The population of schools surveyed derived from relatively disadvantaged areas, with 16 schools reporting a larger proportion of students in the bottom quartile of advantage according to ICSEA scores than the national average (Median difference: 4, range -25, 39). Median school level response rates were 19.9% (Range: 4%, 46%). Response rates were square root transformed to approximate a normal distribution, and linear regression performed to assess the relationship between this score and the excess (or under-representation) of students in the least advantaged quartile. The two were significantly related (Coefficient (95%CI): -0.04 (-0.06, -0.01); p = 0.002), reflecting lower response rates from less advantaged schools.
We received 314 responses from 1,181 (27%) eligible households approached by the 33 participating schools. Of these, 301 primary respondents (96%) provided information regarding the presence or absence of a medically diagnosed case in the household. Reporting households ranged in size from 2 to 9 members (median 4, interquartile range 4 to 5) and contained a total of 1,330 persons (Figure ). The total number of household members in families (n = 13) not reporting case status could not be determined due to missing data. Fifty-one families reported at least one pH1N1-infected individual. Seven of these families reported a secondary case and four reported two secondary cases, for an average secondary household attack rate of 6%. Only one of the 51 primary cases was older than 18 years.
Quarantine and prophylaxis recommendations, by case status of the household.
Four hundred and ninety-six individuals were asked to stay in voluntary home quarantine in association with the school and classroom closures. Quarantine was more likely to be recommended for household members if a child had a confirmed case of influenza. The recommended quarantine periods varied, ranging from 1-14 days (median 7 days, IQR 5-8 days) (Figure ).
Days spent at home relative to the recommended duration of quarantine (size of circles reflects the frequency of reported observations).
Compliance with requirements to stay at home
Individual compliance with the recommendation to stay at home was high, with respondents reporting that individuals stayed at home for more than 94% of the days they were advised to be in quarantine (95% CI 92.8, 95.9). This figure was not associated with the length of quarantine (Figure ) and did not fluctuate over the course of the quarantine period (data not shown). Of the 3,232 quarantined days, respondents reported that they and their family members spent most of their time outside the home during only 177 days. Of these days, 47 were spent in the homes of friends, 44 at school, 18 in the workplace and 68 at 'Other' unspecified locations. The proportion of individuals who remained at home during all days of their prescribed quarantine period was 88% - this lower figure was attributable to the variable length of the recommended quarantine period for any given individual, as shown in Figure .
When compliance was considered at household level, 250 households (84.5%; 95% CI: 79.3%, 88.5%) reported perfect compliance by all family members with quarantine recommendations throughout its duration, regardless of whether there was a case in the household (82.0% compliant) or not (85.0% compliant) (p = 0.57).
We estimated that only one per cent of the variation in this compliance outcome was explained by differences at the school level (level 2 variance), while 99% of variation was due to differences between households (level 1 variance).
Compliance with restrictions on outings
During the quarantine period, 25 reporting households (8.4%; 95% CI: 0.05%, 12.9%) stated that at least one quarantined family member left the home to visit "an outdoor public space with lots of other people around (e.g. playground or market)". A further 36 respondents (12.0%; 95% CI: 0.08%, 17.0%) reported an excursion to an enclosed public space, other than for medical attendance. There was no significant difference in such incidents between families with or without a resident influenza case (data not shown).
Compliance with requirements to avoid social mixing
The main purpose of school closure was to restrict contact between children that may facilitate the spread of infection. Forty-three households reported that a child spent at least one day outside the family home, and mixing with other children occurred on almost half of these occasions (48.8%; 95% CI: 35.7%, 62.1%), whether or not there was a case in the family (p = 0.5). Contact with children who were not immediate family members was far less likely during days spent at home. No child visited a study household in which another child was ill, compared with reported child visitors in 15.9% of 226 homes without a case (p < 0.001). Adult visitors were somewhat more common (31.1%; 95% CI: 25.5%, 37.3%), and again occurred more frequently in households without (33.5%) than with (19.6%) an influenza-infected individual (p = 0.04).
Compared to children in households that complied with recommendations to stay at home, children in households that did not comply with the recommendations were more likely to have been cared for during the quarantine period by an adult from outside the home (28.3% compared with 4.0% for compliant households; p < 0.001), thus also contravening the quarantine recommendation not to mix with adults from outside the household. This distinction was especially marked for households in which there was a confirmed case of influenza, where the difference was 44.4% of children receiving outside care in non-compliant households compared with 2.4% of those that were compliant.
Compliance with antiviral medications
Oseltamivir was prescribed for 313 individuals, more often if there was a case in the household and/or for quarantined persons (Figure ). Compliance with the medication was high, with 75% of respondents stating that the full drug course was completed (95% CI: 68.2, 80.6%). Only 7.1% refused it altogether, 9.9% took up to half, and 5.1% more than half (2.9% were unsure). The presence of a case in the household did not affect adherence to the prophylaxis or treatment regimen, nor did the age of the individual prescribed the medication. Reasons for non-completion of the course did, however, vary by age (data not shown). Where non-compliance was reported, the primary household respondent attributed this to belief that the drug was unnecessary (n = 42), particularly for individuals older than 18 years (p = 0.02). Some children refused to take the medication for unstated reasons (n = 10), but side effects, experienced (n = 12) or anticipated (n = 8), were infrequently reported.