Larson 2010.

Methods	Cluster block‐randomised, controlled trial carried out between 20 November 2006 and 20 June 2008 in an upper Manhattan immigrant Latino neighbourhood (“19 month data collection period”). The study aimed at assessing the effects of eduction versus education and hand sanitiser use versus education and hand sanitiser use and common mask use against upper respiratory infections over a period of under 2 years. Follow up was though an automated telephone system with a small financial incentive (USD 20) for those with 75% or more compliance. Those reporting an ILI received a visit within 48 hours for swabbing  An index case was someone who at the “onset day of illness nobody else in the household had been symptomatic within the previous five days” A secondary case for each episode “was any member of the household who developed symptoms within five days following the index case”, “The secondary attack rate was defined as the number of secondary cases recorded within 5 days of the onset of symptoms in the index case divided by the number of household members minus one”  The text implies that the unit of observation was the episode (“study subjects contributed more than one episode in which they were considered to be the index case”)
Participants	Recruitment and allocation were carried out by household. These had to have at least 3 people living in the household, with at least 1 being a preschool or elementary school child, speaking English or Spanish, having a telephone willingness to complete symptom assessments and having bimonthly home visits and not using alcohol‐based hand sanitiser routinely 617 households were randomised, 211 to the education, 205 to the hand sanitiser and 201 to the hand sanitiser and mask groups. The participants were 2708, mostly adult Latino immigrants to the USA  Intracluster correlation coefficients are reported on page 179 of the manuscript
Interventions	Written Spanish or English language educational materials regarding the prevention and treatment of URTIs and influenza or the same educational materials and hand sanitiser (Purell, J&J), in large (8‐ and 4‐ounce) and small (1‐ounce) containers to be carried by individual household members to work or school or the same interventions as well as regular surgical face masks (Procedure Face Masks for adults and children, Kimberly‐Clark) with instructions for both the caretaker and the ill person to wear them when an ILI occurred in any household member. Replenishment of intervention stocks was done at the bimonthly home visit  Caretakers had to wear a mask for 7 days when within 3 feet of a symptomatic case. These were also encouraged to wear masks within 3 feet of any household member. Reinforcing phone calls were made 3 times in 6 days  The text clearly reports active influenza vaccine promotion during the bi‐monthly visits (“The home visit to each household was made every 2 months to minimise study dropout, reinforce adherence to the assigned intervention, replenish product supplies and record use of supplies, answer questions, and correct ongoing misconceptions. At each visit, new educational materials regarding URTI prevention and treatment and influenza vaccination were distributed.” (PDF page 3). Also just before the Discussion as follows: “Influenza vaccination rates: There was an increase between the baseline and exit interview in all three groups that reported 50% of more of members receiving influenza vaccine (pre‐ versus post‐intervention for each group: 21.1% and 40.8% in the Education group, 19.0% and 57.1% in the hand sanitiser group, and 22.4% and 43.5% in the hand sanitiser and face mask group (P = 0.001). Additionally, those in the hand sanitiser group reported a significantly greater increase than the other 2 groups, controlling for baseline rates (P = 0.002)”  Coverage was unequal across groups, no information on the progressive impact of the vaccine, or indeed the nature of the vaccine(s) is reported. Apparently the first season was mild and the vaccine mismatched, compliance with the trial interventions was low in Arm 3 and a local epidemic of S. aureus meant that the control group started washing hands The authors report no effect on reporting rates of vaccine coverage by arms but with so many confounders who knows?
Outcomes	Laboratory: PCR carried out on samples from deep nasal swabs for influenza and the most common other pathogens (RSV, rhinovirus, enterovirus, parainfluenza viruses etc.). The text describing the results of the swabbing is confusing but in general appears to be non‐random “Households reported 669 episodes of ILI (0 to 5 per individual)". Of the 234 deep nasal swabs obtained, 33.3% (n = 78) tested positive for influenza; 43.6% (n = 34) were influenza A and 56.4% (n = 44) were influenza B. Among the 66.7% who tested negative for influenza, 30.8% (48/156) tested positive for other viruses: 7 for respiratory syncytial virus, 9 for parainfluenza, 11 for enterovirus, 10 for rhinovirus, 6 for adenovirus, and 5 for metapneumovirus. Swabs were not obtained from the remaining 435 reported ILI episodes for the following reasons: 72.0% (n = 313) did not meet the CDC definition of an ILI and were therefore included in the URTI symptom count, 21.4% of episodes (n = 93) were reported after 48 hours of ILI onset or the participant refused to be swabbed, and the research staff were unable to reach the participant in 6.7% of episodes (n = 29) As no definition of URTI is given it is unclear what kind of biases are introduced by the non‐swabbing of the 313/435 “not meeting CDC definition”.  Effectiveness: ILI (CDC definition): “temperature of 37.8°C or more and cough and/or sore throat in the absence of a known cause other than influenza” URTI only referred to as “Viral upper respiratory infections (URTIs)”  Safety: N/A
Notes	The authors conclude that “the Hand Sanitizer group was significantly more likely to report that no household member had symptoms (P,0.01), but there were no significant differences in rates of infection by intervention group in multivariate analyses. Knowledge improved significantly more in the Hand Sanitizer group (P,0.0001). The proportion of households that reported >50% of members receiving influenza vaccine increased during the study (P,0.001). Despite the fact that compliance with mask wearing was poor, mask wearing as well as increased crowding, lower education levels of caretakers, and index cases 0–5 years of age (compared with adults) were associated with significantly lower secondary transmission rates (all P,0.02). In this population, there was no detectable additional benefit of hand sanitiser or face masks over targeted education on overall rates of URTIs, but mask wearing was associated with reduced secondary transmission and should be encouraged during outbreak situations. During the study period, community concern about methicillin‐resistant Staphylococcus aureus was occurring, perhaps contributing to the use of hand sanitiser in the Education control group, and diluting the intervention’s measurable impact”.  The study is at high risk of bias. Randomisation and reasons for dropout are not described. Differentials in cluster characteristics across arms point to randomisation not having worked and the confounding effects of a post‐randomisation staphylococcal scare is difficult to judge. Symptom‐driven follow up gives no idea of the effects on asymptomatic ILI/influenza. Poor definitions (URTI?). There are unexplained dropouts and the analysis plan is unclear. Finally the very small number of cases of influenza and an unclear swabbing attrition may introduce further elements of confounding
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"Cluster block randomised, controlled trial", but sequence generation not reported
Allocation concealment (selection bias)	Unclear risk	"Households were block randomised into one of three groups:" Allocation concealment not reported
Blinding (performance bias and detection bias) All outcomes	High risk	Not possible
Incomplete outcome data (attrition bias) All outcomes	High risk	In 211 control group HH 26 dropped out and 37 did not consent In 205 hand sanitiser group HH 21 dropped out and 36 did not consent In 201 hand sanitiser and face mask group HH 19 dropped out and 35 did not consent Reasons for dropout not described
Selective reporting (reporting bias)	Unclear risk	617 of 772 eligible households were randomised HH in groups comparable