1079
CLINICAL PRACTICE
Prospective controlled study of four infectioncontrol procedures to prevent nosocomial infection with respiratory syncytial virus
To determine the most effective infection control
procedure in preventing nosocomial infection with respiratory syncytial virus (RSV), we did a prospective controlled study of four infectioncontrol strategies in four wards in a large paediatric hospital in the west of Scotland. All children under two years old admitted to four general wards during three winter RSV epidemics (1989-92) were screened for RSV infection (by nasopharyngeal aspirate and direct immunofluorescence) within 18 hours of admission. The main outcome measure was the occurrence of nosocomial infection, defined as the number of children initially RSV negative who became RSV positive 7 days or more after hospital admission (incubation period for RSV infection is
5-8 days). Without special precautions, there was a high rate of nosocomial RSV infection (26%). Nosocomial infection was significantly reduced by the combination of cohort nursing with the wearing of gowns and gloves for all contacts of RSV-infected children (p=0·0022). Neither the use of gowns and
gloves alone nor cohort nursing alone produced a significant reduction in cross-infection. In the final year, general clinical use of a policy of cohort nursing with gowns and gloves resulted in a reduction in the cross-infection rate by two-thirds of its original value (9·5% vs 26%). Combined with rapid laboratory diagnosis, cohort nursing and the wearing of gowns and gloves for all contacts
with
significantly
RSV-infected
children
can
reduce the risk of nosocomial RSV
infection. Lancet 1992; 340: 1079-83.
Introduction In 1971, Ditchburn
et
al identified the hazards for
uninfected children and personnel in hospitals from infants with respiratory syncytial virus (RSV) infections. Nosocomial RSV infection is a serious problem, especially for infants with cardiorespiratory diseases such as congenital heart disease, or immunodeficiency states.2,3 The virus is
spread by close
contact
with infected secretions,
by large-
and by fomites.4-6 Small-particle aerosols be a major route of transmission.6 Hospital personnel are thought to be the main vector, transmitting the virus on contaminated hands and even becoming infected themselves by self-inoculation of mucous
particle aerosols, do
not seem to
membranes.7 Since the mid-1970s a few studies have assessed ways to limit nosocomial spread of RSV among children in hospital.8--12 Since RSV is a lipid-enveloped virus, simple handwashing with soap and water should prevent virus transmission. However, staff do not always wash their hands after every patient contact." Consequently, the main strategy has been to follow procedures that protect staff from being contaminated by infected secretions. Various
strategies include wearing gowns and gloves,lO cohorting RSV-infected patients,l1 and cohorting and handwashing.12 When special care was taken to ensure that gowns and gloves were worn, there was a substantial reduction in nosocomial infection.10 Consideration has also been given to the protective value of face-masks and eye-nose goggles.9,14,15 Masks did not lead to a measurable benefitin terms of infection control. The use of eye-nose goggles was associated with a significant decrease in nosocomial infection, but they are not popular with clinical staff and are frightening to children. Because of differing approaches, which particular infection-control strategy most effectively prevents nosocomial RSV infection remains unclear. Comparisons of previous published studies are difficult because of methodological differences, which include screening criteria, ward design, variations in laboratory diagnostic technique, and ways of assessing cross-infection rates. To address these limitations we carried out a prospective controlled study of four infection-control protocols on children admitted to four wards during two winter epidemics. Our aim was to define the most effective infection-control procedure for the prevention of nosocomial infection on general paediatric wards with limited isolation facilities. In the third year, the most effective strategy was introduced into all ward areas and its efficacy in clinical practice was assessed. ADDRESSES: Departments of Child Health (P Madge, RGN, J. Y Paton, MD) and Statistics (J. H. McColl, MSc), University of Glasgow, and Department of Microbiology (P. L. K. Mackie, PhD), Royal Hospital for Sick Children, Glasgow, UK. Correspondence to Dr J. Y Paton, Department of Child Health, Royal Hospital for Sick Children, Yorkhill, Glasgow G3 8SJ, UK.
1080
Subjects and
methods
TABLE I-INFECTION-CONTROL MEASURES USED IN EACH WARD
Prospective study The prospective study, approved by the ethics committee of the Royal Hospital for Sick Children, Glasgow, was conducted in four medical wards of that hospital during two winter RSV epidemics (1989/90 and 1990/91). Each ward was identical in layout with 8 single cubicles and 3 larger areas of 4, 6, and 6 beds. Each ward admitted new children once every 4 days. When a ward was full, the next ward in rotation took over receiving duties. There was no
separate ward area for managing children with infections.
CN cohort nursing; Go/Gtgowns and gloves for all contact with RSV-positive children; Go/GI, CN =cohort nursing plus gowns and gloves for all contact with
In all 3 years, the study started when 5 infants with symptoms of bronchiolitis were admitted to hospital within 1 week and were shown to have RSV infection. The study continued for 3 months thereafter. Every child less than two years old, irrespective of
RSV-positive children *Basic infection control measures adopted m all wards-ie, hand washmg before and after any patient contact, and wearing of gowns and gloves when dealing with any body fluids tWard 2 only took part in winters 2 and 3
clinical presentation, had respiratory secretions tested for RSV antigen within 18 hours of admission. All those found to be infected with RSV (RSV positive) were tested daily until they were negative, or were discharged home. Children who remained in hospital for more than 7 days, including those who were initially RSV negative, were tested weekly or when signs of respiratory illness developed. Nosocomial infection was assumed ifa child, initially RSV negative, became RSV positive 7 days or more after admission (the incubation period for RSV infection is 5-8 days). We did not study children after discharge from hospital; however, no child was readmitted with RSV infection during the remainder of that winter epidemic.
Virus screening and detection RSV antigen
sought by direct immunofluorescence with pooled RSV-specific monoclonal antibodies (Dako Ltd, High Wycombe, UK) to detect infected exfoliated cells in respiratory secretions obtained by nasopharyngeal aspiration. Before RSV results were available, newly admitted children with one or more of the following clinical features were regarded as RSV positive: upper respiratory tract infection, breathlessness or reluctance to feed, cough or apnoea, and signs of respiratory distress such as tachypnoea. They were nursed according to the individual ward’s policy for RSV-infected patients until RSV status was known. No child initially placed in an infected cohort but shown subsequently to be RSV negative became RSV positive. We initially planned that children at high risk from severe nosocomial RSV infection (eg, with bronchopulmonary dysplasia, cystic fibrosis) should be admitted and nursed in single-bed cubicles; owing to other demands for single rooms this rarely happened. Children with was
cardiac disease were nursed in a separate cardiac ward and were not part of the study population. Whenever possible nurses who
developed a respiratory illness were encouraged to care for babies in the RSV area, but no formal attempt was made to diagnose RSV infection in the staff. While they were RSV positive, children were managed according to an infection-control policy assigned to that ward (table i). Two consecutive negative laboratory results indicated that the patient was a low infection risk and could therefore be nursed with routine procedures outwith the infected cohort. These measures were followed by all staff dealing with RSV-infected children. A research nurse (P. M.) supervised the implementation of the ward policies. However, no special steps were taken to supervise compliance with the measures in use in any ward. In year 3, the policy of cohort nursing with the use of gowns and gloves for all patient contacts was adopted in all four medical wards. Statistics Three statistical methods were used. First, the proportions of infants becoming cross-infected in the four groups were compared by a X2 test of homogeneity.16 The null hypothesis was that the probability of cross-infection was the same under all four conditions. Second, since a X2 test does not exploit relations among the four combinations of infection-control procedures, we used the more appropriate linear logistic regression analysis. 16 The two treatments
(gowns and gloves, cohorting) each appeared at two levels (absent, present). Letting 0 (i, j) denote the probability of cross-infection at the i’th level of gown and gloves (i 0 [absent], 1 [present]) and the j’th level of cohorting (j =0 [absent], 1 [present]) the log-odds on =
RSV cross-infection was modelled as: In
e(ij)
.
...
Oct ——,-o7-:-=-ca-P)-yi) [h Yll
1-0
(’.))
is the reduction (if any) in the log-odds due to gowns and gloves alone,0 the reduction (if any) due to cohorting alone, and y the further reduction (if any) when both procedures were combined-ie, y is an interaction term. Unlike the global X2 test, logistic regression allows direct testing of each of these effects. Forward and backward stepwise selection was used. Third, logistic regression does not take account of the level of risk experienced by individual patients. Individual stays for infants developing nosocomial infection varied from 7 to 63 days. In addition, children admitted to hospital towards the beginning or end of the epidemic were at lower risk than children who were in hospital for the same length of time at the peak of the epidemic when many RSV-infected children were in the wards. This can be considered as a survival analysis problem16,17 in which the interest is in modelling the probability of "survival" (ie, non-infection) as a declining function of time in hospital (the survivor function). In fact, rather than simply using calendar time, we decided to weight each day of an individual’s stay to reflect the fluctuating level of risk in the ward. The weights used were the numbers of RSV-positive infants in the ward each day. For non-infected children, then, "time" of exposure was taken as the sum of these weights over all the days up to and including the day before RSV infection was diagnosed. We also compared the survivor functions across the four infection-control groups with Cox’s proportional hazards model. As in the logistic regression model, covariates were included to denote the individual and joint effects of the two isolation where
a
procedures. All statistical analyses were performed with the BMDP computer
package.
Results Table n shows the number of RSV-positive infants admitted during each winter epidemic. Over the first two winter epidemics the number of RSV-positive children admitted to each study group was broadly similar and thus the burden of infection on each ward was similar. At the end of winter 1 (table Hi), there was a high cross-infection rate in children who were in hospital for more than 7 days (26%) who were given "no special precautions" (handwashing after all contacts and the wearing of gowns and gloves for contact with body fluids, table l). By contrast, the most stringent policy of gloves and a gown for all patient contacts combined with cohort nursing yielded a substantial reduction in infection (cross-infection rate 5%). However, the number of cross-infections was small, and the reduction achieved was not statistically significant. After considering the results, the policy of "no
1081
TABLE II-NUMBERS OF RSV INFANTS IN EACH
INFECTION-CONTROL GROUP
TABLE III-NUMBER OF NOSOCOMIAL INFECTIONS IN RELATION TO INFECTION-CONTROL PROCEDURE USED
*(Mean %, 95% CI).
special precautions" was discontinued. Instead, a gown and glove group was introduced and the other two policies continued. By the end of the second winter, the picture was clearer (table ill), confirming the results of the first year’s study. Over the two years, there was a significant reduction in cross-infection from using gloves and gown combined with cohort nursing (=8-1, p = 0044). Logistic regression showed that the data were consistent with a model in which (x = &bgr; 0 and only the interaction term y was =
The same model resulted from both forward and backward stepping; the p value for retaining y in the final model was 0-0022. This suggested that neither a policy of gown and gloves alone nor cohorting alone produced a significant reduction in cross-infection, but the combination of the two significantly reduced it. non-zero.
Survivor function for Cox proportional hazards model. For the model used, only the interaction term was significant. Neither of the policies alone had an effect on the rate of cross-infection.
TABLE IV-NUMBER OF NOSOCOMIAL INFECTIONS IN EACH WARD IN WINTER 3 WITH ALL WARDS USING GOWN AND GLOVES PLUS COHORT NURSING
In this model, exp (-y) measures the proportionate reduction in the odds on infection due to adopting both infection control procedures. y was estimated to be 2-32 (95% CI for exp (-y) 0-0127, 0-756). Thus the odds on cross-infection was likely to be reduced to between 1-27% and 75-6% of its previous level. In addition, it was possible to calculate an approximate 95% CI for e (1,1), which suggested that, when both procedures were in use, the true cross-infection rate was likely to be between 043% and 18-6%. The survival analysis confirmed the findings of the logistic regression. On fitting a proportional hazards model in a stepwise manner, both forward and backward stepping resulted in a model in which the two procedures individually were found to have no effect on the survivor function, but in combination reduced it significantly (p = 0 0057, figure). In the final year, the combined policy of cohort nursing with the wearing of gowns and gloves was adopted in all wards. By chance, that winter’s epidemic (1991-92) was particularly severe. Table IV shows that there was a significant reduction in cross-infection compared with the previous years’ data (table III), with the cross-infection rate being in the middle of the confidence interval obtained from the logistic regression. This represented a reduction in the cross-infection rate of approximately 16 percentage points (or two-thirds of its original value; table iv).
Discussion In the first year, the rate of nosocomial RSV infection using the simplest infection-control measures was high. Isaacs et al12 showed a similar cross-infection rate in children with congenital cardiac disease who had median hospital stays in excess of 30 days. These results emphasise that nosocomial RSV infection is an important hazard for uninfected children in hospital during an RSV epidemic. Measures to prevent such high rates of cross-infection are
clearly important. The use of rapid diagnostic virological techniques to screen children under two years was critical to this study. Without rapid and accurate ascertainment of which children are infected, any infection-control strategy is likely to be ineffective. In previous studies, screening of newly admitted children has often been confined to those with respiratory symptoms (eg, ref 12). However, this approach will miss children who present with atypical non-respiratory disease. We followed the approach of Krasinski et al," and screened all children under two irrespective of clinical diagnosis or presentation. We found that rapid diagnosis combined with an infection-control policy of cohort nursing and the wearing of gown and gloves for all contacts with RSVinfected children achieved a significant reduction in the risk of nosocomial RSV infection. From the logistic regression analysis, cohort nursing or gowns and gloves had no effect when used alone but in combination interacted to produce a
1082
substantial reduction in nosocomial infection. Although the of cross-infection using this combined strategy appeared higher in the third year, it was in the middle of the confidence interval predicted by the logistic regression model from the first two years’ data. Since our study was prospective and controlled it eliminated many of the problems that have made interpretation of previous studies difficult. Furthermore, the results from the final year, when the most effective policy was introduced into clinical practice, confirmed that the approach leads to an important reduction in cross-infection with RSV. One confounding effect that was not accounted for in the study design was a possible "ward effect". For practical reasons, two wards (3 and 4) continued with the same policy over the first two years of the study. Since it was also necessary to apply policies to whole wards, there is a possibility that ward 4 might have been especially effective at implementing their assigned policy. However, in the final year, the policy of cohort nursing plus gowns and gloves for all contacts with RSV-positive children (as in ward 4) was introduced into all wards. There was a noticeable reduction in nosocomial infection in the three wards where it had not previously been applied. That the resulting rates of nosocomial infection were very similar in all wards, and in the middle of the confidence interval from the logistic regression model, suggests that the observed improvement was due to the intervention rather than to any ward being especially good or bad at implementing their assigned policy. More formally, the logistic regression analysis was repeated with data from all 3 years. Again, only the original interaction term, y (the further reduction, if any, when both procedures were used in combination), was entered in the final model by the stepwise procedure. Neither a time effect nor ward effect was found to be significant. Unlike three earlier studies, we did not find that simpler measures (gowns and gloves alone,lO handwashing with cohorting," screening for RSV with cohorting12) significantly reduced the risks of cross-infection compared with routine measures of handwashing with the wearing of gowns and gloves for all contact with body fluids. However, the numbers of nosocomial infections in these two groups are quite small and the confidence intervals wide. There may be a small effect from cohort nursing, but the weight of evidence from this study is not sufficient to confirm a positive effect from these intermediate procedures. Since nosocomial infection can be associated with significant morbidity and mortality we felt it appropriate to adopt the measure that was clearly effective. One important difference that might provide an explanation for the failure of these simpler measures, found effective in earlier studies 12," relates to steps taken to improve compliance with the infection-control policy used. It is well known that compliance with handwashing is poor." Isaacs et al12 distributed a written handout on the importance of handwashing to try and improve it. Le Clair et apo showed that compliance with measures, especially with wearing gloves, was also poor, being followed in only 39% of contacts with infected children. After a period of education and encouragement this proportion rose to 81 %. It was only after compliance improved that a significant reduction in cross-infection was achieved. We did not include any measures to improve compliance with the infection-control methods used. Thus failures of compliance in our study may have accounted for the failure of the simpler measures to prevent cross-infection. rate
In respect of cohort nursing, the degree of physical separation of the infected and unifected cohort may also be important. In Krasinski’s studyll of rapid screening and cohorting, the infected and uninfected cohorts were separated by 100 feet (34 m) of intervening administrative offices. Our wards consist of a small number of single rooms and 3 contiguous small 4 or 6 bed bays. Consequently, such a high degree of physical separation cannot be achieved. Why, in our study, is there evidence of an interaction between the policies of cohorting and gowns and gloves, with an effect greater than achieved by either measure alone? We noticed that placing children in a cohort and wearing gowns and gloves generated considerable peer, and parental, "pressure" to ensure that the measures were followed. We suspect that this pressure produced a situation in which compliance in the combined group effectively became "self-policing". We speculate that a resultant improvement in compliance led to the significant reduction in crossinfection we observed during the prospective study in winters 1 and 2 in the gowns and glove plus cohort nursing group (table ill). The very slight increase in rate in the third year when this policy was adopted in all wards may, in part, reflect a breakdown in compliance under the pressure of an especially large epidemic. This may have important implications for the prevention of cross-infection when the modes of cross-infection are similar to those for RSV. Although simple procedures should prevent cross-infection, enthusiasm is likely to vary among individuals and wane with time. Education and persuasion do not generally lead to sustained improvement in handwashing.18 Over the longer term, compliance with infection-control procedures might, therefore, best be maintained by attempting to create self-policing, and hence self-enforcing, environments. In this study the rate of nosocomial RSV infection using the simplest infection-control measures was unexpectedly high. Since nosocomial RSV infection can be associated with significant morbidity, paediatricians need to keep this rate under review. In this context, we wonder whether cross-infection rates for RSV and, perhaps, other common viral infections, might provide useful outcome measures for auditing the "quality" of hospital care in paediatric units. In our experience, at the first signs of an RSV epidemic, the introduction of rapid diagnostic screening and a policy of cohort isolation and the wearing of gown and gloves for all patient contacts can significantly reduce the incidence of nosocomial RSV infection. The work was supported by Greater Glasgow Health Board Research Support Group Grant RSG/END/89/90F and RSG/END/90/91/M. Weare particularly grateful for the willing cooperation of the nurses, medical staff, and parents in the mdeical wards of the Royal Hospital for Sick Children, Glasgow, throughout the study.
REFERENCES 1. Ditchburn
RK, McQuillin J, Gardner PS, Court SDM. Respiratory syncytial virus in hospital cross-infection. BMJ 1971; 3: 671-73. 2. MacDonald NE, Hall CB, Suffin SC, Alexson C, Harris J, Manning JA. Respiratory syncytial virus infection in infants with congenital heart disease. N Engl J Med 1982; 307: 397-400. 3. Hall CB, Powell KR, MacDonald NE, et al. Respiratory syncytial virus infection in children with compromised immune function. N EnglJ Med 1986; 315: 4. Hall CB, Douglas virus. J Pediatr 5. Hall CB, Douglas
77-81. RG Jr. Modes of transmission of respiratory syncytial 1981; 99: 100-03. RG, Geiman JM. Possible transmission by fomites of respiratory syncytial virus. J Infect Dis 1980; 141: 98-102. 6. Hall CB, Douglas RG, Schnabel KC, Geiman JM. Infectivity of respiratory syncytial virus by various routes of inoculation. Infect Immun 1981; 33: 779-83.
1083
Chapin M. Infection rate in personnel caring for children with RSV infections. Am J Dis Child 1987; 141:
7. Agah R, Cherry JD, Garakian AJ,
695-97. 8. Hall CB, Geiman
JM, Douglas RG Jr, Meagher MP. Control of nosocomial respiratory syncytial viral infections. Pediatrics 1978; 62:
728-32. 9. Gala CL, Hall CB, Schnabel KC, et al. The use of eye-nose goggles to control nosocomial respiratory syncytial virus infection. JAMA 1986; 256: 2706-08. 10. Le Clair JM, Freeman J, Sullivan BF, Crowley CM, Goldman DA. Prevention of nosocomial respiratory syncytial viral infections through compliance with glove and gown isolation precautions. N Engl J Med 1987; 317: 329-34. 11. Krasinski K, LaCouture R, Holzman RS, Waithe E, Bonk S, Hanna B. Screening for respiratory syncytial virus and assignment to a cohort at admission to reduce nosocomial transmission. J Pediatr 1990; 116: 894-98. 12. Isaacs D, Dickson H, O’Callaghan C, Sheaves R, Winter A, Moxon ER.
and cohorting in prevention of hospital acquired infections with respiratory syncytial virus. Arch Dis Child 1991; 66: 227-31. 13. Goldman D, Larson E. Handwashing and nosocomial infections. N Engl J Med 1992; 327: 120-22. 14. Murphy D, Todd JK, Chao RK, Orr I, McIntosh K. The use of gowns and masks to control respiratory illness in pediatric personnel. J Pediatr 1981; 99: 746-50. 15. Hall CB, Douglas RG Jr. Nosocomial respiratory syncytial viral infections. Should gowns and masks be used? Am J Dis Child 1981; 135:
Handwashing
512-15. 16. Everitt BS. Statistical methods for medical investigations. New York: Oxford University Press, 1989. 17. Cox DR, Oakes D. Analysis of survival data. London: Chapman Hall, 1984. 18. Conly JM, Hill S, Ross J, Lertzman J, Louie TJ. Handwashing practices in an intensive care unit: the effects of an educational program and its relationship to infection rates. Am J Infect Control 1983; 17: 330-39.
VIEWPOINT
A
right to reproduce?
Homo
sapiens has largely escaped from the controls of that limit the population of all other species. Had ecology H sapiens remained a hunter/gatherer, these controls might have worked, but conscious manipulation of the environment brought total release. After a mere ten thousand years or so, human beings have brought the planetary system close to collapse, and all the debate in the world seems unlikely to save it. A combination of breeding and rapacity, both uncontrolled, is propelling us down the slippery slope first envisioned by Malthus, dragging the rest of the planet along too. "Neither an expanding population nor ever-expanding aspirations are tolerable and the two in harness will be fatal."1 Since escape from ecological control has been via consciously directed action, reimposition of controls on breeding would likewise have to be conscious, and most probably voluntary. The major part of humanity is not yet convinced of this necessity; in China, a national policy of family limitation is said to be failing, and even in a totalitarian state it has not been possible to impose it by force. According to the United Nations Universal Declaration of Human Rights (1948, articles 16[I] and 16[III)), "Men and women of full age without any limitation due to race, nationality or religion have the right to marry and to found a family ... the family is the natural and fundamental group unit of society". Is this to be understood as an unfettered right subject only to such controls as each human pair chooses to impose upon itself, for purely individual reasons, or should and can there be other controls? I have argued elsewhere that a rhetoric of rights with no balancing duties or responsibilities is wrong in health care, and I think it is even more wrong in this context. Man’s escape from the controls that bind other animals has been expressed as a mind-set of dominance and exploitation over
the rest of creation, the ecosystem, or whatever term may be applied. Human beings are reluctant to admit that they are only one part of a system in which every part is interdependent. The language of dominance and exploitation is seldom balanced by that of duty and responsibility to the exploited system. Homo is not truly sapiens, the wise, but manipulans, the meddler. Uncontrolled manipulation of the environment has shown itself to be a recipe for disaster, and uncontrolled reproduction has exponentially increased its effects. This rhetoric of rights is meaningless madness unless it is balanced by a language of duty and responsibility. Homo has no ecologically supported right to existence as a species. A right to reproduction can be justified only by its responsible use. And such a right must be balanced by a duty-that the resulting population mass of one species (ours) does not unbalance, damage, or destroy the system upon which it depends. At the very least, this means that ideas of dominion and subjection must be modified by those of interdependence, understanding, and respect. Within a few generations, there is a very strong possibility that "Homo manipulans" will irretrievably wreck the planetary ecosystem by the combination of individual rapacity and numbers of the species, bringing disaster not only on itself as a species but also on the system as a whole. I believe the only solution is voluntary reimposition of the controls from which evolution freed us. We must consciously and deliberately change our mind-set, and totally reorientate our viewpoint towards the system that dominates us. Consciousness enabled us to escape from controls; consciousness must be the means by which we reimpose them. It is long past the time that we discarded every statement of right that is unmatched by equivalent duty and
responsibility. REFERENCES
ADDRESS:
Hospital,
of Pathology, Royal University Saskatoon, Saskatchewan, Canada S7N 0X0 (Prof
Department
H. E. Emson, FRCPC).
1. Editorial. Horse manure after Rio. Lancet 1992; 339: 1515-16. 2. Emson HE. Rights, duties and responsibilities in health care. Philos 1992; 9: 3-10.
J Appl
CLINICAL PRACTICE
Prospective controlled study of four infectioncontrol procedures to prevent nosocomial infection with respiratory syncytial virus
To determine the most effective infection control
procedure in preventing nosocomial infection with respiratory syncytial virus (RSV), we did a prospective controlled study of four infectioncontrol strategies in four wards in a large paediatric hospital in the west of Scotland. All children under two years old admitted to four general wards during three winter RSV epidemics (1989-92) were screened for RSV infection (by nasopharyngeal aspirate and direct immunofluorescence) within 18 hours of admission. The main outcome measure was the occurrence of nosocomial infection, defined as the number of children initially RSV negative who became RSV positive 7 days or more after hospital admission (incubation period for RSV infection is
5-8 days). Without special precautions, there was a high rate of nosocomial RSV infection (26%). Nosocomial infection was significantly reduced by the combination of cohort nursing with the wearing of gowns and gloves for all contacts of RSV-infected children (p=0·0022). Neither the use of gowns and
gloves alone nor cohort nursing alone produced a significant reduction in cross-infection. In the final year, general clinical use of a policy of cohort nursing with gowns and gloves resulted in a reduction in the cross-infection rate by two-thirds of its original value (9·5% vs 26%). Combined with rapid laboratory diagnosis, cohort nursing and the wearing of gowns and gloves for all contacts
with
significantly
RSV-infected
children
can
reduce the risk of nosocomial RSV
infection. Lancet 1992; 340: 1079-83.
Introduction In 1971, Ditchburn
et
al identified the hazards for
uninfected children and personnel in hospitals from infants with respiratory syncytial virus (RSV) infections. Nosocomial RSV infection is a serious problem, especially for infants with cardiorespiratory diseases such as congenital heart disease, or immunodeficiency states.2,3 The virus is
spread by close
contact
with infected secretions,
by large-
and by fomites.4-6 Small-particle aerosols be a major route of transmission.6 Hospital personnel are thought to be the main vector, transmitting the virus on contaminated hands and even becoming infected themselves by self-inoculation of mucous
particle aerosols, do
not seem to
membranes.7 Since the mid-1970s a few studies have assessed ways to limit nosocomial spread of RSV among children in hospital.8--12 Since RSV is a lipid-enveloped virus, simple handwashing with soap and water should prevent virus transmission. However, staff do not always wash their hands after every patient contact." Consequently, the main strategy has been to follow procedures that protect staff from being contaminated by infected secretions. Various
strategies include wearing gowns and gloves,lO cohorting RSV-infected patients,l1 and cohorting and handwashing.12 When special care was taken to ensure that gowns and gloves were worn, there was a substantial reduction in nosocomial infection.10 Consideration has also been given to the protective value of face-masks and eye-nose goggles.9,14,15 Masks did not lead to a measurable benefitin terms of infection control. The use of eye-nose goggles was associated with a significant decrease in nosocomial infection, but they are not popular with clinical staff and are frightening to children. Because of differing approaches, which particular infection-control strategy most effectively prevents nosocomial RSV infection remains unclear. Comparisons of previous published studies are difficult because of methodological differences, which include screening criteria, ward design, variations in laboratory diagnostic technique, and ways of assessing cross-infection rates. To address these limitations we carried out a prospective controlled study of four infection-control protocols on children admitted to four wards during two winter epidemics. Our aim was to define the most effective infection-control procedure for the prevention of nosocomial infection on general paediatric wards with limited isolation facilities. In the third year, the most effective strategy was introduced into all ward areas and its efficacy in clinical practice was assessed. ADDRESSES: Departments of Child Health (P Madge, RGN, J. Y Paton, MD) and Statistics (J. H. McColl, MSc), University of Glasgow, and Department of Microbiology (P. L. K. Mackie, PhD), Royal Hospital for Sick Children, Glasgow, UK. Correspondence to Dr J. Y Paton, Department of Child Health, Royal Hospital for Sick Children, Yorkhill, Glasgow G3 8SJ, UK.
1080
Subjects and
methods
TABLE I-INFECTION-CONTROL MEASURES USED IN EACH WARD
Prospective study The prospective study, approved by the ethics committee of the Royal Hospital for Sick Children, Glasgow, was conducted in four medical wards of that hospital during two winter RSV epidemics (1989/90 and 1990/91). Each ward was identical in layout with 8 single cubicles and 3 larger areas of 4, 6, and 6 beds. Each ward admitted new children once every 4 days. When a ward was full, the next ward in rotation took over receiving duties. There was no
separate ward area for managing children with infections.
CN cohort nursing; Go/Gtgowns and gloves for all contact with RSV-positive children; Go/GI, CN =cohort nursing plus gowns and gloves for all contact with
In all 3 years, the study started when 5 infants with symptoms of bronchiolitis were admitted to hospital within 1 week and were shown to have RSV infection. The study continued for 3 months thereafter. Every child less than two years old, irrespective of
RSV-positive children *Basic infection control measures adopted m all wards-ie, hand washmg before and after any patient contact, and wearing of gowns and gloves when dealing with any body fluids tWard 2 only took part in winters 2 and 3
clinical presentation, had respiratory secretions tested for RSV antigen within 18 hours of admission. All those found to be infected with RSV (RSV positive) were tested daily until they were negative, or were discharged home. Children who remained in hospital for more than 7 days, including those who were initially RSV negative, were tested weekly or when signs of respiratory illness developed. Nosocomial infection was assumed ifa child, initially RSV negative, became RSV positive 7 days or more after admission (the incubation period for RSV infection is 5-8 days). We did not study children after discharge from hospital; however, no child was readmitted with RSV infection during the remainder of that winter epidemic.
Virus screening and detection RSV antigen
sought by direct immunofluorescence with pooled RSV-specific monoclonal antibodies (Dako Ltd, High Wycombe, UK) to detect infected exfoliated cells in respiratory secretions obtained by nasopharyngeal aspiration. Before RSV results were available, newly admitted children with one or more of the following clinical features were regarded as RSV positive: upper respiratory tract infection, breathlessness or reluctance to feed, cough or apnoea, and signs of respiratory distress such as tachypnoea. They were nursed according to the individual ward’s policy for RSV-infected patients until RSV status was known. No child initially placed in an infected cohort but shown subsequently to be RSV negative became RSV positive. We initially planned that children at high risk from severe nosocomial RSV infection (eg, with bronchopulmonary dysplasia, cystic fibrosis) should be admitted and nursed in single-bed cubicles; owing to other demands for single rooms this rarely happened. Children with was
cardiac disease were nursed in a separate cardiac ward and were not part of the study population. Whenever possible nurses who
developed a respiratory illness were encouraged to care for babies in the RSV area, but no formal attempt was made to diagnose RSV infection in the staff. While they were RSV positive, children were managed according to an infection-control policy assigned to that ward (table i). Two consecutive negative laboratory results indicated that the patient was a low infection risk and could therefore be nursed with routine procedures outwith the infected cohort. These measures were followed by all staff dealing with RSV-infected children. A research nurse (P. M.) supervised the implementation of the ward policies. However, no special steps were taken to supervise compliance with the measures in use in any ward. In year 3, the policy of cohort nursing with the use of gowns and gloves for all patient contacts was adopted in all four medical wards. Statistics Three statistical methods were used. First, the proportions of infants becoming cross-infected in the four groups were compared by a X2 test of homogeneity.16 The null hypothesis was that the probability of cross-infection was the same under all four conditions. Second, since a X2 test does not exploit relations among the four combinations of infection-control procedures, we used the more appropriate linear logistic regression analysis. 16 The two treatments
(gowns and gloves, cohorting) each appeared at two levels (absent, present). Letting 0 (i, j) denote the probability of cross-infection at the i’th level of gown and gloves (i 0 [absent], 1 [present]) and the j’th level of cohorting (j =0 [absent], 1 [present]) the log-odds on =
RSV cross-infection was modelled as: In
e(ij)
.
...
Oct ——,-o7-:-=-ca-P)-yi) [h Yll
1-0
(’.))
is the reduction (if any) in the log-odds due to gowns and gloves alone,0 the reduction (if any) due to cohorting alone, and y the further reduction (if any) when both procedures were combined-ie, y is an interaction term. Unlike the global X2 test, logistic regression allows direct testing of each of these effects. Forward and backward stepwise selection was used. Third, logistic regression does not take account of the level of risk experienced by individual patients. Individual stays for infants developing nosocomial infection varied from 7 to 63 days. In addition, children admitted to hospital towards the beginning or end of the epidemic were at lower risk than children who were in hospital for the same length of time at the peak of the epidemic when many RSV-infected children were in the wards. This can be considered as a survival analysis problem16,17 in which the interest is in modelling the probability of "survival" (ie, non-infection) as a declining function of time in hospital (the survivor function). In fact, rather than simply using calendar time, we decided to weight each day of an individual’s stay to reflect the fluctuating level of risk in the ward. The weights used were the numbers of RSV-positive infants in the ward each day. For non-infected children, then, "time" of exposure was taken as the sum of these weights over all the days up to and including the day before RSV infection was diagnosed. We also compared the survivor functions across the four infection-control groups with Cox’s proportional hazards model. As in the logistic regression model, covariates were included to denote the individual and joint effects of the two isolation where
a
procedures. All statistical analyses were performed with the BMDP computer
package.
Results Table n shows the number of RSV-positive infants admitted during each winter epidemic. Over the first two winter epidemics the number of RSV-positive children admitted to each study group was broadly similar and thus the burden of infection on each ward was similar. At the end of winter 1 (table Hi), there was a high cross-infection rate in children who were in hospital for more than 7 days (26%) who were given "no special precautions" (handwashing after all contacts and the wearing of gowns and gloves for contact with body fluids, table l). By contrast, the most stringent policy of gloves and a gown for all patient contacts combined with cohort nursing yielded a substantial reduction in infection (cross-infection rate 5%). However, the number of cross-infections was small, and the reduction achieved was not statistically significant. After considering the results, the policy of "no
1081
TABLE II-NUMBERS OF RSV INFANTS IN EACH
INFECTION-CONTROL GROUP
TABLE III-NUMBER OF NOSOCOMIAL INFECTIONS IN RELATION TO INFECTION-CONTROL PROCEDURE USED
*(Mean %, 95% CI).
special precautions" was discontinued. Instead, a gown and glove group was introduced and the other two policies continued. By the end of the second winter, the picture was clearer (table ill), confirming the results of the first year’s study. Over the two years, there was a significant reduction in cross-infection from using gloves and gown combined with cohort nursing (=8-1, p = 0044). Logistic regression showed that the data were consistent with a model in which (x = &bgr; 0 and only the interaction term y was =
The same model resulted from both forward and backward stepping; the p value for retaining y in the final model was 0-0022. This suggested that neither a policy of gown and gloves alone nor cohorting alone produced a significant reduction in cross-infection, but the combination of the two significantly reduced it. non-zero.
Survivor function for Cox proportional hazards model. For the model used, only the interaction term was significant. Neither of the policies alone had an effect on the rate of cross-infection.
TABLE IV-NUMBER OF NOSOCOMIAL INFECTIONS IN EACH WARD IN WINTER 3 WITH ALL WARDS USING GOWN AND GLOVES PLUS COHORT NURSING
In this model, exp (-y) measures the proportionate reduction in the odds on infection due to adopting both infection control procedures. y was estimated to be 2-32 (95% CI for exp (-y) 0-0127, 0-756). Thus the odds on cross-infection was likely to be reduced to between 1-27% and 75-6% of its previous level. In addition, it was possible to calculate an approximate 95% CI for e (1,1), which suggested that, when both procedures were in use, the true cross-infection rate was likely to be between 043% and 18-6%. The survival analysis confirmed the findings of the logistic regression. On fitting a proportional hazards model in a stepwise manner, both forward and backward stepping resulted in a model in which the two procedures individually were found to have no effect on the survivor function, but in combination reduced it significantly (p = 0 0057, figure). In the final year, the combined policy of cohort nursing with the wearing of gowns and gloves was adopted in all wards. By chance, that winter’s epidemic (1991-92) was particularly severe. Table IV shows that there was a significant reduction in cross-infection compared with the previous years’ data (table III), with the cross-infection rate being in the middle of the confidence interval obtained from the logistic regression. This represented a reduction in the cross-infection rate of approximately 16 percentage points (or two-thirds of its original value; table iv).
Discussion In the first year, the rate of nosocomial RSV infection using the simplest infection-control measures was high. Isaacs et al12 showed a similar cross-infection rate in children with congenital cardiac disease who had median hospital stays in excess of 30 days. These results emphasise that nosocomial RSV infection is an important hazard for uninfected children in hospital during an RSV epidemic. Measures to prevent such high rates of cross-infection are
clearly important. The use of rapid diagnostic virological techniques to screen children under two years was critical to this study. Without rapid and accurate ascertainment of which children are infected, any infection-control strategy is likely to be ineffective. In previous studies, screening of newly admitted children has often been confined to those with respiratory symptoms (eg, ref 12). However, this approach will miss children who present with atypical non-respiratory disease. We followed the approach of Krasinski et al," and screened all children under two irrespective of clinical diagnosis or presentation. We found that rapid diagnosis combined with an infection-control policy of cohort nursing and the wearing of gown and gloves for all contacts with RSVinfected children achieved a significant reduction in the risk of nosocomial RSV infection. From the logistic regression analysis, cohort nursing or gowns and gloves had no effect when used alone but in combination interacted to produce a
1082
substantial reduction in nosocomial infection. Although the of cross-infection using this combined strategy appeared higher in the third year, it was in the middle of the confidence interval predicted by the logistic regression model from the first two years’ data. Since our study was prospective and controlled it eliminated many of the problems that have made interpretation of previous studies difficult. Furthermore, the results from the final year, when the most effective policy was introduced into clinical practice, confirmed that the approach leads to an important reduction in cross-infection with RSV. One confounding effect that was not accounted for in the study design was a possible "ward effect". For practical reasons, two wards (3 and 4) continued with the same policy over the first two years of the study. Since it was also necessary to apply policies to whole wards, there is a possibility that ward 4 might have been especially effective at implementing their assigned policy. However, in the final year, the policy of cohort nursing plus gowns and gloves for all contacts with RSV-positive children (as in ward 4) was introduced into all wards. There was a noticeable reduction in nosocomial infection in the three wards where it had not previously been applied. That the resulting rates of nosocomial infection were very similar in all wards, and in the middle of the confidence interval from the logistic regression model, suggests that the observed improvement was due to the intervention rather than to any ward being especially good or bad at implementing their assigned policy. More formally, the logistic regression analysis was repeated with data from all 3 years. Again, only the original interaction term, y (the further reduction, if any, when both procedures were used in combination), was entered in the final model by the stepwise procedure. Neither a time effect nor ward effect was found to be significant. Unlike three earlier studies, we did not find that simpler measures (gowns and gloves alone,lO handwashing with cohorting," screening for RSV with cohorting12) significantly reduced the risks of cross-infection compared with routine measures of handwashing with the wearing of gowns and gloves for all contact with body fluids. However, the numbers of nosocomial infections in these two groups are quite small and the confidence intervals wide. There may be a small effect from cohort nursing, but the weight of evidence from this study is not sufficient to confirm a positive effect from these intermediate procedures. Since nosocomial infection can be associated with significant morbidity and mortality we felt it appropriate to adopt the measure that was clearly effective. One important difference that might provide an explanation for the failure of these simpler measures, found effective in earlier studies 12," relates to steps taken to improve compliance with the infection-control policy used. It is well known that compliance with handwashing is poor." Isaacs et al12 distributed a written handout on the importance of handwashing to try and improve it. Le Clair et apo showed that compliance with measures, especially with wearing gloves, was also poor, being followed in only 39% of contacts with infected children. After a period of education and encouragement this proportion rose to 81 %. It was only after compliance improved that a significant reduction in cross-infection was achieved. We did not include any measures to improve compliance with the infection-control methods used. Thus failures of compliance in our study may have accounted for the failure of the simpler measures to prevent cross-infection. rate
In respect of cohort nursing, the degree of physical separation of the infected and unifected cohort may also be important. In Krasinski’s studyll of rapid screening and cohorting, the infected and uninfected cohorts were separated by 100 feet (34 m) of intervening administrative offices. Our wards consist of a small number of single rooms and 3 contiguous small 4 or 6 bed bays. Consequently, such a high degree of physical separation cannot be achieved. Why, in our study, is there evidence of an interaction between the policies of cohorting and gowns and gloves, with an effect greater than achieved by either measure alone? We noticed that placing children in a cohort and wearing gowns and gloves generated considerable peer, and parental, "pressure" to ensure that the measures were followed. We suspect that this pressure produced a situation in which compliance in the combined group effectively became "self-policing". We speculate that a resultant improvement in compliance led to the significant reduction in crossinfection we observed during the prospective study in winters 1 and 2 in the gowns and glove plus cohort nursing group (table ill). The very slight increase in rate in the third year when this policy was adopted in all wards may, in part, reflect a breakdown in compliance under the pressure of an especially large epidemic. This may have important implications for the prevention of cross-infection when the modes of cross-infection are similar to those for RSV. Although simple procedures should prevent cross-infection, enthusiasm is likely to vary among individuals and wane with time. Education and persuasion do not generally lead to sustained improvement in handwashing.18 Over the longer term, compliance with infection-control procedures might, therefore, best be maintained by attempting to create self-policing, and hence self-enforcing, environments. In this study the rate of nosocomial RSV infection using the simplest infection-control measures was unexpectedly high. Since nosocomial RSV infection can be associated with significant morbidity, paediatricians need to keep this rate under review. In this context, we wonder whether cross-infection rates for RSV and, perhaps, other common viral infections, might provide useful outcome measures for auditing the "quality" of hospital care in paediatric units. In our experience, at the first signs of an RSV epidemic, the introduction of rapid diagnostic screening and a policy of cohort isolation and the wearing of gown and gloves for all patient contacts can significantly reduce the incidence of nosocomial RSV infection. The work was supported by Greater Glasgow Health Board Research Support Group Grant RSG/END/89/90F and RSG/END/90/91/M. Weare particularly grateful for the willing cooperation of the nurses, medical staff, and parents in the mdeical wards of the Royal Hospital for Sick Children, Glasgow, throughout the study.
REFERENCES 1. Ditchburn
RK, McQuillin J, Gardner PS, Court SDM. Respiratory syncytial virus in hospital cross-infection. BMJ 1971; 3: 671-73. 2. MacDonald NE, Hall CB, Suffin SC, Alexson C, Harris J, Manning JA. Respiratory syncytial virus infection in infants with congenital heart disease. N Engl J Med 1982; 307: 397-400. 3. Hall CB, Powell KR, MacDonald NE, et al. Respiratory syncytial virus infection in children with compromised immune function. N EnglJ Med 1986; 315: 4. Hall CB, Douglas virus. J Pediatr 5. Hall CB, Douglas
77-81. RG Jr. Modes of transmission of respiratory syncytial 1981; 99: 100-03. RG, Geiman JM. Possible transmission by fomites of respiratory syncytial virus. J Infect Dis 1980; 141: 98-102. 6. Hall CB, Douglas RG, Schnabel KC, Geiman JM. Infectivity of respiratory syncytial virus by various routes of inoculation. Infect Immun 1981; 33: 779-83.
1083
Chapin M. Infection rate in personnel caring for children with RSV infections. Am J Dis Child 1987; 141:
7. Agah R, Cherry JD, Garakian AJ,
695-97. 8. Hall CB, Geiman
JM, Douglas RG Jr, Meagher MP. Control of nosocomial respiratory syncytial viral infections. Pediatrics 1978; 62:
728-32. 9. Gala CL, Hall CB, Schnabel KC, et al. The use of eye-nose goggles to control nosocomial respiratory syncytial virus infection. JAMA 1986; 256: 2706-08. 10. Le Clair JM, Freeman J, Sullivan BF, Crowley CM, Goldman DA. Prevention of nosocomial respiratory syncytial viral infections through compliance with glove and gown isolation precautions. N Engl J Med 1987; 317: 329-34. 11. Krasinski K, LaCouture R, Holzman RS, Waithe E, Bonk S, Hanna B. Screening for respiratory syncytial virus and assignment to a cohort at admission to reduce nosocomial transmission. J Pediatr 1990; 116: 894-98. 12. Isaacs D, Dickson H, O’Callaghan C, Sheaves R, Winter A, Moxon ER.
and cohorting in prevention of hospital acquired infections with respiratory syncytial virus. Arch Dis Child 1991; 66: 227-31. 13. Goldman D, Larson E. Handwashing and nosocomial infections. N Engl J Med 1992; 327: 120-22. 14. Murphy D, Todd JK, Chao RK, Orr I, McIntosh K. The use of gowns and masks to control respiratory illness in pediatric personnel. J Pediatr 1981; 99: 746-50. 15. Hall CB, Douglas RG Jr. Nosocomial respiratory syncytial viral infections. Should gowns and masks be used? Am J Dis Child 1981; 135:
Handwashing
512-15. 16. Everitt BS. Statistical methods for medical investigations. New York: Oxford University Press, 1989. 17. Cox DR, Oakes D. Analysis of survival data. London: Chapman Hall, 1984. 18. Conly JM, Hill S, Ross J, Lertzman J, Louie TJ. Handwashing practices in an intensive care unit: the effects of an educational program and its relationship to infection rates. Am J Infect Control 1983; 17: 330-39.
VIEWPOINT
A
right to reproduce?
Homo
sapiens has largely escaped from the controls of that limit the population of all other species. Had ecology H sapiens remained a hunter/gatherer, these controls might have worked, but conscious manipulation of the environment brought total release. After a mere ten thousand years or so, human beings have brought the planetary system close to collapse, and all the debate in the world seems unlikely to save it. A combination of breeding and rapacity, both uncontrolled, is propelling us down the slippery slope first envisioned by Malthus, dragging the rest of the planet along too. "Neither an expanding population nor ever-expanding aspirations are tolerable and the two in harness will be fatal."1 Since escape from ecological control has been via consciously directed action, reimposition of controls on breeding would likewise have to be conscious, and most probably voluntary. The major part of humanity is not yet convinced of this necessity; in China, a national policy of family limitation is said to be failing, and even in a totalitarian state it has not been possible to impose it by force. According to the United Nations Universal Declaration of Human Rights (1948, articles 16[I] and 16[III)), "Men and women of full age without any limitation due to race, nationality or religion have the right to marry and to found a family ... the family is the natural and fundamental group unit of society". Is this to be understood as an unfettered right subject only to such controls as each human pair chooses to impose upon itself, for purely individual reasons, or should and can there be other controls? I have argued elsewhere that a rhetoric of rights with no balancing duties or responsibilities is wrong in health care, and I think it is even more wrong in this context. Man’s escape from the controls that bind other animals has been expressed as a mind-set of dominance and exploitation over
the rest of creation, the ecosystem, or whatever term may be applied. Human beings are reluctant to admit that they are only one part of a system in which every part is interdependent. The language of dominance and exploitation is seldom balanced by that of duty and responsibility to the exploited system. Homo is not truly sapiens, the wise, but manipulans, the meddler. Uncontrolled manipulation of the environment has shown itself to be a recipe for disaster, and uncontrolled reproduction has exponentially increased its effects. This rhetoric of rights is meaningless madness unless it is balanced by a language of duty and responsibility. Homo has no ecologically supported right to existence as a species. A right to reproduction can be justified only by its responsible use. And such a right must be balanced by a duty-that the resulting population mass of one species (ours) does not unbalance, damage, or destroy the system upon which it depends. At the very least, this means that ideas of dominion and subjection must be modified by those of interdependence, understanding, and respect. Within a few generations, there is a very strong possibility that "Homo manipulans" will irretrievably wreck the planetary ecosystem by the combination of individual rapacity and numbers of the species, bringing disaster not only on itself as a species but also on the system as a whole. I believe the only solution is voluntary reimposition of the controls from which evolution freed us. We must consciously and deliberately change our mind-set, and totally reorientate our viewpoint towards the system that dominates us. Consciousness enabled us to escape from controls; consciousness must be the means by which we reimpose them. It is long past the time that we discarded every statement of right that is unmatched by equivalent duty and
responsibility. REFERENCES
ADDRESS:
Hospital,
of Pathology, Royal University Saskatoon, Saskatchewan, Canada S7N 0X0 (Prof
Department
H. E. Emson, FRCPC).
1. Editorial. Horse manure after Rio. Lancet 1992; 339: 1515-16. 2. Emson HE. Rights, duties and responsibilities in health care. Philos 1992; 9: 3-10.
J Appl
