Quick question: in this data, did you include patients who went to hospitals outside of HPN after seeing physicians in HPN? (I.e. patients went to a HPN physician office, for example, in the end of Y1 and in the beginning of Y2 and then went to the hospital outside of HPN)

I.e., for example, patients went to a Cleveland Clinic for a complicated heart surgery after seeing HPN physicians in Y1 and Y2. Are those patient included in the database?

Moreover, what if these patients continued going to the non-HPN inpatient hospitals in Y3 and Y4 (i.e. HospitalDays were outside of the HPN network)?

Local California patients could potentially go to Kaiser or other California hospitals after seeing physician form HPN.

If there is a Leakage in the database (i.e. patients do go to the hospitals outside of the database), is it possible to know the size of the leakage (i.e. % of cases when patients go to hospitals outside of HPN)?

Thank you for the clarification,

Yours truly,

Igor

Although I don't want to imply that I know the answer here, I think hospital leakage isn't a problem.

We're looking at claims data, so even if a person goes to a hospital in Cleveland, that hospital will still want to get paid. When they issue the bill to HPN, the claim & hospital days paid by HPN/IPA is logged. Therefore, I'm assuming that the claim would still be in the database.

From HPN's website, it didn't sound like they actually owned hospitals in CA, so the vast majority of hospital days are "leakage" outside their system. Let me know if this helps.

Thank you!

If I am not mistaken, HPN is not an insurance company, they do not insure patients and they do not pay for the claims to the providers.

HPN is an IDN (integrated delivery network) that include Hospitals/facilities that are part of HPN network (which includes either associated members HPS does not own but provides services to, partially owned facilities or even completely owned facilities, similar to a Partners network in Massachusetts, that owns Mass General & Brigham & Women’s Hospital) .

Probably, HPN has internal CDM software system (charge data master) that processes billing information for all HPN owned and affiliated facilities. If patient goes outside of the HPN facility, CDM does not capture that patient.

I believe, when patient goes to Cleveland Clinic, the patient's insurance company (United, Aetna, Cigna, Medicare, Medicaid, etc.) pays for the claim and HPN has not record of this transaction.

If the vast majority of hospitals a leakage, then we might have a problem - what if more than 50% of the patients that go to HPN outpatient facilities are admitted to non-HPN hospitals ... then "Houston (Baikonur), we have a problem..." .... 

So whether HPN is the direct insurer or not, I think they are taking capitation risk (maybe I'm wrong, I am just looking at their website). Thereby whether they are the intermediary that pays the bill (or not), they would be liable for the costs that a patient incurred in Cleveland. This is actually pretty different from Partners where most patients are not enrolled in HMO/capitation style plans. Actually the differences may be even bigger when you try to compare. Partners is very hospitcal centric style IDN (i.e. they own MGH), whereas Heritage is a physician group practice that has hospital affiliations.

Will be interested to hear what others think...

http://www.chcf.org/~/media/Files/PDF/C/PDF%20CHCMarketReport2006.pdf

Heritage as a limited health plan HMO (see pg 15)?

Thank you! Very interesting document!

I checked the 2009-2010 lists of all Healthcare Insurance Plans in US, and HPN is not there (there is only Heritage IL from United HC).

It seems to me that HPN used to be an HMO in 90'th and till 2005 (as in the report) but they do not offer an HMO anymore but rather became an IDN.

I can check older lists of US Healthcare Insurance Plans to see when HPN stopped offering HMO healthcare insurance.

@rudychev, received an answer from HPN on this. A patient who visits a clinic outside the network should be captured in this dataset. Of course, as Jeremy keeps re-iterating, there is always a disconnect between reality and the contents of a database.

Dear Anthony,

Thank you!

It will definitely increase the quality of the predictive model.

 

Yours truly,

Igor