Cancer in the Mid-West 1994- 2002
Total Page：16
File Type：pdf, Size：1020Kb
Cancer in the mid-west: 1994-2002/ T. Greally, H. Comber Item Type Report Authors Health Service Executive (HSE) Rights HSE Download date 23/09/2021 17:46:42 Link to Item http://hdl.handle.net/10147/44537 Find this and similar works at - http://www.lenus.ie/hse Cancer in the Mid-West 1994- 2002 Introduction Cancer is not a single disease; it is rather a group of diseases of differing impact and cause, which one in three individuals will develop in the course of their lifetime. Over 18,000 new cases of malignant cancer are recorded annually in Ireland with approximately 1,500 in the three counties of Clare, Limerick and North Tipperary. About 3 in 10 registered cancers are non-melanoma skin cancers which are excluded from this report. Background Information has been collected by the National Cancer Registry of Ireland (NCRI) on new cancer registrations in Ireland since 1994; national and regional data are available through NCRI annual reports and on their website (www.ncri.ie). Information on the incidence of cancer and other diseases is not usually available for areas smaller than county level in Ireland, as addresses in Ireland (with the exception of Northern Ireland) do not have local area coding (e.g. postcodes or zip codes.). However, addresses can be assigned, using various data sources, to a small area of residence (electoral district, ‘ED’), and the National Cancer Registry is currently in the process of assigning all cancer cases registered since 1994 to an ED of residence; this has already been done for cases registered up to the end of 2002 for Clare, Limerick and north Tipperary. Because of concerns arising from animal health in Askeaton and its environs during the 1990s a number of studies in recent years have used NCRI data to examine cancer rates in this area, including: • A study of human health indicators which compared health indices in the Askeaton area including cancer incidence for 1994-95 . • A PhD thesis by Dr. Avril Hegarty of the University of Limerick using 1994-95 NCRI data with new methods of mathematical and statistical modelling. • A report on cancer incidence levels across the whole Mid-West region for 1994-99 • Local cancer incidence data for 2000 and 2001 presented by the National Cancer Registry Director, Dr. Harry Comber, to local residents in Askeaton and Rathkeale in 2003 None of these studies showed a significant excess of cancer incidence in the Askeaton area, although significantly elevated rates were noted in parts of Limerick City. This report extends the study of cancer registration in the Mid-West by a further year to cover the 1994-2002 period and describes cancer incidence levels across the whole Mid- West region. Data collection methods The National Cancer Registry of Ireland ( NCRI) is the only source of national cancer incidence in Ireland. Nurses trained in cancer registration methods are employed as tumour registration officers (TROs) by the Cancer Registry and are based in 18 hospitals around the country. Each is responsible for gathering cancer data from a group of 1 hospitals, and from other sources within a designated geographical area. Within their catchment areas, they liaise with hospital pathology and haematology laboratories, special clinics, hospital administrators and medical records staff, Hospital Inpatient Enquiry (HIPE) and casemix staff, public health nurses, hospices, nursing homes and any other persons they consider to be a useful source of registration data. Death certificate information is also used, but represents less than 5% of primary notifications. The Registry carries out internal quality control checks on the data throughout collection and processing and collates and updates information under three headings: patient, tumour and treatment information. The Registry has carried out a number of checks on the completeness of registration of cancer, and all have confirmed that at least 96-97% of all new cancers are picked up by the NCRI; this compares well with international cancer registration standards. However, to ensure that few cancers are missed, an intensive search process, as described above, has to be undertaken, using many sources of information. This process takes some time, so there is an interval of three to four years between the date of diagnosis of a cancer incidence and the time that the Registry is satisfied that all the data are complete. Detailed reports are published annually with geographical information at county or HSE area level. County level data are also available in interactive tables on the Registry’s website. The NCRI was requested to prepare a report on cancer incidence at the level of electoral division (ED)* in Clare, Limerick and north Tipperary (the former Mid-Western Health Board area) between 1994 and 2002. Each cancer case was assigned to an ED using software developed by the Registry to match addresses to two databases—the GeoDirectory, produced by An Post, and the Register of Electors provided by each local authority. Approximately 96% of cases were assigned to a unique ED. The remainder had incomplete addresses, but could be assigned to one or two or three neighbouring EDs. The small number of cases (4%) which could not be allocated to a specific ED were distributed among all EDs according to population, so the number of cases recorded as “observed” in each ED is about 4% higher than the number actually found. This had no impact on the findings. Interpreting the results The results for each ED have been expressed as standardised incidence ratios (SIRs). This figure gives the ratio between the number of cancer cases that were found in an area and the number that would have been expected based on a reference incidence rate. If 5 cases were found and 4 were expected, the SIR was 5/4 or 1.2. An SIR less than 1 is an indication that the actual number of cancers registered was less than expected; an SIR greater than 1 indicates that the number was greater than expected. The number of cancers expected was calculated from two pieces of information—the overall average cancer rate for the entire area (Ireland or the former MWHB area) and the population of the ED. This calculation was done separately for each age group, so the expected value allows for the fact that the average population age may be older for * An electoral division (ED) is an area with a population of, on average, about 1000 persons. It is the smallest area for which full population information is provided by the census and also usually the smallest area to which most addresses can be confidently assigned. Each ED contains, in rural areas, a number of townlands and in urban areas, a number of streets. 2 some EDs than for others. As cancer is mainly a disease of the over-65s, it is essential to make this adjustment for the age composition of the population. The number of cancers diagnosed in most EDs is small and varies considerably from year to year (Table 1). This makes comparison between areas difficult. For Askeaton East, for instance, the number of cases has varied from none to 7 per year. Although the number of cases in Askeaton East, Askeaton West and Croagh was similar for the period 1994- 2000, in the three years 1994-1996 there were more cases in Askeaton West and Croagh, and in the three years 1997-1999 there were more cases in Askeaton East. This is known as “random variation” and is due to the fact that the number of cases arising in an area, or during a specific time, can vary unpredictably. However, this variation can be seen to lie within certain limits. For Askeaton East, Askeaton West and Croagh, in most years the number of cases was between 0 and 4, so we could predict with some degree of confidence that in most future years, if the population size and structure of these areas were to remain unchanged, then the number of cases will continue to be between 0 and 4, and that, for instance, although we see that there were 7 cases in one year in Askeaton East, 20 cases in a year would be very unlikely. This confidence is usually expressed as a “confidence limit” and in most studies of variation we use the “95% confidence limits”. This means that we could predict that in 95 of the next 100 years (or in 19 of the next 20) the number of cases will lie within these limits. Similarly, when we calculate the ratio between the actual number of cases and the number expected (the SIR) this will also have to be given confidence limits based on the confidence limits of the number of observed cases. Only if the confidence limits are both greater than 1 can we conclude confidently that the incidence rate for the area is really larger than expected. In this case, we describe the difference from average as “statistically significant”; otherwise we consider that the difference from average is almost certainly due to random variation. Table 1. Total number of invasive cancers in a number of rural Limerick EDs, 1994- 2002 year of Askeaton Askeaton Aughinish Bally- Ballyna- Castle- Craggs Croagh diagnosis East West allinan carriga town 1994 0 5 0 0 1 0 1 1 1995 0 0 0 2 2 2 1 3 1996 1 2 1 4 2 1 0 3 1997 7 2 0 2 1 2 0 3 1998 4 2 0 2 2 0 0 2 1999 2 4 2 1 1 1 1 3 1994-1999 14 15 3 11 9 6 3 15 3 Findings Proportion of invasive cancers by site in Mid-West residents 1994-2002 colorectal lung oesophagus stomach bladder melanoma pancreas lymphoma leukaemia female breast ovary prostate cervix other Figure 1.