A cross-sectional analysis of symptom severity in adults with influenza and other acute respiratory illness in the outpatient setting

Acute respiratory infections (ARIs) collectively represent the most common seasonal illness, affecting adults three times annually on average [1, 2]. Most ARIs are caused by viruses, including influenza [2]. Studies addressing the burden of illness from viral ARI have historically focused on high-risk populations [3–7] and severe outcomes such as emergency room visits [8], hospitalizations [9–12], and death [13, 14]. However, the severity of ARI symptoms in individuals seeking and receiving outpatient care also contributes to the overall illness burden [15]. Such individuals may not experience severe health complications, but their illnesses nevertheless account for a significant economic burden in lost workplace productivity [16]. Because symptom severity is often a determining factor in seeking healthcare for ARI, it underlies billions of dollars in direct medical care expenditures in the United States [17]. Many large observational studies on ARI and influenza-like illness focus on individuals seeking ambulatory medical care, thus a better understanding of the severity of ARI symptoms may improve the interpretability of results of such studies.

From both an economic and public health perspective, influenza is one of the most important viruses because it may cause more severe symptoms than other respiratory viruses. The severity of seasonal influenza symptoms compared to non-influenza ARI has not been extensively studied though. One study reported an association between increased symptom severity in patients with laboratory confirmed 2009 pandemic influenza compared to seasonal influenza [18]; limited evidence also suggests that influenza ARI may be more severe than ARI due to other causes [19]. In an influenza challenge study, symptom severity was associated with increased magnitude of influenza virus shedding [20]. Given its role in generating an immune response and the production of antibodies, influenza vaccination may mitigate the severity of some influenza symptoms. Evidence of this hypothesized impact is incomplete, however, with one study of hospitalized influenza cases demonstrating fewer deaths and ICU admissions (used as markers of severe clinical complications) among those who received a seasonal influenza vaccination [21].

It is well established that ARIs, including influenza, do not affect all patients uniformly [22], yet little is known about host factors and virulence characteristics that are associated with specific ARI symptoms. We had the opportunity to examine the severity of symptoms, along with demographic and clinical features, in adults presenting to an outpatient facility with ARI. Because influenza may present with more severe features relative to ARI of other etiologies, we hypothesized that symptom severity would be rated higher among adults with RT-PCR confirmed influenza. A secondary hypothesis was that influenza vaccination was associated with reduced symptom severity among participants with confirmed influenza.

There were 2,374 participants who met all eligibility criteria and were included in the analysis. Two-thirds of participants were female, and the majority was under age 50 years (55%), with 29% and 16% in the 50–64 years and ≥65 years groups, respectively (Table 1). Mean (±SD) symptom severity score was 12.3 (±4.1) points overall, with significantly decreased symptom severity in older age groups (p < 0.001). Influenza was significantly less common in older age groups (p = 0.012), with rates of RT-PCR confirmed influenza being 15%, 13%, and 9% in participants age 18–49, 50–64, and ≥65 years, respectively. The most common symptoms were cough (93%), fatigue (91%), nasal congestion (84%), headache (76%), sore throat (76%), muscle aches (70%), fever (69%), and wheezing (49%). The median severity rating for each symptom, when present, was ‘moderate.’ In unadjusted univariate models, higher symptom severity score was significantly associated with younger age group, female gender, a greater number of recent ambulatory visits, not being vaccinated for influenza, borderline high cholesterol, higher BMI, and current smoking (see Table 1). Influenza varied significantly by enrollment season and time interval between symptom onset and enrollment. Compared to those with symptom severity scores below the median, more participants tested positive for influenza among those with symptom severity scores above the median (Figure 1).

The main findings from the final adjusted model in Table 2 indicated that female gender, more ambulatory care visits, borderline high cholesterol, obesity, and smoking (both current and former) were associated with significantly higher symptom severity scores. In addition, there was a significant three-way interaction observed between age group, influenza, and vaccination status. Age modified the association between vaccination status and symptom severity in patients who were influenza positive (Figure 2). Mean symptom severity generally declined with increasing age, but remained highest among non-vaccinated individuals with influenza. Among participants ≥65 years old with influenza, mean symptom severity scores were 5–6 points higher in those who were not vaccinated relative to those who were vaccinated or those who did not have influenza (regardless of vaccination status; p <0.001 for all pairwise comparisons). Significant reductions in the severity of cough and feverishness, as well as a decreased presence of wheezing, largely explained the lower symptom severity scores observed in the older, vaccinated group with influenza (Table 3).

Characteristic	Difference in average adjusted symptom severity score2	Standard error	P-value
Age (years) (ref.: 18–49)
50-64	−0.19	0.35	0.51
≥ 65	−3.08	0.49	< 0.0001
Female gender	0.99	0.18	< 0.0001
Days elapsed between symptom onset and enrollment date (ref.: ≥ 7)
5-6	0.06	0.28	0.82
3-4	0.04	0.26	0.87
1-2	−0.27	0.27	0.31
0	−2.45	0.53	< 0.0001
Number of ambulatory care visits in the past five years	0.01	0.001	< 0.0001
Influenza infection 3	1.73	0.36	< 0.0001
Vaccinated against influenza	−0.39	0.24	0.11
Total cholesterol (mg/dL) (ref.: < 200)
200-239	0.67	0.19	0.0006
≥ 240	0.46	0.32	0.15
Body mass index (kg/m2) (ref.: < 25)
25-29	0.27	0.24	0.27
≥ 30	0.61	0.22	0.005
Smoking (ref.: Never)
Current	0.63	0.23	0.006
Former	0.73	0.19	< 0.0001
Interaction: Age x influenza
50-64 × influenza infection	0.23	0.72	0.75
≥ 65 × influenza infection	4.21	1.37	0.002
Interaction: Age × influenza vaccination
50-64 × vaccinated	−0.48	0.41	0.24
≥ 65 × vaccinated	1.45	0.57	0.01
Interaction: Influenza infection x influenza vaccination	0.92	0.68	0.17
Interaction: Age x influenza infection x influenza vaccination
50-64 x influenza infection × vaccinated	−1.02	1.11	0.36
≥ 65 × influenza infection × vaccinated	−6.75	1.68	< 0.0001

¹Bolded values denote point estimate was significant (p < 0.05). Significant exposures in the crude models were also considered for inclusion in the final reduced model.

²Values are reported as point estimate ± standard error, p-value. Positive values with p < 0.05 indicate a significantly greater symptom severity score relative to the reference category (or 1-unit increase for continuous predictors) and negative values with p < 0.05 indicate a significantly lower symptom severity score relative to the reference category.

³Influenza infection refers to infection with either influenza A or B.

In a secondary analysis, Ct values were examined from RT-PCR confirmed influenza positive samples. Due to the relatively small available sample for this analysis, Ct values were dichotomized as having a high or low viral load based on the median value of 25.7. Among all influenza positive participants, the median severity score was 15 for those with a high viral load and 13 for those with a low viral load (p = 0.09). Among adults ≥ 65 years old, the median severity score was 12 for those with a high viral load and 12 for those with a low viral load (p = 0.27). There was no association between vaccination status and viral load among all participants with influenza (p = 0.62). Among vaccinated adults ≥65 years old, the median severity score was 12 and 10, respectively, for those with a high and low viral load (p = 0.32).

Symptom severity can serve as a useful and complementary measure of respiratory disease impact among individuals seeking care for respiratory illness in the outpatient setting. Perceived ARI symptom severity was greatest in younger adults, women, and individuals with RT-PCR confirmed influenza in this study. As expected, influenza infection was most closely associated with increased symptom severity. The presence of diagnosed CVD, pulmonary disease, or diabetes had little association with symptom severity, though some disease risk factors (e.g., obesity, smoking, dyslipidemia) were modestly associated with higher symptom severity scores.

The subgroup analysis of influenza positive participants identified an association between influenza vaccination and reduced symptom severity in older age groups. Compared to ≥ 65 year old adults who were influenza positive but not vaccinated, perceived symptom severity was 31%-39% lower in older adults who were influenza negative or who were vaccinated. This difference was primarily driven by reductions in the severity of cough and feverishness, as well as reduced presence of wheezing, in those who received the vaccine. However, there were only 10 unvaccinated individuals with influenza who were ≥ 65 years old; therefore these findings could reflect random variation as a function of the small sample size. The biological basis for reduced symptom severity in vaccinated older adults (but not in younger adults) is unclear and largely speculative at this point. The humoral and cell-mediated immune response to natural influenza infection and influenza vaccination declines with increased age, thus the potential vaccine benefits in terms of symptom mitigation would seem to be more likely in younger, rather than older, individuals [28, 29]. It is also possible that the observed reductions in systemic (fever) and lower respiratory (wheezing) symptoms could have resulted from reduced influenza virus replication in vaccinated versus unvaccinated older adults. However, our analysis of RT-PCR Ct values as a surrogate measure of viral load did not reveal any significant associations with either severity score or vaccination status. There may also be perceptual reasons that explain why those in older age groups or who received a prior influenza vaccination perceived their symptoms differently. Further research is needed to understand the relationship between vaccination, influenza severity, and virus shedding in older adults.

Strengths of this study included recruitment from a defined population cohort over multiple seasons, systematic screening and enrollment, standardized reporting of severity at the time of enrollment, use of a validated immunization registry, and confirmation of influenza using a highly sensitive and specific RT-PCR test. From a limitations perspective, individuals who did not receive medical attention for their ARI were not recruited, thus results are not necessarily generalizable to all influenza illness cases in the community. From a measurement perspective, variations of the ARI symptom severity metric used in this study have been used in numerous other investigations [30–38] and have been previously correlated with influenza positive status [19] and viral shedding [20], but have not been formally validated and remain a subjective measure of symptom perceptions that may be subject to recall and self-presentation biases, as is common with some other self-reported metrics [39]. This is a particular concern in some subgroup comparisons where symptom severity expectations may vary as a result of perceptual priming during study enrollment (e.g., different beliefs on how influenza vaccination should impact symptoms). Additional research is also needed in this area to better understand self-reported ARI symptom severity and its relationship to functional disability and serious medical complications. In addition to influenza, future studies should also examine symptom severity for other specific pathogens and virus infections such as RSV, rhinovirus, and human metapneumovirus.

Influenza infection was the strongest predictor of perceived symptom severity among individuals with medically attended ARI over multiple seasons. There was also a modest reduction in perceived severity among vaccinated vs. unvaccinated individuals ≥65 years old with confirmed influenza. However, these findings could be influenced by random effects or measurement biases, thus further research to confirm this finding in other populations and seasons is warranted.