Misleading Conclusion in Taiwan Apartment Dwellers Study
In 1982, a number of buildings were constructed in Taiwan using steel unknowingly contaminated with Co60, exposing thousands of residents to low dose radiation for a number of years. The health of this group has been monitored and the results have been presented in a couple of papers (Hwang et al., 2006; Hwang et al., 2008).
On reading these abstracts, it becomes clear that radiation dose of ~50 mSv increases the risk of cancer. Let us now examine the data presented in these papers to see if such a conclusion is supported by the data.
The paper (Hwang et al., 2006) compared the cancer incidence among the radiated residents of Taiwan apartments with an equivalent matched cohort from Taiwan and determined the Standardized Incidence Ratios (SIRs) for cancer. The SIRs for the exposed population are summarized in Table III on page 854 of the paper.
Whereas the most significant result (with most statistics) was that the overall cancer rate (or solid cancer rate) was significantly lower in the radiated population, the abstract does not even mention it! By slicing the data into smaller subgroups (with poorer statistics), the authors pointed out in the abstract whenever the cancer rates were elevated.
In the followup paper (Hwang et al 2008), SIRs are no longer calculated and shown, but the LNT model is assumed, and a proportional hazards model is used to calculate excess risk of cancer from the low dose radiation. Let us now analyze the results in some detail to see if the observed reduction of cancers in Hwang 2006 paper continues in the data of Hwang 2008 paper.
The average age at the end of the followup period was therefore: 17.2 + 16.1 = 33.3
In Page 144 of the followup paper (Hwang et al., 2008) it says:
In three more years (end of 2002 to end of 2005), average age of the cohort would be 36.3.
Let us now estimate the Expected Cancer incidence in the cohort after 3 additional years of follow up.
To do this, cancer incidence numbers for age groups in 5 year intervals was obtained from Taiwan Cancer Registry:
The average of male and female cancer incidence on the 20 to 45 range was fitted to an exponential function of age, and the cancer incidence for age of 33.3 and 36.3 was calculated, using the fit.
Cancer incidence rates per 100,000 in Taiwan, using above table from Taiwan cancer registry:  
Age Group

20

25

30

35

40

45

Mean Age

22.5

27.5

32.5

37.5

42.5

47.5

All Sites Male

21

32.7

65.8

117

196

307

All Sites, Female

24.4

46.3

86.1

150

232

322

Average Cancer Rate

22.7

39.5

76.0

134

214

314

Using Fit to Power function

21.0

43.3

79.0

132

208

311

From the fit, the cancer incidence at 33.3 and 36.3 was calculated:
Cancer Incidence at age:
 
Age

Cancer Incidence
 
33.3

86.3
 
36.3

117.7

Expected factor increase in 3 years in cancer incidence = 117.7/86.3 = 1.36
All Cancers expected was 114.9 in Hwang 2006 paper (See Table III above)
All Cancers expected 3 years later would be 114.9*1.36 = 156
Page 145 of followup paper (Hwang et al., 2008) says:
All Cancers observed 3 years later was 128 (in Hwang 2008 paper)
Observed/Expected (128/158): 0.82
Thus, there is a 18% reduction in cancers observed in the followup paper
The reduced overall cancer incidence continues, but this information is not mentioned in the paper, as they did not repeat the SIR calculation for cancers. On the other hand, they used proportional hazard (LNT) model to estimate excess cancer risk, as described on page 144 of the Hwang 2008 paper.
The paper's conclusions are misleading in depicting the carcinogenic effects of low dose radiation.
References: