[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Quality of data (cancer rates)



One of the potential problems that may be present in trying to use data about 
cancer incidence and mortality is the quality of the data.  Those who use the 
data in the manner of Gould and Sternglass are just going on faith that the 
data are suitable for their purposes.  I found the following information on 
the Internet about an apparent huge difference in cancer incidence in Wales 
as compared to England.  It was resolved as being due to errors in reporting 
the cases. 

 Note that near the end there is a caution about data quality.  It is hard to 
believe that there will not be differences in reporting techniques, 
especially over long periods of time.  Of course there are many other 
potential problems that can affect reporting of this type, such as migration, 
changes in the age of the population etc.  The figures (graphs) mentioned are 
not included here.  It appears that data such as these are produced for local 
reasons that have nothing do do with trying to find the cause of cancer and 
that the quality of the data is not high enough for those purposes.  I 
suspect that is also the case with the data used by Gould and Sternglass.  

___________________________________________________________

BONE CANCERS IN WALES: AN INVESTIGATION OF
            CASES REGISTERED BY THE WALES CANCER
                          REGISTRY 

             Dr M C Cotter, Mr R T Kilpatrick, Mr G P Davies 

                                                   Summary 

        On the basis of information produced by the Wales Cancer Registry
        (WCR) concern has been expressed that the incidence of bone cancer
        in Wales has been growing. According to WCR data, in 1989 the
        crude rate of registration of the disease was 5 times that of 
England. 

        The epidemiology of cancer would suggest that it is unlikely that such
        rapid and pronounced changes are reflections of changes in the
        incidence of the disease and may rather highlight a problem with the
        data themselves. 

        In order to establish the true level of bone cancer among Welsh
        residents an exercise to validate the data was undertaken. The 
exercise
        was initiated in 1993 and included a detailed assessment of each case
        of bone cancer held by the WCR which was completed in 1995. 

        Registering cancers is a dynamic process in which the original data 
may
        be supplemented and revised by additional information for some time
        after the cancer was first registered. 

        Figure 1 shows the differences between data for Wales published in
        Annual Reference Volumes (ARV) by the National Cancer
        Registration Scheme run by the Office of Population Censuses and
        Surveys (OPCS) [Office for National Statistics (ONS) from April
        1996] and data published by the WCR, both before and after the
        validation exercise. 

        There were differences between the National Cancer Registrations
        Scheme and the WCR publications which became more pronounced in
        the late 1980s. However, both databases contained wrongly registered
        bone cancer cases. 

        The numbers of cases in the revised data - where the erroneous cases
        have been removed - are about a third of the number previously
        registered overall and less than a fifth of the number originally 
published
        for 1989. 

                          end page 1 

        Figure 1: Comparison of the number of cancer registrations 


        Figure 2 shows trends in the crude rate of registration of bone cancer
        per 100,000 population for residents of Wales, comparing data
        published by the National Cancer Registration Scheme with the revised
        data. 

        The bone cancer registration rate published by OPCS was generally
        higher than the WCR figure and in the late 1980s rose rapidly to 
almost
        5 cases per 100,000 population. This compared with a fairly constant
        rate of less than 1 per 100,000 for England. After revision, the 
annual
        registration rate for Wales was also usually below 1 case per 100,000
        population throughout the period 1979 to 1989. 

        Figure 2: Comparison of cancer registration rates 

        The study has confirmed the concerns of the WCR that the number of
        cases of bone cancer on the register was overstated. The revised data
        suggest that the incidence of the disease is, per head of population,
        about the same as in England. 

                          end page 2 

        The most significant factors that contributed to the over 
registration of
        bone cancer were (i) the miscoding of diagnoses from clinical notes, 
in
        particular those where soft tissue tumours of the head and neck were
        mistakenly classified as bone cancers; and (ii) secondary cancers 
being
        recorded as primary cancers. There was no significant problem with
        duplicate registrations. In a small proportion of registered cases 
there
        was no evidence of any reported cancer in the medical records. 

        These findings do not imply that there was any mis-diagnosis of cases
        or that inappropriate treatment was given. 

        The exercise emphasises the importance of those involved in cancer
        registration and the routine production of data for local 
administrative
        purposes producing accurate and timely information. Early validation 
of
        records is essential in order to resolve inconsistencies in the data 
at an
        early stage and while patient records are readily available. 

        There are lessons also for those who use the data. Cancer registration
        is a complex process and those who use the information must be aware
        of this. The data can only be best estimates of incidence at a 
particular
        point in time and those analysing the information should take account 
of
        data quality. 

        Acknowledgements 

        The authors would like to thank all those hospitals and staff who
        participated in the validation exercise. Without this assistance, 
often
        provided in the face of other local clinical and administrative
        demands, it would not have been possible to examine these cases of
        bone cancer and establish the extent of the disease in Wales. Also,
        the authors would like to thank Dr Neil Burnett, consultant
        oncologist at Velindre NHS Trust, for his contribution and advice
        on the aetiology of bone cancer, Dr Gerald Draper and Dr Charles
        Stiller of the Childhood Cancer Research Group, for their help in
        the preparation of this report and all the staff of the Wales Cancer
        Registry for their hard work throughout the exercise. 
____________________________________________________
R. Holloway
holloway3@aol.com
                          
************************************************************************
The RADSAFE Frequently Asked Questions list, archives and subscription
information can be accessed at http://www.ehs.uiuc.edu/~rad/radsafe.html