Early Intervention Services for Schizophrenia: Looking Back and Looking Ahead

Abstract

This invited commentary provides a conceptual history of modern early intervention services, briefly reviews the accomplishments of an international clinical and research community, and offers proposals for how such services might participate in the next generation of progress. In keeping with the theme of this column, we make the argument that such services should orient around bi-directional knowledge translation across basic, clinical and policy domains.

Looking Back: The Pathway to Modern Early Intervention Services (EIS)

A seminal 1998 paper¹ summarized a vast prior observational literature, demonstrating that much of the overall functional decline associated with chronic schizophrenia spectrum disorders occurred within the first 3−5 years after psychosis onset, labeled as a critical period for early intervention. This reduction from premorbid levels of functioning was usually followed not by progressive deterioration, but instead a plateauing and then slow recovery of function over decades. Unlike the chronic illness phase, early course remission was common, but punctuated by frequent relapse. Several modifiable prognostic factors emerged during the critical period, including substance use, medication non-adherence and treatment disengagement. The delay between psychosis onset and effective treatment (Duration of Untreated Psychosis, DUP) has since been robustly associated with poorer outcomes despite wide variability in measurement, illness severity and quality of care.^2,3 Closer attention to this period has revealed a range of more immediate adversities, including higher risks for suicide and aggression.⁴ A growing literature on pathways to care during this period has revealed myriad, interacting sources of delay: the distressing and confusing subjective experiences of the patient (with variable illness awareness),^5–7 the predicament of families and the wider community (with limited knowledge of psychotic illness and how to enable healthcare access),⁸ and healthcare systems that are difficult to navigate.^9,10 This can result in unnecessarily aversive interactions with criminal justice¹¹ and involuntary hospitalization¹² that can further inhibit entry into stable outpatient care.⁹ Also, what awaits patients and families at the end of such tortuous pathways often lacks an adequate range or coordination of services, or is poorly tuned to the needs of young adults.¹³ In summary, while the prognostic gloom associated with early-Kraepelinian dementia praecox had long been challenged,¹⁴ studies of the critical period revealed several specific opportunities for early intervention.

Over the last three decades, published models have demonstrated improvements in both access and care quality. First, the TIPS project¹⁵ pioneered early detection (or the reduction of DUP) at a community level, and has now been replicated in a more fragmented US healthcare context.¹⁶ Aside from the direct relief to patients and their families, such early detection (ED) campaigns can activate community members outside the healthcare sector (e.g. education, criminal justice, clergy) to transform an otherwise harrowing experience of help-seeking.^17–19 Given the myriad and changing sources of delay across healthcare systems, ED remains difficult to accomplish,²⁰ but recent successes in targeting component²¹ and overall²² delay offer specific strategies. Early detection has also paid off in improved short^23,24 and long-term^25,26 outcomes, even without enriching care. Second, the quality of subsequent clinical care has acquired a new international standard. This has involved the use of specialized multidisciplinary teams who can draw from a menu of empirically supported treatments to intervene comprehensively, and in a developmentally appropriate manner. The evidence for these first-episode services (FES, or Coordinated Specialty Care in the United States) has progressed from pioneer clinics,²⁷ to observational studies across varied healthcare systems,¹ to experimental tests of intensively resourced^28,29 and pragmatic³⁰ service models, with demonstrated impact across ecologically relevant settings.³¹ FES has also been implemented as a care standard within national health policies^32,33 with impressive generalization of impact when included in routine care pathways.³⁴ In summary, commissioners of modern early intervention services (EIS) can draw lessons for implementation from a rich international literature, and investments in ED and FES can be made with an expectation of meaningful improvements in population health³⁵ and economic value.^36,37

Looking Ahead

Earl(ier) Intervention and Prevention

“First-episode” psychosis (FEP) has emerged as a useful construct for service delivery and clinical research. The modern syndromal construct of psychosis tags positive symptoms of sufficient severity and conspicuousness to serve both as a reliable target for community members (to assist in early detection), and for clinical services (as a marker of need for professional care). The rest of the term FEP is a misnomer in that care can only rarely begin in the midst of a “first” manifestation of positive symptoms, and “episodes” reflect arbitrary severity thresholds that are met over longer prodromal periods of continuous or intermittent symptoms.³⁸ Nevertheless, in practice, FEP has enabled an intent to engage individuals with predominantly schizophrenia spectrum disorders.³⁹

Given the above, efforts to identify children, adolescents and young adults who are at increased risk for FEP is a natural extension of the project of early intervention. However, studies of populations variously defined to be at higher risk for psychosis have so far revealed significant limitations in our ability to prognose or intervene during this illness phase.⁴⁰ Most such at-risk individuals (~75%) do not “convert” to psychosis over the following 3 years, although many will likely benefit from some form of care.^41,42 Interventions that can delay or prevent progression are yet to established. While these are empirically tractable questions for ongoing research that may deliver earl(ier) interventions for schizophrenia, the overall program has been criticized for overpromising on what should be seen now as a distinct goal: prevention.⁴³ Happily, as in the rest of medicine, specific preventative interventions can be mounted prior to knowledge of pathophysiology (e.g. smoking prevention and lung cancer) and a compelling analogous argument has been made to limit cannabis exposure to reduce the incidence of psychosis.⁴⁴ With current knowledge of risk stratification however, this would however require an effort far in excess of the scale of ordinary clinical services.⁴⁵ A false choice between such preventive efforts and ongoing research with high-risk samples can be avoided. The latter project is better conceived of not as prevention, but the study of early course illness “exemplars”—common in the rest of medical science—where careful examination of idealized prototypes of diseases,⁴⁶ which need not represent any kind of population average, can nevertheless reveal pathophysiologic processes of more universal relevance. A recently initiated multi-national effort to discover biomarkers to guide mechanistically oriented intervention in high-risk states is one example of this program of work.⁴⁷

How can extant EIS (that are currently focused on the care of FEP) contribute to these upstream efforts? Early detection can be deployed within a population health framework that assertively identifies all early course (including high risk) samples across a defined geographic catchment.¹⁹ This could permit better calibration of EIS to the actual needs of community referrers (who often cannot distinguish between prodromal and FEP cases), leverage opportunities to head off aversive pathways (by engaging patients at earlier illness phases with better illness awareness and help-seeking⁴⁸) and reduce DUP. This would also enable more epidemiologically representative sampling of prodromal populations.⁴⁰ However, such an initiative will require care pathways for the majority of at-risk individuals who do not “convert” to FEP. Also, success in this effort could reduce the accuracy of current tools to predict psychosis risk. Given these barriers, ED focused on the prodrome may have to wait for biomarkers that can improve risk stratification.

Improving FES Care

Unlike services targeting prevention or earlier intervention, both of which are yet on the horizon, specific proposals can be considered to fill well-known gaps in extant FES:

• Raise expectations for remission and relapse prevention: Up to 75% of patients can be expected to achieve symptomatic remission over the first year of treatment.⁴⁹ While these results were accomplished with a structured medication algorithm implemented within the context of a clinical trial, they are a reasonable aspiration for FES. There is compelling evidence that high levels of remission can be achieved by targeting a range of modifiable factors already well-known to FES clinicians (e.g. poor engagement with care, substance use, non-adherence).⁴⁹ The relevant approach here is not discovery-oriented research (to test selected interventions with careful controls) but quality improvement (implementing a series of interventions in cycles of measurement driven improvement), which can be learned and implemented within these specialized teams. Such efforts would deploy existing evidence to target one or more of a wide range of published⁵⁰ or locally known factors to improve levels of remission and reduce risk for relapse⁵⁰ soon after FES engagement.⁵¹ This would provide a more stable platform for psychotherapeutic approaches and other elements of FES care—such as supported education and employment services—to optimize functional recovery.

• Target cognitive deficits. Impairments in cognition are present early in the schizophrenias and remain a key limiting factor for functional recovery.^52,53 While several non-pharmacologic interventions have demonstrated promise,⁵⁴ implementation in EIS has been limited. Most clinicians are not trained to provide these interventions, many effective cognitive interventions require significant commitments of time that conflict with school and work schedules for patients, or changes in the workflow and organization of care in most FES. There is thus an urgent need to adapt empirically supported cognitive interventions for integrated delivery within FES, and to assess their added value in improving functional outcomes. Stratifying samples based on behavioral measures or putative biomarkers can help prioritize patients who are more likely to benefit from resource intensive interventions.⁵⁵

• Reduce cardiovascular risk. FES are already oriented toward the important goal of reducing early risks for suicide or accidental death secondary to inadequately treated primary psychotic illness or substance use disorders. However, premature mortality over the lifespan has long been associated with cardiovascular illnesses.⁵⁶ FEP samples while not usually at higher risk than their peers at treatment entry⁵⁷ can show rapid increases in rates of smoking and weight gain within the first year of treatment.⁵⁸ Some of weight and glucose dysregulation can be addressed with better antipsychotic medication choice.⁵⁹ Given efforts already made by FES to engage young adults and their families in close treatment relationships, there is an opportunity to effect both primary prevention (smoking cessation, weight control) and primordial prevention (smoking prevention, healthy lifestyles) of premature cardiovascular mortality.⁵⁸

• Reconceptualize the EIS’s role in recovery. Older distinctions between impairment, disability and handicap⁶⁰ can be revived to organize problem framing before premature problem solving. This can also provide a more explicit, operational approach for an EIS to collaborate with an individual’s personal notion of recovery. The group of schizophrenias have been associated with myriad impairments across several “sub-personal” levels (organ, circuit, cellular, genetic). While these cannot be easily linked to specific subjective experiences or casual observations of behavior, they can underlie reductions in a patient’s ability to think, feel, talk and act as they were accustomed to before the onset of illness. These disabilities that are evident on careful clinical examination, can in turn interact with specific social contexts to engender handicaps that prevent progress towards personally valued goals. The interactions across these levels—of impairments, disabilities and handicaps—can vary across time within an individual, and across individuals at the same age or phase of illness. This means that EIS must anticipate a wide range of needs and draw dynamically from a diverse repertoire of interventions to deliver personalized care. For example, the experience of derogatory auditory hallucinations (impairment) can lead to an avoidance of provocative public gatherings (disability) but this may be irrelevant for recovery unless this ability is necessary for a desired social or occupational role (e.g. teaching in a classroom), where it can manifest as a handicap. This framework allows for tailored approaches within an overall goal of optimizing recovery. FES can directly target sources of impairment, assist patients to cope with residual disabilities, but also consider the socially mediated role of handicaps. The latter may require clinical services to “step out” and engage with local employers, educational institutions and other community organizations to engender inclusion of their patients in ordinary and normalizing sources of community support. While extant supported employment models initially developed for chronically ill and long unemployed patients have been vital to informing first generation EIS, and may remain vital for a minority of early course patients, the shift advocated here is broader and strategic. The work of engendering greater inclusion and thereby greater social functioning early in the critical period cannot be borne only by patients, families and their clinicians, but is a shared responsibility with a wider community that will also benefit from the contributions of these individuals. The pervasive impact of the Americans with Disabilities Act (ADA, 1990) is an instructive example. A visually obvious marker of the ADA was the appearance of ramps at multiple private and public agencies (e.g. banks, post-offices). While this was designed to reduce handicap for the wheelchair bound, it also assisted others with temporary (e.g. sports injuries) or permanent (e.g. age-related) difficulties navigating stairs. Similarly, if college mental health services embrace the accommodations necessary to support the education of individuals with psychotic disorders,⁶¹ this will likely benefit others within a demographic who are at the peak age of risk for several mental illnesses.

Disseminating EIS: Harnessing Learning Health Systems for Research and Quality Improvement

The work of early intervention for schizophrenia has elaborated into an international and pluralistic engagement with the missions of public education, workforce development, healthcare policy, care delivery, and research. Notably, pioneer services modeled close integration of knowledge production with implementation activities, and this ethos is needed more than ever given the broad range of challenges ahead.

Traditional models of dissemination are inadequate for the project of early intervention.³⁵ First, is the implicit and mistaken metaphor of a research to practice pipeline that seeks to deliver identifiable “active ingredients” with high fidelity to community settings. While this makes sense for conceptually simple (albeit, not easy) interventions (e.g. vaccines) it is inadequate for the scaling of complex interventions (that include multiple interacting and dynamic components) within similarly complex systems of care.^62,63 This also glosses over potentially informative sources of heterogeneity in treatment effects.⁶⁴ Variations across implementation sites in patient samples and needs, or clinical and rehabilitative resources, imperil straightforward applications of research-based service models.^65,66 Also, the nuances of delivering team-based, multi-component care, in environments with myriad regulatory demands, inefficient medical record systems and limited reimbursement for psychosocial services should not be seen as mere distractions or obstructions through which an implementation pipeline must be built.⁶⁵ Rather, these are part of the varied ecologies of practice that offer opportunities for adaptation or even levers to enhance the impact of the care model. In this framework, tradeoffs that change the structure and processes of care can be made in service of a shared and overarching goal to improve population outcomes, versus strict adherence to protocols from research studies. Second, is the need to engage services in addressing knowledge gaps. The project of knowledge translation must be seen as bi-directional and operating across all levels from bench to bedside to services to policy. This includes activities such as the recruitment of early course patients for investigations into the causes and mechanisms of impairments and disabilities, clinical trials of investigational medications, brain stimulation or psychosocial interventions; or the development of payment models that can enable transitions from volume to value-based care. All these innovations are necessary and will benefit from a deliberate co-mingling of clinical, research and policy workflows.

The concept of Learning Health Systems⁶⁷ can be harnessed to design such knowledge translation oriented dissemination of EIS.⁶⁸ Networks of such EIS can be designed to attend simultaneously to the four elements of Science (the purposeful collection of data as part of the workflow to support inferences for both discovery-oriented research and quality improvement); Informatics (robust, user-friendly and collaboratively-designed software that enables data collation from all relevant sources, and enables analysis and display of population outcomes and other variables to drive quality improvement and clinical research); Incentives (that value clinicians with meaningful feedback, autonomy and opportunities for growth, and that value population health oriented care models via congruent payment models); and workplace Cultures (that support integration of clinical and research missions⁶⁹). Networks of learning health systems have demonstrated impact in other chronic disorders,⁷⁰ and are an important aspiration for psychiatric disorders.⁷¹ The recently launched US Early Psychosis Intervention Network (EPINET) is a promising effort in this direction.⁷²

In summary, the project of early intervention for schizophrenia is at a promising juncture. Models with demonstrated effectiveness in improving access and quality of care for FEP are available for broader dissemination and can deliver significant improvements in population outcomes. EIS thus present an excellent investment for national healthcare policy. However, these accomplishments have occurred in the face of enduring gaps in knowledge of the causes and mechanisms of psychotic illnesses, and this limits our ability to both target and improve treatments for FEP, and to advance intervention into the prodromal phase. Networks of EIS that are designed as Learning Health Systems can serve the dual agenda necessary for the next phase of progress. Such services can support both dissemination of best practices and quality improvement (“Doing what we know works”) and discovery-oriented research (“Learning what to do next”).

Acknowledgments

The authors have declared that there are no conflicts of interest in relation to the subject of this study.

Funding

This work was supported by National Institutes of Health (R01MH103831) and the Gustavus and Louise Pfeiffer Research Foundation.

References