A decade of norovirus disease risk among older adults in upper-middle and high income countries: a systematic review

Search results

A summary of the literature review process is given in Fig. 1. Publications that provided information on NoV risk by age among older adults are summarized in Table 1. We identified 39 studies conducted in 15 HI and UMI countries (44 % Europe, 28 % in the US/Canada, 23 % in Australia, Asia and the Middle East, and 5 % global reviews which were based predominantly in HI and UMI countries). Data were included in the identified studies from 1966–2013.

Author (Ref)	Country	Study years	Study design	Lab Method	Risk estimation methoda	Age	Data source/population	Outcome of interest
Trivedi [89]	US	2009–2010	Retrospective cohort	NSb	C	NS	NORS and Medicare/LTCFsc in 3 states	All cause hospitalizations and deaths during outbreaks
Hall [51]	US	1999–2007	Retrospective regression analysis	NS	C	All	National Center for Health Statistics/nationwide GEd deaths	Incidence and attributable proportion of unspecified GE deaths
Shahid [109]	US	2007	Retrospective laboratory review	Elisa	B	All	In- and outpatient laboratory samples/city-wide patients tested for NoV	Characteristics of NoV positive specimens
Lopman [70]	US	1996–2007	Retrospective regression analysis	NS	C	All	National Inpatient Sample/nationwide GE hospital discharges	Incidence and attributable proportion of unspecified GE discharges
Rosenthal [38]	US	2003–2006	Prospective surveillance	PCR	A	18–106 y	AGEeoutbreaks/LTCFs in 1 state	Attack rates, case-hospitalization and fatality rates during NoV outbreaks
Chui [71]	US	1991–2004	Retrospective database review	NS	D	65– <85 y	Medicare and Medicaid and US Census/nationwide hospitalizations	Incidence of specific gastrointestinal disease hospitalizations
Gastañaduy [45]	US	2001–2009	Retrospective regression analysis	NS	C	All	MarketScan Insurance Claims/US emergency and outpatient visits among insured	Incidence of emergency and outpatient visits for cause-specified and unspecified GE attributed to NoV
Leshem [35]	US	2010–2013	Retrospective analysis surveillance	PCR	A	All	Sentinel community NOV surveillance (NORS and CaliciNet)/community outbreaks in 5 states	Age-specific NoV distribution during GII.4 outbreaks
Leshem [110]	US	2010–2012	Retrospective analysis surveillance	PCR	A	All	Sentinel community NoV surveillance (NORS and CaliciNet)/community outbreaks in 27 states	Age-specific NoV distribution during GI.6 outbreaks
Hall [111]	US	1979–2010	Literature review	NS	E	All	Published studies with population-based incidence rates of NoV/US studies	Summary NoV incidence estimates
Ruzante [72]	Canada	2001–2004	Retrospective database review	NS	D	All	Canadian Institute for Health Information, Vital Statistics Registry, National Notifiable Diseases database/nationwide hospitalizations and deaths	Age-specific hospitalization and mortality incidence rates based on NoV-specific diagnostic coding
Friesema [87]	NL	2005–2007	Prospective surveillance	PCR	A	42–100 y	NoV outbreaks/LTCFs with nursing wards in 6 local health services	Characteristics of NoV outbreaks by genotype and symptoms
Van Asten [77]	NL	1999–2007	Retrospective analysis surveillance	NS	C	≥65 y	Statistics Netherlands (mortality), weekly laboratory surveillance/nationwide unspecified GE	Attributable proportion of all-cause deaths
Van Asten [68]	NL	1994–2006	Retrospective analysis surveillance	NS	C	≥65 y	Netherlands Information Network of GPs, NoV surveillance, National Medical Register, Statistics Netherland/nationwide unspecified GE	Incidence and attributable proportion of unspecified GE deaths
Verhoef [80]	NL	2009	Retrospective analysis surveillance analysis	NS	C	All	SENSOR study, German notifiable disease data, other literature/nationwide	Age-specific mortality, disease burden and burden due to foodborne NoV GE
Spackova [82]	Germany	2002–2008	Retrospective analysis surveillance	NS	A	All	Hospitalization reports: German notifiable disease data/nationwide	Nosocomial GE due to enteric pathogens including NoV
Werber [63]	Germany	2004–2008	Retrospective analysis surveillance	PCR	A	All	German notifiable disease data, Federal Statistical Office/nationwide	Incidence, deaths and potential life lost due to pathogen-specific GE
Bernard [32]	Germany	2001–2009	Retrospective analysis surveillance	PCR	A	All	German notifiable disease data, Federal Statistical Office/nationwide	Age-specific incidence, hospitalization and mortality due to NoV
Gustavsson [55]	Sweden	2008–2009	Retrospective case–control	PCR	B	≥60 y	Hospital database/hospitalized adults at one hospital	30 and 90 day survival rates
Fernandez [65]	Spain	2000–2007	Retrospective laboratory-based	PCR	B	All	Patient laboratory stool samples	Age-specific prevalence of enteric pathogens in GE cases
Arias [112]	Spain	2004–2005	Prospective surveillance	PCR	A	All	Mandatory physician reported outbreaks/Catalonia region	Age-specific incidence of GE
Manso [67]	Spain	2010–2011	Prospective laboratory-based	PCR	B	All	In- and outpatient laboratory stool samples/hospital complex	Age-specific prevalence of enteric pathogens in GE cases
Huhulescu [64]	Austria	2007	Prospective cohort	PCR	A	All	Outpatients/3 physician clinics	Pathogen-specific prevalence in GE
Harris [78]	England/Wales	2001–2006	Retrospective regression analysis	NS	C	≥65 y	Community pathogen surveillance; mortality statistics: Health Protection Agency, Office of National Statistics/England and Wales	Attributable proportion of GE deaths
Haustein [69]	England	2000–2006	Retrospective regression analysis	NS	C	≥18 y	Hospital Episode Statistics and LabBase Health Protection Agency/nationwide	Incidence and attributable proportion of hospital admissions
Philips [42]	England	1993–1996	Prospective cohort	PCR	A	All	Outpatients/70 GP clinics nationwide	Age-specific NoV incidence
Philips [113]	England	1993–1996	Prospective cohort	PCR	A	All	Community and outpatient non-GE patients/cohort and 70 GP clinics nationwide	Prevalence of asymptomatic NoV
Rovida [76]	Italy	2011–2012	Retrospective laboratory-based	PCR	B	All	Inpatient laboratory samples (stool)/one hospital	Pathogen-specific prevalence of GE
Kirk [83]	Australia	2002–2008	Retrospective surveillance analysis	NS	A	Adults	GE outbreaks OzFoodNet/LTCFs nationwide	Pathogen-specific GE outbreak incidence and characteristics of GE outbreaks
Davis [88]	Australia	2004–2007	Retrospective surveillance analysis	PCR	A	NS	GE outbreaks OzFoodNet/LTCFs in Queensland	Characteristics of NoV outbreaks in LTCFs
Tian [66]	China	2008–2009	Prospective cohort	PCR	B	≥14y	Outpatients with AGE/GE Department one hospital	NoV prevalence by age
Chan [74]	Hong Kong	2012–2013	Prospective cohort	PCR	B	All	Inpatients with NoV+ AGE/one hospital	NoV GII.4 distribution by age
Ho [37]	Hong Kong	2001–2006	Prospective laboratory-based analysis	PCR	A	All	Laboratory stool samples from inpatient, outpatients and outbreak cases with AGE/nationwide	Characteristics of NoV outbreaks by age, genotype and season
Tseng [114]	Taiwan	2005–2007	Prospective surveillance	PCR	B	>15 y	AGE cases at residential psychiatric care institution/one care center	Pathogen-specific GE hospitalization rate
Tang [75]	Taiwan	2011–2012	Prospective cohort	PCR	B	All	AGE symptomatic and asymptomatic in-, out- and emergency patients/one hospital	NoV prevalence by AGE symptom status
Lim [15]	Singapore	2004–2011	Retrospective laboratory	PCR	B	All	AGE stool samples/one hospital laboratory	Age-specific distribution of NoV positive samples
Al-Thani [73]	Qatar	2009	Prospective cohort	PCR	B	All	AGE emergency patients/one hospital	Enteric pathogen distribution by age
Greig [85]	Global	1997–2007	Literature review	NS	E	NS	GE outbreaks in LTCFs/global	Characteristics of enteric outbreaks by pathogen and transmission
Utsumi [86]	Global	1966–2008	Literature review	NS	E	NS	Infectious disease outbreaks in LTCFs/global	Characteristics of infectious disease outbreaks by pathogen

^aMethod: A = laboratory confirmed or epidemiologically-linked population-based surveillance; B = laboratory-confirmed clinic/hospital setting; C = indirect attribution from regression modeling; D = hospital database study select enteric illness codes without extrapolation E = literature review

^bNS not specified; NS diagnostic methods may have included molecular methods; NS age often reported for LTCFs generally serving older adults

^cLTCF long term care facility

^dGE gastroenteritis

^eAGE acute gastroenteritis

Risk estimates based on laboratory-confirmed or epidemiologically-linked population-based surveillance data were provided by 36 % (14/39) of studies, 28 % (11/39) were hospital or clinic-based laboratory-confirmed studies and 23 % (9/39) of the studies indirectly attributed NoV risk based on regression modeling. The remaining studies provided risk estimates based on secondary database review of NoV-specific hospitalization codes (5.1 %, 2/39) and literature reviews (7.7 %, 3/39). The included literature reviews diverged from the current study in scope of research interests and methods. NoV-associated outcomes reported in these studies included infections, illness, outpatient and emergency visits, hospitalizations and deaths. Ten primary studies reported NoV incidence data and 29 studies reported other measures of risk such as proportion of AGE cases positive for NoV or attack rates. Eleven publications focused predominantly on older adults. Seven publications (5 studies and 2 reviews) pertained exclusively to LTCFs (which did not always provide study-specific age distribution) and 1 publication pertained exclusively to nosocomial GE cases. NoV was detected using molecular methods in 22 studies, 1 study used immunoassay, and 16 studies used a variety of methods (which could have included molecular methods) or the method was not specified. Thirty studies considered study periods of at least 1 year.

NoV-associated illness

No studies were identified that reported the community age-specific incidence of non-medically-attended NoV illness among older adults. German notifiable disease surveillance data reported community-level age-specific incidence of NoV-associated cases among individuals ≥65 years, which is presumed to include symptomatic individuals of varying severity [32, 63]. This data provided information on hundreds of thousands of individuals representing both sporadic and outbreak-associated community and institutional NoV cases that were laboratory or epidemiologically-confirmed [32, 63]. Age-specific mean incidence of NoV cases (2001–2009) demonstrated a U-shaped age distribution with the highest rates in the very young and older adults. NoV incidence (per 10,000) estimates from 2004–8 were approximately 54 in those <5 years, 5–15 in those 5–69 years, and 48 in those ≥70 years [63]. A second publication of the German surveillance data (2001–2008) provided a similar age-specific pattern with the highest incidence estimates per 10,000 in those <5 years (43) and individuals ≥75 years (28.9-119.8 depending on the 5-year age strata with highest estimates in those ≥85 years) [32].

NoV-associated outpatient visits

Two studies report age-specific NoV-associated outpatient incidence estimates that include older adults [42, 45]. Both of these studies reported the highest rates in the very young. Model-based rates (per 10,000) from the US were 233 among those 0–4 years, 85 among those 5–17 years, 35 among those 18–64 years, and 54 among those ≥65 years [45]. Community-based age-specific incidence rates per 10,000 from England were 320 in those 0–4 years, 44 among 5–14 year olds, 38 among 15–44 year olds, 26 among 45–64 year olds and 37 among those ≥65 years [42].

Several publications report the mean proportion of outpatient GE visits due to NoV among middle-older aged adults (the lower age for many of these studies was 60 years) which ranged from 4–37 % [45, 64–67]. While most studies did not provide confidence intervals around these estimates, the proportion of GE outpatient visits due to NoV among older adults was not noticeably different from the estimated 6–26 % reported among other age strata [45, 64–67].

While the incidence of NoV-associated outpatient visits in older adults may not be as high as in the very young NoV outbreaks nevertheless result in large increases of unspecified GE outpatient visits in older adults [68]. In the Netherlands, the mean monthly incidence, per 10,000, of unspecified GE outpatient visits increased 70–240 % above average (15.0, 17.1 and 21.7 depending on the season) during NoV outbreaks with an overall mean of 9.1 among those ≥65 years. Taking the mid-point in the estimated proportion of outpatient GE visits due to NoV among older adults of 16.5 % (see above range estimates of 4–37 %), the Dutch monthly incidence data for unspecified GE outpatient visits can be extrapolated ((9.1*0.165)*12)) to an overall mean annual incidence per 10,000, of 18 with increased rates during NoV outbreaks up to 43.0 among those ≥65 years. These estimates depend substantially on the actual proportion of AGE due to NoV and the level/severity of NoV circulation. While it is expected that increased rates would occur across all age groups, a direct comparison by age was not possible due to the study’s focus on older adults.

NoV-associated emergency department (ED) visits and hospitalizations

A single, US model-based study provided mean age-specific incidence estimates per 10,000 for NoV-associated ED visits which were: 38 among those <5 years, 10 among those 5–17 years, 12 among those 18–64 years, and 15 among those ≥65 years [45].

Four studies reported age-specific incidence rate estimates for NoV-associated hospitalizations among older adults [69–72]. Two of these studies were based on indirect model-derived estimates that accounted for under-diagnosis of NoV among hospitalized patient [69, 70] while the other two studies provided estimated NoV hospitalization rates based on NoV-specific diagnostic code(s) that did not include the estimated influence of NoV on non/unspecific GE diagnoses [71, 72].

English model-derived estimates of NoV-associated hospitalization records among adults estimated a rate range, per 10,000, of 0.23–0.48 inpatient admissions among 18–64 year olds and 1.0–4.3 among those ≥65 years [69] depending on the season. These estimates were based on national GE diagnostic codes and laboratory-based pathogen data. Higher estimates were obtained from a US model-based study that estimated the seasonal mean age-specific rate per 10,000 (and range) of NoV-associated hospitalization discharges as follows: 9.4 (5.3–18.5) among those <5 years, 1.1 (0.6–1.9) among 5–17 year olds, 1.0 (0.5–1.4) among 18–64 year olds, 4.7 (2.9–7.7) among 65–74 year olds, 9.2 (5.0–21.5) among 75–84 year olds, and 18.5 (10.6–37.9) among those ≥85 years [70].

Two studies reported estimated incidence rates for NoV-specific hospitalizations based on International Classification of Disease (ICD)-9 and 10 codes [71, 72]. While such specific estimates are considered underestimates of true NoV-associated hospitalization rates, they are useful for further burden of disease extrapolations and provide insight into variable diagnostic practices. One such study, a US-based Medicare and Medicaid Services study of individuals aged 65 to 85 years (representing >90 % of US elderly) reported an annual rate, per 10,000, of NoV-specific hospitalizations (based on ICD-9 code 008.63) of 0.004 among 65–85 year olds [71]. These authors highlighted that a substantial proportion of non-specific GE hospitalizations are likely to be undiagnosed viral GE in this population. Canadian-based national hospitalization data provided higher annual rates, per 10,000, of NoV-specific diagnosis (ICD-9 008.63 or ICD-10 A08.1) among middle aged and older adults (>59 years) of 0.61 [72]. These authors further noted that 63 % of NoV-specific hospitalizations were in this oldest age group.

Differences in the NoV hospitalization rate estimates between model-based indirect estimates [71, 72] and rate estimates based on NoV-specific diagnostic codes [69, 70], exceed 1000 fold when comparing the estimates of hospitalizations above 65 years by Chui et al. [71] to those for individuals 65–74 years reported by Lopman et al. [70]. These large differences highlight the extreme underestimation of risk based on pathogen-specific diagnostic codes. The validity of the model-based estimates is supported by seasonal peaks in non-specific GE hospitalization diagnoses that coincide with specific viral GE such as NoV.

The proportion of non-specific AGE ED/or hospitalized patients due to NoV among older adults has been reported in several model and laboratory-based studies [45, 69, 70, 73–76]. Qatar (laboratory-confirmed) and US-based (model derived) studies estimated that 12 and 17 % (respectively) of GE ED visits were due to NoV among adults aged >60 years [45, 73]. In England, >20 % of GE emergency hospital admissions were due to NoV among those ≥65 years during peak times of NoV activity [69].

The estimated proportion of hospitalized GE cases due to NoV among those ≥65 years, in model-derived US and English-based studies, was 8–13 % depending upon whether all-cause or unspecified caused GE hospitalization codes were included in the analyses [69, 70]. Approximately 50 % of hospitalized GE patients >65 years with stored stool samples were NoV-positive compared to approximately <20 % among younger age groups [76]. This higher estimate may reflect differences in the study population due to more select sampling of sick individuals with NoV-like symptoms, and may be less representative of a broad and diverse GE population.

NoV outbreaks are associated with large increases in unspecified GE hospitalizations in older adults. One study in the Netherlands demonstrated an increase in the mean monthly incidence of unspecified GE inpatient visits >30 % above average during NoV outbreaks (2.4, 2.7 and 2.6 per 10,000 depending on the season) with an overall mean of 1.9 among those ≥65 years [68]. Taking the mid-point in the estimated proportion of hospitalization GE visits due to NoV among older adults of 21.5 %, the Dutch monthly incidence data for unspecified GE inpatient visits can be extrapolated ((1.9 *0.215)*12)) to an overall mean annual incidence, per 10,000, of 4.9 with increased rates during NoV outbreaks up to 7.0 [68]. These estimates are comparable to those estimated in a US model-based study of approximately 5–19 per 10,000 [70].

NoV-associated hospitalizations in those ≥65 years have been reported to be more frequent, more severe, result in longer stays and require transfer to other care facilities following discharge compared to other age groups [32, 70, 72]. In the US, NoV-associated hospitalization rates and associated costs were highest among older adults, with half of the estimated $500 million per year spent on NoV-hospitalizations among those ≥65 years [70].

NoV-associated mortality

Older adults contribute the vast majority of NoV-associated deaths [51, 55, 63]. US and German-based NoV-associated mortality rate estimates demonstrate a sharp increase with age [51, 63]. Specifically, US-based average annual mortality rate estimates, per 10,000, were 0.013 in those <5 years, 0.002 among those 5–64 years, and 0.20 among those ≥65 years (with fluctuations up to 55 % in this age group during non-epidemic and epidemic years) [51]. German NoV surveillance data reported similar NoV-associated mortality rate estimates per 10,000 namely, <0.01 in those <70 years to approximately 0.32 among those ≥70 years [63].

Dutch model-based estimates suggest that seasonal viruses account for nearly 5 % of all-cause mortality in the elderly with influenza accounting for the most deaths (approximately 2 %), followed by respiratory syncytial virus (1.5 %), parainfluenza (1 %) and NoV (0.2 %). The percentage of all deaths due to NoV increased with age to 0.5 % among those ≥85 years [77].

In the US, approximately half of all GE deaths (regardless of age) are coded as ‘cause unspecified’ of which older adults account for an estimated 83 % [51]. Among this population, Clostridium difficile is considered the leading cause of all-cause GE deaths and in the US NoV is the second leading cause (8 %) of all-cause GE deaths [51, 78, 79]. In analyses that excluded deaths due to C. difficile, NoV contributed 16 % of GE deaths among older adults [51]. These findings are comparable to data from England and Wales that modeled seasonal variations in death according to seasonal variations in laboratory reports (excluding deaths due to Clostridium difficile) where NoV accounted for 20 % of older adult deaths due to infectious intestinal disease and 13 % of deaths registered as ‘non-infectious gastrointestinal disease’ [78].

Data from the Netherlands estimated that 14 % of unspecified GE deaths were attributable to NoV among older adults with outbreak-associated increases of approximately 130 % above the mean monthly unspecified GE death incidence per 10,000 (>0.06 compared with 0.026) [68]. Based on an attributable proportion of 14 %, this equates to a mean annual NoV incidence per 10,000 of 0.04 with increases to >0.10 during outbreaks. While these mean estimates are lower than the estimated rates per 10,000 based on German (0.32) and US (0.20) data, the estimated rates during outbreaks reached a comparable level [51, 63, 68].

One publication extrapolated German surveillance data to the Dutch population to obtain estimated mean age-specific NoV case fatality rate per 1000 (95 % CI) of: 0.09 (0.01–0.27) for those <1 year, 0.00 (0–1.12) for those 1–4 years, 0.00 (0–0.04) for those 5–11 years, 0.09 (0–0.34) for those 12–17 years, 0.03 (0.01–0.06) for those 18–64 years and 0.63 (0.55–0.73) for those ≥65 years [80].

In a Swedish-based study of hospitalized adults with community-onset NoV infection, nearly 8 % of individuals died within 30 days of a NoV-positive sample (e.g., an 8 % 30-day NoV mortality rate) and all deaths occurred among those ≥60 years [55]. Individuals with underlying medical conditions had a higher 30-day mortality rate than those without such conditions (10.5 % compared to 5.3 %). The study authors did not specify the relevant underlying conditions. However, another study by Harris et al., noted that circulatory, respiratory and neoplastic conditions were important contributors to NoV-associated deaths [78].

Institutional NoV outbreaks

National surveillance data from HI countries have reported that the majority of NoV outbreaks occur in hospitals and LTCFs [31, 32, 35]. NoV GII.4 has been identified in a majority of institutional outbreaks among older adults [32, 35].

Estimated number of NoV cases in HI and UMI settings

According to 2013 World Bank estimates, there were approximately 209 million individuals ≥65 years living in HI countries (16 % of total HI population) and 193 million individuals ≥65 years living in UMI countries (8 % of total UMI population).

Figure 2 provides estimated rates of NoV-associated health outcomes among older adults and extrapolated annual cases in HI and UMI countries. The wide range of the estimates reflect the differences in 1) study population; 2) detection and attribution methodology of NoV to GE episodes; and 3) NoV activity level and genotype distribution.