Ulster Med J 2007; 76 (2) 83-87 lung cancer patterns in younger age-cohorts in Ireland Recent

83

Paper

Recent lung cancer patterns in younger age-cohorts in Ireland
1,2

Zubair Kabir, 1Gregory N Connolly, 2Luke Clancy

Accepted 6th February 2007 

ABSTRACT Background:  Smoking  causes  85%  of  all  lung  cancers  in  males  and  70%  in  females. Therefore,  birth  cohort  analysis  and  annual-percent-changes (APC) in age-specific lung cancer mortality rates, particularly in the youngest age cohorts, can explain  the beneficial impacts of both past and recent anti-smoking interventions.  Methods: A long-term time-trend analysis (1958-2002) in lung cancer mortality rates focusing on the youngest age-cohorts  (30-49 years of age) in particular was investigated in Ireland. The rates were standardised to the World Standard Population.  Lung cancer mortality data were downloaded from the WHO Cancer Mortality Database to estimate APCs in death rates, using  the Joinpoint regression (version 3.0) program. A simple age-cohort modelling (log-linear Poisson model) was also done, using  SAS software. Results: The youngest birth cohorts (born after 1965) have almost one-fourth lower lung cancer risk relative to those born around  the First World War. A more than 50% relative decline in death rates among those between 35 and 39 years of age was observed  in both sexes in recent years. The youngest age-cohorts (30-39 years of age) in males also showed a significant decrease in  death rates in 1998-2002 by more than 3% every five years from 1958-1962 onwards. However, death rate declines in females  are slower. Conclusions: The youngest birth cohorts had the lowest lung cancer risk and also showed a significant decreasing lung cancer  death rate in the most recent years. Such temporal patterns indicate the beneficial impacts of both recent and past tobacco control  efforts in Ireland. However, the decline in younger female cohorts is slower. A comprehensive national tobacco control program  enforced on evidence-based policies elsewhere can further accelerate a decline in death rates, especially among the younger  generations. Key words: Birth cohort; Lung cancer; Ireland; Smoking Ban INTRODUCTION Lung  cancer  is  currently  the  most  common  cancer  in  the  world1,  accounting  for  almost  1,500  deaths  annually  in  the  Republic  of  Ireland  alone2.  It  has  been  argued  that  both  cigarette  consumption  rates  and  smoking  prevalence  data  are  necessary  to  explain  a  tobacco  epidemic,  especially  when using lung cancer death rate as an index of smokingattributable  mortality 3.  However,  historical  smoking  history  data,  such  as  annual  age  and  sex-specific  cigarette  consumption  rates,  are  not  available  nationwide  for  many  countries4. Also, it has been difficult to quantify the benefits of  large scale, preventive interventions. Therefore, it is necessary  to  have  alternative  approaches  to  explaining  the  beneficial  impacts of both recent and past tobacco control efforts.  Because  85%  of  male  lung  cancer  deaths  are  attributed  to  tobacco  smoking,  any  decline  or  deceleration  in  the  observed lung cancer death rates could be attributed to antismoking  interventions  in  the  past5.  Ireland  does  not  have  a  comprehensive  tobacco  control  program  but  pockets  of  tobacco control efforts were in place over the past 40 years  or  so.  Lung  cancer  trends  in  young  adults  (30-39  years  of  age  in  particular)  have  been  used  as  an  early  indicator  of  progress  in  tobacco  control6,  and  therefore  any  observed  decline among the youngest age-cohorts would indicate the  beneficial impacts of more recent anti-smoking activities. In  addition,  the  relative  change  in  lung  cancer  mortality  rates  between successive time-periods would also signal the need  for additional aggressive anti-smoking strategies. However, in  Ireland, it is premature to use age-specific lung cancer death  rates  to  monitor  the  early  consequences  of  the  nationwide  workplace smoking ban that was only introduced in March  20047. This  study  estimates  the  annual-percent-changes  in  lung  cancer mortality rates from 1958 to 2002 using the Joinpoint  regression  model  (version  3.0)  of  the  US  National  Cancer  Institute’s  Surveillance,  Epidemiology  and  End  Results  (SEER)  program8,  with  special  emphasis  on  the  youngest  age-cohorts (between 30 and 49 years of age). A simple agecohort modelling was also performed to explain the temporal  patterns.
1  Harvard School of Public Health,  Division of Public Health Practice,  401 Park Drive, Landmark Center, 3rd Floor (East),  Boston, MA 02215,  USA. 2  Research Institute for a Tobacco Free Society,  The Digital Depot, The  Digital Hub. Thomas Street, Dublin 8. Ireland. Correspondence to: Dr Zubair Kabir E: zkabir@hsph.harvard.edu

©  The Ulster Medical Society, 2007.

www.ums.ac.uk

20. Farelly MC, Chaloupka FJ, Pechacek TF. The impact of tobacco control program expenditures on aggregate cigarette sales: 1981-1998. NBER Working Paper No. 8691. National Bureau of Economic Research; 2001. 21. Escario JJ, Molina JA. Will a special tax on tobacco reduce lung cancer mortality? Evidence for EU countries. Appl Econ 2004;36(15):1717-22. Figure 1: Age standardised (world population standard) lung cancer death rates in the Republic of Ireland (0-85 + age groups) for both sexes, 1958-2002

84

The Ulster Medical Journal
cohorts, we employed a simple age-cohort modelling  technique. Log-linear Poisson regression modelling  (with an offset) was employed to estimate the effects  across each birth cohorts adjusting for age relative  to  the  youngest  cohorts  for  both  sexes,  using  the  GENMOD  procedure  in  SAS  software  (version  8.0). However, the use of classical age-period-cohort  (APC)  modelling  techniques  could  have  improved  the “fit” of the model (albeit at the expense of extra  degrees  of  freedom),  but  the  random  variation  associated  with  parameter  estimates  might  lead  to  erratic predictions10. The classical APC models are  also limited with the “non-identifiability” problem.  In addition, lung cancer temporal studies consistently  show an age-cohort phenomenon rather than an ageperiod phenomenon in several APC model studies. 

60

50

Rates/100,000 persons

40

30

Males Females

20

For  continuous  changes  in  lung  cancer  death  rates  across  different  time-periods,  log-linear  Poisson  regression models were used to calculate APC and  joinpoint analyses has been extensively used recently  0 1963 1968 1973 1978 1983 1993 1998 for  estimating  such  temporal  effects11,12.  Because  Figure1958Relative Risk (RR) estimates of Lung Cancer deaths in the 1988 2. Republic of Ireland across different birth-cohorts. Calendar Year the  focus  of  this  study  is  on  younger  age-cohorts,  we  employed  jointpoint  regression  analyses  for  Fig 1. Age standardised (world population standard) lung cancer death  estimating temporality only for age-groups between  rates in the Republic of Ireland (0-85 + age groups) for both sexes,  30 and 49 years. Fewer lung cancer deaths per year  1958-2002 for  each  of  these  younger  age-groups  necessitated  to  collapse  every  5-calendar  year  periods  into  an  6 average age-standardised lung cancer death rate for  each of the 5-year age-groups studied (30-34, 35-39,  5 5 40-44, 45-49 years of age).
10
Risks Relative to 1963-1968 4

3 Female 2 Male

1

0

Central year of birth

Fig 2. Relative Risk (RR) estimates of Lung Cancer deaths in the  Republic of Ireland across different birth-cohorts. METHODS Lung  cancer  mortality  data  from  1958  to  2002  were  downloaded from the WHO Cancer Mortality Database9. Agesex specific adjusted lung cancer death rates standardised to  the World Population are also available from the WHO Cancer  Mortality Database9. Age-specific population estimates for the  periods studied were obtained from the Irish Central Statistics  Office website (www.cso.ie). We looked at age-specific lung cancer death rates across the  year  of  birth.  In  other  words,  a  ‘synthetic’  birth  cohort  for  each age group was created based on the year and age of death  of  each  individual,  using  5-year  age  and  5-year  calendarperiod  intervals.  Each  birth  cohort  could  be  identified  by  the central year in the interval. To look at trends across birth 

In brief, the Joinpoint8 analysis fits a series of joined  straight lines on a log scale to the age-specific and  age-standardised  lung  cancer  death  rates.  Line  segments are joined at points called joinpoints. Each  joinpoint  denotes  a  statistically  significant  change  in trend. In joinpoint analysis, the best-fitting points  are  the  years  of  death  that  change  significantly  (increasing  or  decreasing  trends).  The  analysis  starts with the minimum number of joinpoints, and  tests whether one or more joinpoints are statistically  significant  and  should  be  added  to  the  model.  A  maximum  of  three  joinpoints  can  be  added  to  the  final  model.  Because  of  collapsed  5-year  calendar  periods from 1958 to 2002 and not using the single  calendar year death rates for lung cancer trends, the  joinpoint analysis could only test a maximum of two  joinpoints for this particular study design. 

18 90 18 95 19 90 19 05 19 10 19 15 19 20 19 25 19 30 19 35 19 40 19 45 19 50 19 55 19 60

RESULTS Figure 1 shows the peaking of male lung cancer death rates in  the late eighties and the beginning of stabilisation in female  death rates when calendar periods were considered. When we  looked at the effects across birth cohorts, males born ten years  before the First World War had the highest lung cancer risk  relative to the youngest cohorts, and females born around the  First World War had the highest risk of dying from lung cancer  (Figure 2, Table I). While females had a greater risk relative  6 to  the  youngest  cohorts  when  compared  with  males’  lung  cancer risk, those born around and after the Second World  War showed a consistent decline in lung cancer risks, with  little gender variations (Figure 2). So, those born after 1965 

©  The Ulster Medical Society, 2007.

www.ums.ac.uk

Recent lung cancer patterns in younger age-cohorts in Ireland
Table I. Relative Risk (RR) estimates with 95% Confidence Intervals (CI) of Lung Cancer deaths in the Republic of Ireland across different birth-cohorts. Birth-Cohorts Males Females RR (95%CI) 3.78 (1.29, 11.08) 3.69 (1.37, 9.99) 4.65 (1.84, 11.76) 4.95 (2.09, 11.75) 5.20 (2.33, 11.61) 5.44 (2.59, 11.43) 4.85 (2.45, 9.63) 4.57 (2.43, 8.57) 3.89 (2.18, 6.92) 2.65 (1.57, 4.49) 2.19 (1.36, 3.54) 2.07 (1.33, 3.23) 1.55 (1.02, 2.37) 0.79 (0.45, 1.38) * 0.86 (0.54, 1.37) * Reference (RR=1) Females   1958-1962 (Rates)    1998-2002 (Rates)    Relative change   0.7  0.7  2.1  1.0  3.7  6.7  Males   1958-1962 (Rates)    1998-2002 (Rates)    Relative change  2.8  0.9  -68%  7.0  1.3  -81%  13.1  5.2  -60%  Table II.

85

Relative change in lung cancer death rates/100,000 among younger age-cohorts between two five-year time-periods in the Republic of Ireland Age-Groups 30-34 35-39 40-44 45-49

Central year of birth RR (95% CI) 1888-1892  1893-1897 1898-1902 1903-1907 1908-1912 1913-1917 1918-1922 1923-1927 1928-1932 1933-1937 1938-1942 1943-1947 1948-1952 1953-1957 1958-1962 1963-1967 1.42 (0.64, 3.13) *  1.96 (0.94, 4.09) *  2.91 (1.47, 5.81)  3.33 (1.75, 6.37)  3.58 (1.96, 6.58)  3.61 (2.05, 6.38)  3.60 (2.12, 6.13)  3.54 (2.16, 5.81)  3.25 (2.05, 5.16)  2.72 (1.76, 4.19)  2.36 (1.58, 3.53)  2.14 (1.45, 3.15)  1.56 (1.06, 2.28)  0.99 (0.60, 1.60) *  1.40 (0.93, 2.10) *  Reference (RR=1)

32.7 16.9 -48%

9.2 11.0

No change  -52%  +81%  +20%

experiencing  a  rise  in  the  5-year  death  rate,  and  again  the  findings are not statistically significant (Table III).  DISCUSSION In Ireland, the overall lung cancer mortality rates from 1958  to 2002 shows a favourable trend for both sexes, especially  among  the  youngest  cohorts.  This  is  consistent  with  the  recent  lung  cancer  incidence  pattern2,  and  also  with  the  decreasing smoking prevalence in the relatively young adults4.  The  youngest  birth  cohorts  not  only  had  the  lowest  lung  cancer  risk  but  also  showed  significant  decreasing  rates  in  lung cancer death rates in most recent years. Such temporal  patterns  indicate  the  beneficial  impacts  of  both  the  recent  and the past anti-smoking interventions in Ireland. However,  a slower relative decreasing rate among the youngest female  age-cohorts  identifies  the  need  for  additional  and  more  aggressive  tobacco  control  efforts  targeting  at  specific  population groups.   A recent study in Ireland reported a fall in teenage smoking  prevalence  from  20%  in  1995  to  13%  in  200313.  However  it is too soon to estimate the effects on lung cancer rates of  Table III. APC (Annual Percent Changes) with 95% CI (Confidence Intervals) of Lung Cancer Death Rates and Joinpoint Analysis among younger age-cohorts in the Republic of Ireland for both sexes, 1958-2002 Age-Groups 30-34  35-39  40-44  45-49  APC (95% CI) Males -3.7 (-6.7; -0.7)  -3.2 (-4.5; -1.8)  -2.4 (-3.6; -1.1)  -1.4 (-2.8; -0.2)  APC (95% CI) Females -0.6 (-2.4; 1.3) -1.2 (-3.0; 0.7) 0.5 (-1.4; 2.4) 0.2 (-1.0; 1.3)

* Not Statistically Significant did have the lowest lung cancer risk in both sexes. In figures 3 and 4, the age-specific lung cancer death rates  were  relatively  high  among  males  across  all  age-cohorts  (same age groups across different calendar periods of birth),  but were highest among the oldest age-cohorts (80-84 year  olds) for both sexes. Not only lung cancer death rates are low  among the youngest age-cohorts, but those between 30 and  39 years of age in males are also showing a dramatic decline  in death rates across successive cohorts. There has been more  than 80% relative decline in death rates in males between 35  and 39 years of age from 1958 to 2002 (Table II). The females  also have shown a 50% relative decline in death rates among  the same age cohorts (Table II).  When joinpoint modelling was performed for the relatively  young age-cohorts (30-49 years of age), a significant five-year  decline was observed among the males in particular (Table  III).  For  example,  those  male  cohorts  between  30  and  34  years of age had a 3.7% decline every five years from 1958  to 2002, and those between 35 and 39 years of age also had a  significant decline in lung cancer death rates by 3.2% every  five years. The females of the same age cohorts (30-39 years  of age) did show a downward trend in lung cancer death rates  but the findings were not statistically significant (Table III).    In contrast, the female cohorts above 40 years of age were 

©  The Ulster Medical Society, 2007.

www.ums.ac.uk

86

The Ulster Medical Journal
Figure 4. Female age-specific standardised lung cancer death rates across different birth-cohorts in Ireland.

Figure 3. Male age-specific standardised lung cancer death rates across different birth-cohorts in Ireland.

Fig3. Male age-specific standardised lung  cancer death rates across different birthcohorts in Ireland. high teen smoking initiation for the decade before 1995. Non  cancer health gains are also evident following the nationwide  workplace  smoking  ban14,15.  Therefore,  further  health  and  social gains are realistically achievable if target populations,  especially  women,  lower  socio-economic  groups  and  the  youngest  adults,  are  empowered  and  provided  an  enabling  environment. For  further  gains  from  the  ill-health  effects  of  smoking,  especially on the younger generations, an accelerated decline  in smoking prevalence and an increase in smoking quitting  rates  are  essential.  The  Irish  government  is  committed  to  a Tobacco  Free  Society16.  However,  for  a  faster  decline  in  lung cancer rates, a comprehensive tobacco control program  similar to the State of California that showed a 6% decline in  lung cancer incidence within a decade has to be enforced17.  With similar programs in Massachusetts in 199318, smoking  prevalence  in  youths  declined  from  36%  in  1995  to  30%  in 1999 and from 17% in 1993 to 10% in 2000 in pregnant  women19.  Even  smoking  quit  rates  increased  from  18%  in  1993 to 26% in 200219. In addition to smoke-free policies,  both these states also exercised a regular increase in cigarette  price. Evidence shows that a 10% increase in cigarette price  can have a 4% decline in cigarette consumption rates and a  1-2% decrease in smoking prevalence in the developed world,  particularly among youths20. Interestingly, a 10% increase in  tax will also reduce lung cancer mortality rate by 1.2% in the  first year21.
7

Fig 4. Female age-specific standardised lung cancer death  rates across different birth-cohorts in Ireland. In conclusion, current lung cancer death rates in Ireland are  encouraging  but  an  accelerated  further  annual  decline  is  also realistically achievable in both sexes, especially among  the  younger  generations,  if  evidence-based  policies  are  introduced. Youths  are  price-sensitive  and  a  10%  increase  in cigarette price would allow 40,000 Irish smokers to quit  smoking16, and this would save thousands of productive life  years lost due to tobacco-related premature deaths in Ireland.  Future  monitoring  of  the  nationwide  workplace  smoking  ban should assess trends in lung cancer death rates in young  adults’ once long-term lung cancer mortality data are available  post ban. ACKNOwLEDGEMENTS The Royal City of Dublin Hospital (RCDH) Research Trust  funded Dr Zubair Kabir (ZK) to undertake his PhD thesis in  Trinity College Dublin. Currently, ZK is on a cancer research  fellowship  to  the  Harvard  School  of  Public  Health  jointly  funded by the US National Cancer Institute and the Health  Research Board of Ireland. We thank Dr Kathleen Bennett for  statistical advice on age-cohort modelling. Conflict of interest – none declared
REFERENCES 1.   2.   Stewart BW, Kleihues P, editors. World Cancer Report. Lyon: International  Agency for Research on Cancer, World Health Organization; 2003. National Cancer Registry of Ireland (NCRI). Cancer in Ireland, 1994 to 2002: Incidence, mortality, treatment and survival. Cork: National  Cancer Registry of Ireland, 2003.
8

©  The Ulster Medical Society, 2007.

www.ums.ac.uk

Recent lung cancer patterns in younger age-cohorts in Ireland
3.   4.   Lopez AD, Collishaw NE, Piha T. A descriptive model of the cigarette  epidemic in developed countries. Tob Control 1994;3:242-7. Forey  B,  Hamling  J,  Lee  P, Wald  N,  editors.  International Smoking Statistics: a collection of historical data from 30 economically developed countries. 2nd ed.  London: Oxford University Press: 2002. Thun MJ, Jemal A. How much of the decrease in cancer death rates in  the United States is attributable to reductions in tobacco smoking? Tob Control 2006;15:345-7. Jemal A, Cokkinides VE, Shafey O, Thun MJ. Lung cancer trends in  young adults: an early indicator of progress in tobacco control (United  States). Cancer Causes Control 2003;14(6):579-85. K a b i r   Z ,   C l a n c y   L .   P r e v e n t i v e   o n c o l o g y.   L a n c e t  2004;363(9422):1737. Kim HJ, Fay MP, Feuer EJ, Midthune DN. Permutation tests for joinpoint  regression with applications to cancer rates. Stat Med 2000;19(3):33551. Cancer Mondial [homepage on the internet]. Cancer Mortality Database.  Lyon, France: International Agency for Research in Cancer, World Health  Organization. Available  from:  http://www-dep.iarc.fr/  Last  accessed  February 2007. 

87

13.   Goodman P, O’Sullivan A, Manning P, Clancy L. Active smoking and  respiratory symptoms in young Irish teenagers 1995-2003. (Abstract)  Ir J Med Sci 2004;173(Suppl 2):16-7. 14.   Allwright S, Paul G, Greiner B, Mullally B, Pursell L, Kelly A, et al.  Legislation for smoke-free workplaces  and health of bar workers in  Ireland: before and after study. BMJ 2005;331(7525):1117. 15.   Goodman P, Agnew M, McCaffrey M, Paul G, Clancy L. Effects of the  Irish smoking ban on respiratory health of bar workers and air quality  in Dublin pubs. Am J Respir Crit Care Med 2007; Jan 4. [Epub ahead of  print]. Available from: http://ajrccm.atsjournals.org/cgi/reprint/20060810850Cv1. Last accessed February 2007. 16.   Tobacco Research Institute. [homepage on the Internet]. Dublin: Research  Institute for a Tobacco Free Socity (RIFTFS); 2007. Available from:  http://www.tri.ie. Last accessed February 2007. 17.   Barnoya  J,  Glantz  S.  Association  of  the  California  tobacco  control  program with declines in lung cancer incidence. Cancer Causes Control  2004;15(7):689-95.       18.   Koh H, Judge CM, Robbins H, Celebucki CC, Walker DK, Connolly  GN. The first decade of the Massachusetts Tobacco Control Program.  Public Health Rep 2005;120(5):482-95. 19.  Hamilton  WL,  Rodger  CN,  Chen  X,  Njobe  TL,  Kling  R,  Norton  GS.  Independent evaluation of the Massachusetts Tobacco Control Program: eighth annual report January 1994-June 2001.  Prepared for: Massachusetts Department of Public Health. Cambridge, MA: Abt  Associates Inc.; 2003. 20.   Farelly MC, Chaloupka FJ, Pechacek TF. The impact of tobacco control  program expenditures on aggregate cigarette sales: 1981-1998. NBER Working Paper No. 8691.  National  Bureau  of  Economic  Research;  2001. 21.   Escario JJ, Molina JA. Will a special tax on tobacco reduce lung cancer  mortality? Evidence for EU countries. Appl Econ 2004;36(15):171722.

5.  

6.  

7.  8.  

9.  

10.   Dyba T, Hakulinen T. Comparison of different approaches to incidence  prediction  based  on  simple  interpolation  techniques.  Stat Med  2000;19:1741-52. 11.   Weir HK, Thun MJ, Hankey BF, Ries LA, Howe HL, Wingo PA, et al.  Annual report to the nation on the status of cancer, 1975-2000, featuring  the uses of surveillance data for cancer prevention and control. J Natl Cancer Inst 2003;95(17):1276-99. 12.   Cayuela  A,  Rodriguez-Dominguez  S,  Lopez-Campos  JL,  OteroCandelera R, Rodriguez-Matutes C. Joinpoint regression analysis of  lung cancer mortality, Andalusia 1975-2000. Ann Oncol 2004;15(5):7936.

©  The Ulster Medical Society, 2007.

www.ums.ac.uk