From: The effectiveness of champions in implementing innovations in health care: a systematic review
Subcategory (# of studies) | First author, year | Study design | Champion operationalization | Outcome extracted from included study | Statistical analysis/approach | Test statistic (measure of magnitude) | p-value | |||
---|---|---|---|---|---|---|---|---|---|---|
Conceptual innovation use (knowledge/enlightenment) | ||||||||||
Provider (n = 4) Conclusion: Across four studies, there are mixed findings with respect to use of champions and improvement in providers’ conceptual innovation use | ||||||||||
Implementation of new technology or equipment (n = 4) | One study two reports: | Cross-sectional study | Existence of a champion of hip protectors (single item scored on a 5-point Likert scale) | Overall commitment to hip protectors | Bayesian Model Averaging logistic model | Logistic regression coefficient (95% CI) = 0.24 (0.17–0.31) | < .05 | |||
Kabukye, 2020 [61] | Cross-sectional study | Presence of an effective champion (3-item survey scale by Paré et al.[68] | Organizational readiness in a low-resource setting | Structural equation model using a partial least square method | Path coefficient = 0.15 | .0299 | ||||
Paré, 2011 [68] | Cross-sectional study | Presence of an effective champion (3-item survey scale) | Organizational readiness in a large teaching hospital | Structural equation model using a partial least squares method | Path coefficient = 0.23 | < .05 | ||||
Organizational readiness in implementing a mobile computing system for home care | Path coefficient = 0.05 | > .05 | ||||||||
Interrupted time series | Endorsed by champions (three items rated at a 6-point Likert scale based on Mullins et al. [98] | Intention to use transfer boards 2 months post-introduction of transfer boards | Stepwise logistic regression | Partial R2a = 0.036 C(p) = − .041 F = 16.25 | < .0001 | |||||
Structural equation model using a maximum likelihood method | Path coefficients (95 CI) = 0.27 (− .0156–.5556) | > 0.05a | ||||||||
System/Facility (n = 1) Conclusion: There is a study suggesting that the use of champions is related to system/facility’s conceptual innovation use | ||||||||||
Implementation of best practices related to vaccination processes (n = 1) | Tierney, 2003 [75] | Mixed study (generic qualitative and cross-sectional) | Presence of a champion lead (“Yes/No” survey item) | Pediatrician practices’ likelihood or intent to adopt reminder and recall system in their practice in a year | Multivariable linear regression | Test statistic not reported | < .03 | |||
Pediatrician practices’ likelihood or intent to adopt immunization coverage rates assessments in their practice in a year | Test statistic not reported | .002 | ||||||||
Instrumental Innovation Use (adherence in using the innovation (evidence-based practice or technology)) | ||||||||||
Patient (n = 1) Conclusion: There is a study suggesting that the use of champions is related to improving patients’ instrumental innovation use | ||||||||||
Implementation of Kangaroo-Mother Care (n = 1) | Soni,2016 [73] | Interrupted time series | Absence of champions (two champion were present from January 5, 2010–July 31, 2011; transition period from August 1, 2011, to July 31, 2012; champion was absent from August 1, 2012, to October 7, 2014) | Initiation rate of skin to skin by mothers of neonatal intensive care unit (NICU) patients | Competing-risks regression model and observation-weighted linear polynomial test | Subhazard rate ratios (SHR)c (95 CI) = 0.62 (0.47 − 0.82) | < .001b | |||
Overall use of skin to skin by mothers of NICU patients | Multivariate logistic regression and observation-weighted linear polynomial test | OR (95 CI) = 0.49 (0.34–0.70) | .004b | |||||||
Average duration of skin to skin provided by mothers of NICU patients | Multivariate linear regression and observation-weighted linear polynomial test | β (95 CI) = − 1.47 (− 2.07 to − 0.86) | < .001b | |||||||
Initiation rate of breastfeeding by mothers of NICU patients | Competing-risks regression model and observation-weighted linear polynomial test | SHR (95 CI) = 0.88 (0.68–1.14) | .30b | |||||||
Overall use of “breastfeeding” by mothers of NICU patients | Multivariate logistic regression and observation-weighted linear polynomial test | OR (95 CI) = 0.89 (0.55–1.44) | 0.61b | |||||||
Provider (n = 17) Conclusion: Across 17 studies, there are mixed findings with respect to use of champions and improvement in providers’ instrumental innovation use | ||||||||||
Implementation of best practices for smoking cessation (n = 3) | Bentz, 2007 [52] | Cluster randomised trial | Presence of a champion (“Yes/No” item determined through structured interviews with clinic managers or lead nurses) | Monthly rates of documented clients connected by health care providers to the Oregon Tobacco Quitline | Generalized estimating equations | OR (95 CI) = 3.44 (2.35–5.03) | < .05 | |||
Papadakis, 2014 [67] | Cross-sectional study | Presence of physician champion (“Yes/No” survey item) | Frequency of evidence-based smoking cessation treatments delivered by health care providers | Multivariable logistic regression | OR (95 CI) = 2.0 (1.1–3.6) | < .01 | ||||
Strasser, 2003 [74] | Cross-sectional study | Presence of a designated champion (single item rated on a 6-point Likert scale) | Extent that health care providers apply smoking cessation guideline to help parents of cystic fibrosis patients quit smoking | Multivariable logistic regression | β (SE) = − .7570 (0.2110) OR (95 CI) = 0.469 (0.310–0.709) | 0.0003 | ||||
Implementation of best practices related to vaccination processes (n = 3) | Albert, 2012 [47] | Cross-sectional study | Presence of an immunization champion on site (“Yes/No” survey item) | Consistent use of standard orders for influenza vaccines only by non-physician staff | Multivariable logistic regression | OR (95% CI) = 1.12 (0.72–1.76) | > .05 | |||
Consistent use of standard orders for both influenza vaccine and PPV by non-physician staff | OR (95% CI) = 1.67 (1.01–4.54) | .046 | ||||||||
Granade, 2020 [58] | Cross-sectional study | Presence of immunization champions (“Yes/No” survey item) | Primary care clinicians’ adherence to adult vaccination standards | Multivariable logistic regression | APR (95% CI) = 1.40 (1.26–1.54) | < .05 | ||||
Pharmacist’s adherence to adult vaccination standards | APR (95% CI) = 1.20 (0.96–1.49) | > .05 | ||||||||
Slaunwhite, 2009 [72] | Case–control study | 23 champions randomly allocated to 23 hospital units versus 23 matched units with no champion | Difference in overall health care providers vaccination rates between champion and non champion units | t -test | t (22) = 2.86 (11% higher vaccination rate in champion units) | < .03 | ||||
Percentage change in health care provider vaccination rates from previous year in champion units | t (21) = 4.38 (increase from 44 to 54%) | < .001 | ||||||||
Implementation of new technology/equipment (n = 2) | Alidina, 2018 [48] | Cross-sectional study | Presence of an implementation champion for cognitive aids (selected as an important facilitator from a list of facilitators) | Regular use of operating cognitive aids during applicable clinical events | Chi square | Test statistic not reported | 0.8968 | |||
Absence of an implementation champion for cognitive aids (selected as important barrier from a list of barriers) | Regular use of operating cognitive aids during applicable clinical events | Multivariable logistic regression | OR (95% CI) = 0.44 (0.23–0.84) | .0126 | ||||||
Shea, 2016 [70] | Cross-sectional study | Presence of nurse champions (“Yes/No” survey item) | Percentage of providers in a clinic demonstrating Stage 1 meaningful use of electronic health records | Multivariable logistic regression | OR (95 CI) = 0.99 (0.60–1.65) | .983 | ||||
Implementation of best practices related to pain management in neonatal intensive care units (n = 2) | Anand, 2017 [49] | Prospective cohort study | Presence of a nursed champion (“Yes/No” survey item) | Number of continuous pain assessments performed and documented by nurses per day for 1 month in neonatal intensive care units | Generalized estimating equations | OR (95 CI) = 2.54 (1.27–5.11) | 0.009 | |||
Lago, 2013 [66] | Cross-sectional study | Presence of a local champion (single item asking whether a physician champion, a nurse champion, both types of champions, or no champion was present) | Routine use (> 90% of the time) of non-pharmacological and pharmacological pain management interventions during invasive procedures in neonatal intensive care units | Stepwise logistic regression | Six out of 11 interventions: (1) Heel prick: OR (95 CI) = 2.78 (1.2–6.43) (2) Venipuncture: OR (95 CI) = 2.59 (1.13–5.96) (3) PICC insertion: OR (95 CI) = 3.33 (1.38–8.02) (4) Tracheal intubation: OR (95 CI) = 2.68 (1.17–6.16) (5) Mechanical ventilation: OR (95 CI) = 3.74 (1.5–9.32) (6) Chest tube insertion: OR (95 CI) = 3.26 (1.31–8.1) | < 0.05 | ||||
Five out of 11 interventions: (1) Tracheal Aspiration: OR (95 CI) = 1.96 (0.82–4.66) (2) Nasal CPAP: OR (95 CI) = 1.98 (0.87–4.53) (3) Lumbar puncture: OR (95 CI) = 1.99 (0.86–4.59) (4) ROP screening: OR (95 CI) = 2.35 (0.96–5.8) (5) Postoperative pain: OR (95 CI) = 1.58 (0.56–4.43) | > 0.05 | |||||||||
Implementation of best practices related to prevention, identification and management of infections (n = 2) | Campbell, 2008 [54] | Non-controlled before and after study | Appointment of six nurses (two for each shift) champions for 4 weeks | Intensive care unit nurses’ compliance with sepsis-screening protocols | Chi square | χ2 = 30.86 | < .001 | |||
Physician’s initiation of sepsis protocol for patients with severe sepsis | χ2 = 0.563 | .453 | ||||||||
Zavalkoff, 2015 [80] | Interrupted time series | Appointment of a single physician champion to lead projects decreasing catheter associated urinary tract infections | Urinary catheter-use ratio in a pediatric intensive care | Binomial regression (PROC GENMOD, binomial distribution, canonical link) | OR (95% CI) = 0.83 (0.77–0.90) | < .05 | ||||
Generic implementation of best research evidence (n = 2) | Kenny, 2005 [62] | Cross-sectional study | Presence of a champion (“Yes/No” survey item) | Nurses’ direct (instrumental) research use | Pearson’s correlation coefficient | r = .250 | .001 | |||
Goff, 2019 [57] | Cross-sectional study | Presence of a designated quality champion (“Yes/No” survey item) | Average clinical quality scores (adherence of providers to best practices in prescribing treatments for diseases (e.g. asthma, diabetes) | ANOVA | Test statistics not reported (Mean difference = 0.2 favouring presence of a champion) | .03 | ||||
Implementation of diabetes guideline (n = 1) | Ward, 2004 [76] | Cross-sectional study | Presence of champion (single item rated on a 5-point Likert scale) | Provider process measures relative to guideline-based diabetes management | Multivariable predictor generalized estimating equation | β (SE) = 1.24 (0.51) | .02 | |||
Implementation of best practices related to medications prescribed during or after an acute myocardial infarction (n = 1) | Ellerbeck, 2006 [55] | Cross-sectional study | Presence of a physician champion (“Yes/No” survey item) | Aspirin use at admission | Generalized estimating equations | OR (95% CI) = 1.31 (0.87–2.01) | > .05 | |||
Aspirin use at discharge | OR (95% CI) = 1.17 (0.69–2.02) | > .05 | ||||||||
Beta-blockers use at admission | OR (95% CI) = 1.45 (0.91–2.31) | > .05 | ||||||||
Beta-blockers use at discharge | OR (95% CI) = 4.14 (1.66–11.66) | < .05 | ||||||||
Implementation of the findings of a phase III, multicentre randomized control trial (BMT CTN 0201) [88] study (n = 1) | Khera, 2018 [63] | Cross-sectional study | Engagement of local champions (single item scored on a 5-point Likert scale) | Physician reported personal change in preferred unrelated donor graft source for patients with hematologic malignancies from peripheral blood source to bone marrow | Multivariable logistic regression | OR (95 CI) = 1.91 (0.87–4.19) | .11 | |||
Physician reported transplant centre change in preferred unrelated donor graft source for patients with hematologic malignancies from peripheral blood source to bone marrow | OR (95 CI) = 3.18 (1.29–7.85) | .01 | ||||||||
System/Facility (n = 7) Conclusion: Across seven studies, the use of champions was reported to be related to increase in system/facility instrumental innovation use | ||||||||||
Implementation of technology /equipment (n = 3) | Ash, 1997 [50] | Cross-sectional study | Champion scale formulated from existing measures (unknown number of items and lack of detail on items reported (rated on a 5-point Likert scale) | Infusion of electronic mail | Multivariable linear regression | β = 0.09 | .52 | |||
Diffusion of electronic mail | β = 0.34 | .01 | ||||||||
Hsia, 2019 [59] | Cross-sectional study | Presence of leadership's e-health championing behaviour (6-item survey scale) | Extent of hospital medical services and work processes are performed by health care providers using E-health technologies | Structural equation model using a partial least square method | Path Coefficient = 0.280 | < .05 | ||||
Sharkey, 2013 [39] | Non-controlled before and after study | Presence of an internal champion (“Yes/No” question in facilitator reports) | Facility-wide implementation of at least two process improvements focused on using health information technology as a medium for clinical decision support to prevent pressure ulcers in nursing homes (labelled as “Level 2 outcome” by authors) | Nonparametric Spearman correlation | ρ = 0.65 | .013 | ||||
Facility-wide implementation of three or more process improvements focused on using health information technology as a medium for clinical decision support to prevent pressure ulcers in nursing homes (labelled as “Level 3 outcome” by authors) | ρ = 0.75 | 0.002 | ||||||||
Implementation of a depression care programs (n = 2) | Chang, 2012 [40] | Cross-sectional study | Presence of clinical champion (“Yes/No” survey item) | Collocation model implemented | Multivariable logistic regression models | OR (95 CI) = 2.36 (1.14–4.88) | < .05 | |||
TIDES model implemented | Bivariate regression analysis | OR (95 CI) = 0.59 (0.20–1.78) | > .05 | |||||||
BHL model implemented | OR (95 CI) = 0.65 (0.14–2.98) | > .05 | ||||||||
No depression care improvement model implemented | OR (95 CI) = 0.63 (0.31–1.29) | > .05 | ||||||||
Whitebird, 2014 [41] | Mixed study (Generic qualitative and prospective cohort) | Presence of a strong primary care provider champion (“Yes/No” extracted from quality improvement narrative reports) | Average monthly activation rate (patients entering the program per number of full-time health care provider) | Pearson’s correlation coefficient | r (95 CI) = 0.60 (0.10–0.86) | < .05 | ||||
Implementation of patient-reported outcomes collection program (n = 1) | Sisodia, 2020 [71] | Retrospective cohort study | Presence of a clinician champion (“Yes/No” survey item) | Patient-reported outcomes (PRO) collection rate per clinic in the most recent 6 months | Multivariable linear regression | Collection rate change (95 CI) = 11.2 (2.5–20.0) | .01 | |||
PRO successful collection rate (50% or greater) in a 6-month period | Multivariable logistic regression | OR (95 CI) = 3.36 (1.06–10.61) | .04 | |||||||
Implementation of best practices related to vaccination processes (n = 1) | Tierney, 2003 [75] | Mixed study (generic qualitative and cross-sectional) | Presence of a champion lead (“Yes/No” survey item) | Pediatrician practices’ current use of reminder and recall systems | Multivariable logistic regression | OR (95% CI) = 1.85 (1.08–3.18) | < .05 | |||
Public health clinic’s current use reminder and recall systems | Multivariable logistic regression | OR (95% CI) = 3.01 (1.34–6.73) | < .05 | |||||||
Pediatrician practices’ current use of immunization coverage rates assessments | OR (95% CI) = 1.38 (0.89–2.13) | < .05 | ||||||||
Public health clinic’s current use of immunization coverage rates assessments | OR (95% CI) = Not reported | > .05 |