- Email: [email protected]
Estimating age-specific influenza-associated asthma morbidity in Ontario, Canada
Respiratory Medicine 155 (2019) 104–112
Contents lists available at ScienceDirect
Respiratory Medicine journal homepage: www.elsevier.com/locate/rmed
Estimating age-specific influenza-associated asthma morbidity in Ontario, Canada
T
Laura Y. Feldmana, Jingqin Zhua,b, Teresa Toa,b,c,* a
Child Health Evaluative Sciences, The Hospital for Sick Children, Toronto, Ontario, Canada ICES, Toronto, Ontario, Canada c Dalla Lana School of Public Health, University of Toronto, Toronto, Ontario, Canada b
A R T I C LE I N FO
A B S T R A C T
Keywords: Asthma Influenza Health services Health administrative data
Background: There is a need to quantify the potential benefits of influenza-focused interventions in reducing asthma morbidity at a population level. This study aims to estimate age-specific annual excess asthma morbidity attributable to influenza in Ontario, Canada. Methods: Weekly counts of hospitalizations, emergency department (ED) visits and outpatient physician office visits for asthma were obtained from health administrative data in Ontario from 2010 to 2015, for ages 0–14, 15–59 and 60+. Asthma morbidity was modelled as a function of influenza A and B activity using linear regression, controlling for seasonal and long-term trend, mean temperature and respiratory syncytial virus. Excess asthma morbidity attributable to influenza was calculated as the difference between full model predictions and model predictions with influenza A and B variables set to 0. Results: Annually, influenza was associated with the following rates of excess asthma morbidity, per 100,000 people with prevalent asthma: 12.5 hospitalizations for ages 15–59 (95% confidence interval (CI): 1.1–23.5); 35.7 hospitalizations for ages 60+ (95% CI: 3.3–67.1); 114.1 ED visits for ages 15–59 (95% CI: 46.9–181.6); 154.6 ED visits for ages 60+ (95% CI: 86.7–223.3); and 1025.7 outpatient physician office visits for ages 60+ (95% CI: 79.0–1877.3). Conclusions: Influenza was associated with excess asthma hospitalizations and ED visits for ages 15–59 and 60+ and outpatient physician office visits for ages 60+. Individuals with asthma aged 15–59 and 60+ might be important targets for influenza-focused interventions, to reduce asthma morbidity at the population level.
1. Introduction Observational studies have suggested that influenza infection may be an important risk factor for asthma exacerbation in children and adults; however, clinical trials have failed to demonstrate efficacy of influenza vaccination in protecting against asthma exacerbation. In the 1990s, cohort studies by Johnston et al. and Nicholson et al. demonstrated that instances of acute asthma exacerbation in children [1] and adults [2] were very often accompanied by respiratory viruses. While human rhinovirus was the most commonly detected virus in these cohort studies and remains a pathogen of interest for both asthma development and exacerbation [3], influenza A and B were implicated in these early studies as possible triggers of asthma exacerbation. In a recent Cochrane review, Cates and Rowe investigated the efficacy of influenza vaccination in protecting against influenza-associated asthma exacerbation in children and adults [4]. Based on a single clinical trial
*
in children, they found no evidence of a protective effect. However, they noted that uncertainty in this question remains, due to the infrequency of influenza illness among individuals with asthma, as well as the small sample sizes and short follow-up times of existing clinical trials. Population-level health administrative data and virus surveillance data consist of large datasets, long follow-up times and objective, routinely collected outcome measures. Although population-level health administrative data generally lack individual-level data on influenza vaccination status and viral serotyping, and are therefore not suitable data sources to directly investigate the protective effects of influenza vaccination on asthma exacerbation, they can be used as complementary data sources to characterize the relationship between influenza and asthma morbidity and to identify subsets of the population experiencing a greater burden of influenza-attributable asthma morbidity.
Corresponding author. Child Health Evaluative Sciences, The Hospital for Sick Children, 686 Bay Street, Toronto, Ontario, M5G 0A4, Canada. E-mail address: [email protected] (T. To).
https://doi.org/10.1016/j.rmed.2019.07.017 Received 13 March 2019; Received in revised form 16 July 2019; Accepted 17 July 2019 Available online 17 July 2019 0954-6111/ © 2019 Published by Elsevier Ltd.
Respiratory Medicine 155 (2019) 104–112
L.Y. Feldman, et al.
2.3.2. Outcomes The main outcomes in this study were weekly counts of hospitalizations, ED and outpatient physician office visits for asthma. For each week from September 2010 to March 2015, provincial total counts of these health services were collected using International Statistical Classification of Diseases and Related Health Problems (ICD) 9 code 493 and ICD-10 codes J45 and J46. In order to exclude visits for which asthma was merely noted as a comorbidity, only main diagnostic codes were considered in this analysis. For hospitalizations, main diagnostic codes refer to conditions that contributed significantly to the treatment or length of stay. For ED visits, the main diagnostic code refers to the condition that contributed most significantly to the visit. For outpatient physician office visits, only one diagnostic code, which is independent of billing codes, is given for each visit; this was used as the main diagnostic code in this analysis.
Given the universal health care system of the province of Ontario and its regular surveillance of influenza and other viruses, it is an ideal region in which to measure the burden of asthma morbidity attributable to influenza among a cohort of individuals with prevalent asthma. Ecological time series methods have been routinely used to estimate mortality burdens associated with influenza at the population level [5–8] and these methods have, in recent years, been adapted to estimate morbidity burdens [9–12]. Using provincial health administrative data and virus surveillance data, as well as established methodology for measuring influenza-attributable morbidity, we aimed to estimate agespecific annual counts and rates of influenza-attributable asthma morbidity in Ontario, Canada.
2. Methods 2.1. Setting, population & study design
2.3.3. Effect modifier All analyses were stratified by three age groups: 0–14, 15–59 and 60+. This allowed for the relationship between the exposure and outcomes to be analyzed separately, for each age group.
This study was conducted in Ontario, Canada, using provincial health administrative data and respiratory virus surveillance data to investigate the burden of influenza on asthma morbidity in a population-based cohort of individuals with prevalent asthma. The study population included all individuals aged 0–99 living in Ontario who met a validated health administrative definition for asthma: at least 1 hospitalization for asthma and/or 2 ambulatory care visits for asthma in 2 consecutive years [13,14]. Individuals were excluded if they lacked a valid health card number or had missing/invalid information on residence, age or sex. In this ecological time series analysis, weekly counts of health services use (hospitalizations, ED visits and outpatient physician office visits) for asthma were regressed against weekly influenza A and B viral activity, controlling for the effects of seasonality, long-term trend, mean temperature and respiratory syncytial virus (RSV). Three age groups of individuals with asthma were studied: 0–14, 15–59 and 60+. Approval to conduct this study was obtained from ICES and the Research Ethics Board at the Hospital for Sick Children, Toronto, Canada. Individual consent was not required, as this study utilized deidentified health administrative data and analyses were performed at an aggregate, ecological level.
2.3.4. Covariates Covariates included the weekly percentage of positive tests for respiratory syncytial virus (RSV), mean weekly temperature, seasonal trend (using restricted cubic splines with a period of 52 weeks), longterm trend (using polynomial basis functions), the September peak and statutory holidays. Following the methodology of Goldstein et al., one week in July was removed from the 2012–2013 influenza season, which contained 53 weeks; this ensured that all influenza seasons had a duration of exactly 52 weeks [5]. Covariates were chosen to align with existing methodology for estimating influenza-attributable morbidity and to account for patterns of asthma morbidity in Canada, where asthma morbidity among children is known to peak in September [7,10–12,17,18]. Covariates, and the methods used for covariate selection, are described in detail in the appendix.
2.4. Statistical analysis Ordinary least squares (OLS) regression was used to model asthma morbidity (dependent variable) as a function of influenza activity (independent variable). Separate models were fitted for each age group and each measure of asthma morbidity (hospitalizations, ED visits, outpatient physician office visits). Models were selected based on minimization of the Akaike Information Criterion (AIC). Model fitting is described in detail in the appendix. In this time series study, model residuals showed temporal autocorrelation. Therefore, 95% confidence intervals (CI) were corrected for temporal autocorrelation by assuming an autoregressive structure of order 2 (AR2) [5,6,19]. In order to calculate the magnitude of asthma morbidity associated with influenza, two sets of predictions were made for each model: predicted counts of asthma morbidity with influenza A and B levels set to their observed values (“with influenza” scenario) and predicted counts of asthma morbidity with influenza A and B levels set to 0 (“without influenza” scenario). Influenza-associated excess asthma morbidity was calculated as the difference between the “with influenza” scenario and the “without influenza” scenario. Average annual excess influenza-associated asthma morbidity was calculated for influenza seasons 2010–2011, 2011–2012, 2012–2013 and 2013–2014. Annual excess counts were divided by the average size of the asthma prevalent population for each age group during the study period and multiplied by 100,000, to give annual rates of excess influenza-associated asthma morbidity per 100,000 individuals with prevalent asthma. All analyses were conducted in R version 3.4.3 [20].
2.2. Data sources Health administrative data housed at ICES were used to assemble the study cohort and to measure the outcome variables and certain covariates. The following datasets, all housed at ICES, were used: the Canadian Institute of Health Information (CIHI) Discharge Abstract Database (DAD) for hospitalizations; the National Ambulatory Care Reporting System (NACRS) for ED visits; the Ontario Health Insurance Plan (OHIP) Database for outpatient physician office visits; and the Registered Persons Database (RPDB) for individuals’ date of birth. Laboratory testing data were obtained from Public Health Ontario (PHO). PHO operates 11 laboratories across Ontario, which perform approximately 60% of respiratory viral specimen testing across the province [15,16]. These data reflect all specimens tested at the PHO laboratories, which include specimens from patients in outpatient, ED, hospital, intensive care unit and outbreak settings.
2.3. Exposure and outcome measures 2.3.1. Exposure The main exposure in this study was the weekly percentage of positive tests for influenza A and for influenza B (number of positive tests divided by total tests, for a given virus) in Ontario from September 2010 to March 2015. The percentage of positive laboratory tests for a given virus was used as a proxy variable for viral activity [7]. 105
Respiratory Medicine 155 (2019) 104–112
L.Y. Feldman, et al.
Table 1 Observed counts and rates of asthma morbidity in the asthma prevalent population across influenza seasons. Influenza Season 2010–2011 N Asthma Prevalence Total 1,953,018 0 to 14 430,660 15 to 59 1,201,283 60 and above 321,075 Hospitalizations Total 6566 0 to 14 3251 15 to 59 1919 60 and above 1396 ED Visits Total 41,014 0 to 14 15,212 15 to 59 21,739 60 and above 4063 Outpatient Physician Office Visits Total 647,858 239,710 0 to 14 15 to 59 289,435 60 and above 118,713
2011–2012 Ratea
N
2012–2013 Ratea
N
1,997,979 419,465 1,243,142 335,372
2014–2015b
2013–2014 Ratea
2,039,227 407,462 1,281,020 350,745
N
Ratea
2,070,545 391,497 1,313,634 365,414
N
2,083,490 380,273 1,329,859 373,358
336.2 754.9 159.7 434.8
6659 3503 1872 1284
333.3 835.1 150.6 382.9
6816 3207 1941 1668
334.2 787.1 151.5 475.6
6528 3010 1912 1606
315.3 768.8 145.6 439.5
4966 2567 1295 1104
2100.0 3532.3 1809.6 1265.4
39,319 15,278 20,295 3746
1967.9 3642.3 1632.6 1117.0
39,604 14,240 20,982 4382
1942.1 3494.8 1637.9 1249.3
35,160 12,540 18,706 3914
1698.1 3203.1 1424.0 1071.1
23,854 9258 11,902 2694
33,172.1 55,661.1 24,093.8 36,973.6
616,133 232,140 269,010 114,983
30,837.8 55,341.9 21,639.5 34,285.2
629,089 224,875 278,593 125,621
30,849.4 55,189.2 21,747.7 35,815.5
590,577 205,358 262,958 122,261
28,522.8 52,454.6 20,017.6 33,458.2
368,738 139,281 156,197 73,260
ED = emergency department. a Annual rates per 100,000 individuals with prevalent asthma. b Only the first half of the 2014–2015 influenza season is included in this study; therefore, annual rates were not calculated.
3. Results
season. While influenza A generally peaked in late December and influenza B generally peaked in late March/early April, the 2011–2012 flu season was distinct in that influenza A and B peaked around the same time—in mid-March. Although exact levels varied season-toseason, RSV activity tended to increase during winter months (December to March), with relatively broad, rather than sharp, peaks in activity. In general, observed health services use for asthma showed strong seasonality and differed across age groups (Figs. 2–4; dotted blue line). Among individuals aged 0–14, the September peak in asthma hospitalizations and ED visits was apparent in all years and was especially pronounced in 2014, with hospitalization counts almost doubling those
3.1. Descriptive data Approximately 2 million individuals with prevalent asthma were followed in this study: 400,000 aged 0–14, 1.3 million aged 15–59 and 350,000 aged 60 and above (Table 1). Table 1 gives observed counts and rates of asthma morbidity in the study cohort from 2010 to 2015. Viral activity was highly variable by flu season (Fig. 1). Influenza A predominated in the 2010–2011, 2012–2013 and 2014–2015 flu seasons, whereas influenza B predominated in the 2011–2012 flu season; the influenza types were nearly equal in activity in the 2013–2014 flu
Fig. 1. From September 2010 to March 2015, observed virus activity in Ontario, Canada for influenza A (thinner solid black line), influenza B (thicker solid grey line) and respiratory syncitial virus (RSV; dotted blue line). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) 106
Respiratory Medicine 155 (2019) 104–112
L.Y. Feldman, et al.
Fig. 2. From September 2010 to March 2015 in Ontario Canada, observed asthma morbidity (dotted blue line), predicted asthma morbidity using observed influenza A and B levels (“with influenza” scneario; thicker solid red line) and predicted asthma morbidity with influenza A and B levels artificially set to zero (“without influenza” scneario; thinner solid black line) for asthma hospitalizations in ages i) 0–14, ii) 15–59 and iii) 60+. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
107
Respiratory Medicine 155 (2019) 104–112
L.Y. Feldman, et al.
Fig. 3. From September 2010 to March 2015 in Ontario Canada, observed asthma morbidity (dotted blue line), predicted asthma morbidity using observed influenza A and B levels (“with influenza” scneario; thicker solid red line) and predicted asthma morbidity with influenza A and B levels artificially set to zero (“without influenza” scneario; thinner solid black line) for asthma emergency department visits in ages i) 0–14, ii) 15–59 and iii) 60+. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
108
Respiratory Medicine 155 (2019) 104–112
L.Y. Feldman, et al.
Fig. 4. From September 2010 to March 2015 in Ontario Canada, observed asthma morbidity (dotted blue line), predicted asthma morbidity using observed influenza A and B levels (“with influenza” scneario; thicker solid red line) and predicted asthma morbidity with influenza A and B levels artificially set to zero (“without influenza” scneario; thinner solid black line) for asthma outpatient physician office visits in ages i) 0–14, ii) 15–59 and iii) 60+. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
109
Respiratory Medicine 155 (2019) 104–112
L.Y. Feldman, et al.
Table 2 Regression coefficients for association between influenza A and B and asthma morbidity, from regression models, and annual estimates of influenza-associated asthma morbidity. Regression coefficient: Influenza A
Regression coefficient: Influenza B
Number of annual influenza-associated excess visitsa
Ratesb of annual influenza-associated excess visitsa
Estimate
Estimate
95% CI
Estimate
95% CI
Estimate
95% CI
−0.57 0.10 −0.04
(-1.71, 0.49) (-0.39, 0.58) (-0.40, 0.34)
−184.1 157.6 122.5
(-503.1, 132.2) (13.9, 295.8) (11.3, 230.4)
−44.7 12.5 35.7
(-122.0, 32.1) (1.1, 23.5) (3.3, 67.1)
−1.35 2.51 0.56
(-6.32, 3.10) (-0.31, 5.42) (-0.25, 1.37)
−360.4 1437.1 530.4
(-1795.6, 923.6) (590.2, 2287.3) (297.6, 766.1)
−87.4 114.1 154.6
(-435.5, 224.0) (46.9, 181.6) (86.7, 223.3)
−7.75 3.46 4.44
(-41.56, 16.95) (-20.96, 21.20) (-7.26, 14.27)
1908.2 5239.7 3519.8
(-6580.1, 9583.1) (-1289.3, 10,893.9) (271.2, 6442.1)
462.8 415.9 1025.7
(-1596.1, 2324.5) (-102.3, 864.8) (79.0, 1877.3)
95% CI
Hospitalizations 0 to 14 −0.36 (-1.24, 0.62) 15 to 59 0.55 (0.13, 0.94) 60 and above 0.50 (0.19, 0.80) ED Visits 0 to 14 −0.54 (-4.45, 2.93) 15 to 59 4.01 (1.54, 6.35) 60 and above 1.72 (1.09, 2.36) Outpatient Physician Office Visits 0 to 14 12.44 (-8.41, 35.46) 15 to 59 18.27 (2.01, 35.83) 60 and above 10.92 (2.27, 19.60)
95%CI = 95% confidence interval; ED = emergency department. a Estimates and 95% CI obtained using predicted asthma morbidity from September 1, 2010 to August 31, 2014. b Rates per 100,000 individuals with prevalent asthma.
influenza was associated with excess asthma hospitalizations and asthma ED visits for ages 15–59 and 60+, and excess asthma outpatient physician office visits for ages 60+. Asthma morbidity for those aged 0–14 years tended to increase in September and was not significantly associated with influenza activity.
seen in previous years and ED visits being about 50% higher (Figs. 2 and 3). Hospitalizations and ED visits appeared to peak in both September and winter months among those aged 15–59 and in winter months for those aged 60+ (Figs. 2 and 3). Decreases in outpatient physician office visits were noted in the 52nd week of each year (last week of December) for those aged 15–59, 60+ and, to a lesser extent, 0–14 (Fig. 4). All measures of health services use appeared to dip in the summer months.
4.1. Interpretation of findings in broader context of existing literature Our findings demonstrate a differential relationship between influenza and asthma morbidity across age groups, which may shed light on the null findings seen in clinical trials of influenza vaccination. In their recent Cochrane review of the efficacy of influenza vaccination in protecting against influenza-associated asthma exacerbation, Cates and Rowe found no evidence of a protective effect of influenza vaccination [4]. Notably, the only study that could be included in this meta-analysis was a Dutch trial of inactivated influenza vaccine in 696 children aged 6–18, which showed no significant reduction in the frequency, duration or severity of asthma exacerbations, although symptom scores were improved among vaccinated children during weeks with circulating influenza [21,22]. Similar to the findings of Lee et al., who studied asthma ED visits in Korea [23], and Trinh et al., who studied asthma hospitalizations in the state of New York [24], we did not detect any excess influenza-associated asthma morbidity in the 0–14 age group. In this age group, the September peak and holidays appeared to have a substantial impact on asthma morbidity; the findings of Lee et al. suggest that rhinovirus, rather than influenza, is more likely to contribute to asthma morbidity in this age group [23]. Given that we did not detect a signal of influenza-associated asthma morbidity in the 0–14 age group, it is not surprising that the previous trial of influenza vaccination in children with asthma resulted in a null finding. Consistent with the findings of Lee et al. and Trinh et al. [23,24], we detected excess influenza-associated asthma morbidity among individuals aged 15–59 and 60+. While this suggests that influenza-focused interventions among individuals aged 15–59 and 60+ might result in reductions in asthma hospitalizations and ED visits, this has yet to be demonstrated in clinical trials; two clinical trials, one of 328 adults and the other of 25 adults, could not be included in the Cochrane review by Cates and Rowe, as very few cases of influenza infection had been detected [4]. In this population-based ecological time-series analysis, the relationship between influenza activity and asthma morbidity was investigated in a real-world setting—notably, a setting in which a proportion of the study cohort was vaccinated against influenza. In Ontario, influenza vaccination is offered free of charge to anyone 6 months of age or older who is living, working or going to school in the
3.2. Influenza-attributable asthma morbidity In adjusted models, weekly influenza A activity was positively associated with hospitalizations, ED visits and outpatient physician office visits for asthma among individuals aged 15–59 and 60+ (Table 2). Influenza B activity was not significantly associated with asthma morbidity in any age group. The covariates selected in the final models for each health outcome and age group are listed in Appendix eTable 1. Among individuals aged 15–59, influenza was associated with an annual excess of 157.6 hospitalizations (95%CI: 13.9–295.8) and 1437.1 ED visits for asthma (95%CI: 590.2–2287.3), which corresponds to annual rates of 12.5 hospitalizations (95%CI: 1.1–23.5) and 114.1 ED visits (95%CI: 46.9–181.6) for asthma, per 100,000 individuals with prevalent asthma (Table 2). Among individuals aged 60+, influenza was associated with an annual excess of 122.5 hospitalizations (95%CI: 11.3–230.4), 530.4 ED visits (95%CI: 297.6–766.1) and 3519.8 outpatient physician office visits (95%CI: 271.2–6442.1) for asthma (Table 2). This corresponds to excess rates of 35.7 hospitalizations (95%CI: 3.3–67.1), 154.6 ED visits (95%CI: 86.7–223.3) and 1025.7 outpatient physician office visits (95%CI: 79.0–1877.3) for asthma, per 100,000 individuals with prevalent asthma (Table 2). For all other combinations of age groups and asthma health services use types, significant counts and rates of excess asthma morbidity were not detected. These relationships are depicted visually in Figs. 2–4 as the difference in the area under the curve between “with influenza” scenario (thicker solid red line) and “without influenza” scenario (thinner solid black line). Although the point estimates in Table 2 and graphs in Figs. 2 and 3 suggest a decrease in hospitalizations and ED visits for asthma associated with influenza activity for individuals aged 0–14, 95%CI for these estimates were wide and included 0 (no decrease). 4. Discussion In this ecological time series analysis of a population-based cohort of individuals with prevalent asthma followed from 2010 to 2015, 110
Respiratory Medicine 155 (2019) 104–112
L.Y. Feldman, et al.
of this study may not be generalizable to geographic regions with different climates (thus, different conditions for circulation of viruses and possibly different risk factors for asthma exacerbation), nor may they be generalizable to settings where health care or influenza vaccination access differ. In a population less protected by influenza vaccination, the magnitude of influenza-attributable asthma morbidity may be more pronounced than that seen in this study.
province. Influenza vaccination can be delivered in a variety of settings, and since there is no provincial registry recording influenza vaccinations, only a proportion of these are captured in health administrative databases. For that reason, Ontario health administrative data are not optimal for ascertaining individual vaccination status [25,26]. In a subset of the population for whom national health survey data were available, it was determined that 44% of individuals in Ontario who reported having asthma also reported having received an influenza vaccination in the prior 12 months [26]. Vaccination rates ranged in the general Ontario population from around 25% in the youngest age group studied (adolescents 12 years and older) to around 70% in the oldest age group (65+) [25,26]. Despite relatively high rates of influenza vaccination among older adults aged 65+ in the general population, we detected a signal of excess influenza-associated asthma hospitalizations, ED visits and outpatient physician office visits among our cohort of individuals aged 60+ with prevalent asthma. This finding may be explained by the fact that older adults are generally at higher risk of influenza complications, both due to a greater burden of medical risk factors such as diabetes and cardiovascular conditions, as well as risks associated with older age itself [27].
4.3. Conclusions This study demonstrates a substantial burden of influenza-associated asthma morbidity among adults and older adults with prevalent asthma in Ontario, whereas patterns of asthma morbidity among children appear to be more strongly linked to other seasonal trends, such as the annual September peak. Our study sheds light on the null findings of clinical trials of the effectiveness of influenza vaccination in protecting against asthma exacerbation; they suggest that adults and older adults may be important targets of further influenza-focused clinical research and intervention, in order to reduce asthma morbidity. Declaration of interests
4.2. Strengths and limitations The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Strengths of this study include its large sample size, long duration of follow-up, methodological correction for temporal autocorrelation and broad population-level applicability. This study followed a cohort over 2 million individuals with prevalent asthma covered by Ontario's universal health care system across 4 complete influenza seasons. The long duration of follow-up included periods of highly variable influenza activity, thus allowing us to tease out the effects of influenza activity on asthma morbidity, as distinct from seasonal and long-term trends. The methodology used in this study allowed us to adjust our measures of uncertainty, using bootstrap sampling to account for temporal autocorrelation in model residuals. Finally, our use of population-level health administrative data allowed us to study real-world dynamics of asthma morbidity as they relate to measured influenza activity; we are therefore able to provide complementary information to clinical trials, which study these relationships in more highly controlled, selected populations. Interpretation of our study findings is subject to certain limitations, including the use of an ecological level analysis, the possibility of residual confounding and uncertainty in generalizing our results to other geographic regions. In this ecological time series analysis, in which exposures and outcomes are correlated at an aggregate level, it cannot be determined whether individual instances of health services use for asthma were accompanied by influenza infection or not. Therefore, we cannot make inference as to the relationship between influenza infection and asthma morbidity at the individual level. However, we are able to broadly estimate age-specific annual excess asthma morbidity attributable to influenza. As a general limitation of observational studies, there is a possibility of residual confounding. In order to address this limitation, we have controlled for covariates included in a number of national and international studies of the population-level impacts of influenza (e.g., seasonal trend, long-term trend, mean temperature, RSV activity [5,6]), as well as certain covariates known to be particularly important in Canada (e.g., the annual September Peak [17,18] and statutory holidays [10–12]). That said, certain factors, such as meteorological variables apart from temperature (e.g., humidity), may have some impact on patterns of both influenza activity and asthma morbidity. Further, individual-level clinical factors such as asthma severity and duration, use of asthma medications and presence of comorbidities could impact the association between influenza activity and asthma morbidity; these factors were not considered in this study. Lastly, Ontario is located in a temperate region of the Northern hemisphere and has a universal health care system that includes free influenza vaccination for those 6 months of age or older. Therefore, results
Acknowledgement The Ontario Asthma Surveillance Information System (OASIS) is funded by the Ontario Ministry of Health and Long-Term Care (MOHLTC). Data are provided by Public Health Ontario (PHO) and ICES, which is funded by an annual grant from the Ontario MOHLTC. The opinions, results and conclusions reported in this paper are those of the authors and are independent from the funding source. No endorsement by PHO, ICES or the Ontario MOHLTC is intended or should be inferred. Parts of this material are based on data and information compiled and provided by the Canadian Institute for Health Information (CIHI). However, the analyses, conclusions, opinions and statements expressed herein are those of the authors and not necessarily those of CIHI. The authors have no other relevant sources of funding to declare. We would like to thank Kyla Jamieson at the University of Calgary and Jeff Kwong at ICES for their insights during the design of this study, Adriana Peci and Jonathan Gubbay at PHO for helpful commentary and assistance with data acquisition, and Eleanor Pullenayegum at the Hospital for Sick Children and Howard Chang at Emory University for their guidance regarding the statistical methodology used in this study. Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.rmed.2019.07.017. References [1] S.L. Johnston, P.K. Pattemore, G. Sanderson, S. Smith, F. Lampe, L. Josephs, P. Symington, S. O'Toole, S.H. Myint, D.A.J. Tyrrell, S.T. Holgate, Community study of role of viral infections in exacerbations of asthma in 9-11 year old children, BMJ 310 (6989) (1995) 1225–1229. [2] K.G. Nicholson, J. Kent, D.C. Ireland, Respiratory viruses and exacerbations of asthma in adults, BMJ 307 (6910) (1993) 982–986. [3] W.W. Busse, R.F. Lemanske Jr., J.E. Gern, Role of viral respiratory infections in asthma and asthma exacerbations, Lancet 376 (9743) (2010) 826–834. [4] C.J. Cates, B.H. Rowe, Vaccines for preventing influenza in people with asthma, Cochrane Database Syst. Rev. 2 (2013) CD000364. [5] E. Goldstein, C. Viboud, V. Charu, M. Lipsitch, Improving the estimation of influenza-related mortality over a seasonal baseline, Epidemiology 23 (6) (2012) 829–838. [6] A.D. Iuliano, K.M. Roguski, H.H. Chang, D.J. Muscatello, R. Palekar, S. Tempia,
111
Respiratory Medicine 155 (2019) 104–112
L.Y. Feldman, et al.
[7]
[8] [9]
[10]
[11]
[12]
[13]
[14]
[15]
[16] [17]
C. Cohen, J.M. Gran, D. Schanzer, B.J. Cowling, Estimates of global seasonal influenza-associated respiratory mortality: a modelling study, Lancet 391 (10127) (2018) 1285–1300. L. Li, J.Y. Wong, P. Wu, H.S. Bond, E.H. Lau, S.G. Sullivan, B.J. Cowling, Heterogeneity in estimates of the impact of influenza on population mortality: a systematic review, Am. J. Epidemiol. 187 (2) (2018) 378–388. D.L. Schanzer, C. Sevenhuysen, B. Winchester, T. Mersereau, Estimating influenza deaths in Canada, 1992–2009, PLoS One 8 (11) (2013) e80481. W.W. Thompson, D.K. Shay, E. Weintraub, L. Brammer, C.B. Bridges, N.J. Cox, K. Fukuda, Influenza-associated hospitalizations in the United States, J. Am. Med. Assoc. 292 (11) (2004) 1333–1340. D.L. Schanzer, J.M. Langley, T.W. Tam, Hospitalization attributable to influenza and other viral respiratory illnesses in Canadian children, Pediatr. Infect. Dis. J. 25 (9) (2006) 795–800. D.L. Schanzer, J.M. Langley, T.W. Tam, Role of influenza and other respiratory viruses in admissions of adults to Canadian hospitals, Influenza Other Respir. Viruses 2 (1) (2008) 1–8. D.L. Schanzer, B. Schwartz, Impact of seasonal and pandemic influenza on emergency department visits, 2003–2010, Ontario, Canada, Acad. Emerg. Med. 20 (4) (2013) 388–397. A.S. Gershon, C. Wang, J. Guan, J. Vasilevska-Ristovska, L. Cicutto, T. To, Identifying patients with physician-diagnosed asthma in health administrative databases, Can. Respir. J. 16 (6) (2009) 183–188. T. To, S. Dell, P.T. Dick, L. Cicutto, J.K. Harris, I.B. MacLusky, M. Tassoudji, Case verification of children with asthma in Ontario, Pediatr. Allergy Immunol. 17 (1) (2006) 69–76. Public Health Ontario, Public Health Ontario Laboratory Respiratory Pathogen Surveillance Reports, (2016) https://www.publichealthontario.ca/en/ ServicesAndTools/LaboratoryServices/Pages/PHO-Laboratories-surveillanceupdates.aspx , Accessed date: 15 October 2018. Public Health Ontario, Personal Communication, (2018). N.W. Johnston, S.L. Johnston, J.M. Duncan, J.M. Greene, T. Kebadze, P.K. Keith, M. Roy, S. Waserman, M.R. Sears, The September epidemic of asthma exacerbations
[18]
[19]
[20] [21]
[22]
[23]
[24]
[25] [26]
[27]
112
in children: a search for etiology, J. Allergy Clin. Immunol. 115 (1) (2005) 132–138. K. Larsen, J. Zhu, L.Y. Feldman, J. Simatovic, S. Dell, A.S. Gershon, T. To, The annual September peak in asthma exacerbation rates. Still a reality? Ann. Am. Thorac. Soc. 13 (2) (2016) 231–239. J.H. Kim, M. Yeasmin, The size and power of the bias-corrected bootstrap test for regression models with autocorrelated errors, Comput. Econ. 25 (3) (2005) 255–267. R Core Team, R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna, Austria, 2017. H.J. Bueving, R.M. Bernsen, J.C. de Jongste, L.W. van Suijlekom-Smit, G.F. Rimmelzwaan, A.D. Osterhaus, M.P. Rutten-van Mölken, S. Thomas, J.C. van der Wouden, Influenza vaccination in children with asthma: randomized doubleblind placebo-controlled trial, Am. J. Respir. Crit. Care Med. 169 (4) (2004) 488–493. H. Bueving, J. van der Wouden, H. Raat, R. Bernsen, J. de Jongste, L. van SuijlekomSmit, A. Osterhaus, G. Rimmelzwaan, M. Rutten-van Mölken, S. Thomas, Influenza vaccination in asthmatic children: effects on quality of life and symptoms, Eur. Respir. J. 24 (6) (2004) 925–931. T. Jefferson, D. Rivetti, A. Rivetti, M. Rudin, C. Di Pietrantonj, V. Demicheli, Efficacy and effectiveness of influenza vaccines in elderly people: a systematic review, Lancet 366 (9492) (2005) 1165–1174. K. Nichol, K. Margolis, J. Wuorenma, T. Von Sternberg, The efficacy and cost effectiveness of vaccination against influenza among elderly persons living in the community, N. Engl. J. Med. 331 (12) (1994) 778–784. J.C. Kwong, D.G. Manuel, Using OHIP physician billing claims to ascertain individual influenza vaccination status, Vaccine 25 (7) (2007) 1270–1274. K.L. Schwartz, N. Jembere, M.A. Campitelli, S.A. Buchan, H. Chung, J.C. Kwong, Using physician billing claims from the Ontario Health Insurance Plan to determine individual influenza vaccination status: an updated validation study, CMAJ Open 4 (3) (2016) E463–E470. Ontario Agency for Health Protection and Promotion (Public Health Ontario), The Relationship between Influenza Medical Risk Factors and Age, (2017) Toronto, ON.
Contents lists available at ScienceDirect
Respiratory Medicine journal homepage: www.elsevier.com/locate/rmed
Estimating age-specific influenza-associated asthma morbidity in Ontario, Canada
T
Laura Y. Feldmana, Jingqin Zhua,b, Teresa Toa,b,c,* a
Child Health Evaluative Sciences, The Hospital for Sick Children, Toronto, Ontario, Canada ICES, Toronto, Ontario, Canada c Dalla Lana School of Public Health, University of Toronto, Toronto, Ontario, Canada b
A R T I C LE I N FO
A B S T R A C T
Keywords: Asthma Influenza Health services Health administrative data
Background: There is a need to quantify the potential benefits of influenza-focused interventions in reducing asthma morbidity at a population level. This study aims to estimate age-specific annual excess asthma morbidity attributable to influenza in Ontario, Canada. Methods: Weekly counts of hospitalizations, emergency department (ED) visits and outpatient physician office visits for asthma were obtained from health administrative data in Ontario from 2010 to 2015, for ages 0–14, 15–59 and 60+. Asthma morbidity was modelled as a function of influenza A and B activity using linear regression, controlling for seasonal and long-term trend, mean temperature and respiratory syncytial virus. Excess asthma morbidity attributable to influenza was calculated as the difference between full model predictions and model predictions with influenza A and B variables set to 0. Results: Annually, influenza was associated with the following rates of excess asthma morbidity, per 100,000 people with prevalent asthma: 12.5 hospitalizations for ages 15–59 (95% confidence interval (CI): 1.1–23.5); 35.7 hospitalizations for ages 60+ (95% CI: 3.3–67.1); 114.1 ED visits for ages 15–59 (95% CI: 46.9–181.6); 154.6 ED visits for ages 60+ (95% CI: 86.7–223.3); and 1025.7 outpatient physician office visits for ages 60+ (95% CI: 79.0–1877.3). Conclusions: Influenza was associated with excess asthma hospitalizations and ED visits for ages 15–59 and 60+ and outpatient physician office visits for ages 60+. Individuals with asthma aged 15–59 and 60+ might be important targets for influenza-focused interventions, to reduce asthma morbidity at the population level.
1. Introduction Observational studies have suggested that influenza infection may be an important risk factor for asthma exacerbation in children and adults; however, clinical trials have failed to demonstrate efficacy of influenza vaccination in protecting against asthma exacerbation. In the 1990s, cohort studies by Johnston et al. and Nicholson et al. demonstrated that instances of acute asthma exacerbation in children [1] and adults [2] were very often accompanied by respiratory viruses. While human rhinovirus was the most commonly detected virus in these cohort studies and remains a pathogen of interest for both asthma development and exacerbation [3], influenza A and B were implicated in these early studies as possible triggers of asthma exacerbation. In a recent Cochrane review, Cates and Rowe investigated the efficacy of influenza vaccination in protecting against influenza-associated asthma exacerbation in children and adults [4]. Based on a single clinical trial
*
in children, they found no evidence of a protective effect. However, they noted that uncertainty in this question remains, due to the infrequency of influenza illness among individuals with asthma, as well as the small sample sizes and short follow-up times of existing clinical trials. Population-level health administrative data and virus surveillance data consist of large datasets, long follow-up times and objective, routinely collected outcome measures. Although population-level health administrative data generally lack individual-level data on influenza vaccination status and viral serotyping, and are therefore not suitable data sources to directly investigate the protective effects of influenza vaccination on asthma exacerbation, they can be used as complementary data sources to characterize the relationship between influenza and asthma morbidity and to identify subsets of the population experiencing a greater burden of influenza-attributable asthma morbidity.
Corresponding author. Child Health Evaluative Sciences, The Hospital for Sick Children, 686 Bay Street, Toronto, Ontario, M5G 0A4, Canada. E-mail address: [email protected] (T. To).
https://doi.org/10.1016/j.rmed.2019.07.017 Received 13 March 2019; Received in revised form 16 July 2019; Accepted 17 July 2019 Available online 17 July 2019 0954-6111/ © 2019 Published by Elsevier Ltd.
Respiratory Medicine 155 (2019) 104–112
L.Y. Feldman, et al.
2.3.2. Outcomes The main outcomes in this study were weekly counts of hospitalizations, ED and outpatient physician office visits for asthma. For each week from September 2010 to March 2015, provincial total counts of these health services were collected using International Statistical Classification of Diseases and Related Health Problems (ICD) 9 code 493 and ICD-10 codes J45 and J46. In order to exclude visits for which asthma was merely noted as a comorbidity, only main diagnostic codes were considered in this analysis. For hospitalizations, main diagnostic codes refer to conditions that contributed significantly to the treatment or length of stay. For ED visits, the main diagnostic code refers to the condition that contributed most significantly to the visit. For outpatient physician office visits, only one diagnostic code, which is independent of billing codes, is given for each visit; this was used as the main diagnostic code in this analysis.
Given the universal health care system of the province of Ontario and its regular surveillance of influenza and other viruses, it is an ideal region in which to measure the burden of asthma morbidity attributable to influenza among a cohort of individuals with prevalent asthma. Ecological time series methods have been routinely used to estimate mortality burdens associated with influenza at the population level [5–8] and these methods have, in recent years, been adapted to estimate morbidity burdens [9–12]. Using provincial health administrative data and virus surveillance data, as well as established methodology for measuring influenza-attributable morbidity, we aimed to estimate agespecific annual counts and rates of influenza-attributable asthma morbidity in Ontario, Canada.
2. Methods 2.1. Setting, population & study design
2.3.3. Effect modifier All analyses were stratified by three age groups: 0–14, 15–59 and 60+. This allowed for the relationship between the exposure and outcomes to be analyzed separately, for each age group.
This study was conducted in Ontario, Canada, using provincial health administrative data and respiratory virus surveillance data to investigate the burden of influenza on asthma morbidity in a population-based cohort of individuals with prevalent asthma. The study population included all individuals aged 0–99 living in Ontario who met a validated health administrative definition for asthma: at least 1 hospitalization for asthma and/or 2 ambulatory care visits for asthma in 2 consecutive years [13,14]. Individuals were excluded if they lacked a valid health card number or had missing/invalid information on residence, age or sex. In this ecological time series analysis, weekly counts of health services use (hospitalizations, ED visits and outpatient physician office visits) for asthma were regressed against weekly influenza A and B viral activity, controlling for the effects of seasonality, long-term trend, mean temperature and respiratory syncytial virus (RSV). Three age groups of individuals with asthma were studied: 0–14, 15–59 and 60+. Approval to conduct this study was obtained from ICES and the Research Ethics Board at the Hospital for Sick Children, Toronto, Canada. Individual consent was not required, as this study utilized deidentified health administrative data and analyses were performed at an aggregate, ecological level.
2.3.4. Covariates Covariates included the weekly percentage of positive tests for respiratory syncytial virus (RSV), mean weekly temperature, seasonal trend (using restricted cubic splines with a period of 52 weeks), longterm trend (using polynomial basis functions), the September peak and statutory holidays. Following the methodology of Goldstein et al., one week in July was removed from the 2012–2013 influenza season, which contained 53 weeks; this ensured that all influenza seasons had a duration of exactly 52 weeks [5]. Covariates were chosen to align with existing methodology for estimating influenza-attributable morbidity and to account for patterns of asthma morbidity in Canada, where asthma morbidity among children is known to peak in September [7,10–12,17,18]. Covariates, and the methods used for covariate selection, are described in detail in the appendix.
2.4. Statistical analysis Ordinary least squares (OLS) regression was used to model asthma morbidity (dependent variable) as a function of influenza activity (independent variable). Separate models were fitted for each age group and each measure of asthma morbidity (hospitalizations, ED visits, outpatient physician office visits). Models were selected based on minimization of the Akaike Information Criterion (AIC). Model fitting is described in detail in the appendix. In this time series study, model residuals showed temporal autocorrelation. Therefore, 95% confidence intervals (CI) were corrected for temporal autocorrelation by assuming an autoregressive structure of order 2 (AR2) [5,6,19]. In order to calculate the magnitude of asthma morbidity associated with influenza, two sets of predictions were made for each model: predicted counts of asthma morbidity with influenza A and B levels set to their observed values (“with influenza” scenario) and predicted counts of asthma morbidity with influenza A and B levels set to 0 (“without influenza” scenario). Influenza-associated excess asthma morbidity was calculated as the difference between the “with influenza” scenario and the “without influenza” scenario. Average annual excess influenza-associated asthma morbidity was calculated for influenza seasons 2010–2011, 2011–2012, 2012–2013 and 2013–2014. Annual excess counts were divided by the average size of the asthma prevalent population for each age group during the study period and multiplied by 100,000, to give annual rates of excess influenza-associated asthma morbidity per 100,000 individuals with prevalent asthma. All analyses were conducted in R version 3.4.3 [20].
2.2. Data sources Health administrative data housed at ICES were used to assemble the study cohort and to measure the outcome variables and certain covariates. The following datasets, all housed at ICES, were used: the Canadian Institute of Health Information (CIHI) Discharge Abstract Database (DAD) for hospitalizations; the National Ambulatory Care Reporting System (NACRS) for ED visits; the Ontario Health Insurance Plan (OHIP) Database for outpatient physician office visits; and the Registered Persons Database (RPDB) for individuals’ date of birth. Laboratory testing data were obtained from Public Health Ontario (PHO). PHO operates 11 laboratories across Ontario, which perform approximately 60% of respiratory viral specimen testing across the province [15,16]. These data reflect all specimens tested at the PHO laboratories, which include specimens from patients in outpatient, ED, hospital, intensive care unit and outbreak settings.
2.3. Exposure and outcome measures 2.3.1. Exposure The main exposure in this study was the weekly percentage of positive tests for influenza A and for influenza B (number of positive tests divided by total tests, for a given virus) in Ontario from September 2010 to March 2015. The percentage of positive laboratory tests for a given virus was used as a proxy variable for viral activity [7]. 105
Respiratory Medicine 155 (2019) 104–112
L.Y. Feldman, et al.
Table 1 Observed counts and rates of asthma morbidity in the asthma prevalent population across influenza seasons. Influenza Season 2010–2011 N Asthma Prevalence Total 1,953,018 0 to 14 430,660 15 to 59 1,201,283 60 and above 321,075 Hospitalizations Total 6566 0 to 14 3251 15 to 59 1919 60 and above 1396 ED Visits Total 41,014 0 to 14 15,212 15 to 59 21,739 60 and above 4063 Outpatient Physician Office Visits Total 647,858 239,710 0 to 14 15 to 59 289,435 60 and above 118,713
2011–2012 Ratea
N
2012–2013 Ratea
N
1,997,979 419,465 1,243,142 335,372
2014–2015b
2013–2014 Ratea
2,039,227 407,462 1,281,020 350,745
N
Ratea
2,070,545 391,497 1,313,634 365,414
N
2,083,490 380,273 1,329,859 373,358
336.2 754.9 159.7 434.8
6659 3503 1872 1284
333.3 835.1 150.6 382.9
6816 3207 1941 1668
334.2 787.1 151.5 475.6
6528 3010 1912 1606
315.3 768.8 145.6 439.5
4966 2567 1295 1104
2100.0 3532.3 1809.6 1265.4
39,319 15,278 20,295 3746
1967.9 3642.3 1632.6 1117.0
39,604 14,240 20,982 4382
1942.1 3494.8 1637.9 1249.3
35,160 12,540 18,706 3914
1698.1 3203.1 1424.0 1071.1
23,854 9258 11,902 2694
33,172.1 55,661.1 24,093.8 36,973.6
616,133 232,140 269,010 114,983
30,837.8 55,341.9 21,639.5 34,285.2
629,089 224,875 278,593 125,621
30,849.4 55,189.2 21,747.7 35,815.5
590,577 205,358 262,958 122,261
28,522.8 52,454.6 20,017.6 33,458.2
368,738 139,281 156,197 73,260
ED = emergency department. a Annual rates per 100,000 individuals with prevalent asthma. b Only the first half of the 2014–2015 influenza season is included in this study; therefore, annual rates were not calculated.
3. Results
season. While influenza A generally peaked in late December and influenza B generally peaked in late March/early April, the 2011–2012 flu season was distinct in that influenza A and B peaked around the same time—in mid-March. Although exact levels varied season-toseason, RSV activity tended to increase during winter months (December to March), with relatively broad, rather than sharp, peaks in activity. In general, observed health services use for asthma showed strong seasonality and differed across age groups (Figs. 2–4; dotted blue line). Among individuals aged 0–14, the September peak in asthma hospitalizations and ED visits was apparent in all years and was especially pronounced in 2014, with hospitalization counts almost doubling those
3.1. Descriptive data Approximately 2 million individuals with prevalent asthma were followed in this study: 400,000 aged 0–14, 1.3 million aged 15–59 and 350,000 aged 60 and above (Table 1). Table 1 gives observed counts and rates of asthma morbidity in the study cohort from 2010 to 2015. Viral activity was highly variable by flu season (Fig. 1). Influenza A predominated in the 2010–2011, 2012–2013 and 2014–2015 flu seasons, whereas influenza B predominated in the 2011–2012 flu season; the influenza types were nearly equal in activity in the 2013–2014 flu
Fig. 1. From September 2010 to March 2015, observed virus activity in Ontario, Canada for influenza A (thinner solid black line), influenza B (thicker solid grey line) and respiratory syncitial virus (RSV; dotted blue line). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) 106
Respiratory Medicine 155 (2019) 104–112
L.Y. Feldman, et al.
Fig. 2. From September 2010 to March 2015 in Ontario Canada, observed asthma morbidity (dotted blue line), predicted asthma morbidity using observed influenza A and B levels (“with influenza” scneario; thicker solid red line) and predicted asthma morbidity with influenza A and B levels artificially set to zero (“without influenza” scneario; thinner solid black line) for asthma hospitalizations in ages i) 0–14, ii) 15–59 and iii) 60+. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
107
Respiratory Medicine 155 (2019) 104–112
L.Y. Feldman, et al.
Fig. 3. From September 2010 to March 2015 in Ontario Canada, observed asthma morbidity (dotted blue line), predicted asthma morbidity using observed influenza A and B levels (“with influenza” scneario; thicker solid red line) and predicted asthma morbidity with influenza A and B levels artificially set to zero (“without influenza” scneario; thinner solid black line) for asthma emergency department visits in ages i) 0–14, ii) 15–59 and iii) 60+. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
108
Respiratory Medicine 155 (2019) 104–112
L.Y. Feldman, et al.
Fig. 4. From September 2010 to March 2015 in Ontario Canada, observed asthma morbidity (dotted blue line), predicted asthma morbidity using observed influenza A and B levels (“with influenza” scneario; thicker solid red line) and predicted asthma morbidity with influenza A and B levels artificially set to zero (“without influenza” scneario; thinner solid black line) for asthma outpatient physician office visits in ages i) 0–14, ii) 15–59 and iii) 60+. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
109
Respiratory Medicine 155 (2019) 104–112
L.Y. Feldman, et al.
Table 2 Regression coefficients for association between influenza A and B and asthma morbidity, from regression models, and annual estimates of influenza-associated asthma morbidity. Regression coefficient: Influenza A
Regression coefficient: Influenza B
Number of annual influenza-associated excess visitsa
Ratesb of annual influenza-associated excess visitsa
Estimate
Estimate
95% CI
Estimate
95% CI
Estimate
95% CI
−0.57 0.10 −0.04
(-1.71, 0.49) (-0.39, 0.58) (-0.40, 0.34)
−184.1 157.6 122.5
(-503.1, 132.2) (13.9, 295.8) (11.3, 230.4)
−44.7 12.5 35.7
(-122.0, 32.1) (1.1, 23.5) (3.3, 67.1)
−1.35 2.51 0.56
(-6.32, 3.10) (-0.31, 5.42) (-0.25, 1.37)
−360.4 1437.1 530.4
(-1795.6, 923.6) (590.2, 2287.3) (297.6, 766.1)
−87.4 114.1 154.6
(-435.5, 224.0) (46.9, 181.6) (86.7, 223.3)
−7.75 3.46 4.44
(-41.56, 16.95) (-20.96, 21.20) (-7.26, 14.27)
1908.2 5239.7 3519.8
(-6580.1, 9583.1) (-1289.3, 10,893.9) (271.2, 6442.1)
462.8 415.9 1025.7
(-1596.1, 2324.5) (-102.3, 864.8) (79.0, 1877.3)
95% CI
Hospitalizations 0 to 14 −0.36 (-1.24, 0.62) 15 to 59 0.55 (0.13, 0.94) 60 and above 0.50 (0.19, 0.80) ED Visits 0 to 14 −0.54 (-4.45, 2.93) 15 to 59 4.01 (1.54, 6.35) 60 and above 1.72 (1.09, 2.36) Outpatient Physician Office Visits 0 to 14 12.44 (-8.41, 35.46) 15 to 59 18.27 (2.01, 35.83) 60 and above 10.92 (2.27, 19.60)
95%CI = 95% confidence interval; ED = emergency department. a Estimates and 95% CI obtained using predicted asthma morbidity from September 1, 2010 to August 31, 2014. b Rates per 100,000 individuals with prevalent asthma.
influenza was associated with excess asthma hospitalizations and asthma ED visits for ages 15–59 and 60+, and excess asthma outpatient physician office visits for ages 60+. Asthma morbidity for those aged 0–14 years tended to increase in September and was not significantly associated with influenza activity.
seen in previous years and ED visits being about 50% higher (Figs. 2 and 3). Hospitalizations and ED visits appeared to peak in both September and winter months among those aged 15–59 and in winter months for those aged 60+ (Figs. 2 and 3). Decreases in outpatient physician office visits were noted in the 52nd week of each year (last week of December) for those aged 15–59, 60+ and, to a lesser extent, 0–14 (Fig. 4). All measures of health services use appeared to dip in the summer months.
4.1. Interpretation of findings in broader context of existing literature Our findings demonstrate a differential relationship between influenza and asthma morbidity across age groups, which may shed light on the null findings seen in clinical trials of influenza vaccination. In their recent Cochrane review of the efficacy of influenza vaccination in protecting against influenza-associated asthma exacerbation, Cates and Rowe found no evidence of a protective effect of influenza vaccination [4]. Notably, the only study that could be included in this meta-analysis was a Dutch trial of inactivated influenza vaccine in 696 children aged 6–18, which showed no significant reduction in the frequency, duration or severity of asthma exacerbations, although symptom scores were improved among vaccinated children during weeks with circulating influenza [21,22]. Similar to the findings of Lee et al., who studied asthma ED visits in Korea [23], and Trinh et al., who studied asthma hospitalizations in the state of New York [24], we did not detect any excess influenza-associated asthma morbidity in the 0–14 age group. In this age group, the September peak and holidays appeared to have a substantial impact on asthma morbidity; the findings of Lee et al. suggest that rhinovirus, rather than influenza, is more likely to contribute to asthma morbidity in this age group [23]. Given that we did not detect a signal of influenza-associated asthma morbidity in the 0–14 age group, it is not surprising that the previous trial of influenza vaccination in children with asthma resulted in a null finding. Consistent with the findings of Lee et al. and Trinh et al. [23,24], we detected excess influenza-associated asthma morbidity among individuals aged 15–59 and 60+. While this suggests that influenza-focused interventions among individuals aged 15–59 and 60+ might result in reductions in asthma hospitalizations and ED visits, this has yet to be demonstrated in clinical trials; two clinical trials, one of 328 adults and the other of 25 adults, could not be included in the Cochrane review by Cates and Rowe, as very few cases of influenza infection had been detected [4]. In this population-based ecological time-series analysis, the relationship between influenza activity and asthma morbidity was investigated in a real-world setting—notably, a setting in which a proportion of the study cohort was vaccinated against influenza. In Ontario, influenza vaccination is offered free of charge to anyone 6 months of age or older who is living, working or going to school in the
3.2. Influenza-attributable asthma morbidity In adjusted models, weekly influenza A activity was positively associated with hospitalizations, ED visits and outpatient physician office visits for asthma among individuals aged 15–59 and 60+ (Table 2). Influenza B activity was not significantly associated with asthma morbidity in any age group. The covariates selected in the final models for each health outcome and age group are listed in Appendix eTable 1. Among individuals aged 15–59, influenza was associated with an annual excess of 157.6 hospitalizations (95%CI: 13.9–295.8) and 1437.1 ED visits for asthma (95%CI: 590.2–2287.3), which corresponds to annual rates of 12.5 hospitalizations (95%CI: 1.1–23.5) and 114.1 ED visits (95%CI: 46.9–181.6) for asthma, per 100,000 individuals with prevalent asthma (Table 2). Among individuals aged 60+, influenza was associated with an annual excess of 122.5 hospitalizations (95%CI: 11.3–230.4), 530.4 ED visits (95%CI: 297.6–766.1) and 3519.8 outpatient physician office visits (95%CI: 271.2–6442.1) for asthma (Table 2). This corresponds to excess rates of 35.7 hospitalizations (95%CI: 3.3–67.1), 154.6 ED visits (95%CI: 86.7–223.3) and 1025.7 outpatient physician office visits (95%CI: 79.0–1877.3) for asthma, per 100,000 individuals with prevalent asthma (Table 2). For all other combinations of age groups and asthma health services use types, significant counts and rates of excess asthma morbidity were not detected. These relationships are depicted visually in Figs. 2–4 as the difference in the area under the curve between “with influenza” scenario (thicker solid red line) and “without influenza” scenario (thinner solid black line). Although the point estimates in Table 2 and graphs in Figs. 2 and 3 suggest a decrease in hospitalizations and ED visits for asthma associated with influenza activity for individuals aged 0–14, 95%CI for these estimates were wide and included 0 (no decrease). 4. Discussion In this ecological time series analysis of a population-based cohort of individuals with prevalent asthma followed from 2010 to 2015, 110
Respiratory Medicine 155 (2019) 104–112
L.Y. Feldman, et al.
of this study may not be generalizable to geographic regions with different climates (thus, different conditions for circulation of viruses and possibly different risk factors for asthma exacerbation), nor may they be generalizable to settings where health care or influenza vaccination access differ. In a population less protected by influenza vaccination, the magnitude of influenza-attributable asthma morbidity may be more pronounced than that seen in this study.
province. Influenza vaccination can be delivered in a variety of settings, and since there is no provincial registry recording influenza vaccinations, only a proportion of these are captured in health administrative databases. For that reason, Ontario health administrative data are not optimal for ascertaining individual vaccination status [25,26]. In a subset of the population for whom national health survey data were available, it was determined that 44% of individuals in Ontario who reported having asthma also reported having received an influenza vaccination in the prior 12 months [26]. Vaccination rates ranged in the general Ontario population from around 25% in the youngest age group studied (adolescents 12 years and older) to around 70% in the oldest age group (65+) [25,26]. Despite relatively high rates of influenza vaccination among older adults aged 65+ in the general population, we detected a signal of excess influenza-associated asthma hospitalizations, ED visits and outpatient physician office visits among our cohort of individuals aged 60+ with prevalent asthma. This finding may be explained by the fact that older adults are generally at higher risk of influenza complications, both due to a greater burden of medical risk factors such as diabetes and cardiovascular conditions, as well as risks associated with older age itself [27].
4.3. Conclusions This study demonstrates a substantial burden of influenza-associated asthma morbidity among adults and older adults with prevalent asthma in Ontario, whereas patterns of asthma morbidity among children appear to be more strongly linked to other seasonal trends, such as the annual September peak. Our study sheds light on the null findings of clinical trials of the effectiveness of influenza vaccination in protecting against asthma exacerbation; they suggest that adults and older adults may be important targets of further influenza-focused clinical research and intervention, in order to reduce asthma morbidity. Declaration of interests
4.2. Strengths and limitations The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Strengths of this study include its large sample size, long duration of follow-up, methodological correction for temporal autocorrelation and broad population-level applicability. This study followed a cohort over 2 million individuals with prevalent asthma covered by Ontario's universal health care system across 4 complete influenza seasons. The long duration of follow-up included periods of highly variable influenza activity, thus allowing us to tease out the effects of influenza activity on asthma morbidity, as distinct from seasonal and long-term trends. The methodology used in this study allowed us to adjust our measures of uncertainty, using bootstrap sampling to account for temporal autocorrelation in model residuals. Finally, our use of population-level health administrative data allowed us to study real-world dynamics of asthma morbidity as they relate to measured influenza activity; we are therefore able to provide complementary information to clinical trials, which study these relationships in more highly controlled, selected populations. Interpretation of our study findings is subject to certain limitations, including the use of an ecological level analysis, the possibility of residual confounding and uncertainty in generalizing our results to other geographic regions. In this ecological time series analysis, in which exposures and outcomes are correlated at an aggregate level, it cannot be determined whether individual instances of health services use for asthma were accompanied by influenza infection or not. Therefore, we cannot make inference as to the relationship between influenza infection and asthma morbidity at the individual level. However, we are able to broadly estimate age-specific annual excess asthma morbidity attributable to influenza. As a general limitation of observational studies, there is a possibility of residual confounding. In order to address this limitation, we have controlled for covariates included in a number of national and international studies of the population-level impacts of influenza (e.g., seasonal trend, long-term trend, mean temperature, RSV activity [5,6]), as well as certain covariates known to be particularly important in Canada (e.g., the annual September Peak [17,18] and statutory holidays [10–12]). That said, certain factors, such as meteorological variables apart from temperature (e.g., humidity), may have some impact on patterns of both influenza activity and asthma morbidity. Further, individual-level clinical factors such as asthma severity and duration, use of asthma medications and presence of comorbidities could impact the association between influenza activity and asthma morbidity; these factors were not considered in this study. Lastly, Ontario is located in a temperate region of the Northern hemisphere and has a universal health care system that includes free influenza vaccination for those 6 months of age or older. Therefore, results
Acknowledgement The Ontario Asthma Surveillance Information System (OASIS) is funded by the Ontario Ministry of Health and Long-Term Care (MOHLTC). Data are provided by Public Health Ontario (PHO) and ICES, which is funded by an annual grant from the Ontario MOHLTC. The opinions, results and conclusions reported in this paper are those of the authors and are independent from the funding source. No endorsement by PHO, ICES or the Ontario MOHLTC is intended or should be inferred. Parts of this material are based on data and information compiled and provided by the Canadian Institute for Health Information (CIHI). However, the analyses, conclusions, opinions and statements expressed herein are those of the authors and not necessarily those of CIHI. The authors have no other relevant sources of funding to declare. We would like to thank Kyla Jamieson at the University of Calgary and Jeff Kwong at ICES for their insights during the design of this study, Adriana Peci and Jonathan Gubbay at PHO for helpful commentary and assistance with data acquisition, and Eleanor Pullenayegum at the Hospital for Sick Children and Howard Chang at Emory University for their guidance regarding the statistical methodology used in this study. Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.rmed.2019.07.017. References [1] S.L. Johnston, P.K. Pattemore, G. Sanderson, S. Smith, F. Lampe, L. Josephs, P. Symington, S. O'Toole, S.H. Myint, D.A.J. Tyrrell, S.T. Holgate, Community study of role of viral infections in exacerbations of asthma in 9-11 year old children, BMJ 310 (6989) (1995) 1225–1229. [2] K.G. Nicholson, J. Kent, D.C. Ireland, Respiratory viruses and exacerbations of asthma in adults, BMJ 307 (6910) (1993) 982–986. [3] W.W. Busse, R.F. Lemanske Jr., J.E. Gern, Role of viral respiratory infections in asthma and asthma exacerbations, Lancet 376 (9743) (2010) 826–834. [4] C.J. Cates, B.H. Rowe, Vaccines for preventing influenza in people with asthma, Cochrane Database Syst. Rev. 2 (2013) CD000364. [5] E. Goldstein, C. Viboud, V. Charu, M. Lipsitch, Improving the estimation of influenza-related mortality over a seasonal baseline, Epidemiology 23 (6) (2012) 829–838. [6] A.D. Iuliano, K.M. Roguski, H.H. Chang, D.J. Muscatello, R. Palekar, S. Tempia,
111
Respiratory Medicine 155 (2019) 104–112
L.Y. Feldman, et al.
[7]
[8] [9]
[10]
[11]
[12]
[13]
[14]
[15]
[16] [17]
C. Cohen, J.M. Gran, D. Schanzer, B.J. Cowling, Estimates of global seasonal influenza-associated respiratory mortality: a modelling study, Lancet 391 (10127) (2018) 1285–1300. L. Li, J.Y. Wong, P. Wu, H.S. Bond, E.H. Lau, S.G. Sullivan, B.J. Cowling, Heterogeneity in estimates of the impact of influenza on population mortality: a systematic review, Am. J. Epidemiol. 187 (2) (2018) 378–388. D.L. Schanzer, C. Sevenhuysen, B. Winchester, T. Mersereau, Estimating influenza deaths in Canada, 1992–2009, PLoS One 8 (11) (2013) e80481. W.W. Thompson, D.K. Shay, E. Weintraub, L. Brammer, C.B. Bridges, N.J. Cox, K. Fukuda, Influenza-associated hospitalizations in the United States, J. Am. Med. Assoc. 292 (11) (2004) 1333–1340. D.L. Schanzer, J.M. Langley, T.W. Tam, Hospitalization attributable to influenza and other viral respiratory illnesses in Canadian children, Pediatr. Infect. Dis. J. 25 (9) (2006) 795–800. D.L. Schanzer, J.M. Langley, T.W. Tam, Role of influenza and other respiratory viruses in admissions of adults to Canadian hospitals, Influenza Other Respir. Viruses 2 (1) (2008) 1–8. D.L. Schanzer, B. Schwartz, Impact of seasonal and pandemic influenza on emergency department visits, 2003–2010, Ontario, Canada, Acad. Emerg. Med. 20 (4) (2013) 388–397. A.S. Gershon, C. Wang, J. Guan, J. Vasilevska-Ristovska, L. Cicutto, T. To, Identifying patients with physician-diagnosed asthma in health administrative databases, Can. Respir. J. 16 (6) (2009) 183–188. T. To, S. Dell, P.T. Dick, L. Cicutto, J.K. Harris, I.B. MacLusky, M. Tassoudji, Case verification of children with asthma in Ontario, Pediatr. Allergy Immunol. 17 (1) (2006) 69–76. Public Health Ontario, Public Health Ontario Laboratory Respiratory Pathogen Surveillance Reports, (2016) https://www.publichealthontario.ca/en/ ServicesAndTools/LaboratoryServices/Pages/PHO-Laboratories-surveillanceupdates.aspx , Accessed date: 15 October 2018. Public Health Ontario, Personal Communication, (2018). N.W. Johnston, S.L. Johnston, J.M. Duncan, J.M. Greene, T. Kebadze, P.K. Keith, M. Roy, S. Waserman, M.R. Sears, The September epidemic of asthma exacerbations
[18]
[19]
[20] [21]
[22]
[23]
[24]
[25] [26]
[27]
112
in children: a search for etiology, J. Allergy Clin. Immunol. 115 (1) (2005) 132–138. K. Larsen, J. Zhu, L.Y. Feldman, J. Simatovic, S. Dell, A.S. Gershon, T. To, The annual September peak in asthma exacerbation rates. Still a reality? Ann. Am. Thorac. Soc. 13 (2) (2016) 231–239. J.H. Kim, M. Yeasmin, The size and power of the bias-corrected bootstrap test for regression models with autocorrelated errors, Comput. Econ. 25 (3) (2005) 255–267. R Core Team, R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna, Austria, 2017. H.J. Bueving, R.M. Bernsen, J.C. de Jongste, L.W. van Suijlekom-Smit, G.F. Rimmelzwaan, A.D. Osterhaus, M.P. Rutten-van Mölken, S. Thomas, J.C. van der Wouden, Influenza vaccination in children with asthma: randomized doubleblind placebo-controlled trial, Am. J. Respir. Crit. Care Med. 169 (4) (2004) 488–493. H. Bueving, J. van der Wouden, H. Raat, R. Bernsen, J. de Jongste, L. van SuijlekomSmit, A. Osterhaus, G. Rimmelzwaan, M. Rutten-van Mölken, S. Thomas, Influenza vaccination in asthmatic children: effects on quality of life and symptoms, Eur. Respir. J. 24 (6) (2004) 925–931. T. Jefferson, D. Rivetti, A. Rivetti, M. Rudin, C. Di Pietrantonj, V. Demicheli, Efficacy and effectiveness of influenza vaccines in elderly people: a systematic review, Lancet 366 (9492) (2005) 1165–1174. K. Nichol, K. Margolis, J. Wuorenma, T. Von Sternberg, The efficacy and cost effectiveness of vaccination against influenza among elderly persons living in the community, N. Engl. J. Med. 331 (12) (1994) 778–784. J.C. Kwong, D.G. Manuel, Using OHIP physician billing claims to ascertain individual influenza vaccination status, Vaccine 25 (7) (2007) 1270–1274. K.L. Schwartz, N. Jembere, M.A. Campitelli, S.A. Buchan, H. Chung, J.C. Kwong, Using physician billing claims from the Ontario Health Insurance Plan to determine individual influenza vaccination status: an updated validation study, CMAJ Open 4 (3) (2016) E463–E470. Ontario Agency for Health Protection and Promotion (Public Health Ontario), The Relationship between Influenza Medical Risk Factors and Age, (2017) Toronto, ON.