ArticlePDF Available

Problem of small numbers in reporting of cancer incidence and mortality rates in Indian cancer registries

Authors:

Abstract and Figures

The present paper examines the problem of small numbers (<20 cases) associated with many sites of cancers in Indian cancer registries. The cancer incidence data of 14 Population Based Cancer Registries for the periods of 2001-03 and 2004-05 were utilized for the analysis. Nine out of 14 registries had more than 50% of their sites being associated with small numbers while seven registries had 50% of their sites having as low as 5 cases. Sites associated with small numbers showed a lot of variation and significant differences in their incidence rates within two years duration which are not feasible. The percentage age distribution was also found to vary with different periods. The paper has effectively shown the effect of population size on incidence rates. For a registry of population size 300,000, the incidence rate of 6 can very well be unstable. There are many registries in the world with their population size less than 200,000. Even in the case of registries with high population (>or= 500,000) the practice is to report the cancer incidence by different ethnic groups with populations less than 200,000 and thereby introduce the problem of small numbers in reporting the incidences of various cancer sites. To overcome this problem, pooling of data over broad age groups or ten years age groups or 3 to 5 years periods is one immediate solution.
Content may be subject to copyright.
Asian Pacific Journal of Cancer Prevention, Vol 10, 2009 657
Problem of Small Numbers in Reporting Indian Cancer Registry Data
Asian Pacific J Cancer Prev, 10, 657-660
Introduction
The Indian Council of Medical Research (ICMR)
started a National Cancer Registry Programme (NCRP)
in the year 1982 with the main objective of generating
reliable data on the magnitude and pattern of cancer in
India. There are 23 Population Based Cancer Registries
(PBCR) which are currently functioning under the network
of NCRP. The cancer incidence data so collected is
analyzed and reported in standard format from time to
time in the form of one year/two years/three years report.
One of the regular features of these reports is to report
number of incidence and mortality cases, Crude Rate (CR),
Age Adjusted Rate (AAR) as well as rates by five yearly
age groups for selected sites of cancer. The National Centre
for Health Statistics, USA does not publish or release rates
based on fewer than 20 observations, because they feel
these data do not meet their requirement for a minimum
degree of accuracy and such rates are termed as unstable
rates (Chronic Disease Teaching Tools, USA - 1999).
Accordingly, it is expected that whenever the cancer
incidence rates are based on small numbers (<=20 cases)
should not be reported and if reported, at least should be
highlighted. However, no such practice is evident in Indian
Cancer Registries. Thus, the present paper examines the
problem of small numbers associated with many sites of
cancers in reporting from Indian cancer registries. The
objectives of the paper are:
National Cancer Registry Programme, Indian Council of Medical Research, Bangalore, India *For Correspondence:
ramnath_takiar@yahoomail.co.in
Abstract
The present paper examines the problem of small numbers (< 20 cases) associated with many sites of cancers
in Indian cancer registries. The cancer incidence data of 14 Population Based Cancer Registries for the period
of 2001-03 and 2004-05 were utilized for the analysis. Nine out of 14 registries had more than 50% of their sites
being associated with small numbers while seven registries had 50% of their sites having as low as 5 cases. Sites
associated with small numbers showed a lot of variation and significant differences in their incidence rates
within two years duration which are not possible. The percentage age distribution was also found to vary with
different periods. The paper has effectively shown the effect of population size on incidence rates. For a registry
of population size 300,000, the incidence rate of 6 can very well be unstable. There are many registries in the
world with their population size less than 200,000. Even in the case of registries with high population ( 500,000)
the practice is to report the cancer incidence by different ethnic groups with populations less than 200,000 and
thereby introduce the problem of small numbers in reporting the incidences of various cancer sites. To overcome
this problem, pooling of data over broad age groups or ten years age groups or 3 to 5 years periods is one of an
immediate solution.
Key Words: Cancer - AAR - small numbers - least significant incidence rate - data pool - broad age groups
1. To provide the distribution of sites according to their
number of cases to show the extent of problem of small
numbers in Indian Cancer Registries.
2. To highlight the variation occurring in AAR with
period due to small numbers associated with them.
3. To show the variation in percentage distribution of
Age of cancer cases by selected site and different periods,
occurring due to small numbers associated with them.
4. To construct a table providing the least significant
incidence rate associated with the given population size.
5. To provide the possible solutions to deal with the
problem of small numbers in Indian Cancer Registries.
Materials and Methods
For the analysis purposes, the cancer incidence data
of year 2004-05 (NCRP–2008), for the Population Based
Cancer Registries of Bangalore, Barshi, Bhopal, Chennai,
Delhi, Mumbai, Ahmedabad and Kolkata was utilized. In
addition, the cancer incidence data of the year 2005-06
(NCRP–2008), for the North East Population Based
Cancer Registries of Dibrugarh, Kamrup Urban, Silchar
town, Imphal West District, Mizoram State and Sikkim
State was utilized. For males, there are 52 cancer sites
and for females there are 56 cancer sites (C00-C95) for
which five yearly age group distribution of cases and
incidence rates are reported routinely.
In addition, for each site, the total number of cases,
RESEARCH COMMUNICATION
Problem of Small Numbers in Reporting of Cancer Incidence
and Mortality Rates in Indian Cancer Registries
Ramnath Takiar*, Deenu Nadayil, A Nandakumar
Ramnath Takiar et al
Asian Pacific Journal of Cancer Prevention, Vol 10, 2009
658
Crude Rate (CR) and Age Adjusted Rate (AAR) are also
reported routinely.
Each site, based on the number of cases associated
with them, is categorized into following five categories:
5; 6-10; 11-15; 16-19 and 20. Sites having cases more
than 19 were considered as having stable rates otherwise
sites with cases below 20 were termed as having unstable
rates and being associated with small numbers. In order
to show how the sites with small number of cases (<20)
can show variation in their rates from one year to another,
five selected sites, known for having small numbers, were
considered from the incidence data of Bhopal PBCR from
the report of the year 2001-03 (NCRP – 2006) and 2004-
05 (NCRP – 2008). The sites, with their ICD10 Codes,
selected were: Bone (C40-41), Corpus uteri (C54),
Thyroid (C73), Hodgkins disease (C81). The S.E. of these
rates were calculated using the formula S.E= rate/cases.
The difference between the rates for two periods were
tested using the formula; (Rate1-Rate2)/((S.E. of R1)2 +
(S.E. of R2)2)0.5. For females, the percentage age
distribution for the site of Larynx (C17) by five different
periods for the registry of Bhopal was obtained. This was
mainly done to show that if the site is associated with
small numbers and then if we are comparing the
percentage age distribution, thereby incidence rates,
between different periods, it can be misleading.
In order to show how even a high incidence rate of
certain site, in the presence of less populated registries
can be associated with small numbers, a table was
constructed. In table, in rows, different hypothetical
populations ranging from 100,000 to 2000,000 were
displayed. In column an attempt was made to translate
the given incidences from one to ten (1-10) for each
population to an equivalent number of cases. For example,
an incidence rate of 10, for the registry population of
100,000 may give rise to only 10 cases (Cases = Incidence
x Population)/100,000). Thus, an incidence rate when seen
in relation to population of its registry, may give the actual
status of the incidence so as to say whether it is stable or
unstable. In addition, in the last column of the table, the
Least Significance Incidence Rate (LSIR) was calculated
and shown. An Incidence rate is termed as “Least
significant Incidence Rate” for given registry of any
population size if all the incidence rates below it are often
associated with less than 20 cases.
IARC is reporting from time to time, the various rates
related to cancer incidence from different parts of the
world. However, the problem of small numbers in
reporting is even not highlighted in their publication
(Parkin etal., 2002). The problem of a small number is
very often closely associated with the population size of
the registry. Invariably, registries with lower total
populations (say, 200,000) give rise to incidence rates of
various cancer sites being inevitably associated with small
numbers. Therefore, it was thought interesting to ascertain
the number of cancer registries whose cancer incidence
rates have been reported in Cancer Incidence in Five
Continents Vol. VIII (Parkin et al., 2002) and which have
relatively low populations (<200,000). The LSIR values
were also calculated for such registries and are shown in
the Table.
Results
The percentage distribution of sites according to their
number of cases by different Population Based Cancer
Registries for both males and females are shown in Table
1. The registries were arranged according to ascending
order of percentage of sites with adequate numbers. The
registry of Barshi to Aizawl had less 15% of their sites
with adequate number of cases implying more than 85%
of the sites had number of cases less than 20 thereby giving
mainly the incidence estimates as unstable rates. The
registry of Ahmedabad to Bhopal had between 20-40%
of their sites while the registry of Kolkata had around
50% of their sites having adequate number of cases. The
registries of Chennai to Delhi had 70% to 85% of their
sites having adequate number of cases. Similarly, for males
the registry of Barshi to Aizawl including Dibrugarh, more
than 50% of their sites had number of cases as low as 5.
Even the registry of Kolkata, Bhopal and Kamrup urban
had 25-40% of their sites with numbers of cases below 5.
As in the case of males, the registries were arranged
according to ascending order of percentage of sites with
adequate numbers. The registry of Silchar town to
Dibrugarh, less than 20% of the sites only had adequate
number of cases associated with them. Bhopal and Kolkata
registries had 20-40% of their sites having adequate
numbers. In the case of other registries, 65-75% had
adequate number of cases associated with them. Silchar
town to Dibrugarh registries had more than 55% of their
Table 1. Percentage Distribution of Cancer Sites
According to Numbers of Cases
PBCR area 5 6-10 11-15 16-19 20 Total
Males
Barshi 63.5 25.0 7.7 1.9 1.9 52
Silchar town 86.5 7.7 1.9 0.0 3.8 52
Imphal West 61.5 9.6 15.4 7.7 5.8 52
Sikkim 63.5 11.5 11.5 1.9 11.5 52
Aizawl 55.8 19.2 9.6 3.8 11.5 52
Ahmedabad 34.6 21.2 11.5 11.5 21.2 52
Dibrugarh 50.0 11.5 9.6 3.8 25.0 52
Kamrup Urban 38.5 17.3 11.5 3.8 28.8 52
Bhopal 30.8 17.3 5.8 9.6 36.5 52
Kolkata 26.9 5.8 3.8 9.6 53.8 52
Chennai 11.5 5.8 5.8 5.8 71.2 52
Bangalore 15.4 3.8 - 5.8 75.0 52
Mumbai 9.6 1.9 1.9 3.8 82.7 52
Delhi 7.7 5.8 - 1.9 84.6 52
Females
Silchar town 91.1 3.6 0.0 3.6 1.8 56
Barshi 89.3 3.6 0.0 3.6 3.6 56
Aizawl 71.4 12.5 7.1 0.0 8.9 56
Ahmedabad 57.1 26.8 5.4 0.0 10.7 56
Imphal West 67.9 10.7 5.4 5.4 10.7 56
Sikkim 64.3 16.1 3.6 3.6 12.5 56
Dibrugarh 60.7 7.1 8.9 7.1 16.1 56
Kamrup Urban 53.6 17.9 7.1 5.4 16.1 56
Bhopal 46.4 17.9 8.9 3.6 23.2 56
Kolkata 30.4 14.3 14.3 3.6 37.5 56
Chennai 16.1 8.9 7.1 1.8 66.1 56
Bangalore 19.6 5.4 0.0 7.1 67.9 56
Mumbai 12.5 3.6 3.6 5.4 75.0 56
Delhi 10.7 5.4 1.8 3.6 78.6 56
Asian Pacific Journal of Cancer Prevention, Vol 10, 2009 659
Problem of Small Numbers in Reporting Indian Cancer Registry Data
sites having as low as 5 number of cases associated with
them. Bhopal and Kolkata registries had 30-50% of their
sites associated with less than 5 cases.
The variation in AAR by period of reporting for
selected sites, associated with small numbers, is shown
in Table 2. It can be seen that for the cancer sites of Bone
and Thyroid among males, the relative changes in AAR
for the period of 2004-05 as compared to 2001-03 were
more than 200% while for other cancer sites in females,
the relative changes were more than 180% of that seen in
the year 2001-03. So, all the selected sites have either
increased to more than doubled or reduced to one-third in
2004-05 as compared to that seen in the period of 2001-
03. Among males, all the three selected cancer sites
showed significant variation in 2004-05 as compared to
that seen in the year 2001-03. However, in females though
the variation in incidence rates was also observed but it
was not found significant.
The percentage distribution of cancer cases of Larynx
(C32) by selected broad age groups categories and periods
is shown in Table 3. It can be easily seen that going by
the data of 1997-98 and 2001-03, there is no case below
the age of 30 years while by the data of 1990-96, there is
no case below the age of 40 years. However, the data of
2004-05 suggests that there is no case below the age of
50 years. So, different periods suggest different minimum
ages below which the cases are not seen. The percentage
age distribution obviously also seems to be varying.
For individual populations, the Least Significant
Incidence Rates can be obtained easily from Table 4. For
populations up to 150,000, the Least Significant Incidence
Rate was observed to be more than 10. For 200,000
populations, it was 10 while for 250,000 populations it
was 8. The Least Significant Incidence Rate decreases as
we go from low population to high populations. After
750,000 populations, the Least Significant Incidence Rate
will be below 2 per 100,000 populations.
Cancer Registries from different parts of the world
with population below 200,000 along with their Least
Significant Incidence Rate are shown in Table 5. There
were many registries with population below 200.000
whose results are reported in Cancer Incidence in five
Table 2. Variation in AAR by Selected Sites Associated
with Small Numbers and Years - Bhopal PBCR
Site Bone Thyroid Hodgkins disease Corpus
M M M F uteri
ICD10 C40-41 C73 C81 C54
Cases 2001-03 19 6 19 4 19
2004-05 23 12 6 6 25
AAR 2001-03 0.7 0.2 0.8 0.2 0.9
2004-05 1.5 0.7 0.3 0.4 1.6
% change 214 350 37.5 200 178
SE of 2001-03 0.161 0.082 0.184 0.100 0.206
AAR* 2004-05 0.313 0.202 0.122 0.163 0.320
Significance P <0.05 <0.05 <0.05 NS NS
M, males; F, Females; *AAR/cases
Table 5. Cancer Registries with Populations below 200,000 and their Least Significant Incidence Rates (LSIR)
Country Registry name Males LSIR* Females LSIR*
Canada Yukon 14,472 138 13,184 151
Argentina Concordia 71,071 28 74,621 26
Switzerland Neuchatel 79,059 25 84,820 23
Italy Biella Province 90,765 22 99,607 20
Spain Cuenca 100,618 19 101,493 19
Switzerland Graubunden & Glarus 110,407 18 112,872 17
Portugal Vila Nova de Gaia 126,975 15 133,868 14
Switzerland Valais 132,288 15 137,316 14
Iceland - 134,062 14 133,340 14
Italy Macerata Province 142,019 14 150,893 13
Italy Ragusa Province 145,692 13 151,980 13
France Tarn 166,692 11 175,364 11
Austria Vorarlberg 170,225 11 172,264 11
France Martinique 177,763 11 194,584 10
Argentina Bahia Blanca 187,325 10 198,933 10
China Jiashan 191,005 10 186,350 10
*Parkin et al., Cancer Incidence in Five Continent Vol VIII; IARC Scientific Publication No. 155
Table 4. Calculation of Cases and LSIRs with
Populations in 1,000s and Incidence Rates per 100,000
123456 7 8 910
100 1 2 3 4 5 6 7 8 9 10 20.0
150 2 3 5 6 8 9 11 12 14 15 13.3
200 2 4 6 8 10 12 14 16 18 20 10.0
250 3 5 8 10 13 15 18 20 23 25 8.0
300 3 6 9 12 15 18 21 24 27 30 6.7
400 4 8 12 16 20 24 28 32 36 40 5.0
500 5 10 15 20 25 30 35 40 45 50 4.0
750 8 15 23 30 38 45 53 60 68 75 2.7
1000 10 20 30 40 50 60 70 80 90 100 2.0
1500 15 30 45 60 75 90 105 120 135 150 1.3
2000 20 40 60 80 100 120 140 160 180 200 1.0
Table 3. Percentage Distribution of Cancer Cases of
Larynx (C32) by Different Periods - Females
Age group 04-05 01-03 99-00 97-98 90-96
<= 30 - - - - -
31-40 - 20.0 - 33.3 -
41-50 - 20.0 - - 18.8
51-60 33.3 20.0 - 50.0 37.5
61-70 - 20.0 100.0 - 25.0
>=71 66.7 20.0 - 16.7 18.8
Total cases 6 5 1 6 16
Source: National Cancer Registry Programme reports
Ramnath Takiar et al
Asian Pacific Journal of Cancer Prevention, Vol 10, 2009
660
Continents (Parkin etal. 2002), However, only few selected
registries are listed in the Table 6. When the registry
population is less then the LSIR can be very high like in
the case of registry from Canada, the LSIR is 138 for males
and 151 for females. When the population is 191,005 in
the case of Jiashan registry of China, the LSIR is 10.
Discussion
It becomes clear from the Table 1 that for Barshi to
Dibrugarh, all registries listed had a majority of their sites
having small numbers associated with them. From Barshi
to Sikkim, all registries had more than 50% of their sites
being associated with as low as 5 cases. Thus, it can be
said that small numbers associated with various cancer
sites can be in general a problem with majority of the
Indian registries. Further, it was shown clearly that when
the sites are associated with small numbers, they can show
as much as 200% variation in their incidence rates. In
few cases, the incidence can become almost 50% with
the period. Thus, cancer sites when they are associated
with small numbers can show a lot of variation in their
Age Adjusted Rates making it difficult to interpret the
changes occurring in them (Table 2).
The data provided in Table 3, make it clear how the
percentage distribution when seen in relation to small
numbers can be misleading. The minimum age above
which the larynx cases are reported are shown to vary
from around 30 years to 60 years among different
registries. This variation can be attributed to small
numbers. If the incidences of sites are based on adequate
numbers, such a variation in minimum age should not have
been there. While giving most of the results by registries
in percentages like providing the relative percentages of
cancers based on different microscopic diagnosis (Primary
histology, Secondary histology etc.) can be less
meaningful when the numbers associated with them are
particularly very small.
The effect of population on incidence rate is also
brought out clearly in Table 5. It can be argued that when
the population of the registry is around 150,000 then the
incidence rate of even 10 should be viewed with
reservations. Similarly, when the population of the registry
is around 300,000 then even the incidence rate of 6 can
be termed as least significant and should be interpreted
with care. The problem of small numbers in relation to
various cancer sites, particularly, in Indian cancer registries
is never highlighted. While considering the rare cancers
which are often associated with small numbers should be
viewed carefully for assessing the time trend and yearly
fluctuations occurring in them. In addition, the basis of
diagnosis according to different categories (Microscopic,
X-ray etc.) along with their percentages is provided
routinely in each report for each site of cancer. When the
sites are associated with small numbers, the percentage
distribution may not give correct picture and can be
misleading.
As many registries in the world are functioning with
small populations (Table 5), the problem of small numbers
in reporting of cancer incidences becomes inevitable.
LSIR of 138 in the case of males and from the registry of
Canada signifies that all incidences below it should be
viewed with reservations. Similarly, all the incidences
above 10 only should be viewed with reliability in the
case of incidences from Jiashan registry of China. It is
desirable that for all the registries of the world, the LSIR
should be close to one so that all incidences observed can
be viewed as stable rates otherwise an incidence rate of
138 also should be viewed as an unstable rate and in
presence of small numbers such rates can show abnormal
fluctuations from one year to another. Even in the case of
registries with high population ( 500,000) the practice is
to report the cancer incidence by different ethnic groups
like reported from the registries of USA. In such situations
very often the ethnic groups reported end up with small
populations, less than 200,000 and thereby again
introducing the problem of small numbers in reporting
the incidences of various cancer sites. So the problem of
small numbers in reporting can be found to be associated
with various registries of the world from developed as
well as from developing countries.
In view of the above, it is suggested that all such sites
which are associated with small numbers should be
marked and should be highlighted in the report. The
problem of small numbers in cancer registries can be
addressed in more than one ways. First, if the registry
population is less than 300,000, it should be expanded so
as to cover at least 500,000 populations so that meaningful
and stable incidence rates for the given registry can be
provided. Second, the cancer incidence data can be pooled
for 2-5 years and reported. Reporting the incidence data
for broad age groups say every ten years age group can
also be an immediate solution to the problem of small
numbers. In general, It is needed that all registries should
view the problem of small numbers and should make an
attempt to either highlight or should go for any suggested
solution.
References
National Cancer Registry Programme (ICMR). 2006.
Consolidate Report of the Population Based Cancer
Registries: 2001-2004. Bangalore, India.
National Cancer Registry Programme (ICMR). 2008. NE-PBCR
Report: 2005-06. Bangalore, India (http://
www.pbcrindia.org/)
National Cancer Registry Programme (ICMR). 2008. PBCR
Report 2005-06, Bangalore. India (http://
www.pbcrindia.org).
Parkin DM, Whelan Sl, Ferlay J, Teppo and Thomas DB. 2002:
Cancer Incidence in Five Continents Vol. VIII – IARC Press,
Lyon, France.
Rates Based on Small Numbers - Statistics Teaching Tools
USA -1999. (http://www.health.state.ny.us/diseases/
chronic/ratesmall.htm)
... Reporting low numbers can also threaten the confidentiality of individuals (19,20) and small numbers are also more affected by chance. This instability can be addressed by either collapsing categories to increase the count (31) or, as in this article, expanding the time interval. NS is a small province with a population of less than one million residents. ...
... Monthly counts of confirmed pertussis cases by health management zone reported in Nova Scotia between January 1 and December31, 2015 ...
Article
Full-text available
Abstract Background: Pertussis can cause a serious respiratory bacterial infection, especially in infants. Between January 1 and December 31, 2015, there was an increase in the number of reported pertussis cases in Nova Scotia (NS). Surveillance practices for pertussis in NS were challenging because immunization coverage data are not available and rate information was neither timely nor precise with respect to geography. Public health officials in NS decided to adopt a new surveillance technique to inform public health actions across the Province. Objective: To assess the use of a 40-day rolling incidence rate to monitor pertussis activity in Nova Scotia. Intervention: A 40-day rolling incidence rate was calculated for pertussis by age groups and various levels of geography. Public health authorities continued to anticipate new cases of pertussis if the contacts of known cases were still within the incubation period (range between six and 20 days). The 40-day incubation period was chosen to reflect twice the incubation period’s upper range. Rates were calculated using Statistics Canada population projections for 2014 and then compared with traditional case counts and cumulative incidences. The usefulness of the statistics was assessed by public health decision makers. Outcomes: Increased pertussis activity was noted across NS, most notably in the South West region. The use of a 40-day rolling incidence rate as a surveillance tool provided more timely and geographically precise descriptions of ongoing trends in pertussis activity and helped to inform appropriate public health action. Health officials valued the information provided from the rolling incidence because it allowed them to manage activities based on weekly estimates at various levels of geography. Conclusion: Rolling incidence proved to be a useful tool to monitor a cyclical increase in pertussis cases in Nova Scotia and to inform related public health actions. The rolling incidence provided geographically precise and timely information that was useful to estimate new cases in the absence of reliable immunization coverage information. This method could supplement traditional epidemiological surveillance of future communicable disease events, especially those characterized by long incubation periods and low case counts.
... LGAs provide detailed spatial information as well as a sufficient number of cases for calculating accurate and reliable rates. Should the number of cases be lower than 20, estimates may be unreliable, with high variation due to small numbers [37,38]. Alberta LGAs do not exhibit this problem, as over 95% of the 132 ...
Article
Full-text available
Background Knowledge of geospatial pattern in comorbidities prevalence is critical to an understanding of the local health needs among people with osteoarthritis (OA). It provides valuable information for targeting optimal OA treatment and management at the local level. However, there is, at present, limited evidence about the geospatial pattern of comorbidity prevalence in Alberta, Canada. Methods Five administrative health datasets were linked to identify OA cases and comorbidities using validated case definitions. We explored the geospatial pattern in comorbidity prevalence at two standard geographic areas levels defined by the Alberta Health Services: descriptive analysis at rural-urban continuum level; spatial analysis (global Moran’s I, hot spot analysis, cluster and outlier analysis) at the local geographic area (LGA) level. We compared area-level indicators in comorbidities hotspots to those in the rest of Alberta (non-hotspots). Results Among 359,638 OA cases in 2013, approximately 60% of people resided in Metro and Urban areas, compared to 2% in Rural Remote areas. All comorbidity groups exhibited statistically significant spatial autocorrelation (hypertension: Moran’s I index 0.24, z score 4.61). Comorbidity hotspots, except depression, were located primarily in Rural and Rural Remote areas. Depression was more prevalent in Metro (Edmonton-Abbottsfield: 194 cases per 1000 population, 95%CI 192–195) and Urban LGAs (Lethbridge-North: 169, 95%CI 168–171) compared to Rural areas (Fox Creek: 65, 95%CI 63–68). Comorbidities hotspots included a higher percentage of First Nations or Inuit people. People with OA living in hotspots had lower socioeconomic status and less access to care compared to non-hotspots. Conclusions The findings highlight notable rural-urban disparities in comorbidities prevalence among people with OA in Alberta, Canada. Our study provides valuable evidence for policy and decision makers to design programs that ensure patients with OA receive optimal health management tailored to their local needs and a reduction in current OA health disparities.
... Tableau 1 : Nombre de cas et incidence cumulative des cas de coqueluche confirmés pour les zones de gestion de la santé et le Sud-Ouest, déclarés en Nouvelle-Écosse entre le 1 er janvier et le 31 Toutefois, la statistique de l'incidence mobile a permis de contextualiser l'activité de la coqueluche et l'analyse résultante a démontré que l'activité dans la zone du Centre était généralement plus faible et plus sporadique, mais sans présenter les grands pics observés dans les autres zones. ...
... The small number of cases registered annually inevitably led to wide confidence intervals, In their study on cancer registries, Takiar and colleagues argued -that when the population of the registry is around 150,000 then the incidence rate of even 10 should be viewed with reservations‖ ( [33], p660). The patient population of the Intego database has about this order of magnitude. ...
Article
Full-text available
Background The number of newly diagnosed gonorrhoea and syphilis cases has increased in Flanders in recent years. Our aim was to investigate, to which extent these diagnoses were registered by general practitioners (GPs), and to examine opportunities and limits of the Intego database in this regard. Methods Data from a retrospective cohort study based on the Flemish Intego general practice database was analyzed for the years 2009–2013. Case definitions were applied. Due to small case numbers obtained, cases were pooled and averaged over the observation period. Frequencies were compared with those calculated from figures of mandatory notification. Results A total of 91 gonorrhoea and 23 syphilis cases were registered. The average Intego annual frequency of gonorrhoea cases obtained was 11.9 (95% Poisson confidence interval (CI) 9.6; 14.7) per 100,000 population, and for syphilis 3.0 (CI 1.9; 4.5), respectively, while mandatory notification was calculated at 14.0 (CI: 13.6, 14.4) and 7.0 (CI: 6.7, 7.3), respectively. Conclusion In spite of limitations such as small numbers and different case definitions, comparison with mandatory notification suggests that the GP was involved in the large majority of gonorrhoea cases, while the majority of new syphilis cases did not come to the knowledge of the GP.
... The limitations of epidemiological studies on GBC are small sample sizes and problems in quantifying exposure to putative risk factors. 12,13 Other studies relate to prevalence of GST. 14 Gallstones are not the only factor. A 'syndromic approach' was adopted to investigate risk factors in a high-prevalence population. ...
Article
Full-text available
A high prevalence of gallbladder diseases (GBD) in Northern India warranted a population survey into environmental risk factors. In 60 villages of Uttar Pradesh and Bihar from 13 334 households, 22 861 persons aged >30 years were interviewed for symptoms of GBD, diet and environmental factors. Subsequently ultrasonography (US) was performed in 5100 and 1448 people with and without symptoms, respectively. Heavy metal and pesticide content in soil and water were estimated. US revealed a prevalence of GBD of 6.20%. GBD was more common in 5100 persons with symptoms (7.12%) compared with 1448 without (2.99%) (P < 0.05). Adjusted odds ratio (ORs) [95% confidence interval (CI)] revealed a significantly increased risk of GBD in females >50, 1.703 (CI 1.292-2.245); multiparity 1.862 (CI 1.306-2.655) and a genetic history 1.564 (CI 1.049-2.334). An increased risk noted in males with diabetes was 4.271 (CI 2.130-8.566), chickpea consumption 2.546 (CI 1.563-4.146) and drinking unsafe water 3.835 (CI 2.368-6.209). Prevalence of gallstones was 4.15%; more in females 5.59% than males 1.99% (P < 0.05). Cluster analysis identified a positive correlation of nickel, cadmium and chromium in water with a high prevalence of GBD in adjacent villages in Vaishali district, Bihar. A high risk of GBD was observed in older, multiparous women and men with diabetes, intake of chickpeas, unsafe water and villages with heavy metal water pollution.
Article
Choropleth mapping continues to be a dominant mapping technique despite suffering from the Modifiable Areal Unit Problem (MAUP), which may distort disease risk patterns when different administrative units are used. Spatially adaptive filters (SAF) are one mapping technique that can address the MAUP, but the limitations and accuracy of spatially adaptive filters are not well tested. Our work examines these limitations by using varying levels of data aggregation using a case study of geocoded breast cancer screening data and a synthetic georeferenced population dataset that allows us to calculate SAFs at the individual-level. Data were grouped into four administrative boundaries (i.e., county, Zip Code Tabulated Areas, census tracts, and census blocks) and compared to individual-level data (control). Correlation assessed the similarity of SAFs, and map algebra calculated error maps compared to control. This work describes how pre-aggregation affects the level of spatial detail, map patterns, and over and under-prediction.
  • Dm Parkin
  • Whelan Sl
  • J Ferlay
  • Thomas Db
Parkin DM, Whelan Sl, Ferlay J, Teppo and Thomas DB. 2002: Cancer Incidence in Five Continents Vol. VIII-IARC Press, Lyon, France.