By Alan Shafer, PhD
Send correspondence to: Alan Shafer, Texas Health and Human Services Commission, Behavioral Health Services, Office of Decision Support, 909 West 45th Street, Austin, TX 78751; email: Alan.Shafer@hhsc.state.tx.us.
Authors' Note: The views and opinions expressed in this paper are those of the author and do not represent any official view or policy of any government, university, or business.
The primary purpose of the study was to examine the role public sector payers (Medicare and Medicaid) and providers (Texas state mental health hospitals) play in psychiatric hospitalization, using Texas annual hospital discharge files from 1999 to 2010. Psychiatric hospitalization, as defined by a primary behavioral health diagnosis (ICD-9 diagnoses 290-314.99) averaged 146,876 discharges per year, approximately 5.24% of all hospitalizations in the state. Children younger than 18 years accounted for 27,035 discharges per year. The top 4 diagnostic groups were depression (29%), bipolar disorder (22%), schizophrenia (18%), and alcohol-drug disorders (14%). More patients with schizophrenia or other psychotic disorders were served by the public sector, while more patients with depression or alcohol-drug disorders were served by private insurance. Interestingly, patients with bipolar disorder were distributed relatively evenly across both payment groups and ages. Length of stay decreased from 10.5 days in 1999 to 8.1 days in 2010. Most psychiatric discharges (69%) were served by a small group of 42 large psychiatric hospitals.
Although several previous studies have examined national trends and patterns in psychiatric hospitalization, no studies have examined trends in Texas. This study used annual inpatient hospital discharge public use data files1 to examine psychiatric hospitalization across the first 12 years of the Texas hospital discharge file program. Analysis of hospital discharge files provides a valuable public health assessment by capturing most of the population rather than survey samples, thereby yielding useful data for determining incidence rates, health care costs, and utilization rates that can inform policy and planning.
The study reported here has 3 major objectives: to describe the overall patterns and characteristics of psychiatric hospitalization in Texas; to examine the role that the public safety net plays in overall psychiatric hospitalization by examining 3 groups (providers as represented by public mental health state hospitals and payers as represented by Medicaid and Medicare); and to explore differences in demographics, diagnoses, and utilization based on these 3 patient groups and private insurance and other payers.
Although their analyses were different, both Mechanic et al.2 and Blader3 used the National Hospital Discharge Survey to examine psychiatric inpatient care covering a 20-year time frame from 1988 to 1994 (biannual)2 and 1996 to 2007 (annual).3 For adults, discharges per 100,000 persons increased by about 25% from the high 700s and low 800s in the late 1980s and early 1990s, to the high 900s and low 1000s in the early years after 2000 and appear to have stabilized around that point through 2007 (Mechanic used all adults, while Blader divided adults into those younger than 65 years and those older than 65 years). Lengths of stay for adults decreased dramatically, by about 50%, from 10 to 12 days in the beginning of Mechanic’s study (1988-1994, stratified by hospital ownership) to 3 to 6 days at the end of Blader’s study (1996-2007, stratified by age group and payment source).
Mechanic and Blader found consistent decreases in private payments. Both authors appear to have found that publicly funded payments (either classified as government and stratified by age group or classified separately as Medicaid and Medicaid payments stratified by hospital type) were generally increasing to cover the decrease in private payers’ proportion. Although age groups and hospital ownership differ across both studies, the overall trend appeared to be increased public spending and decreased private spending.
Case et al.4 used a different sample, the Agency for Healthcare Research and Quality (AHRQ) National Inpatient Sample, to evaluate changes in inpatient mental health treatment for youth between 1990 and 2000. Their findings for length of stay mirrored those for adults insofar as inpatient time dropped from a median stay of 12 days to 4.5 days. They found no change in the total rate of hospitalization relative to the population, nor did they find any change in the proportion of public versus private payers. However, they did find substantial changes in diagnoses with depression, bipolar disorder, and psychosis all increasing while substance abuse and adjustment disorders decreased. Blader’s analysis of children and adolescents covered a different but overlapping time span (1996 to 2007) and also found decreases for length of stay although not as large. Unlike Case et al., Blader found increases in the overall population rate of hospitalization as well as an increase in government coverage and a decrease in private coverage. For the year 2000, Blader’s child rate approximates Case et al.’s, but Blader also provides an adolescent rate that is higher than the Case rate, which combines children and adolescents.
This study used the Texas Inpatient Public Use Data File (PUDF) that contains data on discharges from Texas hospitals, specifically data from the first 12 years of the program from 1999 through 2010. Discharges in this file include the most hospitalizations in Texas. Hospitals that are exempt from reporting include those located in a county with a population less than 35,000, or those located in a county with a population more than 35,000, fewer than 100 licensed hospital beds, and not in an area that is delineated as an urbanized area by the United States Bureau of the Census. Exempt hospitals also include those that do not seek insurance payment or government reimbursement. Data in the public use file are de-identified in such a way that only discharges, not individuals, can be analyzed, making analyses that would require tracking of individuals across discharges impossible with this file. Discharges selected for analysis were restricted to those with a primary behavioral health diagnosis (ICD-9 between 290 and 314.99) that excluded learning and intellectual disabilities.
The total of 1,762,515 hospital discharges with a primary psychiatric diagnosis from 1999 to 2010 represented 5.24% of all hospital discharges. Psychiatric discharges per year averaged 146,876.
Most of the sample (58%) were non-Hispanic whites; 15% were non-Hispanic blacks, 18% reported Hispanic ethnicity (with any race), and the remaining 8% reported other races and/or ethnicities. As for age, 18% of the sample was younger than 18 years, 48% was aged 18 to 44 years, 25% was aged 45 to 64 years, and 8% was older than 65 years.
In the PUDF, sex was suppressed for persons with a drug or alcohol diagnosis in any of the 25 available diagnosis fields as well as for hospitals with a small number of discharges. Sex was missing from 32% of the records. Of those records that included sex, 55% were females and 45% were males. Males traditionally have somewhat higher alcohol and drug use disorder rates; if 60% of the suppressed portion is assumed to be male, then the actual sex breakdown in the entire sample would be very close to 50/50.
The top 6 diagnostic categories accounted for 91% of the principal diagnoses: depression (29%), bipolar disorder (22%), schizophrenia (18%), alcohol and drug disorders (14%), psychotic disorders not otherwise specified (4%), and other mood disorders not otherwise specified (4%).
Patients were classified into 5 major groups based primarily on their primary payment source and provider. First, payment source was used to classify patients into 4 groups: private insurance, Medicare, Medicaid, and all other payment sources. Second, patients were classified as state hospital patients if they were treated by any state hospital facility regardless of payment source. By far, most payment sources for state hospital patients are state general revenue, which was generally coded in the “other” category so this did not affect the composition of the private insurance, Medicare, or Medicaid groups to any large degree.
Table 1 displays demographics by patient group. With more than 1.7 million patients, every statistical comparison was significant, so only the largest magnitude differences will be discussed. For gender, a significant difference was found among groups (chi square (4) = 15,987, P < .0001). Significantly more males (56%) and fewer females (43%) were treated by state hospitals than any other group, for which the trend was almost exactly the opposite. For race and ethnicity, a significant difference was found among groups (chi square (12) = 95,053, P < .0001). Significantly more blacks (20%) and Hispanics (28%) and significantly fewer whites (46%) were treated both by state hospitals and Medicaid than by the other groups. For age groups, a significant difference was seen among patient groups (chi square (12) = 566,447, P < .0001). Unsurprisingly, Medicaid served more children and Medicare served more seniors than other patient groups. Note that most patients for all groups were adults aged 18 to 64 years. Overall, patients aged 18 to 44 years were the largest proportions for both state hospitals and the “all other payment sources” group. Medicare was nearly the only payment source for patients aged 65 years and older.
Annual discharges increased from 122,533 in 1999 to 175,382 in 2010, accounting for an overall increase of 43%. The average increase in discharges per year was 3.4% per year over the previous year. Considerable variability was seen in growth across the years with 4 years of 7% to 8% growth, 4 years of 1% to 3% growth, and a single year of -4% shrinkage.
Table 2 displays clients’ primary diagnoses by patient group and, as in Table 1, all comparisons are significant, so only the largest magnitude differences will be discussed. For primary diagnosis, the difference among groups was significant (chi square (20) = 173,517, P < .0001). Depression (29%) was the largest diagnostic group overall and the largest percentage both for private insurance (34%) and for other payment (30%) sources. Schizophrenia (29%) was the largest diagnostic group for Medicare (27%) and state hospitals (35%) and the smallest for private insurance (7%) and other payment (11%) sources. Alcohol and drug diagnoses were large for private insurance (19%) and other payment (21%) sources and small for Medicaid (6%) and state hospitals (6%). Interestingly, bipolar disorders (22%) were distributed relatively equally across all patient groups, although Medicaid (26%) did have slightly more.
Length of stay was significantly different among all 5 patient groups (F (4) = 43,019, P < .0001) with state hospitals having the longest stay (41 days) and all others ranging between 6 and 9 days. Length of stay also differed significantly across the years 1999 to 2010 (F (11) = 38,581, P < .0001) with a decrease from 11.8 days in 1999 to 10.7 days in 2010. By patient group between 1999 and 2010, the state hospitals increased their length of stay from 41 days to 44 days, while the other payment sources patient group decreased their length of stay from 10.5 days to 8.1. The other 3 patient groups (Medicare, Medicaid, private insurance) remained generally stable, shifting overall by 0.5 days or less.
Total charges for all stays amounted to $19,147,041,627. While average total charges differed significantly among all 5 patient groups (F (4) = 43,019, P < .0001), the magnitude of differences among the groups was relatively small and had an average cost of $10,863 per discharge. However, total charges are relatively unreliable in that they may bear little relation to actual payments made and are far more illustrative than definitive. But if even only 25% to 50% of the charges were eventually paid, the total represents a very large amount of money.
Specifically for children under age 18 years, discharges average 27,035 per year, rising from 19,653 in 1999 to 34,912 in 2010, which represents a 78% increase that is considerably higher than the adult mental health hospitalization increase. Depression (35%), bipolar disorder (29, and other mood disorders not otherwise specified (13%) represented the 3 largest diagnostic groups and accounted for most of the patients. Although common childhood disorders such as attention deficit hyperactivity disorder (3.8%), oppositional defiant-conduct disorder (2.6%), anxiety (1.6%), and adjustment disorders (1.1%) were all present as children’s primary diagnoses, they accounted for less than 10% of the primary diagnoses for children. Medicaid (37%) and private insurance (36%) represented most child patients. Children’s average length of stay was 12.7 days, somewhat longer than adults, but had decreased from 14.3 days in 1999 to 11.1 days in 2010.
Table 3 summarizes the hospitals providing care. Hospitals were grouped by size in terms of the number of discharges they provided. Although 627 hospitals served psychiatric patients, approximately half of these hospitals (the smallest 2 groups, 1 to 9 and 10 to 99 discharges) were infrequent providers who served less than 0.5% of psychiatric inpatients and who on average served psychiatric inpatients in less than half of the 12 years in the data set. The middle group (100 to 999 discharges) served psychiatric patients on a regular basis, although they served only 4.5% of all patients. The 2 largest groups (1000 to 9999 and 10,000+ discharges), consisting of 137 hospitals, served 95% of all patients, and about 17.5% of these were teaching facilities. The largest group, 42 hospitals, were virtually all specialized psychiatric facilities (93%) and served most of the discharges (69%). Most (69%) of the second largest group, 95 hospitals, were also largely specialized psychiatric facilities that served a sizeable number of patients (25%). Unsurprisingly, most of the state hospitals fell into the largest group of providers.
Overall, the results of this study revealed a substantial increase (43%) in the number of discharges per year from 1999 to 2010 and a small decrease in length of stay (-1.1 days) over the same period. The public sector in terms of Medicare (19.5%), Medicaid (18.1%), and state hospitals (11.1%) served close to half of all psychiatric patients discharged. The public sector served more than half (74%) of patients diagnosed with schizophrenia and most of the not otherwise specified psychotic (54%) patients as well as half of the bipolar patients. For schizophrenia, Medicare actually served the most patients, followed by state hospitals and then Medicaid. The young age and high proportion of schizophrenia and bipolar disorders for approximately two-thirds of the Medicare patients is an important finding in this study. This suggests that a large number of patients aged 18 to 64 years are probably being served through Medicare Supplemental Security Income (SSI) disability programs. However, these patients are also likely to have dual eligibility with Medicaid, so an examination of Medicaid data would be useful in clarifying this. Another interesting finding was the concentrated provider network with a small group of 42 hospitals, who served over 10,000 patients each, serving almost 69% of all patients.
A number of similarities with national research were found. As Mechanic2 and Blader3 reported, overall rates of hospitalization have increased. As in previous studies, for both adults2,3 and children,3,4 the length of stay was generally decreasing over time. However, the size of the decrease over the course of this study was considerably smaller than the older national studies, suggesting that the major decreases in length of stay may have occurred during the late 80s and 90s. Unlike previous studies, no large increase occurred in public programs relative to private insurance. However, the all other payment sources patient category was increasing more quickly than either private insurance or public programs. The largest identifiable subgroup in the other payment sources consisted of self-payment (which was 45% of the all other group) and their share increased dramatically from an average of 9,000 per year in the beginning of the study to an average of 24,000 per year at the end of the study.
This study had a number of limitations. The suppression of sex for any client diagnosed with a substance abuse disorder in any field is regrettable, but its most likely effect was to suppress the proportion of male clients (males generally abuse substances at higher rates than females). A disappointing limitation of this study was that the data elements were insufficient to define a unique patient group based on charitable care provided by the hospitals that would have afforded a very interesting contrast to publicly funded safety net programs. Finally, because the public use data files were used, identifying individual clients for the purposes of assessing readmissions was impossible. Further research to examine cycling of individual patients across provider groups, especially among the arms of the public safety net (Medicaid, Medicare, and state hospitals), would be useful.
The public use data sets were sufficient to provide a reasonably detailed overview of psychiatric discharges in the state. The public programs of Medicaid, Medicare, and state hospitals serve approximately half of all patients and serve most of those diagnosed with severe mental illness (schizophrenia, bipolar disorder, psychosis, and depression), thereby illustrating their importance in providing a public safety net for people who need care. The state hospitals serve an important role in the public safety net because, although they are the smallest sector of the public safety net, they serve the highest proportion of patients diagnosed with schizophrenia and related psychotic spectrum disorders.
1. Texas Health Care Information Council. Hospital inpatient discharge public use data file (PUDF) 1999-2010. Austin, TX: Texas Department of State Health Services.
2. Mechanic D, McAlpine DD, Olfson M. Changing patterns of psychiatric inpatient care in the United States, 1988-1994. Arch Gen Psychiatry. 1998;55(9):785-791.
3. Blader JC. Acute inpatient care for psychiatric disorders in the United States, 1996 through 2007. Arch Gen Psychiatry. 2011;68(12):1276-1283.
4. Case BG, Olfson M, Marcus SC, Siegel C. Trends in the inpatient mental health treatment of children and adolescents in US community hospitals between 1990 and 2000. Arch Gen Psychiatry. 2007;64(1):89-96.