Health IT Safety: A Simplified Sociotechnical Model Uncovers Health IT Safety Risks

‐

By Michael S. Woods, MD, MMM; and Stanley L. Pestotnik, MS, RPh

Concern for the safety of health information technology (HIT) is long-standing, and recent reports have catapulted HIT safety to prominence. The Institute of Medicine (IOM) (2012) notes:

Although definitive evidence is hard to produce, the committee believes poor user-interface design, poor workflow, and complex data interfaces are threats to patient safety. Similarly, lack of system interoperability is a barrier to improving clinical decisions and patient safety, as it can limit data available for clinical decision making (p. 4).

Because safety is an “emergent property of a larger system,” the IOM (2012) proposed using a structured assessment of HIT implementation and operational risk, i.e., a sociotechnical system. However, the  five-domain sociotechnical model (people, process, technology, organization, and external environment) proposed by the IOM (2012) and others (Harrison, Koppel, & Bar-Lev, 2007; Sittig & Singh, 2010) is limited in two critical ways:

  1. It focuses exclusively on HIT, which is only one component required to achieve a safe and reliable system of patient care in a complex sociotechnical environment
  2. The complexity of the five-domain model makes it difficult to apply and teach to frontline caregivers

Based on our experience, current sociotechnical frameworks (IOM, 2012; Harrison et al., 2007; Sittig & Singh, 2010) do not accurately reflect the daily dynamics of frontline patient care—those things done routinely by doctors, nurses, techs, aides, and others—of which technology is just one piece in a very complex, dynamic, and risk-filled system. The complexity of these frameworks makes their practical application difficult. It also is challenging to use the models to teach frontline staff how the dynamics of their daily work can profoundly impact unit-level patient safety—e.g., burnout, resiliency, leadership, teamwork, process improvement and learning, communication practices, and, yes, HIT. Our experience has been affirmed in conversations with chief medical officers and chief medical information officers in organizations attempting to apply these models. As a result of these experiences, we decided to develop a simplified sociotechnical model to elucidate the general state of patient safety on the unit, where technology is just one component to consider.

A New Model
Our simplified sociotechnical model is based on data collected from frontline caregivers, who are the most intensive, frequent, and important users of electronic health records. Our unit-level approach uncovered situations and dynamics that can only be gleaned from direct interaction with these caregivers and is informed by years of clinical practice, consulting, EHR implementation, informatics, and patient safety experience. A simplified sociotechnical model emerged, based on frontline quantitative and qualitative data, resulting in a practical, patient safety–focused model reflective of the frontline caregiver’s day-to-day experience, where culture, processes, and technology interact to produce—or inhibit—patient safety at the unit level.

As noted, we reduced the sociotechnical domains to three—culture, process, and technology—each of which acts independently, interdependently, and intradependently. Domains are composed of sub-domains focused on defining unit-level clinical safety and accurately representing those things that affect the unit’s clinical safety dynamics—the primary driver of our development of this sociotechnical model (Table 1).

The simplified framework consists of three domains and 13 sub-domains focused on elucidating patient safety at the front line of care from those most intimately affected by the technology.

Application
To assess the utility of our model, we applied it to quantitative and qualitative data collected from more than 290 frontline caregivers in eight distinct clinical units in three nonprofit hospitals of  more than 200 beds in three geographic regions in the United States. The quantitative data was collected in 2013 using a validated culture survey instrument based on the University of Texas’ Safety Attitudes Questionnaire (SAQ) (Sexton et al., 2006). Response rates exceeded 60%. Discipline-specific qualitative data were collected using generally accepted, anthropological, field-based research methods (Jordan, 2013) at each facility in 2013 and 2014. Loosely structured ethnographic interviews were guided by the data collected in the SAQ and included questions regarding senior and unit leadership, teamwork, communication, processes, and technology use and functionality (consistent with the sub-domains of the simplified framework).

Qualitative data (verbatim comments) were reviewed by one of the authors (MSW), and de-identified and modified to retain each comment’s meaning. Common themes across groups were identified by triangulation—validating qualitative data themes by reaching the same conclusion from two or more separate data streams collected from different units and/or disciplines and reviewed by the authors for concurrence. Themes were then examined in the context of the three-domain model, and domain interactions were identified. Multiple issues were associated with more than one domain/sub-domain, reinforcing the complex interactions of the environment.

Findings
The initial goal of our exercise was to use our simplified model to assess patient safety within the eight units and the model’s utility, simplicity, and power. The model functioned well, and uncovered patient safety strengths and opportunities in each unit. We were surprised, however, by how powerful the simplified model was at elucidating a number of significant HIT safety issues at the front line of care, despite the fact that we were not focused solely on HIT safety.

Eight key HIT safety findings (and their associated sociotechnical intersections with culture and process) emerged from application of the simplified sociotechnological model:

  1. Quantitative culture data, regardless of score, is insufficient to determine a clinical unit’s safety when HIT is involved. Do not assume that because it has high culture scores,  a unit is able to use HIT safely, or that the HIT itself is safe.
  2. HIMSS Stage levels are not reflective of a clinical unit’s safety. All units in the sample were HIMSS Stage Five or higher, yet frontline staff clearly articulated multiple, proximate, and actionable safety issues within the HIT system. 
  3. Top-down, centralized HIT decision-making reduces unit (and system) safety by reducing the ability to quickly act on HIT-related safety issues identified by frontline caregivers. Miller and Gardner (1997) noted this was an ineffective management strategy for HIT safety oversight, saying, “We recommend local oversight of clinical software systems whenever possible …” This is akin to an HIT rapid response team (RRT)—a concept we developed from our findings.
  4. Frontline, unit-specific usability challenges create cascading inefficiencies. This is where inefficiency creates subsequent inefficiencies that compound upon an individual during a fixed period of time. Cognitive stress caused by poorly functioning HIT hinders work, exacerbates the potential for frontline burnout, and increases safety risk (Boralv et al., 1994). We have, in exploratory work, found a strong correlation between HIT usability and a composite burnout score (exhaustion and lack of accomplishment) with an R2 = 0.66.
  5. Insufficient provider training exacerbates cascading inefficiencies. This compounds on nurses whom ­physicians often rely on to complete orders in the CPOE system, leads to overtime costs, and contributes to burnout.
  6. Cultural components, such as disruptive behavior, profoundly affect the culture, process, and technology domains. Frontline staff must be comfortable reporting any safety or quality concern to anyone at any time. This only evolves in environments driven by principles of civility, where leadership sets the tone and enforces the rules of non-negotiable mutual respect (Woods, 2010).
  7. HIT safety issues identified at the unit level often represent industrialized HIT risk. Industrialization is a word we have coined to describe the extent of penetration of an HIT issue within an organization using an HIT system. For example, an issue with an inaccurate order set that affects only the pediatric oncology ward has a low degree of industrialization, affecting only a small and limited group. A medication reconciliation form completed on all patients in the emergency department that has an error in the underlying structure and is a potential safety risk has a high degree of industrialization, as it would affect every patient being admitted to the organization from the emergency department. The front line in each unit identified such HIT issues, including incorrect/inaccurate protocols, problems with auto-population, inaccurate medication mapping, and many others that create significant, real, ongoing patient risks that are identified by unit-level assessment and replicated across the organization.
  8. Inadequate due diligence prior to go-live results in early derailment of HIT implementation. Basic hardware-software incompatibilities lead to profound unit-level workflow disruption when simple functionality, such as being able to log on or load HIT care modules, is lacking.

Conclusions
A simplified sociotechnical framework of culture, process, and technology that incorporates qualitative and quantitative data in an integrated safety analysis provides an accurate and comprehensive picture of clinical safety, inclusive of HIT, at the front line of care. Periodic sociotechnical reassessment enables the integrated implementation of safety improvement across all three domains. A local HIT RRT can be empowered by leadership to aggressively address HIT safety and usability issues as an important component of the safety effort, as most of the issues identified by the front line in this analysis were proximate and actionable and affected multiple other units.

We believe the approach presented, crafted from direct, on-the-ground, frontline experience, has generally advanced the practical application of sociotechnical assessment for patient safety, has powerful face and content validity, and redefines expectations for practical sociotechnical assessments related to uncovering patient safety-related opportunities. The findings from this simplified assessment enable solutions to be intelligently crafted, intentionally directed, and comprehensively integrated into all domains that impact clinical safety, not just HIT.

A copy of the original peer-reviewed research for this article may be downloaded at www.sociotechnologix.com. ❙

Michael Woods is a board-certified surgeon who has written about and consulted on patient safety for over 15 years. He is a principal of Sociotechnologix, LLC, and the clinical advisor to SafeHIT, its software to identify HIT safety, usability, and workflow issues. He may be contacted at mike@sociotechnologix.com.

Stanley Pestotnik is a clinical pharmacist, bioinformaticist, and patient safety advocate, and is credited with co-founding one of the pioneering real-time clinical decision support software systems, TheraDoc, Inc.  He has served as an advisor to the U.S. Department of Health and Human Services’ Office of the National Coordinator Clinical Decision Support Expert Panel. Pestotnik may be contacted at stan.pestotnik@comcast.net.

 

References

Berg, M., Langenberg, C., Berg, I., & Kwakkernaat, J. (1998). Considerations for sociotechnical design: Experiences with an electronic patient record in a clinical context. International Journal of Medical Informatics, 52, 243–251.

Boralv, E., Goransso, B., Olsson, E., & Sandblad, B. (1994). Usability and efficiency. The HELIOS approach to development of user interfaces. Computer Methods and Programs in Biomedicine, 45, S47–S65.

Harrison, M.I., Koppel, R., & Bar-Lev, S. (2007). Unintended consequences of information technologies in health care – An interactive sociotechnical analysis. Journal of the American Medical Informatics Association, 14, 542–549.

Institute of Medicine. (2012). Health IT and patient safety: Building safer systems for better care. Washington, DC: The National Academies Press. Retrieved June 10, 2015, from http://books.nap.edu/openbook.php?record_id=13269&page=4

Jordan, A. T. (2013). Business anthropology. Long Grove, IL: Waveland Press, Inc.

Karsh, B., Weinger, M.B., Abbott, P.A., & Wears, R.L. (2010) Health information technology: Fallacies and sober realities. Journal of the American Medical Informatics Association, 17, 617–623.

Miller, R. A. & Gardner, R. M. (1997). Recommendations for responsible monitoring and regulation of clinical software systems. Journal of the American Medical Informatics Association, 4, 442–457.

Sexton, J. B., Helmreich, R. L., Neilands, T. B., Rowan, K., Vella, K., Boyden, J.,…Thomas, E. J. (2006). The safety attitudes questionnaire: Psychometric properties, benchmarking data, and emerging research. Baylor Medical Center Health Services Research, 6:44. Retrieved June 10, 2015, from https://med.uth.edu/chqs/surveys/safety-attitudes-and-safety-climate-questionnaire/

Sittig, D. F. & Singh, H. (2010). A new sociotechnical model for studying health information technology in complex adaptive healthcare systems. Quality and Safety in Health Care, 19 (Suppl 3), i68–74.

Woods M.S. (2010) Civil leadership: The final step to achieving safety, quality, innovation, and profitability in health care. Chicago, IL: Joint Commission Resources.