A translation research model (Titler and Everett 2001
), developed from Rogers’ (2003)
diffusion of innovations (DoIs) model, provided the guiding framework for this study (see ). According to this model, adoption of an innovation is influenced by the nature of the innovation and the manner in which it is communicated to users in a social system.
Conceptual Model with Components of Translating Research into Practice (TRIP) Intervention Italicized
The TRIP intervention was based on the translation research model and addressed the characteristics of the innovation (EB acute pain management practices for older adults), the users of the EBPs, the social system (context of care delivery), and the communication strategies (Titler and Everett 2001
). Thus, the multifaceted strategies for promoting use of the EBPs (the TRIP intervention) were organized by the conceptual framework illustrated in (characteristics of the innovation, communication process, users, and social system) and informed by prior research. The specific components of the TRIP intervention are bulleted and italicized in and described further under “Study Intervention.”
A randomized design was employed. Twelve Midwest acute care hospitals that discharged at least 30 patients ≥65 years of age per year with a hip fracture were stratified for size and randomized to an E or C group. Both groups received an EBP guideline on Acute Pain Management in the Elderly
(Herr et al. 2000
). Hospitals identified the principal non-ICU units where adult hip fracture patients were admitted, and the TRIP intervention was implemented on study units at the six hospitals in the E group. Characteristics of the hospitals and study units are in .
Organizational and Unit Characteristics
Practice patterns of nurses and physicians regarding acute pain management for older adults admitted with hip fracture were analyzed before and after completion of the implementation phase of the TRIP intervention. The phases of the study, intervention strategies, and corresponding measures are overviewed in .
Overview of Study Phases and Measures
Dependent Variables and Data Sources
Dependent variables were (a) adoption of EB acute pain management practices for older adults, (b) nurse and physician perceived barriers to EB pain management practices, and (c) mean pain intensity. Adoption of EBPs was defined as (1) nurse and physician adherence to the EBP guideline as measured by acute pain management indicators abstracted from medical records (MRs), (2) nurses’ perceived use of research findings to guide nursing practices for acute pain management (questionnaire), and (3) nurses’ and physicians’ stage of adoption of specific pain assessment and treatment practices (questionnaire). Although the hospital was the unit of randomization, the individual (patient MR/nurse/physician) was the unit of analysis. Data were collected before and after completion of the TRIP intervention implementation phase (see ).
The sample was the MRs of patients ≥65 years old hospitalized with a hip fracture, and nurses and physicians who cared for these patients. MR inclusion criteria were age ≥65, primary diagnosis of hip fracture, admitted to a study unit, and not transferred to an intensive care unit for the first 72 hours following hospital admission. Nurse inclusion criteria were registered nurse, employed by the study hospital, and working at least 50 percent on the study unit. Physician inclusion criteria were M.D. or D.O. caring for patients on the study unit. The study was approved by the Internal Human Subjects Review Board at the PI's institution and corresponding human subjects review boards at participating hospitals.
The translation research model guided the design of the TRIP intervention to address the characteristics of the innovation (EB acute pain management practices), communication, users, and social context (Titler and Everett 2001
). Strategies for each area, outlined in , are described below.
Characteristics of the Innovation/EBP Topic
Characteristics of an innovation affecting adoption include the relative advantage of the innovation (e.g., effectiveness, relevance to the task, social prestige); compatibility with values, norms, work, and perceived needs of users; and complexity of the innovation (Rogers 2003
). EBP topics perceived as relatively simple (e.g., influenza vaccines) are more easily adopted in less time than those that are more complex (acute pain management). Strategies to promote adoption of EBPs related to characteristics of the topic include practitioner's review and “reinvention” of the EBP guideline to fit the local context, use of quick reference guides (QRGs) and decision aids, and use of clinical reminders (Balas et al. 2004
; Bootsmiller et al. 2004
; Bradley et al. 2004
; Fung et al. 2004
; Guihan, Bosshart, and Nelson 2004
; Loeb et al. 2004
; Wensing et al. 2006
This component of the TRIP intervention included QRGs developed by the research team and reviewed by four national pain experts. Six QRGs addressed pain assessment, including standardized pain rating scales, principles of pain treatment, pharmacological treatment recommendations, equianalgesic chart, nonpharmacological treatment, and patient and family education related to pain management. Intervention focus groups (N=18) with nurses, physicians, and senior leaders were used to introduce the acute pain guideline; discuss perceptions regarding the importance, value, and benefits of acute pain management of hospitalized elders; and elicit feedback regarding the QRGs. Each E site then received multiple copies of the guideline, laminated pain rating scales for each patient room, and the QRGs to keep desired pain management practices visible.
Change champions are expert practitioners within the local setting (e.g., patient care unit), committed to improving quality of care, and possessing a positive working relationship with others (Harvey et al. 2002
; Rogers 2003
). Because interpersonal communication with colleagues is preferred rather than Internet or traditional sources of practice knowledge (Estabrooks et al. 2003b
), “change champions” are needed for each patient care area where change is implemented.
Educational outreach promotes positive changes in practice behaviors (Horbar et al. 2004
; Loeb et al. 2004
; Greenhalgh et al. 2005
; Murtaugh et al. 2005
). It is done by a topic expert knowledgeable of the research-base (e.g., acute pain management) meeting with practitioners to inform them of the EBPs, explain the research base, and respond convincingly to challenges and debates.
Strategies of the TRIP intervention to address communication were (a) use of local physician and nurse opinion leaders, (b) use of nurse change champions, (c) outreach visits every 3 weeks (N
=13 visits) by an advanced practice nurse with pain management expertise to consult with the nurses and physicians on acute pain management practices, (d) education of nurse opinion leaders and change champions via a 3-day train-the-trainer program, (e) education of physician opinion leaders by the PI/Co-PI via a 60-minute educational discussion at their respective clinic using principles of academic detailing, (f) education of nursing and medical staff via a web-based course, and (g) provision of resource texts, videotapes, and training manuals. Nurse and physician opinion leaders led organization and unit-level system changes to support use of EB acute pain management practices (e.g., documentation forms, preprinted orders), led education of their peer group, altered group practice norms, and influenced their peers through point of care coaching. The nurse change champions and physician opinion leaders circulated EB facts on acute pain management of older adults, and encouraged their colleagues to align their pain practices with the evidence. Education strategies were based on studies that education alone does little to change practice behavior (Dalton et al. 1996
; Cutler and Davis 2005
), and that interactive and didactic education are more effective when used with other practice-reinforcing strategies (Nieva et al. 2005
The social system (context) has a high degree of influence on adoption of an innovation (Fraser 2004a
; Alexander et al. 2006
). Leadership support is expressed through verbal/written messages, and provision of necessary resources, materials, and time to fulfill assigned responsibilities (Bradley et al. 2004
; Greenhalgh et al. 2005
; Centre for Health Services Research 2006
). The TRIP intervention included a 60-minute continuing education program for senior administrative leaders to discuss their role in promoting adoption of EB pain management practices and foster support for revision of institution-specific documents (e.g., documentation forms, policies, and procedures). Twice during implementation, chief nurse executives were provided brief articles about the project, written specifically for each hospital, to include in organizational publications.
Users of the EB Acute Pain Management Practice
Members of a social system influence adoption of EBPs (Rogers 2003
; Greenhalgh et al. 2005
). Audit and feedback (A/F), performance gap assessment (PGA), and trying the EBP have been tested. PGA and A/F have consistently shown a positive effect on changing practice behavior of providers (Bradley et al. 2004
; Horbar et al. 2004
; Hysong, Best, and Pugh 2006
; Jamtvedt et al. 2006
). Early in the TRIP intervention (engagement phase), study investigators met with physician and nurses at each experimental site to review baseline performance indicators of acute pain management (e.g., avoid meperidine prescription) for patients admitted to their setting with a hip fracture (PGA). Subsequent to provision of these data, A/F of pain data were achieved through concurrent MR abstraction of older patients admitted during the implementation phase and presentation of data to nurses and physicians every 6 weeks for 10 months (six reports). A second TRIP strategy was monthly group teleconferences (N
=11) among nurses working on the project from E sites to discuss issues, strategies for overcoming perceived barriers, progress made in education of staff, and revision of policies and documentation forms. Additionally, a computer was provided for each patient care unit, connected to the Internet, with directions for accessing an acute pain management website developed by the research team. The intervention concluded with a meeting of nurse opinion leaders, change champions, and managers from all the E sites.
Study instruments consisted of a medical record abstract form (MRAF) and a questionnaire administered to nurses and physicians.
Study variables and data elements to operationalize each were determined and used in initial development of the form. Content validity was achieved through review by three investigators with expertise in acute pain. The tool was pilot tested by abstracting 10 MRs of elderly hip fracture patients from a local health care facility resulting in minor modifications. Interrater reliability (r
=.92–.95) was demonstrated through abstraction of 10 records by two individuals trained in use of the instrument. Intrarater reliability was demonstrated by the trained research assistant reabstracting 25 of the same records 6 months following initial abstraction. Intraclass coefficients (for continuous variables) and κ
/tetrachoric values (for categorical data) (Cohen 1960
; Bartko 1966
; Deyo, Diehr, and Patrick 1991
; Seigel, Padgor, and Remaley 1992
; Hutchinson 1993
) ranged from .92 to 1.0.
Dependent variables measured with the MRAF were recommended EBPs for older adults (Herr et al. 2000
) including: every 4-hour pain assessment, reassessment of pain within 60 minutes following administration of analgesics, mean pain intensity, avoiding prescription and administration of meperidine, prescription of patient-controlled analgesia, avoiding prescription and administration of analgesics via the intramuscular route, prescription and administration of around-the-clock opioid and nonopioid analgesics, the parenteral morphine equivalent (PME) of opioids ordered and administered, and administration of therapeutic doses of acetaminophen.
The MRAF was also the source of data for mean pain intensity as documented on a 0 (no pain) to 10 (worst pain) scale and the Summative Index (SI) of Quality Care for Acute Pain Management. The SI was designed to enfold key indicators of EB pain management into a single score that would reflect level of overall adoption of the Acute Pain Management in the Elderly guideline. Using a Delphi approach among study investigators and four pain experts, 18 variables were selected for inclusion in the index, each representing an aspect of EB pain management care that patients might receive. For each variable, the patient received a score of 1 if the patient received the specified care and 0 otherwise. Values on individual indicators were then summed to yield the patient's SI score (possible range 0–18). Construct validity and reliability of the SI are reported elsewhere (Titler in press
Nurse and Physician Questionnaires
Nurse and physician questionnaires included a demographics section and two major tools. The Perceived Stage of Adoption Instrument measures nurses’ and physicians’ adoption of practices that have a research base (Brett 1987
; Rodgers 1994
; Rutledge et al. 1996
; Shively et al. 1997
). Each specific practice is six questions and is scored on a 0 (low adoption) to 4 (implementation) scale. Internal consistency is .95–.75, with test–retest reliability of r
=.83 (Brett 1987
; Rodgers 1994
; Rutledge et al. 1996
; Shively et al. 1997
). For this study, the questionnaire included four sections focusing on the practices of Pain Assessment in the Elderly, Pain Assessment in the Confused Elderly, Prescription/Administration of Analgesics Around-the-Clock, and Avoiding Prescription/Administration of Demerol in Pain Treatment of Elders.
The Barriers to Optimal Pain Management tool, a modification of the Pain Management Activities Questionnaire (Dalton et al. 1995
), addresses the extent that system and practice issues are perceived by nurses and physicians as barriers to pain management and includes (1) suboptimal pain assessment; (2) lack of resources such as equianalgesic conversions, clinical pharmacists, pain experts, and peer consultations; and (3) communication difficulties with patients, between nurses and physicians, and around pharmacological dose and type (see Appendix S1
). Nurses and physicians rated the extent (1=no extent; 4=great extent) that these factors are barriers to optimal pain management. Content validity was established by review of three nurse and physician experts in acute pain and test–retest reliability resulted in r
The nurse questionnaire also included the Use of Research Findings in Practice Scale, a nine-point Guttman-type scale adapted from Meyer and Goes (1988)
that asks respondents to select one statement that best reflects use of research-based acute pain management practices in the organization. Responses can range from 1 (“nurses are learning research findings related to acute pain management in the elderly”) to 9 (“the clinical application of research-based practices for acute pain management in the elderly is used in our unit and other units in the hospital”).
Data Collection Procedures
A list of MRs of eligible patients for each specified time period (baseline=January 1, 2000 to December 31, 2000; completion of the TRIP implementation phase=January 4, 2001 to March 31, 2002) was provided by study sites and up to 75 MRs per site were randomly selected for each period, based on a power of .80, an α of <0.05, and an intraclass correlation of 0.5. MRs were abstracted retrospectively at each site by a single trained research assistant.
Questionnaires were distributed to nurses and physicians at E and C sites before randomization of sites (January 2000) and annually thereafter (January 2001; January 2002). For each study unit, up to 20 randomly selected nurses meeting inclusion criteria (N=198) and all the eligible physicians (N=89) were invited to complete questionnaires.
MR and questionnaire data were double entered by The University of Iowa Data Entry Service. Frequencies, means, ranges, and other basic statistics were generated to check for out of range values and missing data, data were corrected, and revisions reviewed by three research staff for accuracy.
MR and questionnaire data were analyzed using generalized estimating equations (GEE) (Donner 1985
; Liang and Zeger 1986
; Diggle et al. 2002
). Primary statistical analyses compared the E and C groups, treating intrahospital correlation as a nuisance parameter (Liang and Zeger 1986
; Diggle et al. 2002
). An exchangeable correlation matrix (R=CS) was used for the intrahospital correlation structure; the GEE method is robust to this assumption (Donner 1985
; Norton et al. 1996
). For MR data, the effects of baseline values were incorporated into models as control variables, using a hospital's mean preintervention value; for physician and nurse questionnaire data, a subject's preintervention score on a variable was used as the control. Following primary analysis, we examined explanatory statistical models that included other variables beyond our control (e.g., registered nurse [RN] skill mix), but that might impact primary findings. The 5 percent level of significance was used for all the tests.