Cost of Schizophrenia in the Medicare Program

Introduction

Schizophrenia is estimated to affect approximately 1% of the adult population in the United States with similar rates among males and females and across all races. ^1,2,3,4,5 Onset of the disease generally occurs during late adolescence and young adulthood (ages 16–30), and symptoms (eg, hallucinations, thought disorder) often interfere with an individual's ability to live independently. Medications, both injectable and oral, can be effective in managing symptoms, but there is no known cure. Health care costs are generally included in estimates for seriously mentally ill adults in government statistics but are not specifically identified for people with schizophrenia. ^1,2

Most of the very few available studies of health care costs for adults with schizophrenia are based on samples of administrative data taken from Medicaid and other claims data within a single state, and may only include Medicare patients older than age 65 or dual eligible patients (ie, with both Medicaid and Medicare). ^6,7,8,9,10 A few national studies also were conducted using survey data from the Medical Expenditure Panel Survey, and the Medicare Current Beneficiary Survey. ^11,12 Studies based on survey data are limited by the difficulty of surveying people with schizophrenia, and face issues of representativeness and reliability associated with self-reported information about a condition that impairs communication and potentially limits survey access for many people with schizophrenia.

Given current federal budget and general economic pressures, the Medicare program is considering a wide range of reforms designed to lower costs and improve coordination of care, with a focus on beneficiaries with chronic illnesses and high-cost conditions. All of the relevant literature suggests that schizophrenia is a high-cost chronic condition. Understanding the demographic characteristics, co-occurring conditions, and health care utilization patterns for this population informs program and policy design efforts to reform delivery systems and create payment policies that incent the desired reforms. This retrospective observational study describes Medicare beneficiaries with schizophrenia (MBSs) demographics, co-occurring diagnoses, and associated health care costs using nationally representative data from Medicare claims for fee-for-service beneficiaries and makes comparisons to the demographics and costs of the general Medicare beneficiary population (GMB).

Methods

MBSs were selected from the Medicare Standard Analytic Files (SAF) for a 5% fee-for-service beneficiary sample between 2003 and 2007, based on a minimum of 2 physician or hospital claims with schizophrenia diagnoses (International Classification of Diseases, Ninth Revision [ICD-9] codes 295.xx) in 4 consecutive calendar quarters. Beneficiaries were excluded if they had any health maintenance organization coverage (which means the SAFs will not contain all their claims data); if they had fewer than 4 calendar quarters of data after the quarter containing the first schizophrenia diagnosis; or if 75% or more of their claims with schizophrenia diagnoses had schizoaffective diagnoses.

The sample is nationally representative based on use of 2 digits of beneficiary social security numbers and includes 2.5–3 million beneficiaries in any 1 year. The SAFs include final action claims for all services paid by Medicare except for Part D pharmaceutical, including physician, outpatient hospital and clinic, inpatient hospital, skilled nursing facility, home health, hospice, durable medical equipment, and Part B drugs delivered incident to physician services. Medicare Part D was implemented in 2006, and data have not been publicly released in a form compatible to the SAF. A separate file includes demographic information about the beneficiary, including age, race, reason for Medicare eligibility, county and state of residence, and date of death, among other variables. Different SAFs are linked across services and years using an encrypted beneficiary identifier.

The study population was followed through calendar year 2008 claims to describe health care resource utilization patterns and the associated costs of care. A separate analysis was conducted using 5% sample SAFs from 2001 through 2009 to estimate the proportion of the fee-for-service Medicare population accounted for by beneficiaries with schizophrenia using the same study criteria to identify MBSs in each year's data. Costs and counts of GMBs also were calculated for each calendar year. These data were scaled to a national estimate by multiplying by 20, and total Medicare payments and numbers of patients were estimated for each year. Mean per patient per year costs of care were calculated for each year in nominal and standardized 2010 dollars.

Medicare costs are inclusive of Medicare payments as well as the allowed beneficiary share of payments. Beneficiary share of payments vary by type of service and sometimes are covered by Medicaid or by supplemental insurance, though the provider may not collect the full amount. Outlier payments are included in hospital inpatient and outpatient claims. Outlier payments are made for some Medicare services when the costs of care exceed a regulatory threshold, above which the Medicare program and the provider share the cost by a fixed percentage. Payments associated with the study population were all standardized to 2010 dollars by repricing services in 2003 through 2008 using 2010 rates or applying annual updates based on final rate setting regulations for each year.

In the Medicare SAFs, dates of service have been replaced by a calendar quarter prior to 2009. As a result, services cannot be sequenced within a calendar quarter. Per patient year calculations are based on the total patient quarters beginning with the first schizophrenia diagnosis observed in the study period until the patient either died or the claims available for the study period ended in quarter 4 of 2008. In this study, schizophrenia patients were grouped based on the duration for which claims were available. Per patient per year Medicare payments for all Medicare covered services except prescription drugs were calculated for each patient cohort based on duration and for patients who died during the study period compared to those who did not die.

The demographics of the schizophrenia population were compared to those of the general Medicare population during the same study period. Payments for services for this population were compared to payments for all Medicare fee-for-service patients, excluding those enrolled in Medicare Advantage for 2009, the latest year for which data were available. Because the age distribution of schizophrenia patients is so different from the age distribution of the overall Medicare population, mortality rates and age were not age adjusted.

Hospitalizations were analyzed making distinctions between psychiatric and medical hospitalizations. Psychiatric hospitalizations were defined by Medicare provider number coding and diagnosis-related groups (DRGs) into 3 types: psychiatric hospital, psychiatric unit in a short-term acute care hospital, and medical bed in a short-term acute care hospital with a psychiatric DRG. Any hospitalizations that did not meet these criteria were considered “medical.”

Categories of comorbid conditions were developed by creating a frequency distribution of the ICD-9 codes on claims for beneficiaries in the sample, and then matching ranges of ICD-9 codes to related conditions that appear in this distribution. For medical conditions, if the patient had at least 2 diagnoses with codes in the category, then the patient was counted in that category once; for co-occurring psychiatric conditions, if the patient had at least 1 diagnosis in the category, then the patient was counted in that category once. The categories of related conditions with higher frequencies in the study population are described in the results.

Results

Discussion

The study observed that the demographics for MBSs were different than the GMB population. MBSs were more likely to be male, younger than age 65, and originally eligible for Medicare based on disability. The literature on disability and Medicare indicates that mental disorders make up a significant proportion of those who qualify for Social Security disability and eventually become eligible for Medicare. ¹³ With a mean age that is less than that of the general Medicare population (61.3 for women, and 51.5 for men, compared with 70.2), schizophrenia patients enter the Medicare program before most Medicare beneficiaries and remain in the program longer. Medicare patients with schizophrenia represented close to 1.5% of the Medicare fee-for-service population from 2001 through 2009. Given the young age at which adults with schizophrenia enter the Medicare program compared to those qualifying based on age, MBS patterns of service utilization and associated costs may represent a longer duration of overall costs to Medicare.

The symptoms of schizophrenia, when not controlled with medication, can themselves interfere with a patient's ability and willingness to seek appropriate care. Many people with schizophrenia may become detached from family support systems, live in substandard housing, or have transient lifestyles that make it difficult to maintain continuity of care. Most will have low incomes. ¹³ Because the primary treatment for schizophrenia is prescription medication, patients need to have regular contact with psychiatrists, mental health facilities, and/or primary care clinics/physicians if they are to have access to those medications. Patients may function adequately when on medication, but decompensate when medication is interrupted or not available for some reason, often causing disruption to their relationships and living conditions. Periods of time when medication is not available for any reason may lead to situations that result in emergency room visits, psychiatric hospitalization, and medical hospitalizations related to increased vulnerability to infection, injury, and disease associated with transient living conditions. Several studies show significant numbers of adults not receiving ambulatory care services or not receiving medications for their schizophrenia. ^9,14

In instances when patients are treated for psychiatric emergencies in hospital emergency rooms, they may be admitted to medical beds outside of psychiatric units either because no such unit is available or because no bed is available in a specialized unit. Some of these patients are ultimately transferred into psychiatric units or to psychiatric hospitals. Frequent medical hospitalizations for MBSs reflect a vulnerability to health crises that suggests that managing the MBS population to control costs and improve quality will require attention to both coordination and continuity of psychiatric and physical health care services.

Medicare patients with schizophrenia may be less severely ill than schizophrenia patients eligible for Medicaid programs in the states, in that they were functioning at a high enough level at some point in their adult lives to work long enough to qualify for Social Security disability. There is no national database from which to make comparable estimates of the costs of health care for MBSs, and state programs vary widely from each other and over time within the same programs. Many of the Medicare beneficiaries whose experience was studied also may be enrolled in state Medicaid programs. Costs associated with Medicaid program services and other local public mental health services are not included in this study, though in most cases, Medicare would be the primary payer for most health care services for dually eligible patients. Coverage for health care services and access to those services may have varied widely across public and commercial plans and for those intermittently in jails or prisons over the study period, making it impossible to generalize from study findings to the broader adult population living with schizophrenia.

In the current policy environment, the Medicare program is subject to numerous reform proposals to contain costs and improve quality, including efforts to bundle care, limit rehospitalization, develop episode-based payment for chronic conditions, or encourage better coordination of care through development of “medical homes.” Findings from this study indicate that the MBS population costs Medicare significantly more per patient-year than other Medicare beneficiaries on average, and that the bulk of those high costs are related to a combination of psychiatric and medical hospitalization. Although other studies ^11,8 reported much lower hospitalization rates and costs than are seen in the present study, clearly for Medicare beneficiaries, hospitalization costs dominate medical costs and any improvement in hospitalization rates could have a significant impact on overall costs to Medicare. In this study, hospitalization costs accounted for more than 50% of all health care costs, and are concentrated in a subset of the study population, indicating that the highest cost patients could potentially be identified and targeted for intervention to lower hospitalization rates, decrease costs, and improve their quality of life.

The costs of care for Medicare patients with schizophrenia were concentrated in hospital utilization, which was highest among about 30% of MBSs. Further research that examines this high-cost cohort of patients and looks at patterns of physician care outside the hospital in association with hospitalization may suggest how best to identify patients at highest risk for hospitalization. Additional studies would be needed to evaluate how much of the study population actually receives regular psychiatric or primary care or both. Analysis of any association between frequency of contact with psychiatrists/mental health providers and psychiatric hospitalization would be important to understand how much hospitalization may be attributable to lack of care to manage the disease. Analysis of any association between frequency of contact with primary care and other physicians and medical hospitalization would be important to understand the extent to which medical hospitalizations may be caused by lack of routine primary care.

Further research describing patterns of care for Medicare patients may uncover opportunities for policy intervention to improve access to care and decrease costs to the Medicare program. These data may be useful in current federal and state demonstrations to improve primary care services and to better coordinate care for dual eligible Medicare and Medicaid patients, as well as for planning bundled and/or episodic payment strategies for Medicare patients with chronic diseases.