaller towns.

With the potential threat of pandemic influenza, there is growing interest in examining public-health interventions that may temporarily isolate remote communities and use social distancing as prevention or control strategies. Historically the evidence that influenza pandemics had a greater impact in towns and cities compared to rural areas is incomplete.

Some studies of the 1918 pandemic have suggested higher mortality rates in urban settings—e.g. for Nigeria1 and South Africa.2 The picture for the latter is not entirely clear however, given the potential confounding by high mortality rates for particular ethnic groups in cities and the association between high mortality rates and linkages with the railway network. However, higher rural mortality in the 1918 pandemic has been described for Iran.3

Extreme remoteness (combined with some control measures) appears to have protected some populations in towns from infection in 1918 in Canada,4 the continental United States,5 and in Alaska.6 Also some military installations on islands offshore the United States appear to have escaped this pandemic,5 as did the remote island of St Helena in the South Atlantic.7 Remoteness may also have protected some island groups in the South Pacific—e.g. the Lau Islands and the Yasawas in Fiji.8 A detailed study in Yap in the South Pacific also reported that most of the “outer islands” escaped a 1964 influenza epidemic.9

To further explore the impact of rurality for the 1918 influenza pandemic, we examined New Zealand data from this period. We hypothesised that mortality rates (as a proxy for risk of infection) would be higher in towns and cities than counties.

Methods

Data on deaths from pandemic influenza for New Zealand Europeans were obtained from previously published work by one of the authors.10 The period covered was 17 October 1918 (date of first epidemic death) to 27 December 1918 (date of the last epidemic death). These numerator data were collected from the examination of individual death certificates throughout New Zealand (together with Auckland data collected by another historian, Linda Bryder). Denominator data for the New Zealand European population was obtained from the 1916 census.11 To examine the effect of geography on mortality rates, the country was divided into cities (population >20,000), large towns (population >2000), small towns (population <2000) and counties (representing rural areas).

For 11 counties, it was not possible to separate the data for towns and the surrounding county and so these were excluded from the analyses. These counties were: Cook County (and Gisborne), Sounds County (and Picton), Woodville County (and Woodville), Waipukurau County (and Waipukurau), Eketahuna County (and Eketahuna), Wairoa County (and Wairoa), Rodney County (and Warkworth), Waipawa County (and Waipawa), Hobson County (and Dargaville), Waitemata County (and Hellensville) and Whangarei County (and Whangarei). Statistical analyses were performed using Microsoft Excel and OpenEpi (Emory University, 2005) software.

Data was available from the 1916 census11 on mean occupants per dwelling for a total of 81 towns and counties. However, it was considered that this was too crude a proxy for crowding at the individual household level to make further analysis worthwhile.

Results

The cities and towns had significantly higher influenza mortality rates than the counties, which represented rural settings (Table 1). Considering all towns and cities together, the mortality rate ratio (RR) was 2.13 (95% confidence interval (CI) = 2.00–2.27) when compared to counties. However, larger towns (population >2000 people) had a significantly lower mortality rate than smaller towns (RR = 0.81, 95% CI = 0.74–0.88). Similarly, cities had a lower mortality rate than larger towns (RR = 0.89, 95%CI = 0.83–0.95). Within the cities, the mortality rates were significantly higher in the two North Island cities than the South Island cities (eg, the RR for Auckland relative to Dunedin was 1.92 (95% CI = 1.68–2.19).

Within counties there was no statistically significant association between population size of the county and with the mortality rate.

Table 1. Influenza mortality rates in cities, towns and counties for the 1918 pandemic influenza in New Zealand (European only)

Population grouping (N)	Deaths (N)	Population (N)	Mortality rate (per 1000 population per 3 months)**	Mortality rate ratio (95% CI)
Cities (4) Larger towns (26) Smaller towns (85) Counties* (97)	2616 1162 862 1213	404 916 157 554 95 362 367 358	6.5 7.4 9.0 3.3	1.96 (1.83–2.09) 2.21 (2.04–2.40) 2.74 (2.51–2.99) 1.00 (Reference)

*These are counties exclusive of any towns associated with them; **For the 3-month period: October to December 1918.

Discussion

Main findings and interpretation—This analysis identified a statistically significant protective association for mortality from pandemic influenza with rural living in a county relative to towns and cities. We are not aware of these relationships being tested statistically for previous influenza pandemics.

This pattern is consistent with the limited evidence that remoteness provided some protection in past influenza pandemics (as detailed in the introduction). It is biologically plausible that these relationships reflect a reduced risk of infection among rural dwellers with lower levels of person-to-person contact or poorer spread of infection from towns into rural areas (i.e. protection via remoteness). Nevertheless, there is a need for additional morbidity data on this and other influenza pandemics to clarify these possible roles further.

Another possible factor in contributing to lower rural mortality rates could have been that there were higher case-fatality rates in towns and cities relating to poverty (e.g. where these residents faced additional stress from food shortages after days of illness by breadwinners). We did not have the data to test this and indeed any such effect may have been countered by other factors. For example, there is some historical work that suggests that community support for sick people was organised more quickly in towns and cities than for rural areas.10

Furthermore, the protective effect of cities versus large towns, and for large towns over small towns, might possibly have also represented better community care and medical care available with greater levels of civic infrastructure and resources.10 For some small towns there may have been additional factors that compromised the host resistance of their populations such as work in the mining industry. For example, the mining towns of Nightcaps (Southland) and Denniston (West Coast) had extremely high mortality rates of 46.0 and 18.5 per 1000 population respectively. Public health measures to reduce the spread of influenza may have helped one town,12 but these measures were rarely applied.

It is also plausible that some of the differences found in this analysis may relate to differences in the age-sex structure of the population groupings involved. Such differences would probably favour lower mortality rates in the cities given that the European female mortality rate was lower overall in New Zealand10 and the urban population had a female bias (i.e. 115 females to 100 males in the 1916 census in metropolitan areas, versus 99 to 100 nationally11).

Findings from this analysis are also consistent with what we know about the behaviour of influenza in populations. If the whole population is susceptible to infection before the epidemic, then the proportion of an unstructured population infected during the epidemic depends only on the reproduction number (R0).13

When R0=2.0, then about 80% of the population will be infected by the end of the epidemic (cumulative incidence rate) and the R0 for this 1918 pandemic in one setting in New Zealand was estimated to be in the 1.3 to 3.1 range.14 If, as seems likely, a large majority of people in towns and cities in New Zealand became infected, then mortality rate differences would be dependent on factors that influenced survivability, such as access to nursing care. Protection from actual infection probably only became evident from the remoteness and higher levels of social distance experienced in rural areas.

Limitations—As detailed above, there are limitations of using crude mortality and denominator data that were not adjusted by age or sex. Furthermore, there was no adjustment for socioeconomic position which was found to be important in a Norwegian study of this pandemic15 and an Australian study.16 Another limitation was that we did not consider data for the Māori population, given concerns around the under-reporting of Māori deaths and of under-recording in the 1916 census (partly as a protest against officialdom associated with wartime conscription).10

Indeed, Māori had particularly high mortality rates in this epidemic,10,17 even though they were a predominantly rurally-based population at this time. Nevertheless, many Māori settlements would probably have equated more to small towns as opposed to rurally isolated families on farms.

Even for the European population, the denominator data from the 1916 census may not have been fully accurate due to both population growth over the subsequent 2-year period and the population movements associated with World War I (i.e. troops and healthcare workers overseas or in military training camps).

Implications for further research and policy—Detailed statistical examination of the 1918 pandemic can potentially provide insights into the potential behaviour of a future influenza pandemic. It is likely that the severity of the 1918 pandemic was at least partly a function of the low population immunity to this new strain of virus. This situation contrasts with seasonal influenza which infects a population with relatively high levels of full and partial immunity. Consequently, seasonal influenza may provide a less valid model for the behaviour of future pandemics than can be obtained from analysis of past pandemics.

We are planning future work with individual-level data to clarify some of the limitations with this ecological-level analysis. This work may also include an attempt to analysis Māori data in those areas where there is greater confidence around the quality of numerator and denominator data. Others may also want to consider applying modern epidemiological methods to existing mortality datasets from 1918 in other countries to further explore any impacts of rurality and social distancing.

In the meantime, however, this work offers some limited evidence that rurality might have provided some protection from this influenza pandemic. This may support public health efforts in a future pandemic for remote rural populations to consider isolating themselves for a few weeks if they deem the costs of socioeconomic disruption are worth the benefits of either delaying or preventing the arrival of pandemic influenza in their communities.

Preparations such as improved Internet access for rural areas may facilitate home-schooling and working from home during such a period. Nevertheless, all communities of whatever size probably need to be involved in strengthening planning and civic infrastructure given the risk that a future pandemic may not be adequately stopped at borders (even of island nations) or contained within just parts of a country.

Competing interests: None.

Author information: Kirsten McSweeny, Atalie Colman, Nick Fancourt, Melinda Parnell, Sara Stantiall, Medical Students, University of Otago, Wellington; Michael Baker and Nick Wilson, Senior Lecturers, University of Otago, Wellington; Geoffrey Rice, Associate Professor and Head of Department, Department of History, Canterbury University, Christchurch

Acknowledgements: This study had organisational support and some minor financial support from the Department of Public Health, University of Otago, Wellington (as part of medical student training in public health). Some of the ongoing work on this article by two of the authors (NW & MB) was also part of background work for a Centers for Disease Control and Prevention (USA) grant (1 U01 CI000445-01). We also thank Linda Bryder for her past work in collecting mortality data for Auckland and the two anonymous reviewers for their helpful comments.

Correspondence: Dr Nick Wilson, Department of Public Health, University of Otago, Wellington, PO Box 7343, Wellington; Email: nick.wilson@otago.ac.nz

References:

Ohadike DC. Diffusion and physiological responses to the influenza pandemic of 1918-19 in Nigeria. Soc Sci Med. 1991;32:1393–9.
Phillips H. "Black October": The Impact of the Spanish Influenza Epidemic of 1918 on South Africa. Pretoria: Government Printer (Archives Year Book for South African History), 1990.
Afkhami A. Compromised constitutions: the Iranian experience with the 1918 influenza pandemic. Bull Hist Med. 2003;77:367–92.
Sattenspiel L, Herring DA. Simulating the effect of quarantine on the spread of the 1918-19 flu in central Canada. Bull Math Biol. 2003;65:1–26.
Barry J. The Great Influenza: The epic story of the deadliest plague in history. New York: Viking; 2004.
Crosby A. America's forgotten pandemic: The influenza of 1918. Cambridge: Cambridge University Press, 2003.
Killingray D. The influenza pandemic of 1918-1919 in the British Caribbean. Soc Hist Med. 1994;7:59–87.
Herda P. The 1918 influenza pandemic in Fiji, Tonga and the Samoas. In: Bryder L, Dow D, editors. New Countries and Old Medicine: Proceedings of an International Conference on the History of Medicine and Health. Auckland: Pyramid Press; 1995:46–53.
Brown P, Gajdusek DC, Morris JA. Epidemic A2 influenza in isolated Pacific island populations without pre-epidemic antibody to influenza virus types A and B, and the discovery of other still unexposed populations. Am J Epidemiol. 1966;83:176–88.
Rice G. Black November: The 1918 influenza pandemic in New Zealand. Christchurch: Canterbury University Press, 2005.
Department of Statistics. The 1916 Population Census. Wellington: Department of Statistics; 1916.
Wilson N, Rice G, Thomson G, Baker M. Re-evaluating a local public health control measure used in New Zealand for the pandemic influenza of 1918. N Z Med J. 2005;118(1224). http://www.nzma.org.nz/journal/118-1224/1714
Diekmann O, Heesterbeek J. Mathematical epidemiology of infectious diseases: model building, analysis and interpretation. Chichester: Wiley; 2000.
Sertsou G, Wilson N, Baker M, et al. Key transmission parameters of an institutional outbreak during the 1918 influenza pandemic estimated by mathematical modeling. Theor Biol Med Model. 2006;3:38.
Mamelund SE. A socially neutral disease? Individual social class, household wealth and mortality from Spanish influenza in two socially contrasting parishes in Kristiania 1918-19. Soc Sci Med. 2006;62:923–40.
McCracken K, Curson P. Flu downunder: A demographic and geographic analysis of the 1919 epidemic in Sydney, Australia. In: Phillips H, Killingray D, editors. The Spanish influenza pandemic of 1918-19: New perspectives. London: Routledge; 2001.
Pool DI. The effects of the 1918 pandemic of influenza on the Maori population of New Zealand. Bull Hist Med. 1973;47:273–81.