About this Journal

Article Abstract

Online ISSN: 1099-176X    Print ISSN: 1091-4358
The Journal of Mental Health Policy and Economics
Volume 8, Issue 1, 2005. Pages: 15-28

Published Online: 30 March 2005

Copyright © 2005 ICMPE.


Variation in Patient Routine Costliness in U.S. Psychiatric Facilities

Jerry Cromwell1* Edward M. Drozd,1 Barbara Gage,1 Jan Maier,2 Erin Richter,3 Howard H. Goldman4

1Ph.D., Research Triangle Institute, Waltham, MA, USA
2RN, M.P.H., Research Triangle Institute, Waltham, MA, USA
3B.A. Research Triangle Institute, Waltham, MA, USA
4M.D., University of Maryland, Baltimore, MD, USA

* Correspondence to: Jerry Cromwell, RTI, 411 Waverley Oaks Road, Suite 330, Waltham, Massachusetts, 02452-8414, USA.
Tel.: +1-781-7888100 x 131
Fax: +1-781-788 8101
E-mail: jcromwell@rti.org

Source of Funding: This paper was funded in part by the Centers for Medicare & Medicaid Services (CMS) under Contract No. 500-95-0058, TO =//13. The statements contained in this paper are solely those of the authors and do not necessarily reflect the views or policies of CMS. The authors assume the responsibility for the accuracy and completeness of the information contained in this paper.


Medicare’s recent per diem prospective payment system for psychiatric inpatients uses claims-based costing methods based on existing administrative data. Because of data limitations, payment levels in this system cannot account for patient-specific routine costs (thereby compressing relative weights) nor consider key patient characteristics for higher payment. This study tests for compression in the new payment weights using a per diem cost measure augmenting cost reports and claims with primary data on daily patient and staff times from 40 psychiatric facilities (4,149 Medicare days) nationwide. These data include patient characteristics (Activities of Daily Living (ADL) deficits and dangerous behavior) unavailable in existing administrative data. Using a hierarchical 16-group case mix classification system based on diagnosis, age, psychiatric and medical severity, ADLs, dangerousness, and electroconvulsive therapy use, per diem cost regression models are estimated with alternative daily cost measures. Ignoring variation in patient daily resource intensity compresses relative weights for high-cost groups by a factor of 2 or more. Patient characteristics unavailable in administrative data were necessary for identifying the highest-cost groups. Ignoring these further compresses relative costs.  


Background: The Balanced Budget Refinement Act of 1999 included a Congressional mandate to develop a patient-level case mix prospective payment system (PPS) for all Medicare beneficiaries treated in PPS-exempt psychiatric facilities. Payment levels by case mix category have been proposed by the government based on claims and facility cost reports. Because of claims data limitations, these levels do not account for patient-specific staffing costs within a facility's routine units, nor are certain key patient characteristics considered for higher payment.

Aims of the Study: This study uses novel primary data to quantify heretofore unmeasured differences in daily staffing intensity on routine units among Medicare patients. The data are used to test for compression (or narrowing) in case mix payment weights that would result from using only Medicare claims and facility cost reports to quantify daily routine costliness.

Methods: Primary data on patient and staff times in over 20 activities were collected from 40 psychiatric facilities and 66 psychiatric units, nation-wide. Patient times were reported on all inpatients on each shift over a 7-day study period. A resource intensity measure (in Registered Nurse (RN)-equivalent minutes) was constructed on a daily basis for 4,149 Medicare and 4,667 non-Medicare patient days. The routine measure is converted into daily cost using cost report per diems and ancillary costs added using submitted claims. Descriptive tables isolate key cost drivers for Medicare patients. Classification and Regression Trees (CART) clustering identifies 16 potential case mix groups. Multivariate regression is used to compare case mix, day-of-stay, and facility effects using 4 alternative measures of daily routine and ancillary costs.

Results: Patient daily routine intensity of care is found to vary by a factor of 3 or more between the top and bottom 10% of days. Medicare patient days were 12.5% more staff intensive than non-Medicare days, which may have been due to age and other differences. Older dementia and `‡`residual diagnosis'‡' patients are more intensive while schizophrenia and substance-related patients are less intensive. Age, psychiatric and medical severity, deficits in Activities in Daily Living (ADLs), dangerous behaviors, and electroconvulsive therapy (ECT) also contribute substantially to higher staffing intensity. Other patient characteristics were insignificant within broad diagnostic groups. Routine costs based on a single facility per diem produced narrower case mix cost differences - often by a factor of 2 or more - for 10 of 12 groups with significantly higher costs. Adding patient-specific ancillary to uniform per diem costs only marginally decompressed costs. Day of-stay costs were similarly compressed when using only cost reports.

Discussion: Claims-based costing using Medicare cost reports unduly compresses (narrows) estimates of inter-group case mix cost differences. Also, by not capturing ADL deficits and dangerous behaviors, administrative data sets fail to identify small, but very resource intensive, patient groups. ECT treatment regimens, although rare, significantly increase costs on a daily basis.

Implications for Health Policies: Medicare's recently proposed prospective payment system for psychiatric inpatients uses claims-based costing methods based on widely available administrative data. Consequently, fewer high cost groups are identified due to non-reported patient characteristics such as ADL deficits. Moreover, inter-group relative cost differences are likely understated. It is also possible that any standardized dollar amount applied to group relative weights is understated because Medicare patients appear more intensive per day on routine units.

Implications for Future Research: Larger primary samples of special psychiatric units (e.g., med-psych, child/adolescent) could improve estimates of daily routine costliness. Larger samples could also support stronger tests of case mix and cost differences by facility type and teaching status. Medical records information on non-Medicare patients could quantify any systematic differences in average daily costs holding case mix constant. Similar primary studies of psychiatric patients treated outside PPS-exempt units in acute general hospitals could result in a fully integrated payment system for all mentally ill Medicare patients, thereby avoiding payment inefficiencies and inequities.

Received 8 October 2004; accepted 8 January 2005

Copyright © 2005 ICMPE