|Year : 2017 | Volume
| Issue : 3 | Page : 560-565
Quality assessment and improvement of cancer registration system in Kamrup Urban District: A report
Arpita Sharma1, Jagannath Dev Sharma2, Amal Chandra Kataki3, Debanjana Barman4, Ranjan Lahon5, Barsha Roy Deka5, Chinmoy Misra5, Manoj Kalita6
1 Computer Programmer, Population Based Cancer Registry, Guwahati, Assam, India
2 Professor and Head, Department of Pathology and Principal Investigator, Population Based Cancer Registry, Guwahati, Assam, India
3 Director, Dr. B.Borooah Cancer Institute, Guwahati, Assam, India
4 Medical Research Officer, Population Based Cancer Registry, Guwahati, Assam, India
5 Social Investigator, Population Based Cancer Registry, Guwahati, Assam, India
6 Statistician, Population Based Cancer Registry, Guwahati, Assam, India
|Date of Web Publication||24-May-2018|
Mr. Manoj Kalita
Statistician, Population Based Cancer Registry, Guwahati, Assam
Source of Support: None, Conflict of Interest: None
INTRODUCTION: The global burden of cancer incidence and mortality are rising continuously worldwide. As per the GLOBOCAN 2012 estimates, about 14.1 million cancer cases and 8.2 million cancer deaths occurred and 32.6 million people living with cancer (within 5 years of diagnosis) in 2012 worldwide. Reliable data on the magnitude and the pattern of cancer are essential for monitoring the health of the community, assessing the performance of the health care system. Cancer registries should pay great attention to quality of their data. The completeness of cancer registry data- the extent to which all of the incident cancers occurring in the population are included in the registry database- is an extremely important attribute of a cancer registry. There are mainly four aspects influencing the quality of data namely, comparability, completeness, validity and timeliness. MATERIALS AND METHODS: Data regarding incidence and mortality with methods of diagnosis for individual years were obtained from the National Cancer Registry Program database of the Indian Council of Medical Research for 2009 to 2014 periods and recalculated for combined years (2009-2014). RESULTS: In males in 2009-11, 77.1% were microscopically confirmed cases which are improved in the later years and for the year 2012-2014, it is 81.4%. In females also the percentage of microscopically confirmed cases were increased from 80.2% to 82.9%. An improvement in mortality to incidence ratio was observed over the years. MI ratio in males was improved to 32.9%. for the year 2012-14 as compared to 28.6% for the year 2009-11 while in female MI ratio is also increased from 18.8% to 21.8% over the period from 2009-11 to 2012-14. Whereas DCO was decreased from 12% to 10.7% in males and 7.3% to 6.6% in females respectively from the period 2009-11 to 2012-14. CONCLUSION: Although there is a slight improvement in data quality till date, there is an enormous scope for population based cancer registry Guwahati to improve the data quality.
Keywords: Cancer, death certificate only, incidence, mortality, mortality-to-incidence
|How to cite this article:|
Sharma A, Sharma JD, Kataki AC, Barman D, Lahon R, Deka BR, Misra C, Kalita M. Quality assessment and improvement of cancer registration system in Kamrup Urban District: A report. Indian J Cancer 2017;54:560-5
|How to cite this URL:|
Sharma A, Sharma JD, Kataki AC, Barman D, Lahon R, Deka BR, Misra C, Kalita M. Quality assessment and improvement of cancer registration system in Kamrup Urban District: A report. Indian J Cancer [serial online] 2017 [cited 2019 Dec 6];54:560-5. Available from: http://www.indianjcancer.com/text.asp?2017/54/3/560/233152
| » Introduction|| |
Cancer registration is an essential element of any cancer control strategy. Data quality is, however, of paramount importance. The aim of population-based cancer registries is to obtain information from all new cancer cases in a well-defined geographic area to assess the magnitude of cancer burden and its evolution and to provide a basis for research on causes of cancer and outcome in terms of incidence, prevalence, and survival. The completeness of cancer registry data – the extent to which all the incident cancers occurring in the population are included in the registry database – is an extremely important attribute of a cancer registry. There are mainly four aspects that influence the quality of data, namely, comparability, completeness, validity, and timeliness.
Continuous and systematic quality control measures are the characteristic of a smooth-running cancer registry.,, In this article, the experiences of Population-Based Cancer Registry (PBCR) Guwahati covering Kamrup Urban District (KUD) cancer registry are described. The Guwahati Cancer Registry (GCR) is a part of National Cancer Registry Programme (NCRP) of the Indian Council of Medical Research that began full registration of all types of cancer in KUD in 2003. The reporting of cancer is not notifiable in India. However, the GCR makes considerable effort to ensure that there is accurate and complete recording of all cases diagnosed, with active ascertainment and follow-up of cases.
| » Materials and Methods|| |
The aim of the GCR is to register all types of cancer diagnosed in KUD with the highest possible completeness and quality.
Data collection procedures
The registry obtains data on patients with cancer from a variety of sources, primarily through hospitals, pathological laboratory services, and vital statistic departments. Computerized hospital information systems or manual indexes of hospital discharges are the primary sources of data on patients and their diagnoses. The second key source of information are pathology laboratories, which provide definitive histological diagnosis of a tumor. The third main source comprises death certificates supplied to the GCR by the Birth and Death Record Departments. The death documents are important because they enable the identification of cancer cases that may have been missed by the other two sources and make it possible to investigate the survival of patients with cancer. The cancer death certificates are matched against the GCR system database of incidence cancer and the unmatched cases are followed up. If, after follow-up, no patient or cancer can be identified which corresponds to the death certificate, the case is registered as a “death certificate only” (DCO) case.
Validation checks are performed at the point of entry. Internal verifications are done on regular basis. In the preparation of report data set, IARC-CHECK program is implemented. This performs validity checks on both individual data items and consistency between items.
Registry coverage and data sources
GCR covers an area of 336 km 2 of KUD and a total population of 1,179,405, of which 608,844 were males and 570,561 were females (Census 2011, C-14 data).
The combined 6-year (2009–2014) estimated total population is 7,782,202 with 4,100,027 males and 3,682,175 females, with an average annual population of 1,237,034 which is shown in population pyramid [Figure 1].
The primary source of data collection for GCR is Dr. B. Borooah Cancer Institute (BBCI), a regional cancer center in Guwahati; 44.2% of incident cancer cases were collected from this center alone for the period of 2009–2014. Other than BBCI, data were collected from 40 other hospitals and 31 diagnostic centers, along with a pain and palliative care center and a state referral board.
All cancer registries should be able to give some objective indication of the quality of the data that they have collected. The value of any cancer registry relies heavily on the underlying quality of its data, namely, comparability, completeness, validity, and timeliness, and the quality control procedures it has in place.
The comparability of the GCR was addressed by describing the international standards used for classification and coding of neoplasms according to international guidelines. The standard for classification and coding of cancer is International Classification of Diseases for Oncology (ICD-O), published by the World Health Organization (WHO), which provides the standards for coding topography (location of the tumor in the body), morphology (microscopic appearance of the tumor), behavior (whether the tumor is malignant, benign, or in situ), and grade (the extent of differentiation of the tumor).,
The extent to which all the incident cancers occurring in the population are included in the registry database is a very important aspect of data quality. There are numerous ways to examine this, and the methods can be broadly classified into two main categories, depending on whether they can be considered semi-quantitative or quantitative.
Parkin and Bray distinguish between qualitative (or semi-quantitative) methods, which give an indication of the degree of completeness relative to other registries or over time, and quantitative methods, which provide a numerical evaluation of the extent to which all eligible cases have been registered. Among the semi-quantitative methods, the possibility that a relatively high morphologically verified diagnosis (MV%) may represent incompleteness of data collection has already been noted. The other widely used indicators are as follows: mortality-to-incidence (M/I) ratio, stability of incidence over time, and comparison of incidence rates with other (similar) populations. Three methods are available to obtain a quantitative evaluation of the degree of completeness of registration: independent case ascertainment, capture–recapture methods, and death certificate methods.
The accuracy of recorded data is greatly enhanced by consistency checks done at the time of data entry. The validity of cases depends on internal consistency checks' methods, diagnostic criteria methods (histological verification and DCO cases), missing information analyses (e.g., primary site unspecified, age unknown), and population estimation. Most registries will report on three statistics that have a bearing on the accuracy of the recorded data. They are the following: the proportion (or percentage) of cases with missing data, the percentage of cases with MV%, and the percentage of cases for which the information came from only a death certificate (DCO%). A high proportion of cases with missing values generally imply poor diagnostic precision (i.e., low MV% observed among O and U cases) or a failure to specify the site of the primary cancer in cases diagnosed based on tissue obtained from a metastasis.
PBCRs under NCRP network use PBCRDM data management software which helps identify and rectify the quality check errors and perform duplicate checks, different level of consistency checks, and matching before data submission. NCRP does a second level of checks on the data. This ensures faster finalization of the data.
Rapid reporting is often required from cancer registries. However, for cancer registries, a trade-off must be recognized between data timeliness and the extent to which the data are complete. The timeliness depends on the rapidity with which the registry can collect, process, and report sufficiently complete and accurate data. Stability of incidence over time is required to check on the number of cases being registered each year. In the absence of marked changes in the population, this can quickly identify potential defects in case finding. Each PBCR should periodically check data on the number of cases received versus the expected cases (based on previous year) from each major source.
| » Results|| |
A total of 5463 (male = 3071 and female = 2392) patients were diagnosed with cancer in 2012–2014 in KUD compared with 4468 (male = 2536 and female = 1932) cases in 2009–2011. An increase of 21.1% in the number of males and 23.8% in females was observed over the period of observation.
Cancer incidence since 2009 has varied substantially depending on the types of cancer. Overall, age-standardized incidence rates for cancers have increased by about 15.4% over the years in males and 27.4% in females from 2009 to 2014. The number of new cases of all types of cancer was increased from 185.4to 214.0 and 143.9 to 183.3 per 100,000 men and women, respectively, from the years 2009–2014 in KUD.
The top 10 leading sites of cancer for the period of 2009–2014 in males are esophagus (comprising Age Adjusted Incidence Rate (AAR) of 28.26 per 100,000 population), followed by lung (16.95), hypopharynx (16.39), stomach (13.66), prostate (11.59), mouth (9.12), tongue (8.27), gallbladder (8.12), larynx (7.78), and tonsil (5.75).
In females, breast (having AAR of 24.95 per 100,000 population) followed by esophagus (20.45) form the top two positions out of the 10 leading sites. Gallbladder (15.63) occupies the third position, followed by cervix (14.09), ovary (9.74), stomach (8.22), lung (7.98), mouth (7.19), corpus uteri (3.92), and hypopharynx (3.45).
In males, esophagus was the leading site of cancer contributing to about 14.6% of the total cases followed by cancer of hypopharynx (8.4%), lung (7.4%), and stomach (6.8%). In females, breast is the leading site with 16.8% of total cases followed by esophagus (10.4%), cervix uteri (9.1%), gallbladder (8.5%), and ovary (8.1%). Most of the leading sites in KUD in both males and females are cancers related to tobacco usage.
Many individual types of cancer have also shown upward trends in annual rates, for example, lymphomas and prostate, gallbladder, stomach, kidney, thyroid, cervical, uterine, and female lung cancer, although the likely factors involved may differ (e.g., screening, lifestyle factors, or diagnostic improvements).
As per 2012–2014 report of NCRP, International comparison of AAR with that of other Indian PBCRs shows that cancer of hypopharynx of KUD (17.5) in Males has the second highest position after East Khasi Hills district (22.2). Whereas AAR (3.2) in Females in KUD occupies first position internationally.
In our registry, the quality of data improved over the years. An evaluation of the reliability of cancer data is usually measured by the percentage of patient with microscopic diagnosis compared with other methods of diagnosis. The percentage of patients having microscopic confirmation of cancer depends mainly on the accessibility of the site involved. Histopathologically confirmed cases are usually seen in head and neck cancer and cancers of female genital tract than of the tumors of inaccessible sites such as digestive and respiratory systems. A few patients with cancer were found to be diagnosed endoscopically or using other imaging techniques for cancers involving the gallbladder, stomach, pancreas, urinary bladder, and so on.
In males, the percentage of microscopic cases was increased from 77.1% in 2009–2011 to 81.4% in 2012–2014. DCO was decreased from 12.0% to 10.7%. M/I ratio also increased from 28.6% to 32.9% over the study period. In females too, the percentage of microscopic cases was increased from 80.2% to 82.9%, DCO was decreased from 7.3% to 6.6%, and M/I ratio also increased from 18.8% to 21.8% from 2009–2011 to 2012–2014.
| » Discussion|| |
Cancer registries should pay great attention to the quality of their data, both in terms of completeness and accuracy. Cancer incidence and mortality are key measures of cancer burden in a country providing an important basis for implementing public health preventive measures. The incidence rate for overall cancer has been increasing since 2009 in KUD. [Table 1] summarizes an increase in incidence rate of 15.4% in males and 27.4% in females from 2009 to 2014 per 100,000 populations. This shows the burden is growing over the years. Esophagus was the most frequently diagnosed cancer site and the site with the greatest incidence among Assamese population of KUD.
|Table 1: Incidence and mortality rates for males and females per 100,000 populations from individual year 2009 to 2014|
Click here to view
In this study, we compare different quality indicators (completeness, consistency, uniqueness, and accuracy) for two different time periods, 2009–2011 and 2012–2014. From [Table 2], it can been seen that in males, tongue, oropharynx, hypopharynx of the head and neck cancers, and cancers of genital organs and rectum are confirmed microscopically in >90% of cases. From [Table 3], it can be seen that in females, most of the head and neck cancers and cancers of women genital organ have been diagnosed microscopically in >90% of cases, and other sites such as breast, kidney, bladder, lymphomas, NHL, leukemias, and myelomas have also been confirmed microscopically in >90% of cases.
|Table 2: Comparisons of relative proportion of microscopic variations, death certificate only, mortality incidence ratios in males of KUD for the period of 2009-11 to that of 2012-14|
Click here to view
|Table 3: Comparisons of relative proportion of microscopic variations, death certificate only, mortality incidence ratios in females of KUD for the period of 2009-11 to that of 2012-14|
Click here to view
Mortality statistics is an important tool for undertaking epidemiological studies of cancer. Cancer recorded as the underlying cause of death in death certificate and cancer recorded in population-based cancer registries are often compared to assess the quality of registry findings such as M/I ratio.
[Table 2] and [Table 3] show improvement in M/I ratios in males (28.6%) for the years 2012–2014 when compared with males (32.9%) for the years 2009–2011, whereas in females the M/I ratio increased from 18.8% to 21.8% over the study period. There is a 15% rise in males and 16% in females in terms of M/I ratios. The M/I ratio is an important indicator that is widely used to identify possible incompleteness. It is a comparison of the number of deaths, obtained from vital statistics department, and the number of new cases of a specific cancer registered in the same time period. Timeliness plays an important role in collecting mortality data of good quality by obtaining accurate recording of the cause of death.
Because in India cancer is not a notifiable disease, data collection is active. Cancer incidence and mortality registration are under registration in India, and vital registration systems are often absent, unreliable, or unavailable to the registries and many registries in India depend on active follow-up methods. Establishing acceptable and objective criteria for the DCO% has been a contentious issue in registries' comparative studies. A low DCO% may simply reflect efficient trace back of cases initially missed by the normal case finding procedures. The DCO% is also influenced by local circumstances including the availability and accuracy of death certificates. In PBCR Guwahati, in males DCO is decreased from 12% to 10.7% and in females it is decreased from 7.3% to 6.6% for the period from 2009–2011 to 2012–2014. From [Table 2] and [Table 3], it is observed that in male, cancer sites with lowest rate of DCOs are cancers of lip, salivary gland, nasopharynx, small intestine, anal canal, melanoma, Peripheral nerves and autonomic nervous system + connective, subcutaneous and other soft tissues (Conn.+ soft tissue), and male genital organs and cancers of lip, tongue, salivary gland, nasopharynx, small intestine, rectum, anal canal, bone, melanoma, skin, conn soft tissues, female genital organs, kidney, Hodgkin's disease, lymphoid leukemias, etc., are with lowest DCO%.
The factors responsible for low M/I ratios in KUD are inadequate record keeping and diagnosis details in mortality registration centers. Absent of specific cause of death in mortality sources limit the quality of data collection. Records registered manually increase the chances of error, and lack of computerization in the centers consumes a lot of time and energy. Nonavailability of death certificate at the source of information may lead to under-registration. In India, with poorly functioning routine health statistics data systems and unavailable mortality data, cancer survival estimates from PBCRs can sometimes only provide the insight into the status of cancer care in the country.
| » Conclusion|| |
A good cancer registry should follow an active research program. If research and researchers are essential elements of a cancer registry, the data quality aspect becomes a natural part of the daily routines of the registry. The quality of cancer registry data depends partly on the competence and experience of its staff which should include good knowledge on cancer, computer science, and biostatistics. Good relations with practicing physicians, health authorities, and scientists are important.
All cancer registries, both within and outside the NCRP network in India, have made enormous efforts to improve the coverage and the quality of data collected. They have contributed significantly to provide the information base for planning, implementing, and monitoring cancer control inputs and in evaluating cancer control outcomes. The regular annual review meetings of the NCRP and frequent training and re-training courses for the registry staff and the technical collaboration that many NCRP registries have with international partners such as IARC, WHO, IACR, and their networks have substantially improved the completeness, validity, and quality of data.
We would like to thank Dr. P. Mathur, Director, NCDIR and his team for their valuable guidance in making this manuscript. We would also like to thank NCDIR, Bengaluru, and ICMR, India, for financial support. We sincerely acknowledge the cooperation from all the sources of registration and oncologists of Dr. B. Borooah Cancer Institute, Guwahati. We also appreciate the hard work and sincerity of all the staff members at PBCR-HBCR, Guwahati.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| » References|| |
Brewster D. Improving the quality of cancer registration data. J R Soc Med 1995;88:268-71.
Parkin DM. The evolution of the population-based cancer registry. Nat Rev Cancer 2006;6:603-12.
Parkin DM, Bray F. Evaluation of data quality in the cancer registry: Principles and methods Part II. Completeness. Eur J Cancer 2009;45:756-64.
Parkin DM, Muir CS. Cancer incidence in five continents. Comparability and quality of data. IARC Sci Publ 1992;120:45-173.
Saxtn EA, Teppo L, Hakulinen T. Quality control of cancer registry data. In: Nieburg HE, editor. Prevention and Detection of Cancer. Part I. Prevention. Vol. 2. Etiology, Prevention Methods. New York: Marcel Dekker; 1978. p. 2151-64.
Saxén EA. Cancer registry: Aims, functions and quality control. Arch Geschwulstforsch 1980;50:588-97.
Mattsson B, Wallgren A. Completeness of the Swedish cancer register. Non-notified cancer cases recorded on death certificates in 1978. Acta Radiol Oncol 1984;23:305-13.
Parkin DM, Chen VW, Ferlay J, Galceran J, Storm HH, Whelan SL. Comparability and Quality Control in Cancer Registration (IARC Technical Report No. 19). Lyon: IARC; 1994.
Larsen IK, Småstuen M, Johannesen TB, Langmark F, Parkin DM, Bray F, et al.
Data quality at the cancer registry of Norway: An overview of comparability, completeness, validity and timeliness. Eur J Cancer 2009;45:1218-31.
Bray F, Znaor A, Cueva P, Korir A, Swaminathan R, Ullrich A, et al
. Planning and Developing Population-Based Cancer Registration in Low-and Middle-Income Settings. Lyon, France: International Agency for Research on Cancer; 2014.
Teppo L, Pukkala E, Lehtonen M. Data quality and quality control of a population-based cancer registry. Experience in Finland. Acta Oncol 1994;33:365-9.
Torre LA, Bray F, Siegel RL, Ferlay J, Lortet-Tieulent J, Jemal A, et al.
Global cancer statistics, 2012. CA Cancer J Clin 2015;65:87-108.
Coelho KR. Challenges of the oral cancer burden in India. J Cancer Epidemiol 2012;2012:701932.
[Table 1], [Table 2], [Table 3]