Title: What is the optimal period for measuring hand-hygiene compliance: are longer periods better than 20 minute periods? Authors: Sheldon Stone UCL Medical School, University College London Chris Fuller, UCL Research Dept of Infection and Population Health, University College London. Susan Michie, UCL Dept of Clinical, Educational and Health Psychology, University College London. 1 John McAteer, UCL Dept of Clinical, Educational and Health Psychology, University College London. Andre Charlett, Health Protection Services, Health Protection Agency, London. 1 current affiliation Scottish Collaboration for Public Health Research and Policy, Medical Research Council, Edinburgh. Corresponding author: Chris Fuller, UCL Research Dept of Infection and Population Health, University College London (Hampstead Campus), Royal Free Hospital, London, NW3 2PF. (christopher.fuller@ucl.ac.uk). Word count: 900 Clinical Trial identifier: ISRCTN65246961 1
INTRODUCTION Direct observation of hand hygiene is considered the gold standard for measuring healthcare workers (HCW) hand hygiene compliance (HHC) for clinical audit and hand hygiene interventions trials. Many studies and audits observe for 20-30 minutes but systematic review shows that others observe for 1-4 hours or more, without explaining the rationale for this (1). WHO guidance recommends observation for 20 minutes (+10 minutes if necessary) (2). HCWs hand-hygiene may improve when they are aware of being observed (3), but it is unclear whether this reactivity increases or decreases over time or introduces systematic bias. Short periods of observation may not reflect twenty-four hour, seven days a week behaviour (3). In addition short periods may not provide enough observations to meet previously identified criteria for inter-observer reliability as compliance levels differ between observers by over 10% if <15 hand hygiene moments are observed (4). No study has investigated whether there are significant differences between compliance recorded over 20 minutes, one hour, or longer. This study aimed to investigate whether compliance in the first 20 minutes or the first hour differed substantially from that observed over four hours. Findings informed the choice of the optimal observation period for a randomised controlled trial of a hand hygiene intervention (the FIT study ISRCTN65246961) (5). METHOD 2
Observations were carried out using a validated tool (4) by one of three observers, trained as described elsewhere (6). Fifty three, four-hour (10:00-12:00 and 13:00-15:00; October 2006 - January 2007) covert observation sessions were performed on 13 Intensive Therapy Units and 36 Acute Care of the Elderly wards (13 hospitals) across England & Wales during FIT s baseline phase. Data were collected in 20- minute segments. Ethical permission was obtained (05/MREC10/2) Statistical Analysis: Hour to hour variation in compliance was examined by a mixed effects logistic regression model with a binary outcome of HHC, including hospital and ward within hospital as random effects. Ward type (ITU or ACE) and sequential hourly observation period (1 st hour, 2 nd hour, 3 rd hour, and 4 th hour) were included as fixed effects. A similar analysis examined variation in compliance over sequential 20-minute periods. Ward type was excluded after showing no evidence of effect on trend or compliance. RESULTS A total of 3989 hand-hygiene moments and associated behaviours were observed. Overall compliance was 75%. For sequential hour periods (Table), compliance was lowest in the 1 st hour (71%), with the estimated odds ratios for compliance rising significantly (1.32, 1.08-1.61; p=0.007) in the 2 nd hour, remaining stable thereafter. 3
For sequential 20 minute periods (Table ) compliance was lowest (69%) in the 1 st 20 minutes, with the estimated odds ratios rising significantly in the 2 nd 20 minutes (1.42, 1.02-1.96; p=0.04), although not in the third. Odds-ratios then rose and remained stable from the fourth period onwards, although there was fluctuation between the last three 20 minute segments. DISCUSSION Compliance was slightly but significantly lower in the first hour of a four hour observation period. The odds of compliance rose significantly in the 2 nd hour, remaining stable thereafter. This was reflected in the measurement of compliance in 20-minute sequences, where compliance was lowest in the first 20 minutes. No previous study has broken observation periods down into such discrete sequences. Study strengths include size, geographical spread, variety of patient groups and use of a standardised tool. Results are probably representative of English and Welsh practice, and generalisable to acute hospitals. Limitations include use of a convenience sample of wards and difficulties ensuring entirely covert observation. There are several possible explanations of the findings, all speculative, as the study was not designed to explain such variation. The small rise in compliance after the initial observation period may reflect reactivity to observer s presence, a well documented phenomenon in the wider literature (7, 8). The data suggest that the first observation session may be the least reactive and therefore more 4
representative of actual behaviour. Fluctuations during the 20 minute sequences in the fourth hour may be explained by observer fatigue with poorer attention to task. It is unlikely that higher compliance is a result of falling workload (9) since the compliance remained stable in the final three hours despite successive reductions in the number of hand-hygiene moments. Findings indicate that observation periods should be either 20 minutes or one hour, since these periods may be the least reactive. However, the number of hand-hygiene moments observed over 20 minutes (Table) was unlikely to provide enough observations (at least 15) to meet the observational tool s inter-rater reliability criteria (4). The optimal period of observation for the FIT trial was therefore set at one hour for each ward. Subsequent power calculations showed that this would give the trial 98% power to detect a 10% difference, if carried out every 6 weeks (5). The study also has implications for clinical audit, where hour long observation periods may not be pragmatic. The data support the use of 20-minute observation periods, as the least reactive period, provided that at least 15 hand-hygiene moments were recorded to meet inter-rater reliability criteria. This may necessitate more than one 20 minute observation period on separate days but is in line with WHO recommendations (2) that observations be carried out in 20-30 minute segments (10). This observation period may also be suitable for trials, if power calculations based on pilot data show this provides adequate power. 5
Researchers should state and explain the period and frequency of observations in grant applications and publications (1). ACKNOWLEDGEMENTS Financial Support: Funding for the study was received from the Patient Safety Research Programme, the Royal Free Hospital Trustees and GOJO industries. JM received funding for his PhD from the Hospital Infection Society. Potential Conflicts of Interest: None of the authors reported potential conflicts of interest. 6
References 1. Gould DJ, Chudleigh J, Drey NS, Moralejo D. Measuring handwashing performance in health service audits and research studies. J Hosp Infect 2007; 66: 109-115. 2. WHO Patient Safety: A World Alliance For Safer Health Care. SAVE LIVES: Clean your hands. Hand hygiene technical reference to be used by health-care workers, trainers and observers of hand hygiene practices. Geneva: WHO; 2009. 3. Joint Commission. Measuring Hand Hygiene Adherence: overcoming the challenges. Illinois: Joint Commission; 2009. 4. McAteer J, Stone S, Fuller C, Charlett A, Cookson B, Slade R, et al. Development of an observational measure of healthcare worker hand-hygiene behaviour: the handhygiene observation tool (HHOT). J Hosp Infect 2008; 68: 222-9. 5. Stone S, Fuller C & the NOSEC/FIT investigators. Report to the Patient Safety Research Programme. 2010. Last accessed 04-25-2012 http://www.birmingham.ac.uk/documents/collegemds/haps/projects/cfhep/psrp/finalreports/ps029finalreportstone.pdf 6. Fuller C, Besser S, Cookson B, Fragaszy E, Gardiner J, McAterr J, Michie S, Savage J, Stone S. Assessment of blinding of hand hygiene observers in randomised controlled trials of hand hygiene interventions. Am J infect Contr 2010: 38;332-334 7
7. Baer D M, Harrison R, Fradenburg L, Petersen D, & Milla S. Some pragmatics in the valid and reliable recording of directly observed behaviour. Research on Social Work Practice 2005; 15: 440-51. 8. Gittelsohn J, Shankar A V, West K P, Ram R M, & Gnywali T. Estimating reactivity in direct observation studies of health behaviours. Human Organization 1997; 56: 182-189. 9. Boyce, J. M., & Pittet, D. Guideline for hand-hygiene in healthcare settings: recommendations of the healthcare infection control practices advisory committee and the HICPAC/SHEA/APIC/IDSA hand-hygiene task force. Infect Contr & Hosp Epidem: 2002; 23 (suppl), 3-40. 10. WHO. Appendix 4: Monitoring hand hygiene by direct methods. In: WHO Guidelines on Hand Hygiene in Health Care. Geneva: WHO; 2009. 8
Table1: Observed hand hygiene compliance (HHC) and estimated odds ratio (OR) for compliance in each sequential 20 minute and one hour periods of observation. No of observations HHC O.R. 95% CI p. value (%) Hour-period 1st 1212 71.4 reference 2nd 1088 77.0 1.32 1.08 to 1.61 0.007 3rd 997 76.2 1.32 1.07 to 1.62 0.008 4th 692 76.4 1.40 1.11 to 1.76 0.004 20-minute period 1st 417 69.1 reference 2nd 391 74.4 1.42 1.02 to 1.96 0.04 3rd 404 70.8 1.30 0.94 to 1.78 0.11 4th 381 78.0 1.63 1.17 to 2.29 0.004 5th 341 75.4 1.43 1.01 to 2.01 0.04 6th 366 77.6 1.75 1.24 to 2.46 0.001 7th 333 76.6 1.67 1.17 to 2.36 0.004 8th 333 75.1 1.53 1.08 to 2.16 0.02 9th 331 77.0 1.63 1.15 to 2.30 0.006 10th 285 78.9 1.90 1.31 to 2.77 0.001 11th 251 70.1 1.26 0.88 to 1.81 0.2 12th 156 82.1 2.44 1.51 to 3.96 <0.001 9