Context-responsive approaches in occupational safety and health research Ashley Schoenfisch Hester Lipscomb Division of Occupational and Environmental Medicine Duke University Medical Center Durham, NC November 6, 2013 ashley.schoenfisch@duke.edu
Occupational injury epidemiology An applied science of public health Focused on populations of workers Goals Understand patterns of and risk factors for injury Prevent work injuries and their consequences Evaluate prevention interventions Many research challenges
What is context? Noun The circumstances that form the setting for an event, statement, or idea, and in terms of which it can be fully understood and assessed. Synonyms conditions, factors, background, frame of reference to weave together
Why is context important in occupational injury research? The context surrounding the study of workrelated injuries will influence what happens, what we can observe, and the application of our research findings.
Lift Equipment Effectiveness Study Purpose: Evaluate the effectiveness of an intervention in preventing musculoskeletal patienthandling injuries among hospital patient care staff Was the intervention followed by a decrease in rates of injury? Days away from work? Costs? Could change be attributed to the intervention? Context in which intervention took place Intervention implementation and adoption
Intervention Minimal-Manual Lift Environment policy 2004: Medical center; 2005: Community hospital Durham, North Carolina Supported through purchase of lifting devices Train-the-trainer approach As researchers, we were not driving the horse!
Lift Equipment Effectiveness Study Hester Lipscomb 1 Ashley Schoenfisch 1 Douglas Myers 2 Lisa Pompeii 3 Tamara James 1,4 Yeu-Li Yeung 4 Marissa Pentico 4 Ethan Fricklas 4 John Dement 1 Rosa Rodríguez-Acosta 5 1 Division of Occupational and Environmental Medicine; Department of Community and Family Medicine; Duke University Medical Center; Durham, North Carolina 2 Injury Control Research Center; Robert C. Byrd Health Sciences Center; School of Public Health; West Virginia University; Morgantown, West Virginia 3 Department of Epidemiology; University of Texas School of Public Health; Houston, Texas 4 Ergonomics Division; Occupational and Environmental Safety Office; Duke University Medical Center; Durham, North Carolina 5 Division of Safety Research; National Institute for Occupational Safety and Health; Centers for Disease Control and Prevention; Morgantown, West Virginia 5-year study funded by the National Institute for Occupational Safety and Health
From the literature The efficacy of ergonomic devices in reducing hazardous exposures had been demonstrated in controlled environments. However, intervention studies at the workplace showed variability in their effectiveness Often limited to one site, short term Few addressed barriers to implementation, adoption (Hignett, 2003; Fujishiro et al. 2005; Koppelaar et al. 2012; Garg and Kapellusch 2012)
Evaluation considerations Variability in intervention by hospital and unit Implementation and staff training would take time Important institutional happenings 3 years prior: Nursing Ergonomics Committee formed at medical center to address patient-handling concerns At same time as intervention: Major revision to the workers compensation (WC) policy Fiscal responsibility for lost-time injuries and replacement staff shifted from WC to budgets of nursing unit managers WC office focused on claim closure Reporting window imposed (24 hours)
Data sources Duke Health & Safety Surveillance System (DHSSS) 1997-2009
DHSSS, 1997-2009 Study population Dynamic cohort of 11,545 patient care staff Contributed 28,446 full-time equivalents (FTEs) Estimated using employment dates and work schedules A worker employed 40 hours per week and employed the entire year contributed one FTE
DHSSS, 1997-2009 Study population Dynamic cohort of 11,545 patient care staff Contributed 28,446 full-time equivalents (FTEs) Outcomes of interest Musculoskeletal injuries, defined by whether they were attributed to a patient-handling task Nature of injury codes, agent codes, text description Days away from work Costs (medical, indemnity) Rates and rate ratios based on time at work (FTEs)
Additional data sources Monthly walk-through assessments (n=1,282) Equipment use, battery charge, supply availability, etc. Focus groups (n=13; 80 participants) Semi-structured guide: experiences with lift equipment, adequacy of training/support, barriers to adoption, etc. Surveys: 0, 18, 36 months after equipment roll-out Equipment supply purchase data Monthly meetings with hospital ergonomists
Results: Intervention adoption
Proportion of inpatient nursing units with at least three of each size of sling available for the fullbody sling lift by hospital 1.0 Medical center Community hospital 0.8 Proportion 0.6 0.4 0.2 0.0 1 2 3 4 5 Years since powered portable full-body sling lift introduced on unit
Proportion of inpatient nursing units with frictionreducing plastic liners available and accessible by hospital 1.0 Medical center Community hospital 0.8 Proportion 0.6 0.4 0.2 0.0 1 2 3 4 5 Years since equipment introduced on unit
Average monthly full-body sling lift and stand-assist lift use per inpatient nursing unit by hospital over time
Focus groups Participants identified several factors influencing adoption of the lift equipment and policy Time Knowledge/ability Staffing Patient characteristics Institutional environment Cultural aspects of work
Focus groups Injury is just part of the job. Staff were clear about their role to care for the patient, even at the expense of their own safety.
Summary: Implementation and adoption Implementation and adoption varied by hospital. Adoption of lift equipment was complex, limited, and gradual. These characteristics (along with additional institutional happenings) needed to be considered in the analytical approach and interpretation of results related to outcomes of interest.
Musculoskeletal patient-handling injuries, days away from work, and costs surrounding the intervention
Analytical approach Intervention defined as date equipment on unit All analyses initially stratified by hospital Assumed intervention could not produce immediate effect Compared outcome rates post- versus pre-intervention, lagging at 0, 6, 12 and 18 months post-intervention
Lagging Months of lag time Equipment introduced on unit Months since equipment introduced on unit 6 12 18 24 0 (no lag) BEFORE AFTER 6 BEFORE x AFTER 12 BEFORE x x etc. AFTER Expect to see a more protective effect of the intervention as the amount of lagged time increases.
Analytical approach Assumed intervention could reduce risk of experiencing a patient-handling musculoskeletal injury (and related outcomes) Examined and accounted for patterns observed for nonpatient-handling musculoskeletal outcomes Poisson and negative binomial regression Potential confounders: sex, age, race, tenure, job title Controlled for non-intervention trends using non-patienthandling musculoskeletal outcomes as an internal control
Results & Discussion Patient-handling musculoskeletal injury rate, postversus pre-intervention 0 months lagged controlling for non-intervention trends over time Medical center: No change Adjusted rate ratio 1.01, 95% CI (0.79, 1.28) Community hospital: 44% decline Adjusted rate ratio 0.56, 95% CI (0.36, 0.87) Early efforts to address patient-handling concerns diluting medical center s effect? Community hospital s higher rates?
Adjusted musculoskeletal patient-handling injury rate ratios (RR) comparing rates surrounding implementation of the lift equipment_ RR 1.01 0.97 0.90 0.83 RR 0.56 0.54 0.51 0.41
Results & Discussion Additional outcomes, post- versus pre-intervention: 90% decline in rate of days away from work associated with musculoskeletal patient-handling injuries 80% reduction in cost rates associated with musculoskeletal patient-handling injuries Occurred immediately and were sustained Unexpected given time needed for intervention adoption As pronounced for non-patient-handling musculoskeletal injuries Not expected to be affected by policy/equipment
Results & Discussion These observed patterns are more likely reflective of parallel institutional-level policies Such policies would have stimulated: More rapid closure of compensation claims Workplace accommodation of injured workers
Results & Discussion Focus groups suggested schedules for injured staff were re-arranged to provide several days off to recover Patterns related to time to closure of claims Average time to closure 1997-2003 111 days 2004-2009 93 days
Conclusions We do not believe a minimal manual lift environment was adopted completely. We observed a limited effect of the intervention in the first 4 years following implementation. Adoption was occurring gradually (as expected!) Parallel institutional policies likely influenced results An understanding of and response to contextual details was paramount in the conduct of our study.
Recommendations to institution Continued support for the Ergonomics Division to be involved in processes of implementation, adoption and promotion Training/re-training of staff and unit leaders Identify things that do work and promote them Consider alternative approaches and strategies that do not call for such broad culture change Identify uses for the intervention that are in line with the culture of caring seen in patient care Look for opportunities to emphasize patient care aspects
Context matters Yes There are challenges to incorporating contextual details into occupational injury research Difficult and time-consuming to assess and integrate Not captured through traditional epidemiologic techniques Simply documenting them does not provide insight into their magnitude or significance on measures of effect
Context matters Yes There are challenges to incorporating contextual details into occupational injury research Randomized controlled trials still regarded as the gold standard study design in measuring workplace intervention effectiveness Is this the best approach given our research questions and surrounding context? Focus on improving measures of effect Qualitative methods not readily accepted Loss of contextual detail in systematic reviews, meta-analyses, and other researchers interpretations of research findings
Context matters Yes There are challenges to incorporating contextual details into occupational injury research. But remember The context surrounding the study of work-related injuries will influence what happens, what we can observe, and the application of our research findings.
Context matters An active awareness of and meaningful response to context throughout the research process can improve understanding and enhance the quality and reach of occupational injury research. Thus, providing an important opportunity to improve worker safety and health.
Thank you! Context-responsive approaches in occupational safety and health research Ashley Schoenfisch Hester Lipscomb Division of Occupational and Environmental Medicine Duke University Medical Center Durham, NC November 6, 2013 ashley.schoenfisch@duke.edu