UMass Memorial Health Care System

UMass Memorial Health Care system (UMMHC) has been implementing the Zero Suicide framework in a stepped approach since 2016. The effort was precipitated by a grant from the National Institute of Mental Health to study Zero Suicide implementation across a large health care system.

Key Outcomes

  • Different clinical settings required tailored approaches, as well as alignment with accreditation standards.
  • Building screening and intervention tools in the electronic health record (EHR) helped to increase awareness and guide clinical action. Training needed to be continuous and integrated into clinical workflow.
  • Documented screening rates are now over 90% across all emergency departments, with about 4% of our patients screening positive for suicide risk.
  • High rates of documented screening are encouraging, but true fidelity to screening can only be ascertained by direct observation, patient feedback, and improved detection.

About UMass Memorial Health Care System

UMass Memorial Health Care system (UMMHC) is the clinical partner of the UMass Medical School. It is the largest health care system in Central Massachusetts, with five hospitals, a wide range of outpatient services, and 13,000 employees. In addition to the medical centers, the system includes home health and hospice programs, behavioral health programs, and community-based physician practices. UMMHC behavioral health services comprise three inpatient behavioral health units, partial hospitalization, outpatient psychiatry, integrated behavioral health, and a psychiatric emergency department (ED).

Zero Suicide Implementation

In 2016, UMass faculty were successful in obtaining funding from the National Institute for Mental Health to implement and evaluate the Zero Suicide framework in the UMMHC system. The grant facilitated data collection and analysis as well as allowed us to bring a wide range of stakeholders and champions together from across the UMMHC system. The scope of Zero Suicide implementation spans all EDs, inpatient medical and surgical units, behavioral health units, and primary care. To date, we have implemented Zero Suicide primarily in EDs and inpatient settings. Our primary care phase is in its nascent stage, but we anticipate continuing to use the Patient Health Questionnaire-9 (PHQ-9) to detect patients with suicide risk and to empower physicians, nurses, and behavioral health clinicians to respond to emerging suicide risk with evidence-based interventions and enhanced care transitions. We have implemented Zero Suicide’s four clinical elements (identify, engage, treat, transition) and three implementation elements (lead, train, improve) as follows.

“Having system-wider alignment and buy-in has been important and has been facilitated by concerns about compliance to evolving accreditation requirements.”


We use the Patient Safety Screener (PSS-3) to detect and stratify suicide risk in each acute care encounter (so far, ED visits and inpatient admissions). Nurses administer the primary screening items to all patients as part of their primary nursing assessment. Those who screen positive for ideation in the past two weeks or an attempt in the past six months are asked six additional questions to stratify them into mild, moderate, or high risk. Moderate- and high-risk patients receive differential safety precautions in line with Joint Commission requirements. Additional interventions for all patients are outlined below. All screening items are built into the EHR and include prompts, best practice advisory alerts, and reporting capabilities.


Our ultimate goal is that all patients who screen positive for suicide risk on the primary screener will receive the Safety Planning Intervention (SPI) before discharge. This has been difficult to achieve for many reasons, but we have encouraged safety planning by building it into our protocols, templating the safety plan in the EHR, and offering regular SPI and lethal means counseling trainings and train-the-trainer sessions.


We are in the process of implementing Collaborative Assessment and Management of Suicidality (CAMS). The tools are not templated into the EHR, but we engaged leadership and offer twice-yearly in-person trainings to clinicians from across the spectrum of care (ED, inpatient, outpatient) to encourage continuity of CAMS care across settings.


Certain transition components are included for all patients who screen positive. Facilitated by the EHR, all such patients by default receive the Lifeline number and a list of community resources. Our behavioral health service proactively identifies patients who screened positive during their stay, and they reach out with caring contact phone calls and/or postcards. If a patient received a safety plan, that plan and other behavioral health notes are integrated via the EHR to follow the patient through the health system and across visits.


We engaged all levels of leadership to facilitate changes at the system, hospital, unit, and clinician levels. Some changes (such as those to the EHR, forms, and environment) had to be approved and resourced by the system, while others (such as the intervention delivery and role shifts) relied on managerial and frontline buy-in. For each unit, we convened leadership and champions to map the current state and visualize an ideal future state. Screening protocols and safety precautions were mandated by the system, and a new policy and new forms were developed and approved. We also integrated screening, safety planning, and community resource lists into our EHR. We administered a survey of administrators and frontline providers at three linked time points to assess knowledge, attitudes, efficacy, and practice related to suicide prevention and their change over time.


We developed and implemented trainings for different clinical roles on a range of topics and varied modalities. Protocol training on screening and safety precautions for nurses, physicians, and patient care associates comprises a combination of required online training, at-the-elbow coaching, and lectures. Every patient with suicide risk should receive an evidence-based intervention, so we offer quarterly in-person basic and train-the-trainer workshops on the SPI and twice-yearly, day-long CAMS trainings for social workers and behavioral health clinicians.  


We have been implementing Zero Suicide through the existing continuous quality improvement structure, namely lean processes, and adopting a hub and spoke design. Before the go-live for each unit, we gather baseline chart review data that is mapped to the current workflow with each unit. When each unit “goes live,” we gather chart review data, conduct screening fidelity interviews with patients, and support a local team to apply Plan-Study-Do-Act cycles to move closer to the future state we’re aiming to attain.

“We have encouraged safety planning by building it into our protocols, templating the safety plan in the EHR, as well as offering regular SPI and means counseling trainings and train-the-trainer sessions.”

Implementation Data and Key Outcomes

Before we began implementation of Zero Suicide at UMMHC, we had no policy for stratifying patients who had screened positive for suicide risk. Only one of our EDs had been routinely screening all patients for suicide risk, and that ED achieved an increase in suicide risk detection from 3% to 8% of all ED visits during a previous screening implementation study.

Following a significant coordinated effort involving EHR changes, training, policy development, and lean process improvement, documented screening rates are now over 90% across all EDs, with about 4% of our patients screening positive on the PSS-3 (for those with ideation in the last two weeks or an attempt in the last six months).

  • Between October 2017 and April 2019, over 10,000 ED visits involved a positive suicide screening result: about one-quarter of these patients did not have a suicide-related presenting complaint, but most of those patients did have a psychiatric presenting complaint.
  • Only a handful of patients screened positive for an attempt in the past six months only. The vast majority of those who screened positive for an attempt in the past six months also had ideation in the past two weeks.
  • For a subset of patients we interviewed who had negative results documented for their screening, 35% reported not having been asked about ideation, and 50% reported not having been asked about attempt. This means that up to 100,000 ED visits every year may be unscreened, missing the opportunity to detect incidental risk in an estimated 3,000 patients.
  • Approximately 60% of patients with a positive screen were discharged, just over 10% were psychiatrically admitted within our system, and about 25% were medically admitted or transferred elsewhere.
  • Only about 10% of suicide-positive patients had a safety plan documented in the chart.

In the process of implementation, we learned some valuable lessons about data collection and use. Although we were successful in integrating many suicide-related tools into the EHR, we had less success in translating this to real-time feedback and visualization to support clinicians and administrators. The way in which our research team accesses data is not as helpful for those in clinical roles because it requires data manipulation and access to a separate and unwieldy system. We strongly recommend manual data validation to ensure that the reports being pulled are correctly linked to the intended source data in the chart. We changed our EHR provider in October 2017, which brought its own opportunities and challenges, namely the chance to build new tools but the loss of our established reporting capabilities. Conducting fidelity interviews with patients showed us that just because screening is documented in the chart does not mean that it was delivered in real life, and clinicians may have been inferring a lack of risk instead of asking each screening question.

“Just because screening is documented in the chart does not mean that it was delivered in real life, and clinicians may have been inferring a lack of risk instead of asking each screening question.”

Next Steps

As our efforts progress we are gaining additional experience in what works well. Having wider system alignment and buy-in has been important and has been aided by concerns about compliance to evolving accreditation requirements. To that end, we convened a successful system wide Kaizen and accompanying work groups to transform care in the EDs and on the medical/surgical inpatient side. We hope to achieve the same with the behavioral health inpatient units and primary care, obtaining input on the current state and preferred future state from as many stakeholders as possible, as early as possible. Later in the study, we will have access to more longitudinal research-level data from the data lake (a repository of EHR data built for research and quality improvement purposes), as well as longer-term suicide outcomes in our patient population.

We have learned that it takes concerted and coordinated effort to improve fidelity to screening and risk management. Nonetheless, our high rates of documented screening and improved risk detection suggest we are headed in the right direction. We are now exploring multimedia and interactive training approaches to screening

Authorship Details

  • Celine Larkin, PhD, Co-investigator
  • Catarina Kiefe, PhD, Multi-principal Investigator
  • Edwin D. Boudreaux, PhD, Contact Principal Investigator

For more information, contact Celine Larkin: