2.1 SELECCIÓN DE LA METODOLOGÍA DE DESARROLLO DE
2.1.1 DESCRIPCIÓN DE METODOLOGÍAS DE DESARROLLO WEB
In Scotland there have been large increases in the incidence and mortality rates of tongue, mouth and pharyngeal cancer in males (but not females) since 1970 (Swerdlow et a l 1998). Swerdlow et a l (1998) noted that this trend is particularly worrying since it comes against a background o f reduced smoking in this population as recorded by smoking data from 1975 to 1990 and decreasing mortality from lung cancer. They note that the proportions of males aged 16-24 who were smokers diminished from 41% to 28% and that there were decreases in other young adult ages too. They also note that there were large increases in the numbers o f deaths attributable to alcohol during the same period, although it is likely that the increase is in part accounted for by changes in certification practice, for example with regard to the acceptability o f mentioning alcohol on a death certificate.
Hindle et a l (2000a) investigated regional differences in Standardised Incidence Ratios (SlRs) in intra-oral cancer (lCD-9 141, 143-146) between 15 Regional Health Authorities (RHAs) in England and Wales for 1979 to 1983 in relation to surrogate markers o f smoking (incidence o f lung cancer) and alcohol consumption (mortality from liver cirrhosis). They noted that a north-south gradient existed for all three diseases with the southern RHAs having lower rates. For males they noted that there was a stronger correlation between intra-oral cancer incidence and liver cirrhosis mortality (rg=0.75) than between intra-oral cancer incidence and lung cancer incidence (rs=0.63). For females the correlations were positive but non-significant (Hindle et a l 2000a). In other research by the same group, temporal trends in male and female mortality from these diseases were considered for the whole o f England and Wales from 1911 to 1990 for two age groups: 35-64 and 65+ years. The
strongest associations were found in males aged 35-64. For this group, there was a very strong negative correlation between intra-oral cancer mortality and the surrogate measure o f smoking (rs=-0.98) such that a reduction in the mortality of one disease was associated with an increase in the mortality o f the other. While for intra oral cancer mortality and liver cirrhosis mortality there was a strong positive correlation (rg=0.71). This paralleled the findings of Swerdlow et a l (1998) in Scotland and led the authors to suggest that alcohol may be responsible for more intra-oral cancer than smoking (Hindle et a l 2000a).
Swerdlow and dos Santos Silva (1993) produced an atlas o f cancer incidence in England and Wales for the period 1968-1985. They calculated odds ratios for the development o f 37 different types o f cancers based on counties as the geographical areas o f analysis. They also considered the geographical distribution o f 29 different exposures that may cause or prevent cancer. Risks for oral and pharyngeal cancers (combined) in males were greatest in the north-east o f England and in mid- and west Wales. Tongue, oral and pharyngeal cancers (both separately and combined) showed a gradient o f increasing risk with increasing urbanisation o f an area. For females, as with other studies, there was less evidence of an overall pattern in risk o f tongue, mouth and pharyngeal cancer, although several o f the areas with highest risk were found in Wales. These results were attributed to high levels o f smoking and drinking in northern regions o f England and fairly high levels in Wales, along with poor nutrition in these areas. They found no evidence o f increased odds o f these cancers in areas with above average proportions o f the population originating fi"om the Indian sub-continent, but noted that in no county do they constitute more than a very small proportion of the overall population (Swerdlow and dos Santos Silva, 1993).
4.1 .2 Geographical variations in lip cancer
Hindle et a l (2000b) found substantial geographic variations in the incidence (based on RHAs) o f lip cancer in England and Wales for the period 1979-1983. East Anglia had the highest Standardised Registration Ratio (SRR) o f 275 for males (overall mean for England and Wales set to 100). The highest level o f registrations for females was also found in East Anglia (SRR=232). The incidence was raised in both genders in a band stretching from East Anglia to the South West and also in Trent and Wales. The authors concluded that the findings supported the belief that employment in rural occupations was a major risk factor for lip cancer (Hindle et a l
2000b).
Swerdlow and dos Santos Silva (1993) found that risks o f lip cancer in males were twice as great in rural counties as in urban areas. They noted that the geographical distribution o f male lip cancers were “as" striking as fo r any other site exam inedl
The odds in every county in East Anglia were at least two and a half times those of the country overall. High risks were also found in the westernmost parts o f Wales. The distribution could not be solely attributed to the greater proportion o f rural residents in these areas as even within rural areas the odds still showed a twofold increase in East Anglia, south Lincolnshire, and parts o f Wales. Lip cancer is less common in females and although the urban-rural gradient in risk was reduced, it remained in the same direction. The authors also noted that the geographical distribution o f risks does not follow the pattern that would be expected based on levels of exposure to solar radiation (i.e. it does not follow a north-south gradient) and that it does not appear to be an artefact o f the cancer registration process. They
conclude that the reasons for the high risks in East Anglia and western Wales remain unclear (Swerdlow and dos Santos Silva, 1993).
4.1 .3 Socio-economic status/relative deprivation
Globally there is consistent evidence for males at least (and frequently for females too) that people in lower socio-economic/deprived groups experience higher oral cancer incidence rates than those who are more affluent (Faggiano et a l 1997).
Considering specifically population-based studies within England, O ’Hanlon et a l
(1997) examined tongue (ICD-9 141) and intra-oral (ICD-9 143-145) cancer registration data for the Northern Region for two time periods (1976-1983 and 1984- 1991) in order to compare local incidence with national figures for England and Wales. They also used small area statistics to investigate the relationship between oral cancer incidence and area deprivation within 678 local authority (electoral) wards in the Northern Region. This was undertaken using ^-tests to compare the mean Standardised Registration Ratios (SRRs) between only the most affluent quartile and most deprived quartile of local electoral wards. Quartiles o f relative deprivation were based on the ranking o f wards according to their Townsend deprivation score (Townsend et a l 1988). They found higher levels o f SRRs in the Northern Region compared to England and Wales as a whole which they attributed to differences in lifestyles and pointed out that the Northern Region has some o f the worst levels o f material deprivation and long-term unemployment in England and Wales. For the local small area analyses, they found statistically significant differences in the mean SRRs between the highest and lowest quartiles for both tongue and intra-oral cancer for males in both time periods. For females they found a significant difference only for mouth cancer and only in the later time period. In each
case the direction o f the mean difference was always the same i.e. favouring the people who lived in the more affluent quartile (O’Hanlon et a l 1997).
Thome et a l (1997) used data from the South-West Cancer Registry to investigate head and neck cancer incidence in relation to socio-economic status between 1985- 1991. They defined head and neck cancer as including ICD-9 codes 140-149 (‘oral’), 160 (nasal cavities, middle ear, and accessory sinuses), 161 (larynx) and 193 (thyroid gland). They measured socio-economic status using quintiles o f the Carstairs Index recorded at the Enumeration District (ED) level. Using this system they reported that there were a total o f 2,641 head and neck cancers over the 7-year period in the South-West region. Only four registrations included staging information. Tumours were classified as being either ‘smoking-related’ (ICD-9 141, 143-146, 148-149, 161) or not ‘smoking-related’ (ICD-9 140, 142, 147, 160, 193). For males, with smoking-related cancer, they found a socio-economic gradient in age standardised incidence, favouring those who lived in the most affluent areas. They found no association between socio-economic status and non-smoking related cancers in males. They reported that the trends were similar but less clear in females (Thome et a l 1997). However oral cavity, oropharyngeal and hypopharyngeal cancers in combination only constituted 34% o f the total sample. The largest single group was the laryngeal cancers (29%) and much o f the observed relationships may be related to the predominance o f cancers at this site in the study.
Edwards and Jones (1999) used data from four cancer registries (Thames, West Midlands, West o f Scotland and Yorkshire) to investigate the relationship between relative deprivation and the incidence o f Upper Aerodigestive Tract (UAT) cancers (ICD-10: C00-C14, C30-C32). They correlated the world age standardised incidence
rate for each district health authority with the mean Carstairs score (Carstairs and Morris, 1989) for each health authority. They reported a stronger positive association between relative deprivation and UAT incidence in men (r=0.7S, p-value not reported) than women (r=0.60, p-value not reported) (Edwards and Jones, 1999).
4 .1 .4 Summary and basis for the current research
Geographical variations in lip and oral cancer incidence and their correlation with surrogate measures for smoking and excessive alcohol consumption have been studied previously for the period 1979-1983 by Hindle et a l (2000ab). The work of Hindle et a l was restricted to ICD-9 140 (for lip) and to ICD-9 141, 143-146 combined (for intra-oral cancer). The current investigation updates this work for a more contemporary time period and additionally considers the geographical inequalities for other ‘oral cancer’ sites in relation to the same surrogate measures of smoking and excessive alcohol consumption.
There is little robust information in the literature that considers oral cancer incidence and stage at presentation in England in relation to relative deprivation at the small area level. O ’Hanlon et a l (1997) excluded all but the most extreme quartiles from their analyses. Edwards and Jones (1999) undertook their investigation at the level of the district health authority. Health authorities are large geographical units and contain heterogeneous populations. The investigators use of a mean Carstairs score for each health authority therefore masks differences that exist between small areas within in each health Authority. They used regression coefficients to indicate the extent to which there was a linear association between deprivation and cancer incidence, whereas statistical regression could also be used to model the nature of the
relationship between these factors rather than just the degree o f association. None of the studies considered stage at presentation in relation to relative deprivation.
The current investigation explored the relationship between relative deprivation and oral cancer incidence at the small area levels o f both 1991 Enumeration Districts (EDs) and 1998 Wards, thereby providing fine resolution for the analyses. Additionally comparisons were made between the results that are obtained with a commonly used area deprivation index, Townsend and a new index the IMD 2000 (Department o f the Environment Transport and the Regions, 2000). The ‘old style’ deprivation indices such as Townsend and Carstairs include domains that are based on the Registrar General’s classification o f occupations and upon data that are only collected at the decennial census. The ability of the Registrar General’s classification to assign a worthwhile ‘social class’ to individuals is reducing over time. Indices based on census data that are collected only once every ten years soon become out of date. The problems are further compounded by lack of consistency between census based geographies and administrative geographies. The Index o f Multiple Deprivation 2000 (IMD200) is described in detail in Chapter 2. Briefly, it is gaining favour among researchers and governmental agencies because: 1) it avoids the Registrar General’s classification o f occupations; 2) it is based on local administrative areas rather than census areas; and 3) the information that is needed to construct and update the index is part o f the data set that Government requires local authorities to collect routinely.
The potential effect o f deprivation on the stage at which patients present with oral cancer was examined. Relative deprivation is an important factor since it may affect risk behaviour (especially propensity to smoke, chew tobacco, or drink alcohol
excessively) and possibly the extent of delay before presentation and referral to secondary care.
4.2 Aims
The aims o f this part o f the research were:
1. To investigate regional inequalities in oral cancer incidence and their possible relation to the major risk factors of smoking and alcohol consumption.
2. To investigate inequalities in both incidence and stage at presentation in relation to relative deprivation o f area of residence.