Hi Sergi, Table attached. Outbreak 1 was almost certainly increased in apparent magnitude due to an influenza outbreak, see http://www.hcaf.biz/2013/PHE_BJHCM.pdf

Can you do an example calculation? Much appreciated.

These outbreaks can be rather savage, but completely missed by current disease surveillance methodologies due to their assumptions about how disease outbreaks should occur.

Cheers Rod

When I say the outbreaks can be a bit savage, see attached

Perhaps, you can try the Noether test for cyclical trend here.

Hi Guys, Much appreciated. Alas you must speak as to a child. I use Poisson statistics a lot (and use Monte Carlo Poisson simulations to get results that are probably available via statistical tests), but have limited knowledge beyond that, most of my statistics is now too rusty to use.

Obviously the biggest finding is the presence of the outbreaks, the secondary finding may be around a large outbreak preceding a small one in each location - I assume that Sergi's map gives results from the Noether test? B < 0 means?

Hi Guys (and anyone else with an interest), Can you do me a huge favour and ask your wider circle of contacts if they can investigate this phenomena.

I have done the preliminary work identifying its existence but I need others to take the work forward. Deaths are the easiest to analyse because the data is widely available. A long time series of monthly deaths is all that is required (can be split by diagnosis if required). Most of my work can be downloaded from www.hcaf.biz (mainly in the Emergency admissions folder). A couple of  extra papers attached. The book chapter requires you to enter the code 299174 to open a printable version (I purchased the right to distribute).

These events are huge. I demonstrated infectious-like spread of increased medical admissions across a major Australian city during the 2008 outbreak but the Health Board (who supplied the data with restriction on use) was opposed to publication of the findings.

Hence I know these events have a profound effect on human health and requires urgent investigation.

CMV is merely the working hypothesis.

So any assistance on that front would be hugely appreciated. Cheers  Rod

P.S. I gather from the earlier reply that strong high/low behaviour is probably not systematic but a result of randomness.

Ta. The final bit appears to be the problem. Authorities showing no interest, at least here in the UK this is the case.

They range from 1991 to present. 

http://www.hcaf.biz/2014/EandW_spread.pdf

I think you can apply tests for serial correlation to data shown on Fig. 8 to test your hypothesis.

Hi Rodney

It seems that you are dealing With time to event outcome, so in such kind of data a Kaplan Meir plot is recomended 

Hope to be of help

Regards

Ramon 

Could it be possible that the Global Peace Index is itself influenced by these unrecognised outbreaks

Hello Rodney,

I have observed something similar in times series data for infectious disease outbreaks from municipal water supplies. You can use tests for autocorrelation that examine how related one week or months observation is correlated with those + or - n weeks or months from that point. A correlogram can then be used to visualize if s significant increase is followed by a significant decrease and if the data are long enough you can observed these trends repeating in waveforms over months, years, or decades. If you have covariates you can also include them and create s bivariate correlogram to determine the cross correlation between the variables and outcome over time.

Dear Thomas, Many thanks for the reply. Probably a little more complex as there seem to be interactions with influenza, however, seemingly a new type of outbreak. See attached. Cheers Rod