Organization and Outcomes of Integrated Inpatient Medical and Psychiatric Care Units: A Systematic Review

IMPUs are hospital wards that diagnose diseases and disorders and treat hospital inpatients who have concomitant medical and psychiatric disorders or behavioral problems (7).

We comprehensively searched Embase, Medline, Web of Science, PsycINFO, Scopus, CINAHL, Cochrane, and Google Scholar on May 24, 2019. The search strategy was based on use of related terms for IMPUs in the title and abstract fields, and the search syntax is included in an online supplement to this article. We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (17).

The review included English-language, descriptive, evaluative, and comparative studies presenting relevant empirical data on IMPUs and published in peer-reviewed journals. We excluded perspectives, expert opinions, letters to the editor, and meeting abstracts. We also excluded studies reporting on specialized units focusing on a single disease, for example, delirium or dementia; psychosomatic units that primarily focus on the clinical psychotherapeutic treatment for psychosomatic disorders (18); and interventions that mainly consist of embedded consultation (“liaison”) by a psychiatrist or another physician on a medical or psychiatric ward without other staff or ward modifications.

After removing duplicates, the first author (M.A.v.S.) screened all titles and abstracts of the studies identified. The full texts of the included studies were retrieved and checked for eligibility by two authors (M.A.v.S. and J.D.H.v.W.) independently. These authors sought consensus for final inclusion in the qualitative synthesis. The online supplement gives details on the study selection. All studies included were searched for aims, structural characteristics, process characteristics, and outcomes in accordance with Donabedian’s structure-process-outcome framework (14). The two authors independently extracted data from 20 randomly selected studies of the included articles to reach a consensus on the relevance of IMPU characteristics. The structural characteristics considered relevant were hospital type and size, unit size and embedding, ward features (any physical characteristics), medical staff (medical direction, attending and consulting physicians, residents), nursing staff (primary specialty, specialist nurses, and cross-training), and other staff. The procedural characteristics considered relevant were referral sources, age focus (children and adolescents, adults, or elderly persons), population descriptors, collaboration, medical and psychiatric treatment capabilities, LOS, and aftercare. We considered funding to be an external, contextual factor rather than an internal, structural factor. The first author completed data extraction for the complete set of included studies (see the online supplement).

The first two authors (M.A.v.S., J.D.H.v.W.) independently assessed and scored each study’s quality by using the GRADE system (19) and sought consensus in cases where scores differed between them.

We followed an inductive approach relying on each empirical study’s data instead of deductively relying on existing IMPU categorizations. The first data analyses revealed that the existing concepts of unit type and acuity capabilities were beneficial to structure the analysis.

We characterized IMPUs as follows. Units that treated general medical and psychiatric patients in a unified setting were labeled as “nonsegregated.” Units that originated from or were located on a psychiatric unit were labeled “psych-med units.” Conversely, units that arose from or were located on a medical unit were considered “med-psych units.” Units that had an equivalent contribution of medical and psychiatric staff, irrespective of unit origin, were defined as “combined.” Finally, we designated the colocation of two complementary specialist units as “adjacent units.”

The first two authors independently scored each unit for medical and psychiatric acuity capabilities (1, low; 2, medium; and 3, high), following Kathol et al.’s definitions (13). A low level of medical acuity corresponded to chronic medical problems that were well controlled and required minor treatment adjustment. Patients with a medium level of medical acuity had an active, non–life-threatening condition requiring immediate medical intervention. Patients with high medical acuity required direct medical intervention, without which they may have died (9). A low level of psychiatric acuity corresponded to an easily managed behavioral problem or psychiatric illness (13). Patients with a medium level of psychiatric acuity had an active psychiatric condition that posed no acute danger to themselves or others. Patients with high psychiatric acuity required admission to a locked ward or an open ward with close supervision because of dangerous behavior (13). All units received a medical and psychiatric acuity capability score, even when acuity-defining data were incomplete. The authors used a Likert scale (ranging from 1, “not sure at all,” to 5, “completely sure”) to indicate their perceived level of precision in assigning the acuity capability score.

The main findings focused on IMPU aims, structure and process characteristics, outcomes, and study quality. Findings also focused on age group and unit type when considered appropriate. The heterogeneity of the included studies precluded a meta-analysis of outcomes.

Most IMPUs were in the United States (N=23, 59%) or Europe (N=11, 28%). IMPUs were most commonly located in university and teaching hospitals (N=26, 67%). Nine units (23%) were independent, nine (23%) were a subunit, and two (5%) were an annex (“outgrowth”) of the psychiatric hospital into the general medical hospital. Some units reported on their location within the hospital: amid other primary care inpatient units (AE11), contiguous to medical wards (CA3, AE2, AE13), at the core of a center for the elderly (E6), within one block of the hospital (CA4), or half a mile from the hospital’s main center (AE3).

The online supplement lists the aims of IMPUs. IMPUs generally aimed to treat patients with comorbid medical and psychiatric illnesses that were highly prevalent and increasingly common in the hospital environment. Because of their psychiatric illness or behavioral problems, these patients could not be treated satisfactorily on a general medical floor, or, vice versa, their medical needs could not be met on a psychiatric ward (58). Care for these patients was often fragmented, causing multiple ward moves or even refusal of care (5, 12). IMPUs aimed to remedy this fragmentation by providing integrated medical and psychiatric care in an adequate setting with both medical facilities and psychiatric safety features and trained nursing staff (8, 9). Patients with a psychiatric comorbid condition or behavioral disturbances were often regarded as “problem cases” and had unmet health care needs in the medical setting. IMPUs provided an excellent training location for residents and medical students in biopsychosocial diagnosis and treatment. Treatment in a medical setting could improve patient acceptance and reduce stigma. The top five specific aims of IMPUs (Table 1) included integration, decompartmentalization, and continuity of care (N=14, 36%); quality and safety (N=13, 33%); improved patient-related and economic outcomes (N=11, 28%); training professionals (N=10, 26%); and managing disruptive behavior or high health care utilization (N=9, 23%).

Most IMPUs (N=23, 59%) were categorized as psych-med units, eight (21%) were a med-psych unit, three were a nonsegregated unit, one was a combined unit, and one was an adjacent unit (E7). Three units could not be classified.

Of the 16 comparative studies, 14 reported quantitative measures (9, 57–68, 70). Table 1 shows that according to the GRADE system (19), the quality of the included studies was mostly low (57–62, 65–67), often because of indirect comparisons. Table 1 also shows that most studies focused on LOS as the intended outcome, congruent with the aim to improve economic outcomes (see the online supplement). The achievement of the other aims received limited attention.

The findings of this review reveal that a classification based on IMPU types provides information on the units’ origin and embedding. Although they were not strictly IMPUs, we chose to include nonsegregated units because one could think of “nursing in a shared environment” as one end of the integration spectrum and combined units as the other end (i.e., full integration). Mixed or dually trained nursing staff, cross-training, and the active involvement of both medical and psychiatric physicians are typically in place for reaching medium-to-high medical and psychiatric acuity capabilities, in line with findings by Kathol et al. (12). In addition to staffing, medical and psychiatric exclusion reasons and population focus were important determinants of medical and psychiatric acuity capabilities.

Because we followed an inductive approach relying on each study's empirical data, we did not apply Kathol et al.’s (13) categorization of medical-psychiatry units in the data synthesis. On the basis of median medical and psychiatric acuity capabilities, however, the findings of this review suggest that type III units (13, 65), which provide care for patients with low to high psychiatric acuity as well as medium medical acuity, are the most prevalent type among IMPUs. Kathol et al.’s (13) definition of type III units, however, precludes unit embedding outside general medical hospitals because emergency physician coverage and staff training are then much more difficult to provide. We nevertheless found IMPUs with medium medical acuity capabilities located outside the main center of general medical hospitals. Furthermore, seven units had no acute medical capabilities, psychiatric capabilities, or both. One may conclude that in the current health care environment, nongeneral medical hospital units or those without acute discipline–specific capabilities would no longer be included among those calling themselves IMPUs.

Some authors rightly point to the tension between acute medical care and maintaining process-oriented (psychiatric) treatment conditions, such as the therapeutic milieu and delivery of psychotherapeutic interventions. The discussion about the interactions between acute medical care and process-oriented treatment continues. An IMPU, for instance, might exclude or transfer patients whose medical or psychiatric conditions become too acute. However, it might be undesirable to exclude from IMPUs patients with severe mental illness because of general medical causes, because these patients might benefit most from integrated care (5, 16). Solutions to these challenges are the flexible use of extra staff and allowing IMPUs to welcome a differentiated population and focus on acuity.

Location and financing also determine medical acuity capabilities. Some psych-med units were not located closely to other medical or surgical units and to those having medical testing, intervention, and consultation capabilities. This physical distance complicates collaboration among medical specialists. Further, psychiatric payor–financed IMPUs might not be financially profitable because of underpayment of acute medical care, unless patients are transferred to medical units for general medical needs. Of course, this transfer defeats the interdisciplinary purpose of psych-med units. As a result, psych-med units having a segregated psychiatric payment system will need to be rededicated as med-psych units so that medical payors cover the total cost for treating patients with complex conditions (12).

This review found few data on the cost-effectiveness of IMPU designs. Nevertheless, some proposals for cost-effectiveness can be formulated. Kathol et al. argued that medical-psychiatric units should primarily serve patients with complex conditions to maximize health system value and that these patients often have severe and acute (i.e., high-acuity) medical and psychiatric illnesses (12). For these patients, proactive psychiatric consultation-liaison teams in standard general medical health settings are insufficient (12). Relatively simple ways to improve the acuity capabilities of many low-to-medium acuity IMPUs to serve patients with complex conditions better would be active involvement of both psychiatric and medical staff on a single medical unit for patients with comorbid psychiatric and general medical conditions, cross-training of nursing staff, and a focus on acute general medical and psychiatric care capabilities.

However, the studies included here did not contain much information on facility design. Kathol et al. advised on the implementation of facility design (12, 13, 72). Most IMPUs focus primarily on patients with mood disorders. In the future, IMPUs should put greater emphasis on other psychiatric patient groups with significantly increased mortality rates due to illnesses such as delirium, schizophrenia spectrum disorders, substance use disorders, and severe somatic symptom disorders. To further improve value, consideration should be given to what medical and psychiatric care could be replaced by integrated outpatient care (16, 73). At the upper end of the acuity spectrum, clinical experience suggests that IMPUs for high-acuity patients might prevent critical care unit admission, for instance, of severely agitated patients.

From the patient's perspective, their functioning, quality of life, and care continuity are essential goals (74). These aims can often be pursued in a unit with both general medical and psychiatric capabilities, including process-oriented treatment and resocialization. From the perspective of hospitals and insurers, an IMPU should treat patients who have disruptive behavior, thereby better facilitating medical treatment, relieving the referring wards, and potentially shortening LOS and reducing readmissions (16). A short-stay medical-psychiatric unit can achieve these outcomes. Such units admit patients with the most severely disruptive behavior while sending patients with less severe disruptive behavior back to their referring wards. Although specialization of IMPUs might be feasible in some densely populated regions, other regions will need design compromises that would reflect regional needs and alternative facilities available in the local health care network (75). Because patient acuity levels tend to fluctuate (13), some IMPUs may be designed such that they promote care continuity to avoid back-and-forth patient referrals.

The current conceptualization of IMPUs mainly focuses on clinical and horizontal integration of psychiatric and medical services within hospitals. To improve long-term outcomes, integration across primary, community, hospital, and tertiary care services (i.e., vertical integration) might also be promising (76). Such a development would involve other critical elements of integrated care, such as a single point of entry, continuity of care, health promotion and (proactive) prevention, a participatory approach (including shared decision making), case management, training of professionals, risk stratification, and information sharing (74, 77, 78).

Many of the studies included were published >10 years ago, limiting the validity of their findings for present-day IMPUs. The included papers contained limited information on the context of IMPUs, including funding. Therefore, we could not make firm evidence-based statements on the influence of any contextual factors. Furthermore, this review's external validity might have been limited by the overrepresentation of studies done in the United States, university and teaching hospital settings, and psych-med units, as opposed to med-psych or combined units. The extent to which the units included reported structural and procedural characteristics varied widely, which might have introduced a negative bias for units with sparse reporting of acuity capabilities. Besides, the classification scale used to estimate acuity capabilities is coarse and has not been validated. Several of the included studies did not contain enough information to enable statements about the actual acuity of the patients treated in each unit. Although we could classify most included studies, many described medical and psychiatric acuity capabilities in insufficient detail. Many of the included papers offered advice on setting up or running an IMPU, but summarizing these suggestions and guidelines was beyond the scope of this review.

We encourage future studies to explicitly describe the aims, population focus, staffing model, medical and psychiatric exclusion criteria, and population focus of IMPUs (75). Outcome measures related to the top 5 aims of IMPUs deserve much more attention. Reporting of patient-related outcomes and values for the hospital and the health system over more extended periods is needed (16). Research into the costs of various IMPU models in relation to their patient populations’ morbidity can advance the necessary understanding of these models’ cost-effectiveness. Furthermore, research into staff development, such as dual training of physicians, deployment of advanced practice providers, and cross-training methods, is also needed. Eventually, the contextual factor of funding should promote evidence-based IMPU designs and operation. Policy makers could contribute by designing and promoting integrated financial arrangements that promote enhanced medical acuity capabilities. In the United States, especially psych-med units might benefit from such arrangements.