978 I N F E C T I O N CONTROL AND H O S P I T A L E P I D E M I O L O G Y S E P T E M B E R 2 0 1 0 , VOL. 3 1 , N O . 9
Positive Deviance and Hand Hygiene: to have a statistical comparison of the device-associated-in-
fection and the healthcare-associated-infection incidence
More Questions than Answers density changes in the second and third phases, separately,
compared with the first phase (baseline infection rate). We
should expect significant infection incidence density changes
in the east SDU in the second phase and in both SDUs in
To the Editor—Optimizing adherence to hand hygiene rec- the third phase, but no difference between units when com-
ommendations has been problematic in many healthcare set- paring first and third phases.
tings. We read with interest the article by Marra et al1 relating It is not clear whether PD results in a sustained effect. An
their experience with a positive deviance (PD) approach to observation period of only 3-6 months is simply not long
improving hand hygiene adherence. The authors should be enough to judge. As the authors note, the generalizability of
commended for adding to our understanding of hand hygiene this approach is unknown.
and the PD strategy for improving adherence, but a number An extensive list of potential confounders should be noted.
of limitations should be noted and several questions should The overall antibiotic use (number of defined daily doses per
be addressed. 1,000 patient-days) in the first unit to undergo PD interven-
Because actual opportunities for hand hygiene were not tion increased dramatically (by 52%), even though the num-
monitored (use of alcohol-based gel was measured), the level ber of healthcare-associated infections decreased by almost
of hand hygiene adherence in the study is not known. In the 31%. The authors did not further characterize this counter-
east step-down unit (SDU), the rate of alcohol-based gel use intuitive observation. If anything, one would expect antibiotic
increased from 47 dispensations per patient-day to approx- use to decrease in response to successful prevention of health-
imately 60 dispensations per patient-day, whereas in the west care-associated infection. The length of stay decreased dra-
SDU, the rate of gel use initially decreased (before PD was matically in both units (by more than 20%) during the first
introduced) and then returned to baseline (with PD). In our intervention phase. No explanation for this observation was
own study in a critical care unit, we noted opportunities for offered, and it might indicate other forces at work. Severity
hand hygiene at a rate of 295 opportunities per patient-day,2 of illness is an obvious confounder that may have been in-
and Pittet et al3 noted a rate more than twice as high as ours. directly measured by the nurse activity score. This score dif-
In unpublished observations in our SDU, we noted oppor- fered significantly between units in 2 of the 3 observation
tunities for hand hygiene at a rate of 276 opportunities per periods, decreased steadily by 11% in one unit over the course
patient-day. Thus, the rate of gel use observed by Marra et of the study, and fluctuated in the other unit.
al1 at its highest (62 dispensations per patient-day) may have Thus, as is often the case with even the best studies, we
been woefully inadequate; an explanation by the authors for are left with many questions unanswered. A great deal of
the possible reasons for such a low rate would be welcome. work remains to be done to determine how best to measure
It is not clear whether the PD approach worked. It would hand hygiene rates, to improve adherence to hand hygiene
be of interest to know the change (mean difference and stan- recommendations, and to ascertain what impact hand hygiene
dard deviation) in the number of alcohol-based gel aliquots has on healthcare-associated infection.
dispensed between the first phase and the second phase in
the west and the east SDUs separately; these 2 analyses would ACKNOWLEDGMENTS
provide the "over-time baseline changes" (use of gel without Potential conflicts of interest. Both authors report no conflicts of interest
PD) and the "over-time intervention changes" (use of gel relevant to this article.
with PD), respectively. Then, the over-time baseline changes
in the west SDU should be statistically compared with the Mark E. Rupp, MD; Andre C. Kalil, MD
over-time intervention changes in the east SDU to better
determine whether the effect of PD was confounded by the
From the Department of Internal Medicine, University of Nebraska
expected (natural course) over-time baseline changes in the
Medical Center, Omaha (both authors).
use of gel. Also, a comparison analyzing the changes from Address reprint requests to Mark E. Rupp, MD, 984031 Nebraska Medical
the first phase (baseline) to the third phase (PD in both units) Center, Omaha, NE 68198-4031 ().
should produce significant changes (increase in gel use) Infect Control Hosp Epidemiol 2010; 31(9):978-979
within each unit but no significant changes between units, © 2010 by The Society for Healthcare Epidemiology of America. All rights
reserved. 0899-823X/2010/3109-0020$15.00. DOI: 10.1086/656204
because both underwent intervention with PD in the third
phase.
It is not clear what types of healthcare-associated infections
were reduced. The authors showed quite nicely that various REFERENCES
device-associated infections were not significantly impacted, 1. Marra AR, Guastelli LR, Pereira de Araujo CM, et al. Positive deviance:
but they did not relate what types of infections were reduced. a new strategy for improving hand hygiene compliance. Infect Control
Also, similar to our suggestion above, it would be important Hosp Epidemiol 2010;31:12-20.
Positive Deviance and Hand Hygiene: to have a statistical comparison of the device-associated-in-
fection and the healthcare-associated-infection incidence
More Questions than Answers density changes in the second and third phases, separately,
compared with the first phase (baseline infection rate). We
should expect significant infection incidence density changes
in the east SDU in the second phase and in both SDUs in
To the Editor—Optimizing adherence to hand hygiene rec- the third phase, but no difference between units when com-
ommendations has been problematic in many healthcare set- paring first and third phases.
tings. We read with interest the article by Marra et al1 relating It is not clear whether PD results in a sustained effect. An
their experience with a positive deviance (PD) approach to observation period of only 3-6 months is simply not long
improving hand hygiene adherence. The authors should be enough to judge. As the authors note, the generalizability of
commended for adding to our understanding of hand hygiene this approach is unknown.
and the PD strategy for improving adherence, but a number An extensive list of potential confounders should be noted.
of limitations should be noted and several questions should The overall antibiotic use (number of defined daily doses per
be addressed. 1,000 patient-days) in the first unit to undergo PD interven-
Because actual opportunities for hand hygiene were not tion increased dramatically (by 52%), even though the num-
monitored (use of alcohol-based gel was measured), the level ber of healthcare-associated infections decreased by almost
of hand hygiene adherence in the study is not known. In the 31%. The authors did not further characterize this counter-
east step-down unit (SDU), the rate of alcohol-based gel use intuitive observation. If anything, one would expect antibiotic
increased from 47 dispensations per patient-day to approx- use to decrease in response to successful prevention of health-
imately 60 dispensations per patient-day, whereas in the west care-associated infection. The length of stay decreased dra-
SDU, the rate of gel use initially decreased (before PD was matically in both units (by more than 20%) during the first
introduced) and then returned to baseline (with PD). In our intervention phase. No explanation for this observation was
own study in a critical care unit, we noted opportunities for offered, and it might indicate other forces at work. Severity
hand hygiene at a rate of 295 opportunities per patient-day,2 of illness is an obvious confounder that may have been in-
and Pittet et al3 noted a rate more than twice as high as ours. directly measured by the nurse activity score. This score dif-
In unpublished observations in our SDU, we noted oppor- fered significantly between units in 2 of the 3 observation
tunities for hand hygiene at a rate of 276 opportunities per periods, decreased steadily by 11% in one unit over the course
patient-day. Thus, the rate of gel use observed by Marra et of the study, and fluctuated in the other unit.
al1 at its highest (62 dispensations per patient-day) may have Thus, as is often the case with even the best studies, we
been woefully inadequate; an explanation by the authors for are left with many questions unanswered. A great deal of
the possible reasons for such a low rate would be welcome. work remains to be done to determine how best to measure
It is not clear whether the PD approach worked. It would hand hygiene rates, to improve adherence to hand hygiene
be of interest to know the change (mean difference and stan- recommendations, and to ascertain what impact hand hygiene
dard deviation) in the number of alcohol-based gel aliquots has on healthcare-associated infection.
dispensed between the first phase and the second phase in
the west and the east SDUs separately; these 2 analyses would ACKNOWLEDGMENTS
provide the "over-time baseline changes" (use of gel without Potential conflicts of interest. Both authors report no conflicts of interest
PD) and the "over-time intervention changes" (use of gel relevant to this article.
with PD), respectively. Then, the over-time baseline changes
in the west SDU should be statistically compared with the Mark E. Rupp, MD; Andre C. Kalil, MD
over-time intervention changes in the east SDU to better
determine whether the effect of PD was confounded by the
From the Department of Internal Medicine, University of Nebraska
expected (natural course) over-time baseline changes in the
Medical Center, Omaha (both authors).
use of gel. Also, a comparison analyzing the changes from Address reprint requests to Mark E. Rupp, MD, 984031 Nebraska Medical
the first phase (baseline) to the third phase (PD in both units) Center, Omaha, NE 68198-4031 ().
should produce significant changes (increase in gel use) Infect Control Hosp Epidemiol 2010; 31(9):978-979
within each unit but no significant changes between units, © 2010 by The Society for Healthcare Epidemiology of America. All rights
reserved. 0899-823X/2010/3109-0020$15.00. DOI: 10.1086/656204
because both underwent intervention with PD in the third
phase.
It is not clear what types of healthcare-associated infections
were reduced. The authors showed quite nicely that various REFERENCES
device-associated infections were not significantly impacted, 1. Marra AR, Guastelli LR, Pereira de Araujo CM, et al. Positive deviance:
but they did not relate what types of infections were reduced. a new strategy for improving hand hygiene compliance. Infect Control
Also, similar to our suggestion above, it would be important Hosp Epidemiol 2010;31:12-20.
