ISPOR 2023: Holmusk Presents New Real-world Evidence via Three Posters

May 9, 2023

Researchers from Holmusk recently attended ISPOR, a large annual conference focused on health economics and outcomes research. Holmusk presented three posters, all of which analyzed real-world data from the NeuroBlu Database, Holmusk’s continuously growing database that contains clinical data across behavioral health diagnoses, treatments, and assessments. Each of the posters shared new real-world evidence that resulted from these analyses.

Tuesday, May 9 Presentation

Analyzing real-world data to confirm clinical trial results

Another of Holmusk’s studies, which was conducted in collaboration with an external partner, used real-world data from Holmusk’s NeuroBlu Database to assemble a cohort of patients with schizophrenia. Patients within the cohort had been treated with an oral atypical antipsychotic.

This cohort was then compared to participants from a completed single-arm clinical trial that had tested aripiprazole with an ingestible sensor. The study explored how many patients were hospitalized within 3 or 6 months after initiating and being on treatment for a defined period of time.  

Ultimately, the study results suggested that there was a similar trend in the reduction of psychiatric hospitalizations for patients on aripiprazole with an ingestible sensor compared to patients on standard oral antipsychotic treatment. This provided preliminary evidence for the use of antipsychotics with ingestible sensors in reducing psychiatric hospitalizations.

“This poster demonstrated how observational studies using real-world data can be used to supplement data from clinical trials,” said Mayowa Oyesanya, MD, a clinical research scientist at Holmusk and author on the study. “Holmusk’s NeuroBlu Database provides a convenient and rich source of data for this purpose.”  

See the poster here.

Wednesday, May 10 Presentations

Investigating suicidality and healthcare resource utilization in children

In another study, Holmusk researchers used the NeuroBlu Database to explore suicidality prevalence and healthcare resource utilization in a cohort of over 100,000 children younger than 18. Children who had at least one recorded Mental Status Examination were included in the study.

Innovative natural language processing techniques were used to extract information about suicidality from the data included within the Mental Status Examination, together with external stressors. More than 15,000 children (about 14% of the study cohort) were found to have had at least one experience with signs of suicidality suicidal or self-harm during the 20-year period covered by the data. Up to half of those who experienced suicidality reported using emergency department or inpatient healthcare services. Family stressors were the most common external stressor, both for those who experienced suicidal symptoms and those who did not.

“The NeuroBlu Database is an important resource for behavioral health real-world data, and the longitudinal data contained within it helped us to uncover that rates of suicidality among children have been steadily increasing between 2011 and 2019,” said Sherwin Kuah, a senior data scientist at Holmusk who led the study. “In addition, Holmusk’s natural language processing models facilitated discovery of external stressors; this is valuable information that is not readily available in a usable format within typical real-world data sources like electronic health records.”

See the poster here.

Developing transdiagnostic predictors of hospitalization

One Holmusk poster shared findings from a paper which was recently published in The Lancet Psychiatry. The poster findings showed that variability in the Clinical-Global Impressions scale, a scale used to measure clinical severity (CGI-S), can be used as a reliable predictor of psychiatric hospitalizations across different psychiatric diagnoses (e.g. schizophrenia, bipolar disorder, ADHD, major depressive disorder, generalized anxiety disorder).

This is the first publication which has shown that CGI-S variability can operate as a predictor of psychiatric hospitalizations in a clinical care setting. The CGI-S is a standardized, clinician-reported scale which provides a broad rating for the current severity of a psychiatric disorder. It is commonly used as a key outcome measure in randomized-controlled trials and sometimes in routine clinical care.

“The way this study was designed and conducted speaks to the breadth and depth of Holmusk’s NeuroBlu Database,” said Kira Griffiths, PhD, a research scientist at Holmusk who was involved in the research. “Most real-world data sources do not contain many standardized measures, which makes comparison across time difficult. However, the NeuroBlu Database contains frequent collection of standardized measures like the CGI-S, enabling robust research.”

Read more about the research here, or see the poster here.

Printer button icon.Share button icon.
Back to top
Contact us