The great thing about compliance stuff like HIPPA/FERPA/PCI, etc. is
that they're so open to interpretation. My general rule when dealing
with any compliance issue is to make sure my university general
counsel's office is happy with whatever I do. Yes, it can be a pain to
do that, but you'll be better off in the long run. If they're happy and
a legal issue comes up, it's their problem, not mine, which is generally
where I want to be when it comes to compliance issues....

On Thu, Jun 05, 2014 at 12:07:56AM +0000, Sam Kome wrote:
> I'm not up on HIPPA and I am not a lawyer.   
> Years ago I created a system for anonymizing address data that passed muster 
> with the FCC and US Census bureau. In a nutshell we had a third party create 
> a unique hash to identify the record, and geocode to the US Census block 
> group.  
> We never handled let alone stored the name or address ourselves.  We had an 
> independent auditor audit our outsource party and our datasets. Block group 
> is the US Census standard for protecting privacy - it really depends on what 
> other data you retain though as to being able to reconstruct identity.
> 
> Cheers!
> 
> -----Original Message-----
> From: Code for Libraries [mailto:[email protected]] On Behalf Of Simon 
> Spero
> Sent: Monday, June 02, 2014 2:38 PM
> To: [email protected]
> Subject: Re: [CODE4LIB] Anonymizing address data
> 
> This book might be useful (it's a year old)
> 
> Anonymizing Health Data <http://shop.oreilly.com/product/0636920029229.do>
> Case Studies and Methods to Get You Started
> By Khaled El Emam, Luk Arbuckle
> <http://shop.oreilly.com/product/0636920029229.do#tab_03_0>
> Publisher: O'Reilly Media
> Released: December 2013
> Pages: 212
> 
> 
> 
> 
> 
> On Mon, Jun 2, 2014 at 3:40 PM, Kyle Banerjee <[email protected]>
> wrote:
> 
> > HIPPA compliant data cannot include personally identifiable information, a
> > category which includes address. The "safe harbor" approach where
> > geographic subdivisions smaller than states cannot be used frequently
> > renders data useless.
> >
> > The "expert determination" method is always an option and precompiling can
> > work in certain cases, but I was wondering what other methods people have
> > successfully employed? Thanks,
> >
> > kyle
> >

-- 
Thomas L. Kula <[email protected]>
Senior Systems Engineeer, Unix Systems Group
Library Information Technology Office
Columbia University in the City of New York

Reply via email to