First Outpatient Follow-Up After Psychiatric Hospitalization: Does One Size Fit All?
Abstract
Objective:
Claims-based indicators of follow-up within seven and 30 days after psychiatric discharge have face validity as quality measures: early follow-up may improve disease management and guide appropriate service use. Yet these indicators are rarely examined empirically. This study assessed their association with subsequent health care utilization for adults with comorbid conditions.
Methods:
Postdischarge follow-up and subsequent utilization were examined among adults enrolled in North Carolina Medicaid who were discharged with claims-based diagnoses of depression or schizophrenia and not readmitted within 30 days. A total of 24,934 discharges (18,341 individuals) in fiscal years 2008–2010 were analyzed. Follow-up was categorized as occurring within 0–7 days, 8–30 days, or none in 30 days. Outcomes in the subsequent six months included psychotropic medication claims, adherence (proportion of days covered), number of hospital admissions, emergency department visits, and outpatient visits.
Results:
Follow-up within seven days was associated with greater medication adherence and outpatient utilization, compared with no follow-up in 30 days. This was observed for both follow-up with a mental health provider and with any provider. Adults receiving mental health follow-up within seven days had equivalent, or lower, subsequent inpatient and emergency department utilization as those without follow-up within 30 days. However, adults receiving follow-up with any provider within seven days were more likely than those with no follow-up to have an inpatient admission or emergency department visit in the subsequent six months. Few differences in subsequent utilization were observed between mental health follow-up within seven days versus eight to 30 days.
Conclusions:
For patients not readmitted within 30 days, follow-up within 30 days appeared to be beneficial on the basis of subsequent service utilization.
Quality improvement initiatives and quality metrics for general medical conditions have become ubiquitous in health care. Although this proliferation has lagged somewhat in mental health care (1,2), recent attention has yielded an increasing number of administrative quality metrics (3). Yet few of these have been empirically related to patient care utilization (4). Absent universally accepted administrative mental health quality metrics, many researchers default to using metrics implemented by health care payers (5). For example, outpatient follow-up visits within seven and 30 days of psychiatric discharge (National Committee for Quality Assurance Healthcare Effectiveness and Data Information Set [HEDIS]) are a common standard in health care performance management (6). Early outpatient follow-up may facilitate linkage between inpatient and outpatient care plans, including medication reconciliation between multiple records and providers, and may reinforce medication adherence, reduce early readmissions, and match individual care needs with appropriate support services (7–10). Yet the association between early follow-up and subsequent patient care utilization has not been fully explored.
Timely follow-up after psychiatric discharge is particularly important for persons with comorbid serious mental illness and chronic general medical conditions (11,12). Among adults with serious mental illness, 90% have co-occurring general medical conditions such as hypertension, cardiovascular disease, hyperlipidemia, or diabetes (13,14). When combined with serious mental illness, these conditions are more detrimental to overall health than they are to the health of those with the same conditions in the general population (15). In addition to poor access to mental health treatment, individuals with a chronic general medical condition and comorbid serious mental illness have higher hospitalization rates and health care costs than those with the general medical condition alone (16). Furthermore, preventable general medical conditions are the leading cause of premature death among persons with serious mental illness (17,18). For these reasons, appropriately timed follow-up care after hospitalization for serious mental illness is vital for continued monitoring of both mental health and general medical conditions and prevention of hospital readmission.
As part of a larger evaluation, we examined the effect of a Medicaid medical home on health care quality for persons with multiple chronic conditions. In this study, we examined the timing of first outpatient follow-up and subsequent health care utilization for adults with serious mental illness (major depression or schizophrenia) and comorbid general medical conditions.
Methods
Data
We conducted a retrospective analysis using a unique data source, North Carolina Integrated Data for Researchers, which, using probabilistic linkage, connects North Carolina Medicaid claims data with a number of additional administrative health data sources (19). Additional sources included state-funded mental health services, a state psychiatric hospital utilization database, and claims from a five-county behavioral health carve-out program that existed during our study period. Data on state-funded mental health services and state psychiatric hospitalizations that are not covered by Medicaid for persons ages 22–64 provide information about mental health services received that are otherwise unobserved for adults who are not dually eligible for both Medicare and Medicaid. The behavioral health carve-out encounter data describe outpatient mental health utilization for Medicaid enrollees in five participating counties. From this linked database, we extracted data on demographic and clinical characteristics, hospitalizations, pharmacy claims, and service utilization for fiscal years 2008 through 2010.
Setting and Study Design
The larger study examined North Carolina Medicaid’s medical home model, Community Care of North Carolina (CCNC), which has been described in detail elsewhere (20,21). Essentially, it consists of a primary care case management program enhanced with a number of evidence-based programs and protocols, disease management, pharmacy management, case management, and practice performance metrics (annual eye examination for diabetic patients, for example) that are shared regularly with physicians and networks.
This study used a retrospective cohort design to examine the association between timing of first outpatient follow-up visit after psychiatric discharge and subsequent health care utilization among Medicaid enrollees with multiple chronic conditions. We expected heterogeneity of association between follow-up timing and subsequent health care utilization with respect to psychiatric diagnosis (major depression or schizophrenia) because adults with schizophrenia often have follow-up needs different from those of adults with depression. Because patterns of care differ between medical home enrollees and other Medicaid enrollees (22), we also anticipated heterogeneity by medical home enrollment status. Therefore, we created four cohorts stratified by psychiatric diagnosis (depression versus schizophrenia) and medical home status at hospital discharge (enrolled versus not enrolled).
Inclusion and Exclusion Criteria
Our initial population included persons with diagnoses of two or more of eight chronic conditions: major depressive disorder (ICD-9-CM codes 296.2X, 296.3X, 300.4X, and 311.XX), schizophrenia (ICD-9-CM code 295.XX), diabetes mellitus (ICD-9-CM codes 250.XX, 271.4X, 357.2X, 362.0X, 366.41, 791.5X, V4585, V5391, and V6546), hypertension (ICD-9-CM codes 362.11, 401.XX-405.XX, and 997.91), hyperlipidemia (ICD-9-CM code 272.XX), seizure disorder (ICD-9-CM codes 345.XX and 780.XX, excluding 780.31), asthma (ICD-9-CM code 493.XX), and chronic obstructive pulmonary disease (COPD) (ICD-9-CM codes 491.XX, 492.XX, and 496.XX). These conditions were selected according to prevalence and health expenditure within Medicaid and to include both general medical and mental health conditions for which administrative quality measures could be constructed. To classify a person as having a given chronic condition, we required a minimum of two outpatient or emergency department claims or one inpatient claim for the condition during the 36-month study period.
Discharges of adults (age ≥18) with a hospital diagnosis of depression or schizophrenia were selected from the initial population. We excluded observations for which an individual was readmitted during the 30 days after discharge because the readmission prevented us from measuring time to follow-up for those individuals. [A figure depicting the timeline is available in the online supplement to this article.] To ensure complete claims information, we also excluded persons dually eligible for Medicaid and Medicare, because we lacked access to Medicare claims data. Last, because we were unable to observe medication claims outside of Medicaid, persons were excluded if they were not Medicaid enrollees in the month of hospital discharge and the subsequent six-month observation period. [A flow diagram of the selection and exclusion process is available in the online supplement to this article.]
Measures
Our primary explanatory variable was a categorical variable for timing of first outpatient visit with a provider after a patient was discharged from a hospitalization with a diagnosis of depression or schizophrenia. In our primary analyses we limited follow-up to visits with mental health providers to match the HEDIS measure. The first follow-up visit was categorized as occurring within 0–7 days, 8–30 days, or not within 30 days postdischarge. In sensitivity analyses we assessed follow-up with any type of provider because primary care providers are increasingly treating mental health conditions (especially depression) and because our study population included persons with co-occurring mental health and general medical conditions considered likely to also require follow-up with non–mental health care providers.
To separate exposure and follow-up periods, subsequent health care utilization was examined for six months beginning with the first complete calendar month greater than 30 days after discharge (for example, a discharge on January 15th would begin the follow-up period on March 1). We examined subsequent utilization in order to track both reactive health services, such as emergency department or hospital use, and outpatient services, which may indicate whether important opportunities for community-based care were facilitated by earlier contact.
Dependent variables were broadly classified into three categories: outpatient medication utilization, outpatient visits, and inpatient care. Medication utilization variables included a dichotomous indicator representing any claim for either an antidepressant (depression cohorts) or an antipsychotic medication (schizophrenia cohorts) and a measure of medication adherence, the proportion of days covered on psychotropic medication (antipsychotic or antidepressant) during the six-month follow-up period (23). Outpatient dependent variables included number of emergency department visits, outpatient visits (all-cause and mental health), psychotherapy sessions (family or group and individual), and receipt of assertive community treatment (ACT, for schizophrenia only). Inpatient care dependent variables included number of hospital admissions (all-cause and diagnosis specific) and total hospital days (all-cause and diagnosis specific).
Control variables included age, sex, race, Hispanic ethnicity, total number of target chronic conditions (between two and eight inclusive), total number of target conditions for which a patient was receiving medications (between two and eight inclusive), enrollment in a behavioral health carve-out, number of baseline inpatient admissions, number of baseline emergency department visits, and 19 comorbidity indicators based on the Chronic Illness and Disability Payment System (CIDPS) (24). Baseline health status and health care utilization variables were assessed during a six-month period ending the month before the index hospital discharge.
Propensity Score Implementation
To control for potential confounding because of the nonrandomized nature of timing of follow-up visits, we used propensity scores to balance the three groups on a rich set of observed baseline covariates. The propensity score model included variables that were associated with health care utilization, based on theory or prior evidence (25). We used a multinomial logistic regression to estimate the propensity score (26,27) and used inverse probability-of-treatment weights (IPTW) to estimate the average treatment effect (28).
Statistical Analysis
We estimated generalized linear models with a Poisson distribution, log link, and IPTW. For binary dependent variables we used robust standard errors to account for underdispersion of dichotomous data in the context of a count distribution (29). All models included cluster-adjusted standard errors to account for repeated person observations. In additional sensitivity analyses, we examined only the first discharge. [Tables A and B in the online data supplement provide details.] Models were estimated in Stata 11. We conducted hypothesis tests on whether there was an average difference in subsequent utilization between the referent group and each of the other groups (0- to 7-day follow-up and 8- to 30-day follow-up); we also tested whether the effect of having a follow-up within seven days was equal to the effect of having a follow-up 8–30 days posthospitalization.
The study protocol was approved by the University of North Carolina Institutional Review Board.
Results
Descriptive Results
Of 28,006 discharges, 3,072 discharges (11%, representing 2,014 individuals) were followed by a readmission within 30 days and excluded from further analysis. [The flow diagram in the online supplement provides detail.] The remaining 24,934 discharges (representing 18,341 individuals) were stratified by psychiatric diagnosis and medical home status to assess baseline covariates. [Table C in the online supplement shows data for unique individuals and repeat admissions.] Overall unweighted characteristics for each cohort and weighted maximum absolute standardized differences for all pairwise comparisons (0–7 versus 8–30, 0–7 versus >30, and 8–30 versus >30) after IPTW are shown in Table 1. Weighted standardized differences were below the suggested threshold of 10% for all variables in all cohorts, indicating sufficient baseline covariate balance between groups (30).
| Depressionb | Schizophrenia | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Medical home (NT=8,821, N=7,219) | No medical home (NT=9,047, N=7,577) | Medical home (NT=3,572, N=2,473) | No medical home (NT=3,494, N=2,490) | |||||||||
| Characteristic | N | % | Max. std. diff.c | N | % | Max. std. diff.c | N | % | Max. std. diff.c | N | % | Max. std. diff.c |
| Age (M±SD) | 42.7±12.1 | 1.7 | 42.4±12.4 | 2.0 | 40.3±12.7 | 2.1 | 39.0±13.2 | 5.5 | ||||
| Male | 2,064 | 23.4 | .5 | 2,696 | 29.8 | .3 | 1,618 | 45.3 | 3.2 | 1,751 | 50.1 | 2.2 |
| Race-ethnicity | ||||||||||||
| White | 5,363 | 60.8 | .7 | 6,089 | 67.3 | 1.4 | 1,297 | 36.3 | 2.3 | 1,443 | 41.3 | 1.6 |
| Black | 2,849 | 32.3 | 1.4 | 2,515 | 27.8 | .8 | 2,072 | 58 | 1.9 | 1,883 | 53.9 | 1.7 |
| Asian | 35 | .4 | 5.5 | 27 | .3 | .9 | 21 | .6 | 2 | 18 | .5 | 3.2 |
| American Indian | 221 | 2.5 | .7 | 154 | 1.7 | 2.8 | 82 | 2.3 | 7.3 | 73 | 2.1 | 1.0 |
| Hispanic | 159 | 1.8 | 1.1 | 163 | 1.8 | 1.1 | 54 | 1.5 | 2.4 | 49 | 1.4 | .9 |
| Hospital length of current stay (M±SD days) | 4.8±5.9 | 2.3 | 5.5±8.9 | 1.1 | 11.8±26.2 | 2.6 | 14.0±30.4 | 2.8 | ||||
| State hospital admissions (M±SD) | .0±.1 | 1.3 | .0±.2 | 1.0 | .2±.4 | 4.3 | .2 ±.4 | 3.1 | ||||
| Months in medical home (M±SD) | 4.8±1.9 | .7 | .5±1.3 | .7 | 5.1±1.7 | 3.7 | .4±1.2 | 2.3 | ||||
| Chronic Illness and Disability Payment score (M±SD)d | 5.4±2.6 | 2.1 | 4.1±3.0 | 1.4 | 4.4±2.7 | .9 | 3.8±2.9 | 5.3 | ||||
| Months enrolled in Medicaid (M±SD) | 5.8±.7 | 1.0 | 4.6±2.1 | .8 | 5.9±.5 | 1.4 | 5.0±1.9 | 3.5 | ||||
| Enrolled in behavioral health managed care pilot | 759 | 8.6 | .3 | 281 | 3.1 | 1.3 | 179 | 5 | 1.1 | 59 | 1.7 | 1.9 |
| Specified medical conditions (M±SD) | 3.5±2.1 | 1.0 | 3.8±2.4 | .9 | 2.6±2.1 | .7 | 3.2±2.4 | 2.6 | ||||
| Specified psychiatric conditions (M±SD) | 1.6±.6 | 1.2 | 1.5±.8 | 1.1 | 1.6±.6 | .4 | 1.6 ±.7 | 1.7 | ||||
| Specified general medical conditions with medication (M±SD) | 1.1±1.2 | 1.5 | .9±1.1 | .7 | .9±1.2 | 1.6 | .8±1.1 | 4.7 | ||||
| Specified psychiatric conditions with medication (M±SD) | .5±.5 | 2.3 | .4±.5 | .6 | .9±.6 | 5.2 | .8±.7 | 3.2 | ||||
| Individual psychotherapy sessions (M±SD) | .9±2.9 | 3.2 | .6±2.2 | 1.1 | 1.3±3.3 | 9.0 | 1.2±3.5 | 6.0 | ||||
| Number of family or group psychotherapy sessions (M±SD) | .1±.8 | 1.6 | .1±.9 | 1.5 | .2±1.4 | 5.9 | .1±.8 | 8.0 | ||||
| Number of months with ACT visit (M±SD) | .0±.4 | .8 | .0±.4 | 1.1 | .5±1.5 | 4.9 | .4±1.3 | 2.2 | ||||
| Total number of outpatient visits (M±SD) | 9.5±8.8 | 3.8 | 6.3±7.4 | 5.8 | 12.4±16.3 | 4.0 | 8.6±12.9 | 8.8 | ||||
| Number of medical home visits (M±SD) | 2.3±2.7 | 2.2 | .2±.8 | .7 | 1.4±2.1 | 4.4 | .1±.4 | .9 | ||||
| Number of outpatient mental health visits (M±SD) | 1.9±4.6 | 1.9 | 1.3±3.7 | .2 | 3.5±5.5 | 8.2 | 2.8±4.9 | 4.2 | ||||
| Number of emergency department visits (M±SD) | 2.3±3.8 | 2.2 | 1.9±3.7 | .6 | 2.1±3.2 | 1.3 | 1.8±3.2 | 1.8 | ||||
| Total hospital days (M±SD) | 4.8±10.9 | 1.0 | 5.1±12.1 | .4 | 7.7±15.3 | 1.5 | 9.8±21.0 | .6 | ||||
| Total hospital days for depression or schizophrenia (M±SD) | 2.0±6.2 | .4 | 2.2±7.5 | .6 | 6.1±14.1 | 2.0 | 7.9±19.8 | 2.2 | ||||
| Total admissions (M±SD) | .8±1.5 | 1.8 | .8±1.3 | .4 | .8±1.1 | 3.2 | .8±1.2 | 1.8 | ||||
| Total admissions for depression or schizophrenia (M±SD) | .4±.8 | .9 | .3±.8 | .4 | .6±1.0 | 3.1 | .6±1.0 | 4.9 | ||||
| Not living in community independently | 247 | 2.8 | .8 | 561 | 6.2 | 1.4 | 700 | 19.6 | 1.3 | 989 | 28.3 | 2.9 |
a
All utilization covariates except current length of stay are for previous 6 months. NT, total discharges; N, total unique individuals. Means±SDs are unweighted person-discharge characteristics within each cohort.
b
Number of state hospital admissions and months with assertive community treatment (ACT) are included for depression cohorts for completeness and because state hospital admissions reflect greater disease severity.
c
Weighted maximum standardized difference between all pairwise comparisons (a–b, a–c, and b–c) according to inverse probability of treatment weights
d
Possible scores range from 0 to 19, with higher scores indicating greater number of comorbidities.
As expected, the four cohorts differed on variables such as race, gender, and service utilization history. Individuals discharged with a depression diagnosis were older than those with a schizophrenia diagnosis (medical home, 42.7 versus 40.3, respectively; no medical home, 42.4 versus 39.0; p<.001), were more likely to be female (medical home, 77% versus 55%; no medical home, 70% versus 50%; p<.001), and were more likely to be white (medical home, 60.8% versus 36.3%; no home, 67.3% versus 41.3%; p<.001) and experienced shorter hospital stays (medical home, 4.8 versus 11.8 days; no home, 5.5 versus 14.0; p<.001). Yet persons with a schizophrenia diagnosis versus a depression diagnosis at discharge had fewer of the eight study conditions for inclusion in the sample (medical home, 2.6 versus 3.5, respectively; no home, 3.2 versus 3.8; p<.001) and fewer total CIDPS comorbidity indicators (medical home, 4.4 versus 5.4; no home, 3.8 versus 4.1; p<.001). Although the two diagnostic groups had similar numbers of emergency department visits and all-cause hospitalizations during the baseline period, persons with depression diagnoses at discharge had fewer total hospital days than those with a schizophrenia diagnosis (medical home, 4.8 versus 7.7; no home, 5.1 versus 9.8; p<.001) and fewer diagnosis-specific hospitalizations (medical home, .4 versus .6; no home, .3 versus .6; p<.001).
The cohorts with and without a medical home also exhibited baseline differences. Medical home enrollment was associated with more outpatient visits (depression, 9.5 versus 6.3; schizophrenia, 12.4 versus 8.6; p<.001) during the baseline period regardless of discharge diagnosis. Number of outpatient mental health visits (depression, 1.9 versus 1.3; schizophrenia, 3.5 versus 2.8; p<.001) was also higher among medical home cohorts. Although total all-cause admissions did not differ between cohorts with and without a medical home, the medical home cohorts experienced fewer total hospital days and fewer diagnosis-specific hospital days than cohorts without a medical home.
[Unweighted category-specific (individual-level) follow-up time characteristics for all cohorts are available in Tables D and E of the online supplement , and weighted category-specific follow-up time characteristics (person discharge level) and absolute standardized differences for each pairwise comparison are shown in Tables F–I.]
Regression Results
Table 2 enables comparison of outcomes in the six months after discharge by follow-up (within 0–7 days of discharge or 8–30 days of discharge or no follow-up within 30 days of discharge). For example, in the medical home cohort with a discharge diagnosis of depression, an outpatient visit within 0–7 days compared with no outpatient visit within 30 days after discharge was associated with an 11.8-percentage point greater chance of having any claim for an antidepressant during follow-up (p<.01). Among those with depression with no medical home who had a follow-up within 0–7 days versus follow-up within 8–30 days of discharge, an earlier first outpatient visit compared with 8–30 days after discharge was associated with a 5.8-percentage point greater chance of having any claim for an antidepressant during follow-up (p<.01).
| Medical home | No medical home | |||||
|---|---|---|---|---|---|---|
| Dependent variable | 0–7 versus >30 days | 8–30 versus >30 days | 0–7 versus 8–30 days | 0–7 versus >30 days | 8–30 versus >30 days | 0–7 versus 8–30 days |
| Depression | (NT=8,821, N=7,219) | (NT=9,047 N=7,577) | ||||
| Any claim for antidepressant (%) | 11.8** | 11.2** | .6 | 17.1** | 11.3** | 5.8** |
| Percentage of days covered by claim (%) | 1.5 | 1.8 | –.3 | 3.8 | 1.1 | 2.7 |
| N of outpatient mental health visits | 2.59** | 2.54** | .05 | 3.11** | 2.26** | .85* |
| N of outpatient visits | 2.06** | 2.31** | –.25 | 3.67** | 2.88** | .79 |
| N of family or group psychotherapy sessions | .20** | .12** | .08 | .02 | –.04 | .06 |
| N of individual psychotherapy sessions | 1.10** | 1.15** | –.05 | .99** | .91** | .09 |
| N of emergency department visits | –.07 | .05 | –.12 | .24 | –.04 | .28 |
| N of admissions | –.02 | –.08 | .06 | .15 | –.07 | .22 |
| N of admissions for depression | .05 | .01 | .04 | .07 | .03 | .04 |
| Total hospital days | –1.85 | –2.41* | .56 | .42 | –.51 | .93 |
| Total hospital days for depression | –.10 | –.24 | .14 | .01 | –.15 | .16 |
| Schizophrenia | (NT=3,572, N=2,473) | (NT=3,494, N=2,490) | ||||
| Any claim for antipsychotic (%) | 7.8** | 5.8** | 2.0 | 8.0** | 10.3** | –2.3 |
| Percentage of days covered by claim (%) | 3.3 | 2.1 | 1.2 | 4.0 | 3.9* | .1 |
| N of mental health visits | 3.32** | 2.48** | .84* | 2.86** | 3.07** | –.21 |
| Total number of outpatient visits | 3.20** | 1.60 | 1.5 | 2.37** | 2.38** | –.01 |
| Any assertive community treatment (%) | –7.6** | –10.6** | 3.0 | –10.4** | –8.8** | –2.6 |
| Months receiving assertive community treatment | –.32* | –.54** | .22 | –.46** | –.39** | –.07 |
| N of family or group psychotherapy sessions | .07 | .08 | –.01 | .09** | .17** | –.08 |
| N of emergency department visits | .06 | .22 | –.18 | –.22 | –.03 | –.19 |
| N of admissions | –.00 | .14* | –.14* | –.24** | –.08 | –.16 |
| N of admissions for schizophrenia | .03 | .14* | –.11 | –.14** | –.03 | –.11 |
| Total hospital days | –.14 | –.05 | –.09 | .80 | 2.41 | –1.61 |
| Total hospital days for schizophrenia | –.40 | –1.69 | 1.29 | 2.60 | 5.15 | –2.55 |
a
All measures reflect utilization during subsequent 6 months. Example interpretation: First outpatient visit with a mental health specialist within 0–7 days versus >30 days is associated with an 11.8-percentage point increase in the probability of having any claim for an antidepressant during 6-month follow-up among persons discharged with a depression diagnosis in a medical home. NT, total discharges; N, total unique individuals
*
p<.05, **p<.01
After discharge with a diagnosis of depression or schizophrenia, earlier follow-up visits were generally associated with greater medication utilization. However, the difference between 0–7 and 8–30 days for initial follow-up was relatively small and rarely statistically significant.
Generally positive associations were observed with early (0–30 days) outpatient follow-up and subsequent outpatient health care utilization. For example, receipt of an early outpatient visit in the depression medical home cohort was associated with 2.06 (p<.01, 0–7 days) and 2.31 (p<.01, 8–30 days) more outpatient visits than no outpatient follow-up within 30 days. Earlier visits were also associated with more outpatient psychotherapy visits. For people with schizophrenia, earlier outpatient visits were associated with lower participation in ACT teams. In general, effect sizes were relatively similar between depression cohorts with and without a medical home, but somewhat more divergent for schizophrenia cohorts with and without a medical home. In the depression cohorts, no association between timing of follow-up and emergency department visits or number of hospitalizations was observed, but early follow-up was associated with fewer hospital days (−2.41, p<.05) for those enrolled in a medical home. In the schizophrenia cohorts, small but statistically significant associations between timing of follow-up and subsequent admissions were observed.
Sensitivity Analysis
We found generally similar results when timing of follow-up visit to any provider, as opposed to a mental health provider, was considered the treatment variable (Table 3). Compared with no follow-up within 30 days for those with a medical home, those with a 0- to 7-day follow-up (9.6%, p<.01) had greater probability of any antidepressant medication. Early outpatient follow-up (within 0–7 or 8–30 days) was correlated with greater subsequent mental health and total outpatient visits in all cohorts. The probability of receiving ACT services for persons with a diagnosis of schizophrenia was generally higher among people with early follow-up visits, diverging somewhat from the primary analysis. We also observed a small increase in emergency department visits for those with a depression diagnosis without a medical home (.36, p<.01) and for those with a schizophrenia diagnosis in a medical home (.40, p<.01).
| Medical home | No medical home | |||||
|---|---|---|---|---|---|---|
| Dependent variable | 0–7 versus >30 days | 8–30 versus >30 days | 0–7 versus 8–30 days | 0–7 versus >30 days | 8–30 versus >30 days | 0–7 versus 8–30 days |
| Depression | (NT=8,821, N=7,219) | (NT=9,047, N=7,577) | ||||
| Any claim for antidepressant (%) | 9.6** | 6.5 | 3.1** | 10.8** | 9.2** | 1.6 |
| Percentage of days covered by claim (%) | 4.8** | 5.1** | .3 | 5.2** | 5.0** | .2 |
| N of mental health visits | 1.45** | 1.11** | .34 | 1.46** | .86** | .60** |
| N of outpatient visits | 4.25** | 3.54** | .71** | 4.51** | 3.42** | 1.09** |
| N of family or group psychotherapy sessions | .11** | .08** | .03 | .02 | –.01 | .03 |
| N of individual psychotherapy sessions | .76** | .62** | .14 | .67** | .39** | .28** |
| N of emergency department visits | .20 | .14 | .06 | .36** | .29** | .07 |
| N of admissions | .17** | .12** | .05 | .05 | .07 | –.02 |
| N of admissions for depression | .09** | .03 | .06 | .00 | .01 | –.01 |
| Total hospital days | –.73 | –.95 | .22 | –2.45** | –1.47** | –.98 |
| Total hospital days for depression | .12 | –.30 | .42 | –2.30** | –1.67** | –.63 |
| Schizophrenia | (NT=3,572, N=2,473) | (NT=3,494, N=2,490) | ||||
| Any claim for antipsychotic (%) | 8.3** | 7.1** | 1.2 | 14.8** | 14.2** | .6 |
| Percentage of days covered by claim (%) | 8.4** | 8.2** | .2 | 7.3** | 7.6** | –.3 |
| N of outpatient mental health visits | 2.36** | 1.41** | .95** | 2.20** | 1.86** | .34 |
| Total number of outpatient visits | 9.29** | 6.59** | 2.70** | 5.85** | 4.68** | 1.17* |
| Any assertive community treatment (%) | 10.0** | 7.4** | 2.6 | 3.1* | 2.5 | .6 |
| Months receiving assertive community treatment | .61** | .46** | .15 | .28** | .24** | .04 |
| N of family or group psychotherapy sessions | .04 | .01 | .03 | .09** | .13** | –.04 |
| N of emergency department visits | .40** | .48** | .08 | .27 | .19 | .08 |
| N of admissions | .15** | .15** | .0 | –.03 | .02 | –.05 |
| N of admissions for schizophrenia | –.10* | –.01 | –.09 | –.01 | .03 | –.04 |
| Total hospital days | –1.74 | –3.18 | 1.44 | 2.22 | .66 | 1.56 |
| Total hospital days for schizophrenia | 2.43 | –1.75 | 4.18 | 1.04 | .29 | .75 |
a
All measures reflect utilization during subsequent 6 months. Example interpretation: First outpatient visit (any provider) within 0–7 days versus >30 days is associated with a 9.6-percentage point increase in the probability of having any claim for an antidepressant during 6-month follow-up among persons discharged with a depression diagnosis in a medical home. NT, total discharges; N, total unique individuals
*
p<.05, **p<.01
The associations with subsequent inpatient utilization exhibited some differences between cohorts with and without a medical home. In the cohorts with a medical home, there was a small increase in all-cause (.15–.17, p<.01) admissions for early first outpatient visits (0–7 and 8–30 days) compared with no visits within 30 days. For these cohorts, early first outpatient visit was not associated with total hospital days (all cause or diagnosis specific). In contrast, in the non–medical-home cohorts, we did not find associations between timing of first outpatient visit and number of hospital admissions. We observed a meaningful reduction in total all-cause and depression-specific hospital days (−2.45, p<.01, and –2.30, p<.01) in the depression non–medical home cohort only.
Discussion
Policy makers, providers, and researchers are increasingly interested in improving mental health care quality. At the same time, the prevalence of multiple chronic conditions is increasing, and individuals with serious mental illness often have comorbid general medical conditions. However, evidence for appropriate quality metrics for the treatment of co-occurring mental health and general medical conditions is limited. Intuition suggests that early outpatient follow-up after psychiatric discharge may be beneficial, and the HEDIS measures of follow-up within seven and within 30 days reflect this idea. We found that early (within 30 days) outpatient follow-up was associated with increased medication utilization and outpatient encounters.
Given the inadequate treatment of persons with chronic comorbid psychiatric and general medical conditions (31–34), this greater utilization may reflect increased opportunities to engage individuals in outpatient settings. In addition, although medical home enrollees and nonenrollees are different populations (Table 1), our findings were generally similar for both groups. Although rates of outpatient follow-up after hospital discharge were not reported to medical home clinicians treating patients in our sample, other performance metrics (such as an annual eye examination for patients with diabetes) were regularly reported, which may contribute to unobserved differences in practice patterns between medical home clinicians and other clinicians. In addition, within the medical home cohorts, individuals with highest readmission risk were probably more likely to receive follow-up within seven days because of a CCNC clinical initiative to provide transitional care services to high-risk individuals that began during our study period (8). Because the highest-risk individuals for readmission were more likely to receive early follow-up, if we did not adequately control for this risk with our CIDPS comorbidity indicators and other covariates, our results could incorrectly suggest that earlier follow-up was associated with greater subsequent inpatient utilization, an association observed in our sensitivity analysis.
This study had several limitations. First, our follow-up period of six months did not allow us to generalize to long-term patterns of adherence, outpatient utilization, and readmissions, which may be important for the chronic conditions examined here. Second, we derived measures of adherence from claims data and did not observe actual medication use. Third, in order to capture all medication claims, we required six months of continuous Medicaid enrollment and excluded Medicare enrollees; therefore, findings may not generalize to patients who are eligible for both forms of insurance or who have short-term Medicaid enrollment. Finally, although our propensity score model included a rich array of service use measures before admission, we cannot rule out the possibility that unmeasured differences between follow-up groups (such as homelessness, incarceration, care preferences, motivation, or need for services) may have contributed to our findings.
Conclusions
After psychiatric hospitalization of persons with multiple chronic conditions that included depression or schizophrenia, among those not readmitted within 30 days, outpatient follow-up within 30 days may facilitate access to additional services during the next six months. We did not find strong evidence for systematic differences in service use between individuals receiving follow-up within 0–7 days after psychiatric discharge and those receiving follow-up within 8–30 days, with a few notable exceptions. Therefore, a universal seven-day threshold for first outpatient follow-up may not be meaningful for all adults with multiple chronic conditions discharged from inpatient treatment with a psychiatric diagnosis.
Supplementary Material
File (appi.ps.201400081.ds001.pdf)
- View/Download
- 195.45 KB
References
1.
Pincus HA, Page AE, Druss B, et al: Can psychiatry cross the quality chasm? Improving the quality of health care for mental and substance use conditions. American Journal of Psychiatry 164:712–719, 2007
2.
Pincus HA, Spaeth-Rublee B, Watkins KE: The case for measuring quality in mental health and substance abuse care. Health Affairs (Project Hope) 30:730–736, 2011
3.
Herbstman BJ, Pincus HA: Measuring mental healthcare quality in the United States: a review of initiatives. Current Opinion in Psychiatry 22:623–630, 2009
4.
Hermann RC, Leff HS, Palmer RH, et al: Quality measures for mental health care: results from a national inventory. Medical Care Research and Review 57(suppl 2):136–154, 2000
5.
Hermann RC, Palmer RH: Common ground: a framework for selecting core quality measures for mental health and substance abuse care. Psychiatric Services 53:281–287, 2002
6.
HEDIS Quality Measures. Washington, DC, National Committee for Quality Assurance, 2014. Available at www.ncqa.org. Accessed Feb 21, 2014
7.
Naylor MD, Aiken LH, Kurtzman ET, et al: The care span: the importance of transitional care in achieving health reform. Health Affairs 30:746–754, 2011
8.
Jackson CT, Trygstad TK, DeWalt DA, et al: Transitional care cut hospital readmissions for North Carolina Medicaid patients with complex chronic conditions. Health Affairs 32:1407–1415, 2013
9.
Enderlin CA, McLeskey N, Rooker JL, et al: Review of current conceptual models and frameworks to guide transitions of care in older adults. Geriatric Nursing 34:47–52, 2013
10.
Logue MD, Drago J: Evaluation of a modified community based care transitions model to reduce costs and improve outcomes. BMC Geriatrics 13:94, 2013
11.
Viggiano T, Pincus HA, Crystal S: Care transition interventions in mental health. Current Opinion in Psychiatry 25:551–558, 2012
12.
Vigod SN, Kurdyak PA, Dennis CL, et al: Transitional interventions to reduce early psychiatric readmissions in adults: systematic review. British Journal of Psychiatry 202:187–194, 2013
13.
Boyd CM, Leff B, Weiss C, et al. Full Report: Clarifying Multimorbidity to Improve Targeting and Delivery of Clinical Services for Medicaid Populations. Baltimore, Johns Hopkins University, Center for Health Care Strategies, Dec 2010. Available at www.chcs.org/media/Clarifying_Multimorbidity_for_Medicaid_report-FINAL.pdf
14.
Nasrallah HA, Meyer JM, Goff DC, et al: Low rates of treatment for hypertension, dyslipidemia and diabetes in schizophrenia: data from the CATIE schizophrenia trial sample at baseline. Schizophrenia Research 86:15–22, 2006
15.
Hennekens CH, Hennekens AR, Hollar D, et al: Schizophrenia and increased risks of cardiovascular disease. American Heart Journal 150:1115–1121, 2005
16.
Wang PS, Lane M, Olfson M, et al: Twelve-month use of mental health services in the United States: results from the National Comorbidity Survey Replication. Archives of General Psychiatry 62:629–640, 2005
17.
Osborn DP: The poor physical health of people with mental illness. Western Journal of Medicine 175:329–332, 2001
18.
Parks J, Svedsen D, Singer P, et al: Morbidity and Mortality in People With Serious Mental Illness. Alexandria, Va, National Association of State Mental Health Program Directors, 2006
19.
Jaro MA: Probabilistic linkage of large public health data files. Statistics in Medicine 14:491–498, 1995
20.
Dobson LA Jr, Hewson DL: Community Care of North Carolina: an enhanced medical home model. North Carolina Medical Journal 70:219–224, 2009
21.
Willson CF: Building primary care medical homes within the Community Care of North Carolina program. North Carolina Medical Journal 70:231–236, 2009
22.
Lichstein JC, Domino ME, Beadles CA, et al: Use of medical homes by patients with comorbid physical and severe mental illness. Medical Care 52(suppl 3):S85–S91, 2014
23.
Martin BC, Wiley-Exley EK, Richards S, et al: Contrasting measures of adherence with simple drug use, medication switching, and therapeutic duplication. Annals of Pharmacotherapy 43:36–44, 2009
24.
Kronick R, Gilmer T, Dreyfus T, et al: Improving health-based payment for Medicaid beneficiaries: CDPS. Health Care Financing Review 21:29–64, 2000
25.
Brookhart MA, Schneeweiss S, Rothman KJ, et al: Variable selection for propensity score models. American Journal of Epidemiology 163:1149–1156, 2006
26.
Spreeuwenberg MD, Bartak A, Croon MA, et al: The multiple propensity score as control for bias in the comparison of more than two treatment arms: an introduction from a case study in mental health. Medical Care 48:166–174, 2010
27.
Feng P, Zhou XH, Zou QM, et al: Generalized propensity score for estimating the average treatment effect of multiple treatments. Statistics in Medicine 31:681–697, 2012
28.
Robins JM, Hernán MA, Brumback B: Marginal structural models and causal inference in epidemiology. Epidemiology 11:550–560, 2000
29.
Zou G: A modified Poisson regression approach to prospective studies with binary data. American Journal of Epidemiology 159:702–706, 2004
30.
Austin PC: Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples. Statistics in Medicine 28:3083–3107, 2009
31.
Lawrence D, Kisely S: Inequalities in healthcare provision for people with severe mental illness. Journal of Psychopharmacology 24(suppl):61–68, 2010
32.
Lord O, Malone D, Mitchell AJ: Receipt of preventive medical care and medical screening for patients with mental illness: a comparative analysis. General Hospital Psychiatry 32:519–543, 2010
33.
Druss BG, Zhao L, Cummings JR, et al: Mental comorbidity and quality of diabetes care under Medicaid: a 50-state analysis. Medical Care 50:428–433, 2012
34.
Kaufman EA, McDonell MG, Cristofalo MA, et al: Exploring barriers to primary care for patients with severe mental illness: frontline patient and provider accounts. Issues in Mental Health Nursing 33:172–180, 2012
Information & Authors
Information
Published In
Cover: Breakfast in the Garden, by Frederick Carl Frieseke, circa 1911. Oil on canvas, 26 x 325/16 inches. Daniel J. Terra Collection, 1987.21. Terra Foundation for American Art, Chicago/Art Resource, New York City.
History
Received: 2 March 2014
Revision received: 12 June 2014
Accepted: 21 August 2014
Published online: 15 December 2014
Published in print: April 01, 2015
Authors
Competing Interests
Mr. Ellis has received research grant funding from Amgen. Dr. Farley has received consulting support from Daiichi-Sankyo for unrelated research. Dr. Domino has received salary support for unrelated projects from CCNC. The other authors report no financial relationships with commercial interests.
Funding Information
National Institute of Nursing Research10.13039/100000056: 2T32NR008856
Agency for Healthcare Research and Quality10.13039/100000133: R24 HS019659-01
This work was supported by grant R24 HS019659-01 from the Agency for Healthcare Research and Quality (AHRQ) and North Carolina Community Care Networks and by AHRQ grant 5T32-HS000032. Additional funding was provided for Dr. Beadles by grant TPP 21-023 from the U.S. Department of Veterans Affairs and for Dr. Lichstein by grant 2T32NR008856 from the National Institute of Nursing Research.Metrics & Citations
Metrics
Citations
Export Citations
If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.
For more information or tips please see 'Downloading to a citation manager' in the Help menu.
View Options
View options
PDF/EPUB
View PDF/EPUBLogin options
Already a subscriber? Access your subscription through your login credentials or your institution for full access to this article.
Personal login Institutional Login Open Athens loginNot a subscriber?
PsychiatryOnline subscription options offer access to the DSM-5-TR® library, books, journals, CME, and patient resources. This all-in-one virtual library provides psychiatrists and mental health professionals with key resources for diagnosis, treatment, research, and professional development.
Need more help? PsychiatryOnline Customer Service may be reached by emailing [email protected] or by calling 800-368-5777 (in the U.S.) or 703-907-7322 (outside the U.S.).