Inpatient Claims Explored

Posted by Ashay Thakur on July 31, 2012

Identifying inpatient claims in healthcare data is centerpiece to most Healthcare data analysis. Inpatient claims is one of the more expensive claims over all claim groups. Therefore, identification of Inpatient claims in claims data is vital. Unfortunately, there is no standard algorithm out there to identify Inpatient admission claims. They are rarely alluded to as such in the data in a direct manner.

We, at Deerwalk, have worked out logics to define how we go about performing in-patient identification in a dataset. The simplest case for such identification is when admission days are present in the data; the most difficult case would be when procedures need to be used; and in cases when both admission days and procedures are missing, Inpatient Days needs to be calculated from blocks of contiguous claims. We apply this logic in our reporting and analysis tool Makalu.

We discuss the case of identification through Room and Board charges in this article.

In many instances, hospitals use revenue codes instead of admit and discharge codes. If revenue codes are given, Room and Board revenue codes (100-169, 200-229, 720-722, 800-804, 987) need to be checked for. In the claim lines with Room and Board revenue codes, if service units are given, the service units can be counted as Inpatient Days.

However, service units in all the lines should not be counted as Inpatient days indiscriminately, as some lines may just as well be change in rooms. In such cases, if the service date difference between the first Room and Board code and the next is less than 30 days, and the provider is the same, multiple Room and Board codes can be considered as one admission. Thus the sum of those service units gives Inpatient days.
If service units do not represent Inpatient days, blocks with admit related claims need to be defined. In the block, first date can be taken as admit date and last date as the discharge date. We need to limit these claims to POS 21, 51 or 61, as long as they are given.

Suppose we do have room and board charges. In such a case, we take a block of claims with the Room and Board codes as follows.

Member ID Claim ID Claim Line ID Procedure Code Service Units Service From Date
123654798798 7431031291795 1 360 10 2010-10-26
123654798798 7431031291795 2 252 10 2010-11-05
123654798798 7431031291795 3 122 5 2010-11-15
123654798798 7431031291796 30 252 5 2010-12-28
123654798798 7431031291796 36 122 10 2011-01-02
123654798798 7431031291796 37 761 15 2011-01-17

In the table above, there are two claim blocks. The first block starts from the line with service date 2010/10/26 to the line with service date 2010/11/15, as the differences in service dates in adjacent lines are less than 30.

The second block starts with the service date 2010/12/28. Note that the service date difference between the start of second block and the end of first block is more than 30 days, hence the delineation. The second block ends in the last line.

The total Inpatient days in this case is 55 – the sum of all service units.

The application of Inpatient logic helps us with Quality metrics and Utilization metrics.