Comparison of Mental Health Treatment Adequacy and Costs in Public Hospitals in Boston and Madrid.

AN0137097058;2ws01jul.19;2019Jun22.05:18;v2.2.500

Comparison of Mental Health Treatment Adequacy and Costs in Public Hospitals in Boston and Madrid 

Analyses of healthcare expenditures and adequacy are needed to identify cost-effective policies and practices that improve mental healthcare quality. Data are from 2010 to 2012 electronic health records from three hospital psychiatry departments in Madrid (n = 29,944 person-years) and three in Boston (n = 14,109 person-years). Two-part multivariate generalized linear regression and logistic regression models were estimated to identify site differences in mental healthcare expenditures and quality of care. Annual total average treatment expenditures were $4442.14 in Boston and $2277.48 in Madrid. Boston patients used inpatient services more frequently and had higher 30-day re-admission rates (23.7 vs. 8.7%) despite higher rates of minimally adequate care (49.5 vs. 34.8%). Patients in Madrid were more likely to receive psychotropic medication, had fewer inpatient stays and readmissions, and had lower expenditures, but had lower rates of minimally adequate care. Differences in insurance and healthcare system policies and mental health professional roles may explain these dissimilarities.

Previous Presentation A preliminary analysis of this data was presented at the 23rd European Congress of Psychiatry, March 28–31, 2015.

Introduction

Despite the high worldwide prevalence of mental disorders, rates of treatment do not match the need for care and mental health policies are a low priority in most countries.[1]–[3] Most persons with mental disorders receive no treatment, which is a major public health concern given the impact of mental health on quality of life and subsequent expenditures.[2],[4]–[6] To address this, there have been calls from international health agencies for initiatives to improve access to treatment and to compare efficiency and quality of system models in order to implement evidence-based policy.

Cross-national comparative analyses of healthcare system efficiency have been carried out in several prior studies to evaluate system differences. One study compared the National Health Service of the UK and Kaiser Permanente, a large staff-model managed care organization in the United States of America (USA), finding that Kaiser patients had fewer acute days per capita and received overall higher quality of care at the same costs.[7] International comparative efforts have also been made specifically in the mental health field. After comparing 32 Organization for Economic Cooperation and Development countries, Moran and Jacobs concluded that Slovenia, Korea, Poland, and Denmark were the most efficient in mental healthcare provision.[8] The study provided an important step in cross-national comparisons but was limited by its exclusive focus on inpatient mental health care and lack of case-mix adjustment. There remains a dearth of international comparisons of mental health treatment quality and expenditures that incorporate the full continuum of mental health services (inpatient, outpatient, emergency care, and psychopharmacological treatment). Such analyses are needed to identify policies and practices that improve quality, particularly given the complexity of mental health disorders and the systems developed to treat them.[9],[10]

Much like other public hospitals in Spain, Madrid hospitals in this study are part of the public health system and serve a socio-demographically diverse population of over 700,000 people.[11] The Boston system also serves a diverse linguistic and racial/ethnic minority population, with a relatively high percentage of patients covered by some form of public insurance, similar to other safety net settings in urban areas of the USA.[12] In both locations, psychiatric inpatient and emergency care is provided mainly by the hospital, while entrance to outpatient mental health treatment predominantly relies on referrals from primary care physicians (PCPs), specialists, or emergency room (ER) physicians.

One difference between sites is that outpatient care is predominantly delivered by either a psychologist or a psychiatrist in Madrid, whereas psychiatric nurses and social workers are also integrally involved in counseling services in Boston.[13],[14] There are also major differences between the USA and Spain with regard to payment and insurance schemes. Spain has a single payer system in which copayments for mental healthcare visits are not common, and an unlimited number of mental health visits are allowed.[15] The USA has a multiple payer system (Medicaid, Medicare, private insurance, and safety net programs) with wide variation in the number of co-payments billed to the patient, and in which many insurance plans limit the total number of mental health visits.[16]

It is important to acknowledge that these differences in healthcare financing and system organization across the two sites may result in different patterns of usual treatment and different characteristics of patient populations that are treated at the two hospital systems. For example, cost per unit of healthcare services in both sites varies greatly because of negotiations between payers, providers, and government regulators. Accordingly, differences in usual care and patient characteristics and diagnoses are described across sites. Additionally, regression models of service use variables are estimated, adjusting for differences in cost per unit, purchasing power, and patient demographics and diagnoses using statistical analysis to isolate differences in usual care. In these analyses, the cost per unit is standardized (i.e., each specialty mental health outpatient visit is worth $207 USD, the US average per authors' own analysis of US Medical Expenditure Panel Survey data). In prior studies, this has been referred to as a "quantity index," a metric of utilization that allows comparisons of the total value of services used by different individuals.[17] Furthermore, adjusting for World Bank Purchasing Power Parity (PPP) allowed for differences in the costs of services.[18] While not a perfect comparison, these adjustments allow for a more standardized comparison of the resources spent.

To improve knowledge of the relative merits of the two different mental healthcare systems, this retrospective study has three aims: (1) to compare any use of mental health care and per capita mental healthcare expenditures by site (Boston, Madrid) and treatment setting; (2) to compare the sites on two quality measures, "minimally adequate episodes of care," and 30-day inpatient readmissions; and (3) to describe healthcare system differences and how they may help explain expenditure and quality differences.

Methods

Data

Following procedures approved by Institutional Review Boards of both sites, cross-sectional data was analyzed from 2010 to 2012 electronic health records (EHRs) from a public hospital system with three hospitals in the Boston metropolitan area (n = 14,109 person-years) and three hospitals in Madrid (n = 29,944 person-years). These hospitals serve a high percentage of Latino patients and were included in the current study as part of their collaboration in the International Latino Research Partnership (ILRP). The ILRP connects research institutions, hospitals, and community clinics in the USA and Spain to conduct comparative research on Latinos' behavioral health service needs. Working with researchers, clinicians, and administrators, all variables used in this study were harmonized to ensure their comparability. To best harmonize the available data in the two sites, outpatient and inpatient treatment, psychotropic medication, and any emergency department (ED) use among patients receiving treatment in hospital psychiatric departments were analyzed. "Any ED use" was used, instead of a variable denoting ED use for psychiatric illness specifically, because of the difficulty in disentangling physical and psychiatric illness as the primary reason for a patient visit in the Madrid ED. Current engagement with substance abuse treatment, but not patients who might have a substance use disorder, was excluded since treatment services are predominantly offered by another provider network in Madrid.

A quantity index, or adjusted expenditures, was determined by multiplying the quantity of visits/uses by the cost per unit of treatment. In both sites, the cost per unit was estimated using the nationally representative 2012 US Medical Expenditure Panel Survey (MEPS) average cost for psychiatric inpatient, outpatient and ER services, and psychotropic medication expenditures. Comparing the sites on this weighted service use variable allows for an approximation of the overall resources spent in each site and removes the potentially confounding variability that arises due to negotiations between payers, providers, and the government. Additionally, not only adjusting for the PPP adjusts for differences in currency exchange rates, but also underlying price differences in medical inputs and other goods and services in the domestic market.[7] In exploratory analysis, the study team compared expenditures incorporating site differences in both price and quantity, comparing MEPS average cost expenditures for the Boston site to fee for service reimbursement levels (also called official healthcare cost sheets) published yearly by the city of Madrid.

The main dependent variable is annual mental healthcare expenditures adjusted for site differences as described above. Utilization was disaggregated into (a) any access to treatment and (b) expenditures conditional on access to treatment (a continuous variable > $0) and analyzed expenditures by setting (outpatient, inpatient, pharmacy, and emergency). Mental health events include the following: (a) treatment provided by a specialist (psychiatrist, psychologist, counselor, or social worker) for disorders covered by ICD-9 codes 291, 292, or 295–314 in Boston and based on the diagnosis recorded in text fields in Madrid or (b) prescriptions of medicines in Boston or Madrid considered a psychotropic drug per the Multum drug classification system.[19],[20]

The first dependent variable related to quality of care is minimal adequacy of care, defined as having ≥ 4 outpatient visits in the last year and a psychotropic medication fill, or ≥ 8 outpatient visits (with or without a medication fill). This definition has been used in prior studies and represents the minimum of what clinicians and scientists have considered to be adequate care for depression, anxiety, and other mental illnesses.[6],[21],[22] The second quality variable is inpatient hospital readmission within 30 days used in prior literature to evaluate systems performance.[23] Hospital readmissions are likely due to inadequate treatment, inadequate care coordination and follow up care and/or complications from hospital treatment.[24] In the USA, the Affordable Care Act has authorized Medicare to provide incentives to reduce re-hospitalizations, reaffirming the measure as an important marker of quality.[25]

The primary independent covariate of interest was the site indicator. Other covariates included sex, age, primary mental health diagnosis (depression, anxiety, bipolar, psychosis, PTSD, other), and primary type of substance use (cocaine, marijuana, alcohol, other). Identification of mental health diagnoses and substance use type was possible for all Boston patients and approximately 75% of the population in Madrid. To account for this missingness, mental health diagnosis and substance use type for the other 25% of Madrid patients were imputed by multiple imputation via the Stata 14 MI procedure. Standard errors were derived by incorporating standard rules to account for the uncertainty due to imputation.[26]

Statistical analysis

First, usual psychiatric care in each site was determined to establish the context, health systems, and providers involved with care. To describe site characteristics, age- and gender-adjusted expenditures by site, and unadjusted rates of the independent variables were compared. Comparisons for continuous variables were conducted using t tests; categorical variables were compared using chi-squared tests.

For expenditures analyses, a two-part modeling strategy was implemented, separately modeling the probability of any expenditures (logit model) and the level of expenditure conditional on positive expenditures (generalized linear models (GLM)), adjusting for sex, age, and type of mental health and substance use disorder.[27] The GLM estimates expected expenditures E(y|x, y > 0) directly as f(x′β) where f is the link between the observed raw scale of expenditure, y, and the linear predictor x′β, where x is a vector of the predictors. GLMs allow for heteroskedastic residual variances related to the predicted mean.[28] A one-part GLM of total mental healthcare expenditures was estimated, since all sampled individuals had positive mental healthcare expenditures. For all GLMs, using diagnostics in Manning and Mullahy and Buntin and Zaslavsky, the optimal GLM to have a log link and residual variance proportional to mean squared was identified.[28],[29]

Minimal adequacy of care and 30 day inpatient readmissions were compared using logit models, adjusting for site (the covariate of interest), sex, age, and type of mental health and substance use disorder.

Results

Demographics

There are significant differences between the Boston and Madrid sites (Table 1). In Boston, patients were more likely to be female, older by approximately 3 years, and more likely to receive services with diagnoses of depression, anxiety, bipolar disorder, PTSD, marijuana use, and alcohol use compared to Madrid. In Madrid, patients were more likely to receive services with diagnoses of psychosis, other diagnosis, and cocaine use disorder.

Sample characteristics (n = 44,053) in Boston and Madrid (2010–2012)a

*Difference between sites is significant at p