- Tom Jefferson, coordinator1,
- Ruth Foxlee, trials search coordinator2,
- Chris Del Mar, dean3,
- Liz Dooley, review group coordinator4,
- Eliana Ferroni, researcher5,
- Bill Hewak, medical student3,
- Adi Prabhala, medical student3,
- Sree Nair, professor of biostatistics6,
- Alex Rivetti, trials search coordinator1
- 1Cochrane Vaccines Field, Alessandria, Italy
- 2Cochrane Wounds Group, Department of Health Sciences, University of York
- 3Faculty of Health Sciences and Medicine, Bond University, Gold Coast, 4229, Qld, Australia
- 4Cochrane Acute Respiratory Infections Group, Faculty of Health Sciences and Medicine, Bond University
- 5Public Health Agency of Lazio Region, Rome
- 6Department of Statistics, Manipal Academy of Higher Education, Manipal, India
- Correspondence to: C Del Mar
- Accepted 23 October 2007
Objective To systematically review evidence for the effectiveness of physical interventions to interrupt or reduce the spread of respiratory viruses.
Data extraction Search strategy of the Cochrane Library, Medline, OldMedline, Embase, and CINAHL, without language restriction, for any intervention to prevent transmission of respiratory viruses (isolation, quarantine, social distancing, barriers, personal protection, and hygiene). Study designs were randomised trials, cohort studies, case-control studies, and controlled before and after studies.
Data synthesis Of 2300 titles scanned 138 full papers were retrieved, including 49 papers of 51 studies. Study quality was poor for the three randomised controlled trials and most of the cluster randomised controlled trials; the observational studies were of mixed quality. Heterogeneity precluded meta-analysis of most data except that from six case-control studies. The highest quality cluster randomised trials suggest that the spread of respiratory viruses into the community can be prevented by intervening with hygienic measures aimed at younger children. Meta-analysis of six case-control studies suggests that physical measures are highly effective in preventing the spread of SARS: handwashing more than 10 times daily (odds ratio 0.45, 95% confidence interval 0.36 to 0.57; number needed to treat=4, 95% confidence interval 3.65 to 5.52); wearing masks (0.32, 0.25 to 0.40; NNT=6, 4.54 to 8.03); wearing N95 masks (0.09, 0.03 to 0.30; NNT=3, 2.37 to 4.06); wearing gloves (0.43, 0.29 to 0.65; NNT=5, 4.15 to 15.41); wearing gowns (0.23, 0.14 to 0.37; NNT=5, 3.37 to 7.12); and handwashing, masks, gloves, and gowns combined (0.09, 0.02 to 0.35; NNT=3, 2.66 to 4.97). The incremental effect of adding virucidals or antiseptics to normal handwashing to decrease the spread of respiratory disease remains uncertain. The lack of proper evaluation of global measures such as screening at entry ports and social distancing prevent firm conclusions being drawn.
Conclusion Routine long term implementation of some physical measures to interrupt or reduce the spread of respiratory viruses might be difficult but many simple and low cost interventions could be useful in reducing the spread.
Although respiratory viruses usually cause minor disease, epidemics can occur. Mathematical models estimate that about 36 000 deaths and 226 000 admissions to hospital in the United States annually are attributable to influenza,1 and with incidence rates as high as 50% during major epidemics worldwide, respiratory viruses strain health services,2 are responsible for excess deaths,2 3 and result in massive indirect costs owing to absenteeism from work and school.4 Concern is now increasing about serious pandemic viral infections. In 2003 an epidemic of the previously unknown severe acute respiratory syndrome (SARS) caused by a coronavirus affected about 8000 people worldwide, with 780 deaths (disproportionately high numbers were in healthcare workers), and causing a social and economic crisis, especially in Asia.5 A new avian influenza pandemic caused by the H5N1 virus strain threatens greater catastrophe.6
High viral load and high viral infectiousness probably drive virus pandemics,7 hence the need for interventions to reduce viral load. Mounting evidence suggests, however, that single measures, particularly the use of vaccines or antivirals, will be insufficient to interrupt the spread of influenza. Agent specific drugs are also not available for other viruses.78910
A recent trial found handwashing to be effective in lowering the incidence of pneumonia in the developing world.w1 Clear evidence has also shown a link between personal (and environmental) hygiene and infection.11 We systematically reviewed the evidence for the effectiveness of combined public health measures such as personal hygiene, distancing, and barriers to interrupt or reduce the spread of respiratory viruses.12 13 We did not include vaccines and antivirals because these have been reviewed.4 10 1415161718
We considered trials (individual level, cluster randomised, or quasirandomised), observational studies (cohort and case-control), and any other comparative design in people of all ages provided some attempt had been made to control for confounding.
We included any intervention to prevent the transmission of respiratory viruses from animals to humans or from humans to humans (isolation, quarantine, social distancing, barriers, personal protection, and hygiene) compared with no intervention or with another intervention. We excluded vaccines and antivirals.
The outcome measures were deaths; numbers of cases of viral illness; severity of viral illness, or proxies for these; and other measures of burden, such as admissions to hospital.
We searched the Cochrane Central Register of Controlled Trials (Cochrane Library issue 4, 2006), Medline (1966 to November 2006), OldMedline (1950-65), Embase (1990 to November 2006), and CINAHL (1982 to November 2006). See bmj.com for details of our search terms for Medline and the Cochrane register (modified for OldMedline, Embase, and CINAHL). We applied no language restrictions. Study design filters included trials; cohort, case-control, and cross-over studies; and before and after and time series. We scanned the references of included studies to identify other potentially relevant studies.
We scanned the titles and abstracts of potentially relevant studies: when studies seemed to meet our eligibility criteria (or when information was insufficient to exclude them), we obtained the full text articles. We used a standardised form to assess the eligibility of each study, on the basis of the full article.
We analysed randomised and non-randomised studies separately. Randomised studies were assessed according to the effectiveness of the randomisation method, the generation of the allocation sequence, allocation concealment, blinding, and follow-up. Non-randomised studies were assessed for the presence of potential confounders using the appropriate Newcastle-Ottawa Scales19 for case-control and cohort studies, and a three point checklist was used for controlled before and after studies.20
Using quality at the analysis stage as a means of interpretation of the results we assigned risk of bias categories on the basis of the number of items judged inadequate in each study: low risk of bias, up to one inadequate item; medium risk of bias, up to three inadequate items; and high risk of bias, more than three inadequate items.
Two authors (TJ, CDM) independently applied inclusion criteria to all identified and retrieved articles. Four authors (TJ, EF, BH, AP) extracted data from included studies and checked their accuracy on standard field forms used by Cochrane groups for vaccines, supervised and arbitrated by CDM.
Aggregation of data depended on study design; types of comparisons; sensitivity; and homogeneity of definitions of exposure, populations, and outcomes used. We calculated the statistic I2 for each pooled estimate to assess the impact on heterogeneity.21 22
When possible we did a quantitative analysis and summarised effectiveness as an odds ratio with 95% confidence intervals, expressing absolute intervention effectiveness when significant as a percentage using the formula: intervention effectiveness=1−odds ratio. For studies that could not be pooled we used effect measures reported by the authors (such as relative risk or incidence rate ratio, with 95% confidence intervals or, when not available, relevant P values). We calculated numbers needed to treat (NNT) using the formula 1/absolute risk reduction whenever we thought the data were robust enough to allow it.
Overall, 2300 titles of reports of potentially relevant studies were identified and screened. In total, 2162 were excluded and 138 full papers retrieved, totalling 49 reports of 51 studies (fig 1⇓).
The quality of the methods of included studiesw1-w51 varied (tables 1-5⇓⇓⇓⇓⇓). Considerable loss of information resulted from incomplete or no reporting of randomisation,w3 blinding,w5 numerators and denominators,w4 w6 interventions, outcomes,w39 attrition of participants,w34 confidence intervals,w33 and cluster coefficients in the relevant trials.w4 The impact of potential biases (such as cash incentives given to participantsw39) were not discussed. Some authors confused the cohort design with a before and after design, which provided conclusions unsupported by the data.w34 The quality of methods was sometimes eroded by the need to deliver behavioural interventions in the midst of service delivery.w37 Even when suboptimal designs were selected, authors rarely articulated potential confounders. A common confounder specific to this area is the huge variability in viral incidence over time, commonly ignored.w19 w41 Sometimes this was tackled in the study design,w30 even in controlled before and after studies (one attempted correlation between admissions for respiratory syncytial virus and respiratory syncytial virus circulating in the communityw21; another attempted linking exposure—measured as nasal excretion—and infection rate in the periods before and after interventionw14). Inadequate blinding or adjustment for confounders is a well known factor in exaggerating the effects of an intervention.23
Inappropriate interventions for comparison caused problems with study designs: in some studies these probably carried sufficient effect to dilute the intervention outcomew9; in two studies blinding may have failed because placebo handkerchiefs were impregnated with a dummy compound that stung the users’ nostrils.w5
Some interventions were tested under impractical and unrealistic situations: participants allocated to the intervention hand cleaner (organic acids) were not allowed to use their hands between cleaning and challenge with virus, so the effect of normal use of the hands on the intervention remains unknown,w3 and 2% aqueous iodine is a successful antiviral intervention when painted on the hands but it stains and is impractical for all but the highest risk of epidemic contagion.w51
Compliance with interventions—especially educational programmes—was problematic for several studies, despite the importance of many such low cost interventions.
The most impressive effects came from high quality cluster randomised trials in preventing the spread of respiratory virus into the community using hygienic measures aimed at younger children. One study reported a significant decrease in respiratory illness in children up to age 24 months (relative risk 0.90, 95% confidence interval 0.83 to 0.97), although the decrease was not significant in older children (0.95, 0.89 to 1.01).w11 Another study reported a 50% (95% confidence interval 65% to 34%) lower incidence of pneumonia in children aged less than 5 years in a developing country.w1 Additional benefit from reduced transmission to other household members is broadly supported by the results of other study designs although the potential for confounding is greater.
Six case-control studies assessed the impact of public health measures to curb the spread of the SARS epidemic in China, Singapore, and Vietnam in 2003. Homogeneity of case definition, agent, settings, and outcomes made meta-analysis possible, using a fixed effects model because no comparisons showed significant heterogeneity (fig 2⇓ and table 6⇓). Only binary data were pooled despite the availability of continuous data because the variables differed or were measured in different units, and standard deviations were usually missing. The data suggest that implementing barriers to transmission, isolation, and hygienic measures are effective and relatively cheap interventions to contain epidemics of respiratory viruses, such as SARS, with estimates of effect ranging from 55% to 91%: washing hands more than 10 times daily (odds ratio 0.45, 95% confidence interval 0.36 to 0.57, NNT=4, 95% confidence interval 3.65 to 5.52); wearing masks (0.32, 0.25 to 0.40, NNT=6, 4.54 to 8.03); wearing N95 masks (0.09, 0.03 to 0.30, NNT=3, 2.37 to 4.06); wearing gloves (0.43, 0.29 to 0.65, NNT=5, 4.15 to 15.41); wearing gowns (0.23, 0.14 to 0.37, NNT=5, 3.37 to 7.12); and handwashing, masks, gloves, and gowns combined (0.09, 0.02 to 0.35, NNT=3, 2.66 to 4.97). All studies selected hospital cases, except onew45 in which the cases were people with probable SARS reported to the Department of Health in the territory of Hong Kong up to 16 May 2003. Evidence was limited for the superior effectiveness of barrier devices to droplets such as the N95 masks (respirators with 95% filtration capability against non-oily particulate aerosolsw48) over simple surgical masks. An incremental effect was found for decreased burden of respiratory disease by adding virucidals or antiseptics to normal handwashing in atypical settings, but the extra benefit may have been, at least partly, from confounding additional routines.
Studies on interventions to prevent the transmission of respiratory syncytical virus and similar viruses in more typical settings suggested good effectiveness, although doubt was cast on the findings because of method quality inherent in controlled before and after studies, especially different virus infection rates.
Few studies reported on resource consumption for the physical intervention evaluated. One case-control studyw45 concluded that handwashing needs to be carried out more than 10 times daily to be effective. One study,w25 in a military training setting, reported a need to wash hands more than four times daily. During one month of the respiratory syncytical virus “season” on a ward containing 22 cribs, one study reported that 5350 gowns and 4850 masks were used.w18
Proper evaluation of global and highly resource intensive measures such as screening at entry ports and social distancing was lacking. The handful of studies (mostly done during the SARS epidemic) did not allow firm conclusions to be drawn.
In this systematic review we found that physical barriers such as handwashing, wearing a mask, and isolation of potentially infected patients were effective in preventing the spread of respiratory virus infections. It is not surprising that methods of the included studies were at risk of bias as these types of interventions are difficult to blind, are often set up hurriedly in emergency situations, and funding is less secure than for profit making interventions. Hasty design of interventions to minimise public health emergencies, particularly the six included case-control studies, is understandable but not when no randomisation (not even of clusters) was done in the several unhurried cohort and before and after studies, despite randomisation leading to minimal disruption to service delivery. Inadequate reporting often made interpretation of before and after studies difficult.
The settings of the studies, carried out over four decades, were heterogeneous, ranging from suburban schoolsw4 w37 w29 to military barracks,w25 intensive care units, paediatric wardsw14 w16 in industrialised countries, slums in developing countries,w1 and day care centres for children with special needs.w22 Few attempts were made to obtain socioeconomic diversity by, for example, involving several schools in the evaluations of one programme.w29 We identified few studies from developing countries where the most burden lies and where cheap interventions are needed. Even in Israel, the decrease in acute respiratory tract infections subsequent to school closure may have been related to atypical features: the high proportion of children in the population (34%) and limited access to over the counter drugs, which together with the national universal comprehensive health insurance means that symptomatic treatment is generally prescribed by doctors.w19
Compliance with interventions—especially educational programmes—was a problem for several studies, despite the importance of such low cost interventions. Routine long term implementation of some would be problematic—particularly maintaining strict hygiene and barrier routines for long periods, probably only feasible in highly motivated environments such as hospitals without the threat of an epidemic.
Global and highly resource intensive measures such as screening at entry ports and social distancing lacked proper evaluation. The handful of studies (mostly done during the SARS epidemic) did not allow us to reach any firm conclusions, although a recent analysis of historical and archival data from the 1918-9 influenza pandemic in the United States suggests an effect of social distancing measures such as school closures and bans on public gatherings.24
Nevertheless our systematic review of available research does provide some important insights. Perhaps the impressive effect of the hygienic measures aimed at younger children derives from their poor capability with personal hygiene.w1 w11
Simple public health measures seem to be highly effective at reducing the transmission of respiratory viruses, especially when they are part of a structured programme including instruction and education and when they are delivered together. Further large pragmatic trials are needed to evaluate the best combinations. In the meantime we recommend implementing the following interventions combined to reduce the transmission of respiratory viruses: frequent handwashing (with or without antiseptics), barrier measures (gloves, gowns, and masks), and isolation of people with suspected respiratory tract infections.
What is already known on this topic
People are increasingly concerned about pandemics of virus infections such as avian influenza and SARS
Preparation against pandemics includes developing vaccines and stockpiling antiviral agents—interventions that are virus specific and of unknown effectiveness in epidemic disease
What this study adds
Several physical barriers, especially handwashing, masks, and isolation of potentially infected people, were effective in preventing the spread of respiratory virus infections
Such interventions should be better evaluated and given higher priority in preparation for pandemics
We thank Peter Doshi, Anne Lyddiatt, Stephanie Kondos, Tom Sandora, Kathryn Glass, Max Bulsara, and Allen Cheng for commenting on the draft protocol; Jørgen Lous for translating a Danish paper and extracting data; Taixiang Wu for translating Chinese text; and Ryuki Kassai for translating Japanese text.
Contributors: RF and AR constructed the search strategy. TJ, CDM, and LD drafted the protocol. LD, CDM, and RF incorporated the referees’ comments. TJ, FE, BH, and AP extracted study data. SN carried out the analyses. TJ and CDM wrote the final report and are the guarantors for the paper. All authors contributed to the final paper.
Funding: Cochrane Collaboration Steering Group, UK, and each author’s institution.
Competing interests: None declared.
Ethical approval: Not required.
Provenance and peer review: Not commissioned; externally peer reviewed.