County-level heat vulnerability of urban and rural residents in Tibet, China

This is the first study assessing and mapping heat vulnerability of urban and rural
populations, in the Tibet Autonomous Region of China. Our analysis was based on 10
factors which may increase human vulnerability to heat according to previous epidemiologic
research, and we applied similar statistical methods to those used by Reid et al.
16]. Overall, our study shows that the relative vulnerabilities of the two study populations
varies across counties, with generally higher (and more adverse) scores for vulnerability
of rural populations in the central Tibet, and for urban residents in the West and
Southeast. Particular attention should be paid to those high-elevation areas in South
Tibet. They not only display relatively higher population vulnerability (due to higher
proportions of relevant demographic risk factors), but also are warming more quickly
than parts of Tibet at low altitude.

There is a growing body of work on spatial heat vulnerability indices. However, most
of the studies carried out in the past were conducted in cities of developed countries
such as the UK 15], the United States 16], 31], Australia 32] and Canada 30]. Little is known about spatial variability of heat vulnerability in areas of developing
countries like China and there have been no relevant reports previously from a high-elevation
setting. The Tibetan plateau may not experience such extreme heat events as other
inland provinces of China, but the temperature in Tibet is increasing more quickly
than other areas of China 1], 4]. Moreover, we have already noted that some sub-populations including the elderly,
men and illiterate persons are at higher risks of dying or being sick during high
temperature days 12], 13]. The vulnerability approach could be a useful tool for a rapid developing Tibet to
quickly identify potential areas for heat-health action plans. This kind of literature-based
assessment does not require us to wait until all determinants of human vulnerability
are described comprehensively in regions which lack reliable long-term health data.

The factor structure of derived heat vulnerabilities indices is generally the same
for the two study populations in Tibet. One exception which should be noted is the
configuration of factor loadings for the illiteracy variable. It loaded with small
living spaces to form an independent factor in rural populations, but with variables
of advanced age and those who lost labor abilities in the index of urban residents.
For those living in rural villages in Tibet, higher proportions of elderly residents
and those living in small dwellings tended to occur in the same counties. This may
be explained by a tendency for the oldest sections of the population to live in old
and crowed dwellings in poor and remote neighborhoods in China. In contrast, most
of the urban residents in small living spaces are likely to be younger people who
have been drawn to the cities or towns and are sharing rental dwellings. In the urban
residents’ index, variables for illiteracy, the elderly and loss of labor ability
loaded on the same factor, suggesting that illiterate residents living in cities and
towns tend to be older and experience mental or physical disabilities.

In this study we created cumulative heat vulnerability indices for urban and rural
populations separately. We suggest that this approach may enable better targeted assignments
of adaptive interventions and resources, because conditions for urban and rural residents
in the same county in Tibet may be quite different. For example, in Ngari Area, we
observed the highest vulnerability score of rural populations, but the lowest for
urban residents (Additional file 1). Ngari Area, with an average altitude of 4500 m, has one of the lowest population
densities in the world due to its extreme altitude and very harsh natural environment.
Rural residents in this area tend to live in remote villages, and are further disadvantaged
by limited access to education, low income, poor housing quality and living conditions.
The area with the lowest vulnerability values of urban residents is also found to
be Ngari Area, possibly explained by the fact that it has the lowest illiteracy rate
among urban people in all Tibet. Educational facilities and conditions have been improved
dramatically over recent decades in Ngari by the Chinese government’s substantial
investment. However, most of the new schools are located in towns, and greater investments
in education in rural and remote areas in Tibet are needed to minimize educational
inequality and vulnerability to poor health.

Chinese governments have designed and implemented a variety of research projects,
policies and plans to better build capacity to cope with meteorological disasters
and long-term climate change in Tibet, while the health implication of climate change
are relatively neglected. An Assessment Report of Climate Change Impacts in Tibet
Autonomous Region has been published by the National and Tibet Climate Centers to
better understand current and future adverse effects of climate variability and change.
This report recognizes the adverse impacts of a rapidly warming climate in Tibet including
snow line rising, glacial recession, changes in river levels, frozen soil layer movements
towards the north, grassland degradation, increasing plant diseases and pests, decreasing
biological diversity and more meteorological disasters. However, there are no human
health data nor is an assessment of health risks included in the report. Another example
is the development and implementation of the “Programme to address climate change
in Tibet Autonomous Region” by the Development and Reform Commission and Meteorological
Administration of Tibet. The Plan did not consider the public health challenges of
a changing climate, although it identified systematic strategies to deal with climate
change effects on agriculture, livestock, forest, water resources, industry and natural
ecosystems. We suggest that better knowledge of the range and magnitude of temperature-sensitive
health outcomes is the first important step to increase awareness at the local government
level of health implications of climate change.

The limited capacity of public health departments in Tibet creates further challenges.
The local Center of Disease Control and Prevention (CDC) is the major institution
working in the field of diseases control and public health managements, and is supposed
to be the lead agency when it comes to developing public health components of heat
preparedness measures in Tibet. There are a number of barriers that make it difficult
for Tibet CDC to manage the health risks related to rising temperatures. Chronic staff
shortage, insufficient expertise and limited technical skills of current staff are
the major constraints. According to Tibetan Statistics in 2008, despite dramatic increases
in income levels, savings, educational attainment and dwelling conditions over past
two decades, the number of public health technical personnel per 1000 persons has
fallen from 3.39 in 1990 to 3.02 in 2008. Options for building capacity in the health
sector include increasing the number of public health personnel, and improving education
and training of the current work force. The capacity to develop adaptation policies
and measures in health sections is restricted also by limited information on health
impacts of climate variability due to an absence of reliable health data in Tibet.
At present, there are only five counties that have been selected to carry out long-term
surveillance of death and prevalence of chronic diseases in Tibet. Long-term hospital-based
data are also limited due to very uneven spatial distributions of hospitals and lack
of well-established electronic medical record systems. More surveillance locations
are required to collect more valid and comprehensive health data for future research
and intervention planning. Also, sharing access to existing relevant data sets should
be improved to meet needs of sustainable intersectoral collaboration and actions.

The heat vulnerability indices we developed have a number of limitations that users
need to keep in mind when using the results for decision-making. Our analysis was
limited by the lack of data for other important determinants of heat-health vulnerability.
We did not include exposure variables which are frequently used in other assessments
of this kind, such as land cover and surface temperature. Extreme urban heat is still
not very common in Tibet except in a few locations such as Chengguan District of Lhasa
and uneven spatial resolution of surface temperature datasets permits only intra-urban
heat vulnerability assessments 33]. Furthermore, surface temperature and green spaces are not the only factors which
determine air temperature and indoor temperatures. Unlike other studies, we did not
include a measure of home air-conditioning (AC) although it is known access to air-conditioning
may protect against heat-related illnesses and deaths 22], 34], 35]. The reasons are that, first, these data are not available at county-level and, second,
air-conditioning at home is uncommon in Tibet (in 2008 it was estimated that only
3 % of households had AC) 36]. Similarly, we did not include a measure of use of cooling fans, as this information
is not available. Besides, although many studies have indicated that individuals in
occupations that entail exposure to high temperatures outdoors are more likely to
develop heat-induced diseases, we were unable to include a measure of occupation in
this study. Housing factors (e.g. dwelling age, the number of floors, construction
types) may also increase the likelihood of exposure to severe heat and influence heat-health
risks. However, these data were not included in our analysis, as there are too many
missing values for most of the remote counties. Another important limitation related
to data unavailability in this study is lack of sensitivity analysis using alternative
variables.

Given the lack of information on other domains of vulnerability, our indices were
mainly based on demographic variables. We observed different demographic patterns
associated with heat vulnerability between urban and rural populations. For instance,
illiteracy is strongly associated with household size in rural settings but matches
closely with age and disability in urban settings. These findings indicate that the
population variables which can best represent heat vulnerability are different in
urban and rural areas. Nevertheless, the 10 variables in this study were based on
our epidemiological studies in Tibet as well as validated vulnerability indices in
previous literature. Despite their imperfections we suggest the results in this study
provide important and timely information to policy makers and can be used along with
local meteorological records in developing heat adaptation plans.

Following Reid et al. 16], we created heat vulnerability indices without weighting factors. The use of a composite
index without weights is not ideal, and future studies may explore differential weighting
of the variables based on closer understanding of heat-health vulnerability pattern
in Tibet. The composite indices in this analysis have not been validated with local
health data since this information is not currently available. Validation is certainly
an important component in vulnerability assessment studies 37], but the next step, in the absence of comprehensive local data sets, could be to
consult local Tibetan stakeholders and relevant professionals, obtain feedback on
the heat vulnerability maps, and modify as appropriate variable selection and other
aspects of the analysis.