Evaluating the Proactive C-L Model: Insights and Unanswered Questions from the UK HOME Study

Abstract

The HOME Study is the largest and among the most important trials in the history of consultation-liaison psychiatry, and its study team is to be commended for this landmark study. In this article, we provide context for the HOME Study and consider several aspects of the study critical to understanding the nature of the intervention and interpreting its results. First, we compare the model of proactive integrated consultation-liaison psychiatry as implemented in the HOME Study with versions of proactive consultation-liaison psychiatry as commonly practiced in the US. Key distinctions of proactive integrated consultation-liaison psychiatry include a direct assessment of all study arm participants by a consultation-liaison psychiatrist rather than the use of initial screening for acute psychiatric issues, the unique inclusion of an occupational therapist as a member of the proactive integrated consultation-liaison psychiatry team, and patient-level randomization as opposed to unit-based approaches as commonly practiced in the US. Next, we consider several characteristics of the HOME Study sample relevant to its generalizability. These include an average age of 82 years, limited ethnic and racial diversity, and a high prevalence of both cognitive and functional impairment. Third, we review how study methodology informs study interpretation. These include early trial termination due to COVID, which limits power to detect a 1-day reduction in hospital, a mean 3.5-day delay from hospital admission to study enrollment, and the exclusion of patients who had already received a psychiatric consultation, which likely reduced the acuity of mental health issues addressed in this study. Despite these considerations, the HOME Study is a truly remarkable contribution to the literature, and its results will be discussed for years to come. In view of the aging global population, the HOME Study set about to tackle an especially ambitious and forward-looking question by focusing exclusively on older hospitalized adults. Like all good studies, this trial raises many important questions. As the first randomized trial of proactive consultation-liaison psychiatry of any form, the HOME Study is an encouragement to the field to consider the range of potential benefits of providing proactive, integrated mental health care to medical and surgical inpatients with mental health needs.

The model of care known as proactive consultation-liaison (C-L) psychiatry continues to expand, with growing implementation across hospitals in academic and community settings (1), adults and children (2), as well as general medicine and critical care units (3). A growing body of evidence has found this model of care a financially viable means of improving and expanding access to mental health care among medical-surgical inpatients (4). Across inpatient settings, proactive C-L has been shown to improve a variety of metrics, including length of stay, staff satisfaction, and time to consultation; however, prior studies have been largely designed as quality improvement projects—without randomization (4, 5).

The September issue of The Lancet Psychiatry features the results of the HOME Study,(6) the first randomized trial of any form of proactive C-L psychiatry and, with 2744 participants, the largest randomized clinical trial in the field of C-L psychiatry to date. This study was truly ambitious in scope, and its successful conduct is a testament to the investigators’ extraordinary investment in improving hospital-based mental health care for medically admitted patients around the globe.

Foremost, Sharpe et al. demonstrated that this intervention, which they describe as proactive, integrated consultation-liaison psychiatry (PICLP), can be implemented at scale and in a way that is highly regarded by non-psychiatric providers receiving support, the patients themselves, as well as the mental health clinicians delivering PICLP services (7, 8). The HOME Study is a landmark clinical trial that will reverberate across the field of C-L psychiatry for years to come. PICLP, as studied in this trial, is an offshoot of the model of care known in the US more commonly by the monicker proactive C-L. Rather than focusing chiefly on meeting mental health needs, PICLP sought to help wards provide biopsychosocial care for older adult patients with complex problems, in many ways as an inpatient corollary to outpatient models of collaborative care.(9)

In this parallel-group trial conducted across 24 medical wards of three UK hospitals, participants were randomized at the patient level to receive PICLP services or care as usual. For the primary outcome, participants in the PICLP study arm spent 0.45 fewer days in hospital within 30 days of randomization than usual care participants; however, this difference did not reach statistical significance (95% confidence interval −1.11, 0.21 days). Across secondary outcomes, the intervention group had a statistically higher daily discharge rate over the whole index admission (relative risk 1.09; 95% confidence interval 1.00, 1.17) and a lower mortality rate prior to the onset of the COVID-19 pandemic (relative risk 0.88; 95% confidence interval 0.76, 1.00). PICLP was likely cost-effective up to 1 and 3 months, but not 12 months.

Comparing PICLP with proactive C-L in the US

Although most proactive C-L services share broadly similar aims, care delivery across individual services can differ in important ways. How does PICLP in the HOME Study compare with models of proactive C-L as commonly practiced in the US? Let’s consider the four aspects of proactive C-L as operationalized in the recent American Psychiatric Association Resource Document.(9) Even though there is diversity across proactive C-L services, Table 1 illustrates several key distinctions between the model of care delivered in the HOME Study relative to its usual implementation in the US.

Table 1:

Comparing PICLP to proactive C-L as commonly practiced in the US

	PICLP	Proactive C-L
Systematic screening	In the HOME Study, patients were not screened for acute mental health needs. Rather, older adults with biopsychosocial complexity were regarded as “patients who might benefit” from proactive psychiatric involvement. All patients randomized to PICLP were evaluated in-person by a psychiatrist.	The initial step is screening for patients with active mental health conditions. Screening can be manual, automated, or a hybrid of the two.(11) For the rare proactive C-L team that screen all unit patients in person, this is done by a social worker.
Proactive care engagement	Initial engagement in care is by in-person psychiatric evaluation. Universal assessments by an attending psychiatrist may limit the ability to prioritize patients with more acute concerns.	Initial engagement in care is by way of reviewing positive screens with a primary medical team/practitioner, nursing, or as part of interdisciplinary care rounds to understand active concerns and confirm whether there is an indication for consultation.
Interdisciplinary approach	PICLP teams comprised psychiatrists, psychiatrists in training, and psychiatric occupational therapists.	Interprofessional teams, including psychiatrists, psychiatric nurse practitioners, social workers, recovery coaches, nurses, trainees, et al., tailored to the clinical population and goals of the service.
Integrated care delivery	Participants were randomized at the patient level. As such, care interventions were delivered to individual patients rather than as unit-level support. All participants received a detailed biopsychosocial formulation and systematic management plan. Per the study authors, there “may have been contamination of usual care.”	Care is delivered as a unit-based model with the goal of benefitting from clinical relationships and point-of-care education, resulting in a spillover or halo effect.(9, 15) Care services, including which team members are involved with a given patient, are tailored to the patient’s needs.

Of these distinctions, perhaps the one that stands out most is that a psychiatrist personally evaluated every patient randomized to the PICLP arm of the HOME Study. One wonders what proportion of these patients had an indication for a psychiatric evaluation or would have screened “positive” with proactive screening (i.e., indicating a potentially active psychiatric issue) as is generally performed in proactive C-L. Across the US, C-L psychiatrists all but uniformly serve as consultants, providing recommendations to primary practitioners or medical-surgical teams rather than directly to patients as one might in a co-management model.(10)

It would be hard to overemphasize the importance of screening, as it is impracticable for a psychiatrist or, as in this study, a C-L psychiatrist, to complete a biopsychosocial assessment on every patient admitted to every medical unit. Whereas this Cadillac approach can provide insights about the true scope of mental health issues across the study sample, it does not reflect clinical practice and sacrifices efficiency for the sake of comprehensiveness. We suspect that optimized approaches of screening for mental health issues, by way of direct clinical contact with primary services or as aided by electronic medical records (11), to be a much more feasible and scalable approach across a broad healthcare landscape.

We also consider how services were integrated in the HOME Study. The team composition implemented in the HOME Study included psychiatric clinicians and occupational therapists as integrated members of the ward team. To our knowledge, this is the only study of proactive C-L to investigate this team composition. More commonly, proactive C-L teams include psychiatrists, psychiatric nurse practitioners, and psychiatric social workers, with other clinicians as dictated either by the needs of the clinical population or, practically speaking, availability. The goal is to develop relationships with unit staff of corresponding clinical backgrounds and relevant expertise. It is understandable that occupational therapists were included in this study that enrolled only older adults, but this decision does constrain generalizability of the study results.

Finally, whereas PICLP provided clinical recommendations to teams, care delivery was done strictly at the patient level. Evaluating proactive service outcomes poses methodological challenges, as there are tradeoffs when randomizing individual patients vs randomizing ward units. By randomizing individual patients, the HOME Study sought to isolate the effectiveness of PICLP, recognizing that patients and staff often move across wards. As a result, though, the HOME Study was unable to benefit from one of the more important value propositions offered by proactive C-L—unit-wide benefits or, alternatively, either halo or spillover effects. In our experience of proactive C-L, unit teams find that regular exposure and access to a multidisciplinary team of mental health specialists to be among the most rewarding aspects of this model of care (12). Rather than being concerned about potential “contamination of usual care,” as noted by the HOME Study authors, quality improvement projects of proactive C-L seek to capitalize on the prospects of improving mental health awareness across unit staff. Ultimately, this kind of unit-based integration is a means of advocacy for mental health care.

Generalizability of the HOME Study sample to clinical populations in the US

Beyond the model of PICLP itself, how well does the HOME Study sample reflect the population of patients in US hospitals who might receive proactive C-L services? The HOME Study enrolled adults 65 years and older, which means that young and middle-aged adults were excluded from participating. The average age of 82 years in the HOME Study is roughly two to three decades older than the samples receiving proactive C-L services in our own published work (13–15) (Table 2), as well as most other proactive C-L samples (5). Next, whereas the male to female ratio is nearly 1 to 1 across the samples shown in Table 2, there is far less racial and ethnic diversity in the HOME Study sample than in analogous samples from our experiences in US hospitals, which raises questions about the external validity to such diverse settings (16). Additionally, the average length of stay in the HOME Study was longer than in these comparison samples, likely reflecting the higher average age of the HOME Study participants and the related complexity of discharge readiness.

Table 2:

HOME Study sample compared with proactive samples^* published by the authors

	HOME Study	Oldham et al. (13)	Triplett et al. (14)	Sledge et al. (15)
Age	82.3 yrs.	64.2 yrs.	52.6 yrs.	51.2 yrs.
Male	51%	54%	45%	51%
White, non-Hispanic	93.4%	66.4%	45%	70%
Length of stay	10.4 days	9.2 days	7 days	6.7 days
Institutions	Royal Devon and Exeter, John Radcliffe, Addenbrooke’s	Strong Memorial Hospital	Johns Hopkins Hospital	Yale-New Haven Hospital

^*The values included in this table are restricted to those receive proactive C-L services

The HOME Study sample’s average score on the Montreal Cognitive Assessment was 13, which in the dementia range,(17) and the average Barthel Index was 46, indicative of marked functional dependence.(18) In view of such a striking prevalence of dementia in the HOME Study, one wonders how effective a C-L psychiatry service using a biopsychosocial approach might be during the brief period of a single index hospitalization. For example, medication options for managing dementia and its neuropsychiatric disturbances are extremely limited, and one would hardly expect that medication adjustments would have a quick enough effect to facilitate discharge in more than a fraction of the participants in this sample. Moreover, such a high prevalence of dementia means that most clinical issues would not have been the more traditional primary psychiatric disorders such as schizophrenia, posttraumatic stress disorder, or major mood disorders. One suspects that many of the barriers to discharge involved availability of community services rather than simply clinical stability, as is the bread and butter of proactive C-L.

Methodological considerations

It is also important to consider how study methodology might inform our interpretation of the HOME Study results. The HOME Study was stopped early due to the COVID-19 pandemic, and this reduced the power to detect a target 1-day between-group difference from 90% to roughly 80%. Although it is unlikely that a modestly larger sample, as originally planned, would have increased the point estimate of effect from roughly half a day to a full day, it could have altered the statistical significance of the half-day reduction.

Participants in the HOME Study had been hospitalized for an average of 3.5 days prior to study enrollment, both due to trial procedures and the delay in transferring from the ED to the medical ward. This is important because time to consultation has been consistently associated with hospital duration.(19) Additionally, participants who had already received psychiatric consultation prior to evaluation for study eligibility, presumably due to acute, clearly evident mental health concerns, were excluded from participating in the HOME Study. This likely further reduced the overall acuity of mental health needs in the study sample.

Interpreting the HOME Study results

These challenges and methodological limitations only underscore this remarkable contribution by the HOME Study team, which sets the stage for future trials.(20) Beyond demonstrating that this version of proactive C-L can be delivered at scale, this trial provides crucial lessons to be learned by future investigators regarding sample sizes and outcomes. Although the primary outcome was statistically insignificant, even a half-day reduction, if real, would be momentous.(21, 22) The clinical and administrative benefits of shorter hospitalizations include more efficient bed utilization and improved access to inpatient care due to increased turnover. It can also be financially advantageous, especially in fee-for-service reimbursement settings across the US. Small changes in length of stay across a large inpatient population add up.

Despite the uncertain generalizability of the HOME Study results to many other hospital populations, especially those in the US, one could argue that this study set about to answer an even more ambitious public health question given the aging global population and growing rate of cognitive disorders.(23) Quoting the HOME Study authors, “Growing evidence suggests that psychiatric and psychosocial problems, including cognitive impairment, depression, anxiety and the social implications of functional dependency, are critical factors in prolonging hospital stays.”(24) Hospital-based C-L psychiatry must be prepared to provide care for older patient populations.(13)

These capstone results also build on prior analyses from the HOME Study, which have shown that participants and medical staff appreciate the value of PICLP.(7, 8) Further, the HOME study serves as a call for pragmatic trials of proactive C-L as it is commonly practiced in diverse settings, under real-life circumstances. Though this study’s target of a 1-day length-of-stay reduction was not achieved, there are other signs of the program’s success: for instance, the compelling suggestion that PICLP reduced 1-year mortality merits studied attention. We would love to see a generation of creative trials that seek to define the benefits of the innovative and intuitively valuable approach of proactive C-L not only for patients and their families but also to clinicians and entire healthcare systems.