A Statistical Breast Cancer Cluster?

[BillyJoe]

http://www.theage.com.au/news/natio...-scare-building/2006/12/21/1166290663237.html

Excerpts:

The ABC building in the Brisbane suburb of Toowong will be abandoned without knowing the cause of a high incidence of breast cancer cases among women staff members.

A five month investigation by an independent review panel was conducted after 12 women were diagnosed with breast cancer in the past 11 years.

The panel has been able to answer one of the two key questions we put to it - namely is this a significant atypical spike in the incidence of breast cancer?
The answer to that question is yes.
The panel has been unable to answer the second question, which is what has been the cause of this high incidence of breast cancer?

"But on the basis of the panel's findings regarding the unusually high incidence of breast cancer among Toowong staff, the ABC has decided to relocate staff away from the site," he said.

A television report stated that the chance of 12 women developing breast cancer over an 11 year period, considering the ages of the women and the presence/absence of other risk factors, is about 1 in a million.

Is this likely to be just a statistical cluster (like the expected run of 8 heads in a 100 coin flip)?
Without finding an underlying cause, should they be abandoning the building?
If the twelve women now sue, will they have a case?

BillyJoe

 

[a.real.girl]

You know, I understand being better safe than sorry, especially as a corporate parent.

However, I'm not sure the "television report" is an accurate assessment of odds. There are many factors not even mentioned, to include type of cancer (BC is not a fully descriptive term,) history of cancers in the affected population and their families, overall rate of BC in the town...

As for your questions... it's likely that if the demography is considered, this is just a statistical cluster. I work in clinical research, including BC, and do not know of an environmental trigger that affects folks regardless of other factors.

Keep in mind I'm no expert, but I think it's a reactionary little story, unless there's more data they've just not published.

 

[ponderingturtle]

Also would they have a real case even if it was caused by the enviroment? Wouldn't the company have had to know about it, or have a reasonable expectation to know about it to be liable?

Not that it really matters in a jury trial, but it does seem that as they do not even have a mechanism to show, how can they be responcible for it with out that?

 

[a.real.girl]

I guess they could sue the building owners, no?

 

[casebro]

I wouldn't think that an "11 times higher spike" would be out of the ordinary for a random cluster. Billions of women, hundreds of thousands get breast cancer. Million to one odds of a 12 patient cluster seem right in line to me.

 

[ponderingturtle]

I guess they could sue the building owners, no?

How? I think it would have to be the result of a known risk, or a risk that they should have known about. If it was an unknown danger how can you be responcible for not preventing it?

 

[BillyJoe]

You know, I understand being better safe than sorry, especially as a corporate parent.

I guess so, but, along the way, are we promoting ignorance about statistics and the expected occurence of clusters? Isn't it like assuming an underlying cause for the cluster of 8 consecutive heads in a 100 coin flip, rather than explaining it as a statistical cluster?

However, I'm not sure the "television report" is an accurate assessment of odds. There are many factors not even mentioned, to include type of cancer (BC is not a fully descriptive term,) history of cancers in the affected population and their families, overall rate of BC in the town....

This was mentioned as alluded to in my original post. I don't have acopy of the original report, but apparently all confounding factors were taken into account. I guess experts will be dissecting the report in due course and we may even hear some dissenting voices. I hope so. I'd like to see this nutted out.

As for your questions... it's likely that if the demography is considered, this is just a statistical cluster. I work in clinical research, including BC, and do not know of an environmental trigger that affects folks regardless of other factors.

Yes, interesting. So you would be pretty doubtful of there being a connection. But you would pull out just to be on the safe side? This is not a criticism really but I don't really understand that fully.

regards,
BillyJoe

 

[BillyJoe]

I wouldn't think that an "11 times higher spike" would be out of the ordinary for a random cluster. Billions of women, hundreds of thousands get breast cancer. Million to one odds of a 12 patient cluster seem right in line to me.

I wonder if there is a sort of graph or something that could put this into perspective. How unusual really is this cluster of 12 cases over 11 years amongst women working in that news office.

 

[BillyJoe]

I think it would have to be the result of a known risk, or a risk that they should have known about. If it was an unknown danger how can you be responcible for not preventing it

And if the danger is unknown is there actually a danger at all?
The women were pretty emotional. I wouldn't be surprised if legal action is taken. I just hope they have an expert panel to nut out the statistical issues. A jury aint gonna understand it and will just give these poor women the money. But science and reason will be the real losers if it's nothing but a statistical cluster.

 

[ponderingturtle]

And if the danger is unknown is there actually a danger at all?

Mabey, things are found to be dangerous that where not thought to be on occasion. The issue is did they know it was dangerous and act in an inapropriate maner.

The women were pretty emotional. I wouldn't be surprised if legal action is taken. I just hope they have an expert panel to nut out the statistical issues. A jury aint gonna understand it and will just give these poor women the money. But science and reason will be the real losers if it's nothing but a statistical cluster.

Quite possibly.

 

[a.real.girl]

But you would pull out just to be on the safe side? This is not a criticism really but I don't really understand that fully.

Sorry, I was unclear. I was simply saying that I could understand the position of the management. This might be the path of least resistance, make them look concerned most of all with the employee's comfort and safety, etc. Indeed, it's possible this was just a great reason to get out of a bad lease.

Me personally? I would not move based on this info. In fact, with environmental factors from the office seemingly ruled out, I would work that angle instead: "We confirm this office is a safe workplace" sort of thing

 

[BillyJoe]

Sorry, I was unclear. I was simply saying that I could understand the position of the management. This might be the path of least resistance, make them look concerned most of all with the employee's comfort and safety, etc.

I still think it's wrong.

They must stick with the facts:
- Yes, it is a 1 in a million chance, but, given the number of women and the number of office blocks, a cluster like this is going to happen somewhere. It happened here.
(Assuming this is the case of course - I would like to see this confirmed first)
- And, we have also looked for causes in the workplace and none were found. The workplace has been found to be safe. There is no reason to move.

Indeed, it's possible this was just a great reason to get out of a bad lease.

I'm not sure if the office is leased but what about the equipment?

Me personally? I would not move based on this info. In fact, with environmental factors from the office seemingly ruled out, I would work that angle instead: "We confirm this office is a safe workplace" sort of thing

Yes, if the facts of the matter support that conclusion, they should stick with this through to the end. If the women with the cancer and all those who work with them refuse to return, they can then move whilst still denying the actual need to. That should continue to stick with the facts of the case: The office is safe. This is a just a random cluster. We are moving because we understand how you feel, but there still really is no need to move.
(Again, provided those are the facts)

 

[Skeptic Ginger]

The first thing I do when reading a report like this is to find a non-news source for the information.

Here is the Queensland health department report on the investigation. While 12 cases were reported only 8 met the criteria for being included. There was no environmental cause identified. They will reopen the case if anything new develops. So at this time, it is an aberrant cluster. With 8 cases over a short period of time, the cause may never be revealed.

BTW, there are two threads on this topic. I put this post in both.

 

[BillyJoe]

The first thing I do when reading a report like this is to find a non-news source for the information.

Here is the Queensland health department report on the investigation.

Thanks for that. I am reading it.

While 12 cases were reported only 8 met the criteria for being included. There was no environmental cause identified. They will reopen the case if anything new develops. So at this time, it is an aberrant cluster. With 8 cases over a short period of time, the cause may never be revealed.

Here are the conclusions of the report (I haven't read the whole lot yet)

Eight breast cancer cases met the case definition for inclusion in the analysis where a case was defined as a female employed at ABC studios Toowong in the period 1995 to 2005 inclusive, who self reported breast cancer during that period of time and whose diagnosis could be verified on Queensland Health records to the satisfaction of the investigation team.​

After a series of meetings, interviews, site visits, and statistical analyses, the investigators concluded that:​

• The occurrence of eight cases of proven breast cancer in a small workplace over this time period raises the likelihood that this is a statistical excess of cases.

• Applying standard epidemiological techniques where the workplace per se rather than aspecific environmental factor is considered the risk, proved to be difficult, as it was not possible to fully identify and accurately quantify the exposed population. A definitive, valid statistical conclusion about the expected number of cases in the workplace over the period could not be drawn from the data.​

• Therefore, in this instance, where the epidemiological investigation was unable to definitively answer the initial question related to excess of cases an environmental appraisal of the workplace was undertaken.​

• The subsequent environmental appraisal did not indicate any plausible exposures that would explain a significantly elevated number of breast cancers.

Considering the two questions posed by concerned ABC staff, this investigation has shown that theapplication of standard epidemiological investigations does not answer the question of whether there are more cases of breast cancer than would be expected.

However based on environmental assessments of the workplace, in combination with current knowledge of causes of breast cancer,there is no evidence to support the hypothesis that there is an environmental exposure at the workplace that could account for the concentration of breast cancer cases.​

​

 

[BillyJoe]

Quite a different picture form the one portrayed in the media.

Eight breast cancer cases met the case definition for inclusion in the analysis

8 cases not 12

A definitive, valid statistical conclusion about the expected number of cases in the workplace over the period could not be drawn from the data.

So, where did they get the 1 in a million chance from?

the epidemiological investigation was unable to definitively answer the initial question related to excess of cases

Again, where did the 1 in a million chance come from.

the application of standard epidemiological investigations does not answer the question of whether there are more cases of breast cancer than would be expected.

Ditto.

---------------------------------------------------------------------

On the other hand, what does this mean:

The occurrence of eight cases of proven breast cancer in a small workplace over this time period raises the likelihood that this is a statistical excess of cases

What is a "staistical excess of cases"?
Is a 'statistical excess of cases" equivalent to "a statistical cluster"?

Perhaps I should read on...

 

[BillyJoe]

Notice how the conclusion in the report differs markedly from that contained in the media report;

The official report:

...a definitive, valid statistical conclusion about the expected number of cases in the workplace over the period could not be drawn from the data...​

...the epidemiological investigation was unable to definitively answer the initial question related to excess of cases...​

...the application of standard epidemiological investigations does not answer the question of whether there are more cases of breast cancer than would be expected...​

...the results of the various analyses showed that statistically robust epidemiological conclusions were not feasible. This is because even minor changes in estimates of the size of the workplace population had major impacts on the epidemiological results....​

...it is very difficult to fully identify and accurately quantify the exposed population and thus draw a valid conclusion with respect to the expected number of cases in the workforce population....​

...in this instance, an epidemiological investigation does not provide a valid answer to the initial question of whether the incidence of breast cancer was higher than expected...

No less than six times they state that they could not draw a valid statistical conclusion about whether or not there were more cases of breast cancer then would be expected.​

The media report:

...the panel has been able to answer one of the two key questions we put to it - namely is this a significant atypical spike in the incidence of breast cancer. The answer to this question is yes...

...on the basis of the panel's findings regarding the unusually high incidence of breast cancer among Toowong staff, the ABC has decided to relocate staff away from the site...

Six times NO must equal YES!

 

[a.real.girl]

Using the logic that two wrongs make a right, the media report is right three times over! How 'bout them stats!

I don't think that logic is correct, however.... it's three lefts make a right, right?

 

[Skeptic Ginger]

If you look at some of the other links and you go to the Queensland Health Department link there is a lot of additional information on breast cancer rates. A statistical excess case is one that is above the expected rate. In other words, the background rate of cancer serves as the control.


Now I'm getting confused between the two threads. I just posted some links to a 62 page report on this and a bunch of additional material. I'll edit in a link to my post in a sec.

There's a much longer discussion here.