5. METODOLOGÍA
5.3. Objetivo específico 3: Establecer la influencia de los contaminantes en los receptores
5.3.7. Análisis multiespectral
A number of data sources were used in the component of objectively measured health impacts within the three selected study areas, as well as evidence of CSG health impact. These data sources were the resource development data, population data, and hospital admissions data. The basis for matching Statistical Local Areas (SLAs) and the given populations within the environmental settings has been described in Chapter 3. Further details and analyses specific to this section will be presented in Chapters 6 and 7.
2.2.1 Resource development data
CSG well data, including the corresponding number of wells per SLA within the CSG area, were obtained from the Queensland Government (Queensland Government, 2014b) to calculate the numbers of wells as they accumulated for each month of development over the study period. The number of wells in each SLA of the CSG study area was combined to obtain the increasing monthly numbers for the whole of the CSG area. This was done to align the CSG well data with the hospital admissions data. Figure 4.1 shows the cumulative numbers of wells over the study period.
Figure 4.1. Cumulative coal seam gas (CSG) well counts by month and year for the study time period, 1995-2011.
The production values (discussed in Chapter 3) can serve as environmental proxies, indicating the potential environmental change for each setting over time. While the production values for the coal mining (CM) and rural/agricultural (RA) areas could serve as environmental proxies for the alternative
settings, CSG well numbers were of primary interest for this study and were considered to be the environmental proxy for the CSG area. Well development and production are the two main stages of UNGD. The well development stage includes road construction, pad preparation, well drilling, hydraulic fracturing, and well completion; and the production stage includes gas collection, processing, and distribution (McKenzie et al., 2012; Moore, Zielinska, & Jackson, 2013). The well development stage would provide the greatest indication of environmental change in an UNGD area due to these activities.
For this thesis, well counts were considered an indicator for CSG development activity; therefore, this count served as a proxy for the development activities (e.g., well pad preparation, well drilling, well completion, and related construction activities).
While the focus of this work was on CSG impact, saleable coal production figures were obtained from several Queensland Government departments to understand coal production trends over the study period and confirm this setting (Department of Employment Economic Development and Innovation, 2010; Department of Mines and Energy, 2007; Queensland Government, 2013c). Figure 4.2 shows the coal production values for Queensland to provide context for this alternative environmental setting.
Figure 4.2. Queensland saleable coal production, 1996-2013.
The Queensland saleable coal production values show a linear increase in saleable coal production since 1996; however, the Queensland coal industry in the Bowen Basin has grown steadily since the early 1980s (Rolfe, Miles, Lockie, & Ivanova, 2007). Production figures for the CM area were considered to be a proxy for coal mining activity in the CM area, providing an indication of the continuing environmental impact in one of the selected alternative settings.
In terms of mines that previously operated or are currently operating in the CM area, the
Blackwater open-cut mine was the first to start operating in 1967, with a number of mines (e.g., Curragh (1983), Jellinbah (1989), Kestrel (1992), Yarrabee (1994)) following thereafter (Blackwater International Coal Centre, n.d.; Rio Tinto, n.d.; Yancoal Australia, 2014). The industry has had a steady increase in exporting coal since 1997 (Department of Natural Resources and Mines, 2014b). Additional mines have also begun operations in more recent years, with the Lake Lindsay operation starting coal processing in
2008 and the Kestrel Mine Extension opening in 2013 (AngloAmerican, 2013; Rio Tinto, n.d.). For the year 2012-13, a total of 43 open-cut mines and 13 underground mines contributed to Queensland coal production (Department of Natural Resources and Mines, 2014b). The CM study area has 22 coal mines within or along the border of the delineated geographic area. However, CM-related environmental proxy data were not examined alongside hospital admissions data as with the CSG environmental proxy because this programme of work focused on CSG.
2.2.2 Hospital admissions data
Routinely collected data, such as data from hospital databases, are useful for a variety of applications, including injury surveillance, public health planning, health policy applications, resource distribution, informing legislation, clinical research, and monitoring the effects of environmental
conditions on human health (Burns et al., 2012; Schoenman, Sutton, Kintala, Love, & Maw, 2005). While hospital data can be used for several purposes, there are strengths and weaknesses associated with using these data as a source of information.
Databases offer a cost-effective and time saving approach to data collection and may be more accurate and reliable than other data sources, such as surveys, where recall bias is a limitation
(Schoenman et al., 2005). Schoenman et al. (2005) noted that routinely collected hospital admissions data may be more reliable than passive physician reporting for certain conditions or diseases, which can result in underreporting. In addition, while survey sampling needs representativeness of the population,
hospitalisation databases are typically for the entire population of an area, making it representative of the population (Schoenman et al., 2005).
Hospital databases also have weaknesses. These include data quality problems (inaccuracies in diagnosis and procedure codes and hospital-specific errors), excluded populations (those who have outpatient procedures and patients admitted outside of their state), and missing data (race and ethnicity, which makes it difficult to adequately examine disparities) (Schoenman et al., 2005). Due to
misclassification of diseases in diagnosis coding, there is also the potential for disease rates to be over- or underestimated (Farzandipour, Sheikhtaheri, & Sadoughi, 2010). Also, as noted in Chapter 2, variation in hospitalisation rates in certain areas can be reflective of lack of access to primary care as opposed to being solely reflective of morbidity in the population (New South Wales Health, 2010).
For this thesis, admitted patient data (defined as admission to hospital for a period of 24 hours or longer) were obtained from the Queensland Hospital Admitted Patient Data Collection (QHAPDC).
QHAPDC data are from public and private hospitals (defined as declared public hospitals and licensed private hospitals), as well as day surgery units (Queensland Health, 2015). Data were obtained for each calendar year, from 1 January 1995 through to 31 December 2011, for admission to any hospital in Queensland for any resident of one of the study areas described in Chapter 3. Thus, data were obtained only for residents of the study areas and were not obtained for people who were hospitalised in one of the study areas, but who were not residents in one of those areas (e.g., fly-in, fly-out workers, tourists, etc.).
This was to ensure accurate calculation of hospitalisation rates for residents of the area, using the population data described in Chapter 3.
The lowest level of data that could be requested was at the grouped SLA level, so data were categorised according to the groups of SLAs outlined in the previous chapter (Table 3.1). Hence, data were grouped by Queensland Health into the CSG, CM, or RA study areas as these were the smallest aggregations of geographic areas allowed by Queensland Health. Data were not identifiable by SLA, only by the three study areas. The variables requested from the QHAPDC are shown in Appendix D.
Hospitalisation data are episode-based, not person-based; therefore, each hospitalisation episode may not represent unique individuals.
The main variable of interest was the primary diagnosis code for each admission, provided in the form of an International Classification of Diseases (ICD) code. Two versions of ICD coding were used for primary diagnosis codes during the study period: ICD-9-Australian Modification (ICD-9-AM), which was used for cases from 1995 to July 1999, and ICD-10-Australian Modification (ICD-10-AM), which was used from July 1999 through the remainder of the study period (Roberts, Innes, & Walker, 1998).
There are 22 main ICD chapter headings (discussed in Chapter 6). Each main ICD chapter has a number of sub-chapters and codes that fall below the main chapter heading. Each record was classified according to the main ICD chapter heading, followed by the highest level sub-chapter within that (e.g.,
‘Chapter 10’ - ‘Diseases of the respiratory system’, followed by ‘J00-J06’ - ‘Acute upper respiratory infections’).
3 Indicators
As discussed in Chapter 1, a suite of indicators was selected for use in this study. While it would have been preferable to develop a comprehensive set of environmental health indicators, the suite of indicators for this study was limited by the available data. Therefore, demographic indicators were used to assess the make-up of the population and socioeconomic indicators were used to act as a proxy to
consider various social determinants of health. An indicator to measure sense of place was also included.
Health indicators were included, along with a limited number of environmental indicators. The broad indicator categories and corresponding indicators are shown in Table 4.1.
Table 4.1. Indicator categories and the corresponding indicators used in this study.
Broad indicator category Specific indicator
Demographic Age
Gender
Relationship status
Employment status
Number of people in household
Socioeconomic Education
Housing status
Personal income
Household income
Employed by coal seam gas (CSG) company
Employed by coal mining company Sense of place Length of time in community
Environmental Number of CSG wells in the CSG setting over time
CSG production (Queensland) over time
Number of coal mines in the coal mining setting
Coal production (Queensland) over time
Gross value agricultural production (crops, livestock, total) (Queensland) over time
Health
‘Subjective Health Outcomes’
Mean scores of general health within each study area and compared to the Australian general population
Prevalence of numerous symptoms groupings (e.g., digestive, skin)
Proportion of population in the ‘above normal’ range for levels of depression, anxiety, and stress
Mean scores for a measure of depression, anxiety, and stress within each study area and compared to the Australian general population
Mean scores of environmental distress within each study area
‘Objective Health Outcomes’
Crude all-cause hospitalisation rates
Crude cause-specific hospitalisation rates for each International Classification of Diseases (ICD) chapter
Age-standardised, all-cause hospitalisation rates
Age-standardised, cause-specific rates for each ICD chapter
Age-specific, all-cause hospitalisation rates
Age-specific, cause-specific rates for each ICD chapter