Theses Doctoral

The Structure of Mental Health in Haiti: A Latent Class Analysis of Common Mental Disorders, Severe Mental Disorders, Neurological Conditions, Clinical Symptoms, and Functional Impairment

Portnoff, Larissa

The experience of mental disorders while part of humanity, reveal inequities that are inhumane due to a lack of quality clinical service provisions globally. In Haiti, a formalized mental healthcare infrastructure developed after the 2010 earthquake where emerging dissemination and implementation studies demonstrated the potential for treatment utilization within recently established primary care. Partners in Health (PIH) and Zanmi Lasante (ZL) the frontline healthcare team have coordinated with the Haitian Ministry of Health to lead this initiative. A community-based mental healthcare system has proven to be sustainable through a task-sharing model, which delivers mental healthcare for common mental disorders (CMDs), severe mental disorders (SMDs), and neurological conditions (NCs)–with specific care pathways for major depression, psychotic disorders, and epilepsy. The extent to which patient mental healthcare are evaluated in lower-middle income countries (LMICs) like Haiti, however, have been limited. The primary aim of this study was to therefore evaluate patterns of mental disorders and to assess current patient care priorities in Haiti.

The present study, builds upon previous literature by examining the continuum of mental disorders. A latent class analysis provides a data-driven approach to examine features of mental disorders to inform clinical treatment and best practices. EHR data from PIH and ZL were obtained from patients (N=914) who met criterion for a primary diagnosis and had completed mental health evaluations that were assessed at 13 sites in Haiti from 2016-2018. Known characteristics of mental disorders include the patient’s primary diagnosis, mood symptoms such as depression and suicidality, and the level of functional impairment. Accordingly, each were included as an LCA model indicator. Post-hoc multinomial logistic regression (MLR) models predicted mental health class selection and correlates based on the descriptive and clinical symptom variables. Results suggested there are six distinct mental health subgroups, that were distinguished by functional impairment: class 1a “common mental disorders– none to low functional impairment” (11.5%), class 2a “severe mental disorders–none to low functional impairment” (4.9%), class 3a “neurological conditions–none to low functional impairment” (11.1%), class 4b “common mental disorders–high functional impairment” (38.62%), class 5b “severe mental disorders–high functional impairment” (13.02%), and class 6b “neurological conditions–high functional impairment” (20.9%).

MLR model 1 revealed CMDs were 2–3 times more likely female and received psychosocial interventions more often, and by comparison SMDs and NCs typically received psychiatric medication. MLR model 2 included patient’s clinical symptoms, that suggested severe CMDs with high functional impairment were somewhat more likely depressed when compared to other LCA subgroups. Although, in all likelihood this finding was probably attributed to CMDs including mild to severe forms of major depression, whereas SMDs were mostly psychotic disorder and bipolar disorder. Taken together, the most frequent primary diagnosis included: 1) major depressive disorder (60.3%) and generalized anxiety disorder (27.2%) for CMDs, 2) psychotic spectrum disorders (47.6%) and bipolar disorder (23.7%) for SMDs, and 3) epilepsy (88.8%) for NCs. Patients were infrequently diagnosed with co-occurring psychological disorders. The varied mental health disorder subgroups that participated in psychotherapy and psychiatric medication management, demonstrate such mental health treatments for Haitian’s are feasible and acceptable. While the present analysis was exploratory, LCA provides potential tools for treatment specification and best practices.

The WHODAS, a measure of functional impairment, may be useful as a screening tool for triage, and primary outcome to determine patient improvement. Mental healthcare pathways based on results should expand to include women’s mental health and bipolar disorder. These findings are generalizable due to the data being from a community sample and directly from EHRs with inclusion criterion that was not limited by diagnostic specification, symptom severity, or co–occurring disorders. Overall, there is a vast need for mental health services that are broadly accessible for CMDs, SMDs, and NCs. This study highlights, specific clinical training and supervision needs, and the necessity for increased nursing, psychiatry, and neurology collaboration in Haiti. There is hope that healthcare expansion will strengthen and continue to empower communities in Haiti.

Geographic Areas


  • thumnail for Portnoff_columbia_0054D_16538.pdf Portnoff_columbia_0054D_16538.pdf application/pdf 3.77 MB Download File

More About This Work

Academic Units
Thesis Advisors
Verdeli, Helen
Ph.D., Columbia University
Published Here
May 19, 2021