Experimental Determination of Excess Lung Cancers in Pulp Mill School Districts Versus Control No Mill Districts
By: Jorma Jyrkkanen. Sept 1990.
I addressed the simple null (Ho) hypothesis that living in a pulp mill town would have no impact on the incidence of lung cancer (LC) mortality. The research hypothesis (Hi) is that yes there is an effect air pollutants and that it would be elevated leading me to select a one-tailed test.
Major assumptions are that the populations being drawn upon are similar in demographics and life-styles, and that sampling is random, and that there are no other major sources of lung carcinogens. These assumptions are unexplored and should be examined in greater depth if one wishes to improve certitude.
I therefore decided to use the parametric Students t-test with alpha=.05 (the risk that chance alone could cause the observed effect in 5% of cases) and 1-tailed, unequal sample size model. Data (Tables 1, 2) were obtained from Gallagher et. al. 1986. Occupational mortality in BC.. Health and Welfare Canada. BC Research.
Table 1. LC DEATHS AMONG MALES AND FEMALES AMONG TOTAL DEATHS WHERE THERE ARE NO PULP MILLS IN THE SCHOOL DISTRICT
| No Mills in School District | Male LC deaths | Total Male deaths | Female LC deaths | Total Female deaths | ||||
| Nelson | 112 | 2882 | 33 | 1938 | ||||
| QCI | 15 | 445 | 4 | 234 | ||||
| Merrit | 38 | 922 | 9 | 535 | ||||
| Smithers | 29 | 797 | 7 | 448 | ||||
| Terrace | 56 | 2530 | 8 | 1049 | ||||
| Stikine | 5 | 246 | 0 | 122 | ||||
| Keremeos | 21 | 551 | 2 | 260 | ||||
| Summerland | 46 | 981 | 15 | 688 | ||||
| Central Okanag. | 326 | 6702 | 89 | 4415 | ||||
| Vernon | 168 | 3957 | 49 | 2693 | ||||
| Penticton | 147 | 3259 | 40 | 2153 | ||||
| Delta | 164 | 2871 | 64 | 2133 | ||||
| Mission | 86 | 2016 | 25 | 1183 | ||||
| N=13 | ||||||||
Table 2. LC DEATHS AMONG MALES AND FEMALES AMONG TOTAL DEATHS WHERE THERE ARE PULP MILLS IN THE SCHOOL DISTRICT
| Yes Mills in School District | Male LC deaths | Total Male deaths | Female LC deaths | Total Female deaths |
| Prince George | 230 | 4874 | 63 | 2226 |
| Castlegar | 47 | 1126 | 9 | 744 |
| Prince Rupert | 74 | 2164 | 12 | 1091 |
| Campbell River | 98 | 1836 | 37 | 999 |
| Kamloops | 250 | 5669 | 68 | 3090 |
| Port Alberni | 119 | 2959 | 21 | 1718 |
| Harmac | 359 | 6327 | 105 | 4313 |
| Powell River | 110 | 1919 | 31 | 1214 |
| Howe Sound | 50 | 862 | 16 | 443 |
| N=9 |
Analysis
Table 3. Statistical comparisons of percentage Lung Cancer mortality between mill school districts and non-mills school districts for males.
MALE LC MORTALITY
| N | No Mills LC% | Mills LC% |
| X2 | X1 | |
| 1 | 3.88619 | 4.781917 |
| 2 | 3.370787 | 4.174067 |
| 3 | 4.121475 | 3.419593 |
| 4 | 3.638645 | 5.337691 |
| 5 | 2.213439 | 4.409949 |
| 6 | 2.03252 | 4.021629 |
| 7 | 3.811252 | 5.674095 |
| 8 | 4.689093 | 5.732152 |
| 9 | 4.86422 | 5.800464 |
| 10 | 4.245641 | |
| 11 | 4.510586 | |
| 12 | 5.712295 | |
| 13 | 4.265873 | |
| Mean | 3.950924 | 4.80984 |
| SE | 0.27927 | 0.28829 |
| Median | 4.121475 | 4.718917 |
| SD | 1.006922 | 0.864871 |
| Var | 1.013891 | .748 |
| CL | 0.688477 | 0.665 |
The results of the Student’s-t test appear in Table 4 below.
Table 4. Results of Student’s-t test
| Students t-test assuming Var1=Var2 | (Using Microsoft Software in Excel) | |
| Mean | 3.956319 | 4.80984 |
| Var | 1.105651 | 0.85656 |
| Pooled Variance | 1.007604 | |
| Hypothetical mean diff | 0 | |
| Calculated t-statistic | 1.88771 | Significant |
| P(T<=t), one-tailed | 0.037648 | |
| t-critical, one-tailed | 1.734063 | |
| P(T<=t), two-tailed | 0.075296 | |
| t-critical, two-tailed | 2.100924 |
Conclusion
Since the calculated t 1.88771 falls within the region of rejection, exceeding the critical tabular value 1.734063 (Table 4.), the null hypothesis must be rejected, that is Ho is rejected and therefore the research hypothesis (Hi) must be accepted, implying that there is an elevated lung cancer mortality amongst males in the Mill school districts (.025
<.05).
To conclude, the Mill districts exhibited statistically significantly higher lung cancers in males (4.81% Lung Cancer mortality) compared to No Mill districts (3.95%LC mort) (P(T<=t), one-tailed = 0.037648) as I had predicted. This test alone is not proof however of causality but is highly suggestive. This is especially so when considered along with the other evidence I have examined (see Fenni-excerpts in jormabio/archives).
When the analysis was conducted for females, the same pattern was found for females though the magnitude of the effect was about half.
Warning! When I did this cursory analysis for BONE CANCER, I found that it was very highly linked to pulp mill school districts and should be investigated for specific causal agent.
© 2010 Jorma Jyrkkanen. All rights reserved.
Tags: cancer, pulp mills, males, females, British Columbia
Tags: British Columbia, cancer, females, Jorma Jyrkkanen, males, pulp mills
