Results
A total of 654 articles were identified in the literature search. Three-hundred and fourteen articles were excluded after reviewing abstracts, and 303 articles were excluded after full text review (Figure 1). Thirty-seven articles described 60 outbreaks occurring between 1980 and 2011 (Table 1), and 8 articles (21%) reported more than one outbreak. The reported outbreaks occurred in Australia (n = 4), Belgium (n = 1), Canada (n = 7), England (n = 3), France (n = 3), Japan (n = 4), Singapore (n = 1), Taiwan (n = 1), and the United States (n = 36). Fifty-one outbreaks (85%) consisted of only influenza A cases, seven (12%) consisted of only influenza B cases, and two (3%) consisted of cases of influenza A and B. The median number of cases in each outbreak was 28 (range: 7, 139) and a median of 128 individuals was at-risk of influenza (range: 28, 729).
(Enlarge Image)
Figure 1.
Article selection process. n = number of articles.
Influenza Case Definitions
Fifty-seven (95%) of 60 outbreaks reported a case definition for identifying influenza in the facility (Table S1). The Centers for Disease Control and Prevention define ILI as a temperature of at least 100°F (37·7°C) and cough or sore throat in the absence of a known cause other than influenza. Seven (12%) of 57 outbreaks with case definitions used a variation of this definition by also including individuals with coryza, sneezing or rhinorrhea, while 18 outbreaks (32%) had less specific definitions requiring a fever and at least one additional sign or symptom such as malaise or pneumonia. Twenty-one outbreaks (37%) did not require fever but included fever among possible signs defining an influenza case. While all studies used laboratory testing to establish influenza as the cause of the outbreak, eleven outbreaks (19%) required diagnostic confirmation of each case through the use of rapid diagnostic testing, fourfold seroconversion, viral culture or polymerase chain reaction (PCR). Three outbreaks required both laboratory confirmation and a consistent clinical presentation to identify cases.
Epidemic Thresholds
Fourteen outbreaks (23%) reported a facility policy defining an influenza outbreak (Table S1). Three outbreaks defined an influenza outbreak as the detection of two or more cases of ILI within 72 hours in a single residential unit. Two of these outbreaks required at least one positive rapid influenza diagnostic test among the ILI cases. The remaining 11 outbreaks declared an outbreak when an attack rate of at least 10% was observed within a 7-day period, with nine outbreaks requiring influenza viral isolation during the previous 7 days.
Antiviral Prophylaxis
Chemoprophylaxis was defined as offering antiviral drugs to asymptomatic individuals in the facility. Forty (67%) of the 60 outbreaks used prophylactic antiviral drugs (Table 1), 19 of which used at least two drugs. Of the 40 outbreaks, 34 (85%) consisted of only influenza A cases, two (5%) included cases of influenza A and influenza B, and 4 (10%) consisted of only influenza B cases.
Amantadine was used in 19 (56%) of the 34 influenza A only outbreaks and one (50%) of the two influenza A/B outbreaks. Amantadine was the only antiviral drug used in nine outbreaks and was used alongside rimantadine (n = 4), oseltamivir (n = 5) and zanamivir (n = 1) (Table S2). Rimantadine was used in 14 (40%) of the influenza A only outbreaks and one (50%) of the two influenza A/B outbreaks. Rimantadine was used alone in two outbreaks and was used with zanamivir in eight outbreaks (Table S2). None of the influenza B only outbreaks used amantadine or rimantadine, as is consistent with lack of pharmacologic activity by adamantanes for influenza B.
Compliance and side effects were typically reported by article. Compliance with amantadine use was reported by one article, which described staff prophylaxis as less than half of staff members taking ≥70% of the prophylactic regimen. Resident compliance was not reported by any articles. Five articles (13%) reported discontinuation of amantadine by 11/111 (10%) residents reporting side effects. Side effects were wide ranging with agitation, anorexia, depression, fatigue, and gastrointestinal symptoms most frequently reported.
Twenty-five (63%) of the 40 outbreaks offered neuraminidase inhibitors (Table S2): 14 used oseltamivir (11 influenza A outbreaks, 1 A/B, 2 B only) and 11 used zanamivir (nine influenza A, two B only). Only two outbreaks used zanamivir as the sole antiviral drug. One influenza A outbreak used amantadine, rimantadine, and zanamivir. Oseltamivir and zanamivir were never used in the same outbreak.
Refusal of oseltamivir prophylaxis was reported for 3/89 (3%) and 2/129 (2%) residents in two articles, and two articles each reported refusal by one patient after 2 and 3 days of prophylaxis without explanation. Side effects of oseltamivir resulting in discontinuation included difficulty swallowing (n = 1), nausea (n = 2), and vomiting (n = 2). Zanamivir was refused by 11/130 (8%) residents in one study and inhalations were difficult for 29/130 (22%) residents.
Vaccination
Prophylactic vaccination was defined as offering seasonal trivalent inactivated or live intranasal influenza vaccine to residents and/or staff prior to the identification of influenza cases in the facility. Institutions had prophylactically vaccinated a median of 89% of residents (IQR: 72%, 92%) and 48% of staff (IQR: 28%, 70%) during the most recent influenza season among 47 and 21 outbreaks, respectively. Thirty-four outbreaks (72%) reported ≥75% of residents vaccinated against seasonal influenza, while only five outbreaks (21%) reported this proportion of vaccinated staff. Over time, vaccination rates of staff and residents increased by 2·3% and 0·84% per year (P = 0·016 and P = 0·018) (Figure 2a), but increased staff and resident vaccination rates were not associated with decreased attack rates (P = 0·46 and P = 0·07) (Figure 2b).
(Enlarge Image)
Figure 2.
(A) Percentage of vaccinated long-term care facility residents (open circles) and staff (dark circles) over time in 47 and 21 influenza outbreaks reporting vaccination rates, respectively. (B) Percentage of vaccinated residents (open circles) and staff (dark circles) and influenza-like illness attack rates.
Reactive vaccination was defined as offering influenza vaccine to asymptomatic individuals after the identification of influenza cases in the facility. Only four (7%) of the 60 outbreaks reported reactive vaccination as an outbreak control measure: in two outbreaks, vaccination was offered or recommended to staff and residents; in one outbreak, all unimmunized residents were vaccinated; and in one outbreak, vaccination was offered to staff. The number or percentage of individuals receiving reactive vaccination was not reported in any outbreak.
Non-pharmaceutical Interventions (NPI)
Physical measures attempting to reduce influenza transmission that did not require drugs or vaccines such as PPE and social distancing were defined as NPI (Table 1). Glove and mask use, hand hygiene, and droplet precautions were considered PPE and were reported by 19 (32%) of 60 outbreaks.
Social distancing was reported by 23 outbreaks (38%) and included patient isolation and restrictions on staff, visitors, admissions and ward transfers. Isolation was defined as restriction of movement within the facility by symptomatic individuals through the use of room, unit, or ward quarantine or cohorting and was reported by 13 outbreaks (22%). Two outbreaks (3%) reported the length of required isolation as a period of 5 days or for the duration of symptoms.
Visitor and staff restriction was reported by 14 of 60 outbreaks (23%). Four outbreaks (7%) reported policies for symptomatic staff during influenza outbreaks, which emphasized taking sick leave until symptom resolution or 5 days post-symptom onset (Table S3). Eleven outbreaks (18%) reported policies restricting visitors during influenza outbreaks (Table 1), including six outbreaks advising visitors of ongoing outbreaks or asking visitors to postpone visits (Table S3).
Attack Rates
Influenza attack rates for the 60 outbreaks ranged from 1·3% to 65% with an unadjusted mean of 28%. Among 17 outbreaks that did not implement intervention measures, an unadjusted mean attack rate of 41% (95% CI: 24, 51) was observed, which did not differ from three outbreaks using only PPE [30% (95% CI: 19, 37)] or from 21 outbreaks using a single antiviral drug [25% (95% CI: 14%, 29%)] (Figure 3). Among 21 outbreaks using a single antiviral drug, unadjusted mean attack rates did not differ significantly between drugs (P = 0·44) and ranged from 17% (95% CI: 7, 34) for zanamivir to 29% (95% CI: 25, 42) for oseltamivir (Figure 3).
(Enlarge Image)
Figure 3.
Unadjusted mean attack rates (95% CI) from influenza outbreaks in long-term care facilities by intervention method. Outbreaks do not sum to 60 as 19 outbreaks used ≥2 antiviral drugs for prophylaxis. Non-pharmaceutical interventions were used in conjunction with amantadine (n = 4 outbreaks), oseltamivir (n = 6 outbreaks) and zanamivir (n = 2 outbreaks). CI = confidence interval; n = number of outbreaks.
After adjusting for NPI, use of any antiviral drug halved attack rates among outbreaks consisting of influenza A cases [OR: 0·52 (95% CI: 0.29, 0·93)] and influenza A and B cases [OR: 0·48 (95% CI: 0·28, 0·84)] compared with outbreaks that did not implement influenza control measures (Table 2). More specifically, the use of adamantanes produced statistically significant protective effects among outbreaks with cases of influenza A [OR: 0·33 (95% CI: 0·17, 0·62)] and influenza A and B [OR: 0·27 (95% CI: 0·14, 0·48)]. Combined use of adamantanes and neuraminidase inhibitors consistently demonstrated protective effects, although not statistically significant, but surprisingly neuraminidase inhibitors alone did not (Table 2). Social distancing and PPE were not associated with significant changes in influenza attack rates (Table 2).