Applying Concept Mapping to Solving In-Patient Mental Health Recidivism

Abstract

Concept Mapping is a powerful research tool using visual representation to expose the complementary impact and synergy of factors affecting a specific process. this article outlines an example, in the domain of mental health, of concept mapping applied to the specific issue of readmissions or recidivism of mental health patients. Concept mapping is used to unify the diverse perspectives found across the existing literature and in mental health care delivery so that patient input and engagement in the care process can be maximally applied to improvement in the delivery of local inpatient mental health care and, penultimately, to transformation of an inefficacious care delivery model.

Introduction

The National Institute of Mental Health (NIMH) strategic plan addressed the need to tackle the irreversible cost and health implication of unsuccessful mental health interventions.¹ The strategic plan proposed in part to promote and disseminate the evidence based of successful mental health interventions, and in part the need to develop new and better interventions.¹ NIMH stressed that these interventions should take into consideration the diverse needs and circumstances of people with mental health.¹ In fact, NIMH, like other studies, highlighted the importance of developing clear and strong intervention designs.^1–3 The ideal design should identify and define the complementary impact of each mental health driver and the synergy between them that creates the undesired mental health outcome as well as defining those drivers whose co-action leads an appropriate care delivery cycle. This paper presents a case study abstracting the use of a concept mapping approach as a research methodology to remedy the gaps health providers’ ability to achieve the patient driven intervention design needed to change patient and provider behavior.

Concept mapping (CM) is a tool for organizing and representing knowledge and concepts and the relationships that can exist between concepts.⁴ CM provides a strong, multifaceted strategy to assess the basic building blocks of data that underpin successful intervention design. Beyond this, using a CM in a clinical setting allows for a wider view of etiological factors that more adequately consider the complexity of the patient’s situation as a whole rather than a piecewise function of patient situational characteristics. The complexity inherent in mental health treatment continues to pose a major research challenge that is complicated by the need to evaluate success on the basis of outcomes that are often complexly related to process changes. Paul, beginning in 1967, discusses the complexity of psychotherapy research and treatment evaluation.³ He argues that measuring the outcome(s) cannot be accurately achieved without defining, beforehand, the outcome variables, and the many other latent variables that may indirectly confound measurement of the outcome variables.³ Therefore, Paul suggests that, during the process of study design, researchers need to question the treatment, the person who is providing it, and whether the proposed treatment is most effective for a particular patient with a specific mental illness, and the other circumstances surrounding the patient that may impact treatment effectiveness. CM assists in delineating processes as multilevel sources of impact at the individual patient level, at the treatment or interventional viability level, and at the level at which care providing organizations can understand the efficacy of certain constellations of drivers that could or do interrupt the recidivist cycle. Ultimately, understanding and measuring the effects of changes to these levels of processes creates the potential to affect the underlying taxonomy of care delivery.

To begin this kind of measurement and expand on Paul’s strategy, an exhaustive list of drivers associated with mental health related behaviors is required to depict the patient’s situation accurately. These drivers form the basis for developing well-founded interventions and experimental procedures, and as well as suggesting outcomes for evaluation. From this perspective, several studies and scholarly articles embraced developing a CM prior implementing any treatment. van Manen et al. (2012) compiled a comprehensive list of patient characteristics that were found pertinent to treatment selection with personality disorders.⁵ Later, the authors recruited 29 experts to rank those characteristics based on their level of influence on patient health outcomes. Similarly, Saha (2010) developed an educational CM to increase awareness of mental sickness and mental hygiene.⁶ Corcoran (2005) developed a mental referral checklist to identify mental health needs of youth in the juvenile justice system.⁷ To do so, the author used a CM to help identify and cluster those needs into meaningful sets. Considering these examples, this research team developed a strategy whereby concept mapping played an integral role in developing and guiding the clinical research strategies to be used, especially the complex multi-dimensional analyses which are required to study mental health.^8–10 The material reported in this article is part of an extensive pre-study. Going forward, extensive research is planned in which CM is a strong qualitative method through which relationships among drivers are defined and additional drivers are added to close any remaining gaps, thereby increasing understanding of the causality effect on patient outcomes. Furthermore, the research will utilize concept mapping to increase medical staff and patient critical thinking and to improve clinical preparedness.¹¹ The comprehensive and relational aspects of concept mapping allows patients to be seen more holistically because the entirety of a patient’s situation and how this affects their clinical status can be visualized in one complete model rather than a segmented collage of risk factors spread across various clinical notes. Such visceral understanding is required if patients and care providers alike are to move beyond an intellectual understanding of the problem of recidivism to care to long term changes in behavior. The involvement of medical staff and patients in developing more refined concept maps increases the patient centeredness of research efforts and the interventions created and trialed. CM also allows both patient and care givers to look at factors in the patient/provider situation that are positive or protective in nature, rather than relying only those factors which are supporting a deficit.

Concept Map Approach

As part of a pre-study done in support of a patient generated research project around recidivism, four research assistants conducted a content analysis on a list of systematically retrieved articles related to recidivism or contextually similar situations, from multiple domains, including health care, criminal justice and anthropology. The intent of this content analysis was to document the relationships among different drivers found in previous research to affect mental health status, length of stay, and recidivism to in-patient care by mentally ill patients. A CM was then developed to summarize these relationships. During the literature review process, 586 drivers where collected from 600+ articles. These drivers were partitioned into 55 subject derived sub domains, and these sub domains were grouped into 9 major domains (see Table 1). Most of the drivers reported in this literature review are related to diagnosis and treatment with 177 and 135 drivers respectively. The economic and legal domains are the second densest domain with 53 and 52 drivers respectively.

Table 1.

Mental Health Drivers Domains and Subdomains

Domains	Subdomains	Number of Drivers
Demographic Domain (D1)	Gender, Age, Immigration, Relationships, Community Related Drivers, Education, and Health Related Drivers	16
Economic Domain(D2)	Health Insurance, Income Level, Employment Status, Treatment Payment Methods, and Other Economic Related Resources	47
Cultural Domain (D3)	Region, Religion, Customs, and Family, Individual, Organization Culture, and Community	52
Criminal Domain (D4)	Sexual Crime, Criminal History, And Participation In Criminal Prevention Program, Criminal Behavior, Number Of Criminal Offenses, and Type of Criminal Offenses	34
Diagnosis Domain(D5)	Type Of Mental Disease, Symptom Screening Tool Scores, Comorbid Conditions, Substance Abuse Disorders, and Uncategorized	177
Legal Domain (D6)	Court Related, Driving Related, Family Related Drivers, Criminal Arrests Related Drivers, Substance Abuse Related Drivers, and Legal Issues Related To Violence	53
Treatment Domain(D7)	Hospital Stay Characteristics, Pre-Hospitalization Characteristics, Post Care Post Outpatient Treatment, Counseling Behavioral, Pharmacotherapy Treatment, Readmission Drivers, Criminal Justice, Clinical Assessment Tools, and Other Treatment	135
Living Arrangement Domain(D8)	Family Related Issues Causing Housing Issues, Housing Type, Community Issues Related To Housing Access, Employment Problem Causing Restricted Access To Stable Housing, Relationships Issues Related To Housing Issues, and Location Related Housing Issues	36
Prescription Drug Domain (D9)	Antipsychotics, Antidepressants, Mood Stabilizers, and Other Uncategorized Prescription Drugs	36

The complexity of predicting mental health status, and thus the possibility of recidivism, requires a detailed understanding of the magnitude of effect of each driver and the connections between them. The number of relationships or associations that exist among drivers can grow exponentially. For instance, this literature review developed 586 drivers from which 2⁵⁸⁶-587 sets of driver associations can be developed. These associations are best understood when represented at the major domain level by allowing more naïve individuals to clearly visualize the type of relationships to expect between drivers (see Figure 2).

Figure 2.

Concept map showing drivers that impact a mental health case and the level for an intervention to break the mental health recidivism cycle.

Concept mapping is also useful to help patients and providers understand the relationships between and among individual drivers. Figure 2 shows the two types of associations that exist between different drivers: a causal and a simple relationship. For instance, drivers from D1 can have a causal relationship with drivers from D3, and D8. Similarly, drivers from D7 have a causal relationship with drivers from D9, and D6. Furthermore, the CM shows the magnitude of the relationship that can be found between drivers. This magnitude ranges from one to five, where one is a weak relationship and five is a strong relationship. For example, D1 and D3 have a strong causal relationship. Therefore, cultural and demographics drivers can proxy each other.

This CM exercise identified five locations where an intervention could be implemented. These potential interventions target patient characteristics, environmental factors, and the overall patient situation (in which both patient and environmental factors are targeted simultaneously), as well a critical events that can cause unplanned hospital visit, and the scenario at the hospital visit level. From these targets an intervention designed to break the recidivism cycle can be developed. However, as implementation cost, effectiveness, and length of the intervention are major factors in the design of cost effective interventions, their effect on the likelihood of long-term behavior change must be considered. Interventions, in this case, that target environmental or patient factors may prove to be most cost and care effective, because they supersede the critical event state that involves an extensive value chain of care provision.

Application of Concept Mapping to Recidivism Preventing Interventions

CM can help target where an intervention is warranted. Figure 3 illustrates an application of CM to a recidivism case developed for this research from a real patient situation (see Box 1). To demonstrate how CM can be used to develop intervention points, this study trialed the possible implementation of an intervention by targeting the cause that led to patient readmission through the use of an in-depth case developed from an actual patient situation (see Box 1).

Box 1. Excerpt Overview of Patient M’s Current Situation.

M is a 50-year-old white, English speaking male. He was recently incarcerated for domestic abuse and assault. M is currently receiving repeat assessment and treatment for depression, drug and alcohol abuse, agoraphobia, and self-inflicted wounds about the hands and arms. M has received in-patient care on 15 occasions in the last 12 months, each time accepting care when his wife or child brought him to the ER. His care has been received from 3 inpatient facilities. The primary diagnosis for each of these previous hospitalizations is drug and alcohol abuse, usually resulting in him being disorientated and verbally abusive. Each time, his visit was lasted 3–6 days, usually until he was fully detoxified and coherent. M also uses marijuana multiple times per week and when he has access to it, misuses both prescription oxycodone and Xanax. M has been married twice. M usually lives with his wife of 26 years in a home he built and owns. The home is located in a very rural area and 30 minutes travel is needed to reach a highway… (More details were available in the full case and will be provided upon request).

Learning Experience

The application of CM to the present research study as part of a successful recidivism prevention effort is broad if compared with the findings from van Manen et al. (2012), where the authors suggest that there are two major criteria (patient tolerance to emotional pressure and to support and expressive continuum ) when empirically selecting treatment for personality disorder patient.⁵ van Manen’s work serves this project well as a foundation upon which the analysis of drivers allows the apportioning of drivers into comprehensive domains. It does so by narrowing the selection of drivers to specific domains. Supported by this domain allocation, concept mapping can improve understanding of the patients’ situation, increase health providers knowledge about the causes of recidivism, and as well as aid in the development of patient accepted interventions that are cost effective and improve the delivery of care through long term behavior change.

Conclusion

The results of this research pre-study aimed at developing a patient centered approach to recidivism interventions suggests that mental health professionals and patients alike can benefit from the results of the application of concept mapping to mental health drivers as part of the process of reducing recidivism to in-patient mental health care.

Figure 1. High Level CM Summary of Mental Health Drivers.

Note: the number in parenthesis shows the type of association and the effect size of the association that can exist between domains. The type of the association takes one in case of bi-directional relationship, and two for bi-directional relationship. The effects size or the expected size of the relationship ranges between one and five, where one refers to weak relationship and five refers to strong relation.

Biography

Patricia Alafaireet, PhD, (above) Adam Bouras, MHA, Howard L. Houghton, MD, Beau J. Lavoie, PharmD, Jaie P. Lavoie, PharmD, Beth Cressman, MHA, and Saumil Modi, MHA are in the Department of Health Management and Informatics, University of Missouri School of Medicine.

Contact: AlafaireetP@health.missouri.edu