Estimation of Cancer Pattern by Means of Hospital-based Cancer Registries and Pathology Laboratory Data
Arvo building, Lecture room F025, address: Arvo Ylpön katu 34
Doctoral defence of M.Sc. Murali Dhar
The field of science of the dissertation is Epidemiology.
The opponent is professor Jaakko Kaprio (University of Helsinki). Professor Emeritus Matti Hakama acts as the custos.
The language of the dissertation defence is English.
Utilization of hospital based cancer registry in assessing cancer pattern
Cancer registration is one of the necessary ingredients of any cancer control programme. The requirement of cancer registration would be that it has to be population-based with diagnostic, clinical and treatment details of all the cases. Even in developed countries, there is the possibility of clinical data being incomplete for a
considerable proportion of the cancer cases. In the setup of developing countries, the registries suffer from the lacunae of not being able to cover the whole population of the country and non-availability of reliable clinical and treatment details for the majority of cases. Therefore, the principle of two different registries, namely; population-based cancer registries (PBCRs) and hospital-based cancer registries (HBCRs), evolved and are in place. Another concept that has come up recently is that of the cancer atlas (ATLAS) which collected data from all the pathology centres and tabulated it according to their place of residence. Therefore, in the scenario of developing countries, there is a need to explore the potential population-based uses of HBCR and ATLAS data and to test their validity. The overall aim of the study was to explore the feasibility of using HBCR and ATLAS data in the evaluation of cancer patterns in India.
With respect to the place of residence, there are two sets of patients that account for qualitative and quantitative differences between HBCR and PBCR. They are INPTSOUT (Set of patients who are actually from INside the HBCR area, but do not get registered in HBCR) and OUTPTSIN (the set of patients who are actually from OUTside the HBCR area but get registered in the HBCR). Thus, PBCR data = HBCR data – OUTPTSIN + INPTSOUT. Any population-based interpretation of HBCR data makes basic assumption that INPTSOUT and OUTPTSIN are quantitatively equal and qualitatively similar.
As far as ATLAS is considered, data were obtained from all the sources of microscopic diagnosis of cancer, who gave their consent for the participation in the project. Utilizing the data on place of residence, the cases were divided to arrive at the number of incident cancer cases by district of residence. Subsequently, these numbers were divided by the population of the district to obtain ‘minimum cancer incidence rate’ (MCIR) for that district.
To compare the pattern of cancer according to HBCR and PBCR, three places with both types of registries were selected. Similarly, another three places for which PBCR and ATLAS both reported pattern of cancer were also selected. Ten leading sites by sex based on PBCR data were selected for each of the registries separately. Proportion of leading sites of cancer according to PBCR and that according to HBCR or ATLAS were obtained from published reports and compared by calculating absolute and relative differences. To compare the differences between HBCR/ATLAS and PBCR across the registries, we calculated the eighted average of the absolute values of differences (WAAD) and relative differences (WARD). These indicators were used also to compare HBCR data minus OUTPTSIN (i.e. RHBCR) with PBCR. In addition, trend in consistency was examined by plotting a scatter diagram between PBCR and HBCR proportions for different time periods. Present study utilized the data from consolidated reports of the national Cancer Registry Programme in India for PBCR, HBCR and ATLAS during 1984-2006, and the report of Tata Memorial Hospital in 2000.
Present study found that there was gross over/under representation of different sites of cancer in the data from HBCRs or ATLAS in India. Therefore, any assessment of risk of even leading sites of cancer based on HBCR or ATLAS data may be biased. By and large, easily accessible sites like mouth, tongue and hypopharynx were overrepresented in HBCR data, whereas, the sites not so easy to diagnose, like stomach, lung, prostate and brain were underrepresented. Among female specific sites, cancer of the breast and ovary were underrepresented whereas that of the uterine cervix was overrepresented. Reduction of HBCR data to RHBCR did not improve the consistency with PBCR. Differences of HBCR with PBCR were not consistent over a period of time for most of the leading sites. The differences in ATLAS were on the lines of that in HBCR data. It was to be expected, because ATLAS data are conceptually similar to a multi-centric HBCR data.
Since cancer registry is the backbone of cancer control, an erroneous assessment of cancer burden and pattern can have long term negative implications on the health resources of a country, especially in a limited resource country. HBCRs are valuable institutions having vital role in patient care assessment and conduct of epidemiological and survival studies with different environmental, clinical and treatment features. Population-based interpretation of HBCR or ATLAS data on occurrence of cancer, however, may not be valid due to the diversity in the availability of health services and prevalence of etiological factors.
It may be concluded that estimates on cancer occurrence or risk of cancer based on HBCR data or use of ATLAS as an alternative to PBCRs cannot be taken as granted. This attracts the attention of policy planners and administrators towards opening up more PBCRs, especially in rural areas and the areas considerably distant from existing PBCRs.
The dissertation is published in the publication series of Acta Universitatis Tamperensis; 2313, Tampere University Press, Tampere 2017. The dissertation is also published in the e-series Acta Electronica Universitatis Tamperensis; 1817, Tampere University Press 2017.