Table 3.

Data collected, interventions implementation, and facilitators and barriers to implementation by site

	Site Aa	Site B	Site C	Site D
Data collectionb
Documents	26	17	13	42
One-to-one interviews	–	3	10	2
Group interviews	5	2	–	5
Observations, including informal conversations	–	2	2	3
Field survey	–	2	–	–
Presentations	5	2	3	2
Demonstrations of innovations	–	1	1	1
Interventions implemented
Electronic patient information/communication portals	Portal with information about the ICU, care team, and option for patient/family to upload information about themselves for viewing by care team. Accessed via bedside iPads.	Portal with information about the ICU, care team, and option for patient/family to upload information about themselves for viewing by care team. Accessed via bedside iPads.	Portal for communication with care team, care plan information, and educational tools. Accessed via bedside iPads.	Portal to enhance patient/family engagement and educate patients/families about care in the ICU. Version 1 accessed via bedside iPads; version 2 via any personal device.
Interactive provider IT tools	Care team portal to display harms status at ICU level; Failed implementation: sensors to integrate ICU devices with care team portal.	Care team portal to display harms status at ICU level.	Tool included: care plan summary, nursing care plan, safety checklist, and communication with patients/families and other providers.	In development: Predictive algorithm for identifying harm; electronic patient safety checklist.
Interventions to improve unit culture, provider behavior, and/or workflow	Standardized program to escalate safety issues to management.	Standardized program to escalate safety issues to management.	Structured, paper-based tool for guiding communication with patients/families.	Redesigned rounds to include nurses; standardization of room entry; standardization of policies and practices.
Facilitators and barriers to implementing the interventions
Implementation facilitators	History and prior experience within the unit with research and innovation in patient safety meant clinicians were willing to implement changes. Transdisciplinary implementation team utilizing skills and expertise from a wide range of people working as one, integrated group. Co-location of key project personnel from different disciplines enabled transfer of important innovation development techniques and information. Used an innovation prototype for clinicians for test outside of the ICU.	Interprofessional work culture enabled equitable participation of different professional groups (e.g., doctors, nurses, pharmacists, physical therapists) in innovation development and implementation. Belief among clinicians in the value of the provider IT tools for creating situational awareness needed to reduce harms in the ICU. Clinical “super-users” of the provider IT tool to support adoption and use among clinicians in the ICU. Site B was able to work with Site A to learn from their implementation experiences prior to implementing in their organization. The “zero harms” goal of the project was aligned with the institutional priorities which helped to increase adoption and potential for maintenance of changes.	Strong medical and nursing leadership provided top-down support for implementing changes. Development of innovations followed observations of workflows to improve integration into existing workflows. Existing checklist culture in the ICU made adoption of an electronic patient safety checklist easier. Previous team experience developing and implementing IT innovations, and established relationships with enterprise IT developers. Patient portal was initially tested as a prototype in a separate phase for refinement prior to implementation which allowed an improved tool to be implemented.	Patient engagement culture embedded at all levels of the organization led to easy acceptance of patient engagement efforts among clinicians. Engaged frontline staff in innovation design across all ICUs to ensure integration with unit-specific workflows. Common governance structure and alignment of processes across critical care led to consistent adoption. Acceptance and embrace of innovation development and implementation as a learning process.
Implementation barriers	Lack of application program interface (API) for integrating provider IT tools with medical devices to produce a “smart” ICU. Lack of fit of prototype into real-world clinician workflows. Lack of clinician readiness for workflow changes. IT glitches from enterprise clinical system updates corrupted the outputs from the provider IT tools. Lack of relationship with enterprise IT needed to integrate IT innovations with the EHR. Mismatch in timescales to achieve the scope of vision for change and produce outcomes resulting from that change. Building relationships among experts who had not worked together before took time to develop. Complex IRB consent processes for patients with high acuity reduced patient portal adoption.	Cost of implementing and maintaining the IT interventions was high and may prohibit spread across the organization. IT glitches (e.g., loss of connectivity to WiFi, software updates, etc.) slowed uptake due to poor user experience. Cycle time for adapting the IT interventions’ software for further refinement was felt to be too slow and expensive. Mismatch in timescales to achieve the scope of vision for change and produce outcomes resulting from that change.	Established methods of communication and workflow used by providers led to low adoption of provider IT communication tool. Open unit configuration in which physicians rotated in and out meant it was hard to get provider adoption of IT communication tools. Instability and turnover in the workforce were disruptive to the unit. Static, unit-based hardware devices limited access by patients/families to the patient portal. Complex IRB consent processes for patients with high acuity reduced patient portal adoption.	Lack of alignment between timeframe associated with project grant and expectations for health service innovation to produce measurable impact. Regulation of protected health information limited accessibility of patient portal.

^aSite A was implementing a new electronic health record at the time of the visit and therefore no interventions were operational for observing.

^bData sources varied by site as specific features of the three categories of interventions across sites varied and reflect the availability of implementers/personnel at the time of the site visit