Supplementary Online Content Köpke S, Mühlhauser I, Gerlach A, et al. Effect of a guideline-based multicomponent intervention on use of physical restraints in nursing homes: a cluster randomized controlled trial. JAMA. doi:10.1001/jama.2012.4517. etable 1. Overview of guideline recommendations etable 2. Components of the intervention etable 3. Steps of process evaluation etable 4. Baseline characteristics of nursing homes etable 5. Estimated institutional costs of the intervention efigure 1. Percentage of residents with any physical restraint in each nursing home at baseline and after 6 months eappendix. Post-hoc repeated-measures analysis This supplementary material has been provided by the authors to give readers additional information about their work.
etable 1. Overview of guideline recommendations Type of intervention Grade of recommendation a Quality of evidence b 1. Educational programmes Intervention is recommended Low 2. Special dementia units Intervention can be considered Low 3. Person-centred care Intervention can be considered Very low 4. Environmental modifications Intervention can be considered Low 5. Electronic alarm systems Intervention can be considered No evidence 6. Specific occupational programmes Intervention can be considered Very low 7. Active and passive music interventions Intervention can be considered Very low 8. Animal-assisted therapy Intervention can be considered Very low 9. Advanced geriatric nursing practice No recommendation possible Very low 10. Special residential concepts No recommendation possible No evidence 11. Milieu therapy No recommendation possible Very low 12. Visual barriers No recommendation possible No evidence 13. Geriatric rehabilitation and exercise No recommendation possible Low programmes 14. Specific night care No recommendation possible No evidence 15. Continence training No recommendation possible No evidence 16. Therapeutic touch, massage and acupressure No recommendation possible Low 17. Aromatherapy No recommendation possible Moderate 18. Basic stimulation No recommendation possible No evidence 19. Validation therapy No recommendation possible Low 20. Snoezelen No recommendation possible Low 21. Reality orientation and cognitive stimulation No recommendation possible No evidence 22. Reminiscence therapy No recommendation possible Very low 23. Specific bright lighting Intervention is not recommended Moderate 24. Light therapy Intervention is not recommended Low a The guideline development slightly modified the GRADE system 1 to the following pattern: Intervention is recommended; Intervention can be considered; No recommendation possible; Intervention is not recommended; Intervention is advised against. b Following the GRADE system 1 quality of evidence, if available, was graded as high, moderate, low and very low.
etable 2. Components of the intervention Intervention Description Basis of rationale Declaration Declaration confirming the nursing home s dedication to the Proven strategy in previous studies 2;3 intervention s objectives, i.e. the avoidance of physical restraints signed by head nurses and/or directors of each nursing home Structured 90-minute information program for all nursing staff External structured 1-day intensive training workshop for nominated key nurses from different nursing homes Structured support for key nurses Practice guideline Printed supportive material Definition of physical restraints Desired and unwanted effects of physical restraints Legal aspects of physical restraint use Guideline development and recommendations Nurses subjective attitudes and experiences Alternative approaches focusing on physical restraints reduction as most important alternative Advanced version of 90-minute session In-depth work with the guideline Exchange and discussion between nurses from different nursing homes Group presentation and discussion of individual barriers and facilitators of physical restraints reduction Individual discussion of nursing homes prevalence data Individual discussion of results of nurses knowledge and self efficacy assessment after educational sessions for nursing staff Development, presentation and documentation of nursing home specific agendas of physical restraints reduction Key nurses received diaries to document activities and problems of daily intervention s implementation In the first three months of the intervention project staff contacted key nurses monthly via telephone or in person to discuss important issues and offer advice Provision of one printed version of the guideline for each key nurse and for the nursing home s head nurse Provision of the guideline s 16-page short version for all nursing staff Provision of the guideline s 16-page short version for legal guardians and relatives focusing on legal aspects Provision of a leporello style flyer for relatives and other visitors with information about the project s main objectives Cochrane review 4 Theory of planned behavior 5;6 Acknowledging perceived barriers, current practice culture, and concerns and emotional responses of nursing staff by using different educational strategies, e.g. working with case vignettes and small work 7;8 Structured and easy to understand education on guideline content and background 9;10 Key nurses to support reduction of physical restraints 11-13 All aspects referred to in the above box Key nurses to support reduction of physical restraints 11-13 Guideline implementation recommendations Publications on guideline implementation 9;10;14 Involvement of significant others 15
Other supportive material Publicity Provision of posters with the intervention s logo and slogan ( Dare more Freedom ) Provision of pencils and post-its with the intervention s logo for all nurses attending the educational session Provision of mugs with the intervention s logo for key nurses Key nurses were encouraged to post the declaration in the nursing home s foyer together with information about the project and the names of contact persons (i.e. key nurses) Posters with the project's logo and slogan were to be displayed throughout nursing homes Individual activities as e.g. information evenings for relatives were encouraged Guideline implementation recommendations 9;10;14 Involvement and support of peers and superiors 9
etable 3. Steps of process evaluation 16 Step Method (1) Recruitment Documentation of recruitment and reasons for (not) participating (2) Reach Attendance in structured information program Attendance in intensive training workshop (3) Fidelity Nurses' knowledge after participation in information program Nurses' self-efficacy after participation in information program (4) Satisfaction Documentation of investigators' monthly contact to key nurses within the first three months of the study concerning satisfaction with the intervention In-depth interview with one key nurse and one head nurse each per cluster immediately after follow-up Structured interview with one randomly chosen staff nurse per cluster (5) Dose delivered Documentation of teachers' impression concerning conduct of education programs Documentation of investigators' monthly contact to key nurse concerning implementation of the intervention (6) Dose received Documentation of monthly contact to key nurses to explore barriers and facilitators of the intervention's implementation (using key nurses diaries) Structured interview with one randomly chosen staff nurse per cluster In-depth interview with one key nurse and one head nurse each per cluster after follow-up
etable 4. Baseline characteristics of nursing homes Intervention (n=18) Control Group (n=18) Participants per nursing home 107 ± 44 (47-184) 99 ± 40 (53-215) Ownership of nursing home Non-profit 8 (44) 5 (28) Affiliated to church 5 (28) 9 (50) Private 5 (28) 4 (22) Proportion of trained nurses among nursing staff a 58 ± 7 (49-76) 57 ± 11 (40-76) Nursing homes with specialized dementia care unit 9 (50) 8 (44) Participants per caregiver 3.1 ± 1.3 (1.7-7.0) 3.7 ±.7 (1.0-4.5) Standardized documentation of physical restraints 15 (83) 16 (89) In-house nursing standard on physical restraints 10 (56) 17 (94) Values are numbers (%) or mean ± standard deviation (range). a In Germany nursing home nurses can basically be divided into trained with 2 or 3 years vocational training and untrained nurses with little or onthe-job training.
etable 5. Estimated institutional costs of the intervention Item of resource Unit Units used Cost per unit ( /$) Total cost ( /$) Ward nurses salary Hour 853.5 a 15.61/21.07 17 13323/17986 Key nurses salary Hour 400 a 15.61/21.07 17 6244/8429 Guidelines Item 50 20/27 1000/1350 Short versions for Item 800 3/4 2400/3200 nurses Short versions for Item 400 3/4 1200/1600 legal guardians and relatives Flyers for relatives Item 500 1/1.35 500/675 Mugs Item 100 5.52/7.45 552/745 Posters Item 100 2.97/4 297/400 Pencils Item 1000.58/.78 580/780 Post-its Item 1000.27/.36 270/360 Investigators salary Hour 22 a 31/42 682/924 (personal visits) Investigators salary Hour 7.75 a 31/42 240/326 (telephone calls) Total intervention cost 27288/36839 a Total amount of time during the study period for all staff in all intervention clusters. Source of information Documented time spent by nurses Invoice by print service Invoice by manufacturer Real time calculation according to protocol
efigure 1. Percentage of residents with any physical restraint in each nursing home at baseline and after 6 months Intervention clusters Control clusters Cluster-adjusted prevalence of residents in each cluster with any physical restraint at baseline and after 6 months (size of circle proportional to size of cluster). Diagonal line represents no change in prevalence of residents with any physical restraint from baseline to 6 months. Clusters above line had an increased prevalence of residents with any physical restraint and clusters below line decreased. Clusters near line show little change.
References (1) Guyatt G, Oxman A, Vist G et al. GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ 2008;336:924-926. (2) Meyer G, Köpke S, Haastert B, Mühlhauser I. Comparison of a fall risk assessment tool with nurses' judgement alone: a cluster-randomised controlled trial. Age Ageing 2009;38:417-423. (3) Meyer G, Warnke A, Bender R, Mühlhauser I. Effect on hip fractures of increased use of hip protectors in nursing homes: cluster randomised controlled trial. BMJ 2003;326:76. (4) Möhler R, Richter T, Köpke S, Meyer G. Interventions for preventing and reducing the use of physical restraints in longterm geriatric care. Cochrane Database Syst Rev 2011;CD007546. (5) Ajzen I. The theory of planned behavior. Organizational Behavior and Human Decision Processes 1991;50:179-211. (6) Ceccato N, Ferris L, Manuel D, Grimshaw J. Adopting health behavior change theory throughout the clinical practice guideline process. J Contin Educ Health Prof 2007;27:201-207. (7) Curran S. Staff resistance to restraint reduction: identifying & overcoming barriers. J Psychosoc Nurs Ment Health Serv 2007;45:45-50. (8) Dunbar J, Neufeld R, White H, Libow L. Retrain, don't restrain: the educational intervention of the National Nursing Home Restraint Removal Project. Gerontologist 1996;36:539-542. (9) Francke A, Smit M, de Veer A, Mistiaen P. Factors influencing the implementation of clinical guidelines for health care professionals: a systematic meta-review. BMC Med Inform Decis Mak 2008;8:38. (10) Grimshaw J, Thomas R, MacLennan G et al. Effectiveness and efficiency of guideline dissemination and implementation strategies. Health Technol Assess 2004;8:iii-72. (11) Capezuti E, Wagner L, Brush B, Boltz M, Renz S, Talerico K. Consequences of an intervention to reduce restrictive side rail use in nursing homes. J Am Geriatr Soc 2007;55:334-341. (12) Huizing A, Hamers J, Gulpers M, Berger M. A cluster-randomized trial of an educational intervention to reduce the use of physical restraints with psychogeriatric nursing home residents. J Am Geriatr Soc 2009;57:1139-1148. (13) Evans L, Strumpf N, Allen-Taylor S, Capezuti E, Maislin G, Jacobsen B. A clinical trial to reduce restraints in nursing homes. J Am Geriatr Soc 1997;45:675-681. (14) Bero L, Grilli R, Grimshaw J, Harvey E, Oxman A, Thomson M. Closing the gap between research and practice: an overview of systematic reviews of interventions to promote the implementation of research findings. The Cochrane Effective Practice and Organization of Care Review Group. BMJ 1998;317:465-468. (15) Moore K, Haralambous B. Barriers to reducing the use of restraints in residential elder care facilities. J Adv Nurs 2007;58:532-540. (16) Driessen M, Proper K, Anema J, Bongers P, van der Beek A. Process evaluation of a participatory ergonomics programme to prevent low back pain and neck pain among workers. Implement Sci 2010;5:65. (17) Statistisches Bundesamt. Earnings and work costs 2010 [in German]. Wiesbaden: Statistisches Bundesamt, 2011.
eappendix "Post-hoc repeated-measures analysis" A post-hoc analysis of the primary outcome was performed for the sub of residents with assessment of physical restraints at all three dates of physical restraint assessment (at baseline, after three and after six months). Methods First, prevalences of physical restraints were estimated for each assessment point including cluster adjusted 95% confidence intervals (CI). Generalized (logistic) mixed models adjusted for repeated measurement and for clusters were fitted to the sub of residents with assessment of physical restraints at all three dates of physical restraint assessment (at baseline, after three and after six months). Assessments at three months and six months with the binary indicator for physical restraints were used as dependent variable and intervention versus control was used as fixed effect [1]. A random effect for clusters was included. Adjustment for repeated measurement was performed using a covariance pattern model with compound symmetry structure [1]. Other models were fitted to investigate adjustment for time effects and interaction between intervention and time. Adjustment for baseline prevalence of physical restraints was performed by including the corresponding indicator as additional fixed effect in the model. Generalized mixed models were fitted using SAS PROC GLIMMIX version 9.3 (TS1M0) on Windows 7 64 Bit. Prevalences of physical restraints and odds ratios of intervention effects were calculated separately for assessments at three months and six months stratified by baseline physical restraint status to illustrate the effect of baseline adjustment on raw data (with raw confidence intervals, not cluster adjusted). Results For the post-hoc sub analysis of residents available for assessment of physical restraints at all three assessment dates, 2985 residents were included (intervention : 1531, control : 1454). Physical restraints use in the sub of residents available at all three assessment dates Baseline 3 months 6 months Intervention (n=1531) Control (n=1454) Intervention (n=1531) Control (n=1454) Residents with any 473 (30.9 433 (29.8 377 (24.6 459 (31.6 physical restraint [25.1-36.7]) [24.6-34.9]) [19.8-29.5]) [27.4-35.7]) Values are numbers of residents (cluster-adjusted percentages [95% CI]). Intervention (n=1531) 379 (24.8 [19.9-29.6]) Control (n=1454) 460 (31.6 [27.1-36.1]) Results are comparable to results for the primary analysis (Table 2). The generalized (logistic) mixed model including only the intervention as fixed effect results in an estimated odds ratio adjusted for cluster correlation and repeated measurement (at three months and six months) of 0.73 (95% CI 0.53-1.02, p = 0.062). An extended model including time and interaction of time and effects showed no significant effects (time: p = 0.861, interaction: p = 0.947). Adjusting for baseline physical restraints by including
individual restraint indicators as fixed effects in the model resulted in a lower odds ratio of 0.44 (CI 0.29-0.69, p < 0.001) compared to the primary analysis and a significant effect for physical restraint status at baseline (OR 62.36, CI 49.91-77.90, p < 0.001) with no significant interaction between intervention and physical restraint status at baseline (p = 0.323 for interaction in an extended model). All results are adjusted for cluster correlation and repeated measurement. Results for raw data stratified by baseline physical restraint status support the model based estimations. The next two small tables show that the use of physical restraints during follow up is determined by physical restraint use at baseline. Here, adjustment for physical restraint use at baseline indicates a pronounced intervention effect for both residents with and without physical restraints at baseline. Physical restraints use in residents available at all three assessment dates without physical restraints at baseline 3 months 6 months Intervention Control Intervention Control (n=1058) (n=1021) (n=1058) (n=1021) Residents with any physical restraint 40 (3.8%) 83 (8.1%) 57 (5.4%) 99 (9.7%) Odds ratio (95% CI) 0.44 (0.30-0.65) 0.53 (0.38-0.74) Values are numbers of residents (percentages) and raw odds ratios (95% CI without cluster adjustment) Physical restraints use in residents available at all three assessment dates with physical restraints at baseline Residents with any physical restraint 3 months 6 months Intervention (n=473) Control (n=433) Intervention (n=473) Control (n=433) 337 (71.3%) 376 (86.8%) 322 (68.1%) 361 (83.4%) Odds ratio (95% CI) 0.38 (0.27-0.53) 0.43 (0.31-0.58) Values are numbers of residents (percentages) and raw odds ratios (95% CI without cluster adjustment) Reference [1] Brown H, Prescott R: Applied Mixed Models in Medicine. Statistics in Practice. 2nd Edition. John Wiley & Sons, 2006.