Abstract
Objective. This study was performed to determine predictors of clinical influenza diagnosis among patients with laboratory-confirmed influenza.
Methods. Prospective, laboratory-confirmed surveillance for influenza was conducted among patients of all ages who were hospitalized or presented to the emergency department with fever and respiratory symptoms during 2009–2013. We evaluated all enrolled persons who had influenza confirmed by viral culture and/or polymerase chain reaction and received any discharge diagnosis. The primary outcome, clinical influenza diagnosis, was defined as (1) a discharge diagnosis of influenza, (2) a prescription of neuraminidase inhibitor, or (3) a rapid test positive for influenza virus. Bivariate analyses and multiple logistic regression modeling were performed.
Results. Influenza was diagnosed for 29% of 504 enrolled patients with laboratory-confirmed influenza and for 56% of 236 patients with high-risk conditions. Overall, clinical influenza diagnosis was predicted by race/ethnicity, insurance status, year, being hospitalized, having high-risk conditions, and receiving no diagnosis of bacterial infection. Being diagnosed with a bacterial infection reduced the odds of receiving an influenza diagnosis by >3-fold for all patients and for patients with high-risk conditions.
Conclusions. Many influenza virus–positive patients, including those with high-risk conditions, do not receive a clinical diagnosis of influenza. The pattern of clinical diagnoses among influenza virus–positive patients suggests preferential consideration of bacterial diseases as a diagnosis.
Keywords: influenza, human, clinical practice patterns, diagnosis, antiviral agents
The World Health Organization estimates that 5%–10% of adults and 20%–30% of children are infected with influenza virus each year [1]. Influenza-attributable illnesses are costly, with an annual estimated direct medical cost of approximately $10.4 billion in the United States [2]. Outpatient and emergency department (ED) visits for influenza are more common than hospitalizations, but all settings contribute to the annual cost [3–5].
Only a fraction of all persons with influenza receive a clinical diagnosis of influenza [6–11]. One contributor is that accurate diagnosis of influenza can be challenging [12, 13]. Clinical symptoms of influenza often overlap with symptoms associated with other respiratory viruses, as well as with symptoms associated with other illnesses [14, 15]. Predicting influenza virus infection by using clinical decision rules has been hampered by trade-offs between sensitivity and specificity for inclusion of various symptoms and number of symptoms [13, 16–18]. Both the timing and duration of the influenza season vary each year. Rapid influenza diagnostic tests can aid timely diagnosis of influenza, although false-negative results are possible, given the moderate specificity of these tests [19, 20].
A specific influenza diagnosis is important for persons for whom such a diagnosis would alter the diagnostic evaluation or treatment plan. Examples include persons with influenza who are hospitalized or have high-risk conditions and, in countries such as the United States, are recommended to receive antiviral treatment [21, 22].
We evaluated data from prospective, population-based influenza surveillance in the ED and inpatient settings over 4 seasons to determine predictors of a clinical influenza diagnosis among enrolled patients who had laboratory-confirmed influenza. The objective was to provide insight as to which patients with influenza disease as determined by culture or molecular testing were clinically diagnosed with influenza at time of presentation or hospitalization.
METHODS
Study Setting and Population
Over 4 influenza seasons (2009–2013) prospective influenza surveillance was conducted among adults and children presenting to the ED and inpatient settings at 3 hospitals. These hospitals included a large community hospital and an academic medical center with a children's hospital, which together care for >94% of the residents of Forsyth County, North Carolina. Eligible persons were enrolled ≥4 days a week at each setting throughout each season. Eligible persons resided in Forsyth County or a contiguous county and presented to the ED or inpatient setting with a history of fever and/or acute respiratory symptoms, irrespective of symptom duration. These symptoms included cough, nasal congestion, difficulty breathing, earache, sore throat, or wheezing. Persons with fever explained by an identified nonrespiratory source prior to enrollment, such as urinary tract infection or abscess, and without acute respiratory symptoms were excluded. Most eligible patients were enrolled within 24 hours of presentation, although some patients admitted over weekends were enrolled within 36–48 hours of presentation.
Approvals
Written informed consent and, when appropriate, child assent were obtained for all enrolled persons. The Wake Forest School of Medicine Institutional Review Board approved this study with an authorization agreement between the institutional review boards of Forsyth Medical Center and Wake Forest School of Medicine.
Study Period
Surveillance was systematically performed over 4 consecutive influenza seasons, defined as November through April for all seasons except 2009–2010, when the season spanned September through mid-April. Each enrollment season was divided into a peak influenza period and a nonpeak period. The 2009–2010 season peak was defined on the basis of the dates of the first two thirds of all positive samples (1 September 2009 through 24 November 2009), reflecting the early peak for this season. The peak periods of all other seasons were defined as the middle two thirds of all positive samples and spanned 12 January through 24 February 2011, 18 January through 23 March 2012, and 28 November 2012 through 27 February 2013.
Demographic and Clinical Information
Each enrolled person or guardian was administered a standardized questionnaire to obtain demographic characteristics and medical history, including symptoms and their duration. Because headaches, myalgias, and malaise are difficult to ascertain in young children or noncommunicative persons, these symptoms were collected among communicative persons ≥5 years of age. Race/ethnicity was determined by self-report and was categorized as white, black, non-Hispanic white, other, or unknown. Influenza vaccination status (H1N1 monovalent vaccine in 2009–2010 and seasonal vaccine during 2010–2013) was determined by vaccine verification from the North Carolina Immunization Registry, the primary care provider, or the reported location of vaccine administration.
A trained research assistant completed a systematic chart review for the enrollment visit. For those enrolled in the ED and then hospitalized, the chart review included both that ED visit and the hospitalization. Age, sex, clinician-ordered clinical testing, insurance status, and antiviral prescription were determined by chart review. Age was considered equal to the enrollment date minus the birth date. Timing in season (before peak, peak, and after peak) was determined on the basis of the enrollment date relative to the influenza peak for that season.
High-risk conditions were identified through standardized questionnaire and chart review. High-risk conditions included all medical conditions associated with increased risk of severe influenza disease and a condition-specific influenza vaccine recommendation [23]. They include chronic cardiac or pulmonary diseases, diabetes, cancer, immunodeficiency, human immunodeficiency virus infection, transplant receipt, chronic renal or hepatic diseases, cognitive or neurological disorders, hemoglobinopathies, and pregnancy.
Influenza Virus Testing
One nasal swab and 1 throat swab were collected from each enrolled person, placed in a single vial of transport medium, and refrigerated. Specimens were cultured for influenza virus. Influenza virus was identified using direct fluorescent antibody assays. RNA was extracted from specimens and tested by reverse transcription–polymerase chain reaction (RT-PCR) for influenza A virus, influenza B virus, influenza A(H1N1)pdm09 virus, influenza A(H3N2) virus, and seasonal influenza A(H1N1) virus. Human Rnase P gene RNA was detected to confirm adequate sample collection. The RT-PCR protocols and the probe and primer sequences were kindly provided by the Centers for Disease Control and Prevention (Virus Surveillance and Diagnostics Branch, Influenza Division, Atlanta, Georgia). Some patients also had physician-ordered molecular testing for influenza virus, using the FilmArray Respiratory Panel (Idaho Technologies). Patients who had influenza virus detected by RT-PCR in the research laboratory or in the clinical laboratory were defined as having laboratory-confirmed influenza.
Categorization of Diagnosis
All discharge diagnoses for each enrolled person were systematically collected from chart review of the ED visit, review of the discharge summary from the hospitalization, or both. Thus, the discharge diagnoses were determined by that written in the ED visit note or hospitalization discharge summary and not from International Classification of Disease, Ninth Revision or International Classification of Disease, Tenth Revision diagnoses or from billing codes. For those enrolled in the ED and then hospitalized, the diagnoses in the inpatient setting were evaluated. A clinical influenza diagnosis was defined as (1) an explicit discharge diagnosis of influenza, (2) prescription of antiviral medication (oseltamivir or zanamivir), or (3) a rapid test positive for influenza virus. Each medical chart diagnosis was categorized a priori as an influenza diagnosis, a viral diagnosis, a bacterial diagnosis for which antibiotics were prescribed, a diagnosis of an exacerbation of a chronic condition, a diagnosis of a symptom, or other.
For each enrolled person, all discharge diagnoses were categorized in the following hierarchical order: influenza and bacterial diagnoses, influenza diagnosis, bacterial diagnosis, chronic condition exacerbation, viral illness, symptom, or other. To be classified as a bacterial diagnosis, the clinical diagnosis had to involve a bacterial disease for which antibiotics were prescribed. For example, persons who received a diagnosis influenza and pneumonia and were prescribed antibiotics were classified as having influenza and bacterial diagnoses. Persons who received a diagnosis of influenza and viral pneumonia and were not prescribed antibiotics were classified as having an influenza diagnosis.
Statistical Analysis
Bivariate analysis of the clinical influenza diagnosis outcome, by demographic and clinical characteristics, and of the hierarchical discharge diagnosis categories, by age group, was conducted by χ2 or Fisher exact tests. Bivariate analysis of continuous data was performed using the Student t test. A P value of <.05 was considered statistically significant.
Multiple logistic regression modeling was performed to identify demographic or clinical predictors of clinical influenza diagnosis among patients with laboratory-confirmed influenza. Potential predictors as determined by the literature [6, 24] or clinical judgment were entered as covariates into the initial model, and those with the highest P values were eliminated sequentially until all of the remaining covariates had a P value of <.20. Each eliminated covariate was added back into the model to determine whether the variable had a P value of <.20; no covariates were eligible for being added back into the model.
Because persons with high-risk conditions are recommended for treatment with antiviral medications, a secondary analysis was performed to identify predictors of clinical influenza diagnosis among patients with high-risk conditions only. To avoid model overfitting, the covariates were limited to 4 characteristics that could be used clinically to identify patients prone to a missed diagnosis and chosen a priori--age group (0–17 years, 18–49 years, and ≥50 years), setting, and bacterial diagnosis, as well as an interaction between age group and setting. Two post hoc sensitivity analyses were performed. One sensitivity analysis excluded data from 2009–2010 because it was an atypical influenza season. A second sensitivity analysis defined all patients with a clinician-ordered rapid influenza test, indicating sufficient suspicion to order a test, as having a clinical influenza diagnosis. All analyses were performed using SAS, version 9.3 (Cary, North Carolina).
RESULTS
Overall, 4689 (78%) of 6011 eligible persons approached for enrollment during the 2009–2010 through 2012–2013 influenza seasons were enrolled. A total of 505 (11%) of 4689 enrolled patients had laboratory-confirmed influenza. One person who left the ED prior to receiving a discharge diagnosis was excluded. The remaining 504 patients composed the study population. Of 504 patients with laboratory-confirmed influenza, 97% tested positive in the research laboratory; almost all of the remaining 3% who tested positive only in the clinical laboratory were adults who had received oseltamivir for ≥36 hours prior to collection of the study nasal/throat swab. Among the 4365 patients without laboratory-confirmed influenza, 228 (5%) received a clinical diagnosis of influenza.
Patients with laboratory-confirmed influenza represented all age groups (Table 1). Approximately half were female and had public insurance; self-reported race/ethnicity was 48% black, 40% non-Hispanic white, and 11% Hispanic white, with 1% consisting of other or unreported races/ethnicities. The number of patients with laboratory-confirmed influenza varied each year. Over two thirds reported symptoms for ≥2 days. Over one fourth were hospitalized, almost one half had high-risk conditions, and three fourths were unvaccinated. Just under one third had a bacterial diagnosis and were prescribed antibiotics. Of the 103 patients prescribed antiviral medication, 48 (47%) were discharged from the ED, and 55 (53%) were hospitalized; thus, a higher proportion of patients in the inpatient setting, compared with the ED setting, received an antiviral prescription (37% vs 14%; P < .001). The majority (70%) of all patients with a clinical diagnosis of influenza received antiviral medication.
Table 1.
Demographic and Clinical Characteristics by Influenza Diagnosis Among All Enrolled Patients With Laboratory-Confirmed Influenza and the Subset With High-Risk Medical Conditions
| Demographic or Clinical Characteristic | Influenza Diagnosis for All Subjects, No. (Row %) |
Influenza Diagnosis for High-Risk Subjects, No. (Row %) |
||||
|---|---|---|---|---|---|---|
| Yes (n = 147) | No (n = 357) | P Value | Yes (n = 133) | No (n = 103) | P Value | |
| Age, y | ||||||
| 0–4 | 23 (23) | 78 (77) | .02 | 9 (53) | 8 (47) | .31 |
| 5–17 | 24 (28) | 61 (72) | 12 (41) | 17 (59) | ||
| 18–49 | 49 (26) | 140 (74) | 27 (32) | 57 (68) | ||
| ≥50 | 51 (40) | 78 (60) | 45 (42) | 61 (58) | ||
| Sex | ||||||
| Male | 65 (28) | 166 (72) | .64 | ND | ND | ND |
| Female | 82 (30) | 191 (70) | ||||
| Race/ethnicity | ||||||
| White, non-Hispanic | 69 (34) | 134 (66) | .005 | ND | ND | ND |
| Black, non-Hispanic | 54 (23) | 186 (78) | ||||
| Hispanic/other/unknown | 24 (39) | 37 (61) | ||||
| Insurance | ||||||
| Public only | 69 (26) | 199 (74) | .17 | ND | ND | ND |
| Any private | 54 (34) | 104 (66) | ||||
| None/unknown | 24 (31) | 54 (69) | ||||
| Study year | ||||||
| 2009–2010 | 36 (53) | 32 (47) | <.001 | ND | ND | ND |
| 2010–2011 | 41 (20) | 167 (80) | ||||
| 2011–2012 | 8 (31) | 18 (69) | ||||
| 2012–2013 | 62 (31) | 140 (69) | ||||
| Timing in influenza season | ||||||
| Before peak | 20 (28) | 51 (72) | .62 | ND | ND | ND |
| Peak | 96 (28) | 244 (72) | ||||
| After peak | 31 (33) | 62 (67) | ||||
| Setting | ||||||
| ED | 71 (20) | 284 (80) | <.001 | 32 (28) | 84 (72) | <.001 |
| Hospitalized | 76 (51) | 73 (49) | 61 (51) | 59 (49) | ||
| High-risk condition(s) | ||||||
| Yes | 93 (39) | 143 (61) | <.001 | ND | ND | ND |
| No | 54 (20) | 214 (80) | … | … | ||
| Influenza vaccinationa | ||||||
| Yes | 24 (27) | 65 (73) | .58 | ND | ND | ND |
| No | 91 (30) | 212 (70) | ||||
| Symptom duration, d | ||||||
| ≤2 | 39 (28) | 100 (72) | .75 | 17 (37) | 29 (63) | .72 |
| >2 | 107 (29) | 256 (71) | 75 (40) | 113 (60) | ||
| Bacterial diagnosis | ||||||
| Yes | 27 (18) | 124 (82) | <.001 | 20 (23) | 68 (77) | <.001 |
| No | 120 (34) | 233 (66) | 73 (49) | 75 (51) | ||
| Antiviral medication | ||||||
| Yes | 103 (100) | 0 (0) | <.001 | 70 (100) | 0 (0) | <.001 |
| No | 44 (11) | 355 (89) | 23 (14) | 142 (86) | ||
Abbreviations: ED, emergency department; ND, not done.
a Children <6 months of age are not included because they are too young to receive the influenza vaccine.
Most enrolled patients (≥70%) had cough, nasal congestion, and fever (Table 2). Most adults reported fatigue/malaise. The prevalence of reported symptoms varied by setting. More patients in the ED than in the inpatient setting reported cough, nasal congestion, fever, fatigue/malaise, headache, poor appetite, sore throat, and myalgias/muscle aches. In contrast, hospitalized patients were more likely to report shortness of breath and wheezing than patients discharged from the ED. The variation in the prevalence of these symptoms by setting reinforced the clinical challenge of making specific influenza diagnoses.
Table 2.
Frequency of Symptoms Reported Among All Enrolled Patients With Laboratory-Confirmed Influenza, by Age Group
| Symptoms | Emergency Department, by Age, No. (Column %) |
Hospitalization, by Age, No. (Column %) |
||||||
|---|---|---|---|---|---|---|---|---|
| Total | 0–17 y | 18–49 y | ≥50 y | Total | 0–17 y | 18–49 y | ≥50 y | |
| Enrolled patients | 355 | 156 | 166 | 33 | 149 | 30 | 23 | 96 |
| Cough | 341 (96) | 148 (95) | 160 (96) | 33 (100) | 133 (89) | 25 (83) | 21 (91) | 87 (91) |
| Fatigue/malaisea | 172 (86) | ND | 142 (86) | 30 (91) | 92 (77) | ND | 17 (74) | 75 (78) |
| Nasal Congestion | 306 (86) | 134 (86) | 144 (87) | 28 (85) | 105 (71) | 25 (86) | 15 (65) | 65 (68) |
| Fever | 301 (85) | 147 (94) | 126 (76) | 28 (85) | 109 (75) | 27 (93) | 18 (78) | 64 (69) |
| Poor appetite | 266 (75) | 110 (71) | 130 (78) | 26 (79) | 88 (60) | 17 (59) | 15 (65) | 56 (59) |
| Headachea | 218 (79) | 49 (64) | 139 (84) | 30 (91) | 81 (58) | 5 (25) | 16 (70) | 60 (63) |
| Shortness of breath | 205 (58) | 63 (40) | 116 (70) | 26 (79) | 115 (77) | 19 (63) | 18 (78) | 78 (81) |
| Sore throat | 215 (63) | 78 (55) | 114 (69) | 23 (70) | 68 (48) | 6 (25) | 16 (70) | 46 (48) |
| Wheezing | 181 (51) | 58 (37) | 95 (57) | 28 (85) | 98 (68) | 17 (57) | 15 (65) | 66 (73) |
| Myalgias/muscle achesa | 38 (57) | 36 (47) | 139 (84) | 28 (88) | 38 (27) | 6 (30) | 11 (48) | 21 (22) |
| Posttussive emesis | 145 (41) | 57 (37) | 71 (43) | 17 (52) | 65 (44) | 15 (50) | 10 (43) | 40 (42) |
| Earache | 90 (26) | 28 (19) | 47 (28) | 15 (45) | 23 (16) | 4 (15) | 4 (18) | 15 (16) |
Abbreviation: ND, not done.
a History of myalgias/muscle aches and history of headaches were collected among 97 children 5–17 years of age and all adults (277 subjects in the emergency department and 139 who were hospitalized). Four noncommunicative children had missing responses, of whom 1 child received a diagnosis of influenza.
Among 504 persons with laboratory-confirmed influenza, 147 (29%; 95% confidence interval [CI], 25%–33%) received a clinical influenza diagnosis, defined as an explicit clinical diagnosis, antiviral prescription, or positive rapid influenza test result. The mean days between symptom onset and enrollment was similar for patients with and those without a clinical influenza diagnosis (4.4 vs 4.8; P = .42). Among 147 patients with a clinical influenza diagnosis, 86% had an explicit clinical diagnosis of influenza, 70% received an antiviral prescription, and 41% had a positive rapid influenza test result. Altogether, explicit clinical diagnosis alone captured 86%, explicit clinical diagnosis and/or antiviral prescription captured 98%, and explicit clinical diagnosis and/or antiviral prescription plus positive results of rapid influenza tests captured the rest. More adults than children had a rapid test ordered (42% vs 20%; P < .001). A total of 86 of 504 persons (17%) with laboratory-confirmed influenza had a negative rapid influenza test result.
Overall, the pattern of discharge diagnoses in this hierarchical order (Table 3) differed by age group (P < .001). For children and adults aged 18–49 years, the most common discharge diagnosis category was viral diagnosis in the ED and influenza in the inpatient setting. In the ED, adults aged ≥50 years most commonly received a diagnosis with bacterial diseases alone (51%) followed by influenza diagnosis with or without a bacterial disease (21%) and viral diagnoses (21%); in the inpatient setting, 46% received a diagnosis of influenza with or without a bacterial diseases, and 31% received a diagnosis of bacterial diseases alone. Altogether, influenza diagnosis with a bacterial diagnosis increased by age group (P < .001), whereas influenza without a bacterial diagnosis remained similar across age groups.
Table 3.
Discharge Diagnosis Categories in Hierarchical Order Among All Patients With Laboratory-Confirmed Influenza, by Age Group and Setting
| Diagnosis or Diagnoses | 0–17 y, No. (Column %) | 18–49 y, No. (Column %) | ≥50 y, No. (Column %) | Total, No. (Column %) |
|---|---|---|---|---|
| Emergency department | ||||
| Influenza and bacterial diagnosesa | 1 (1) | 5 (3) | 2 (6) | 8 (2) |
| Influenza diagnosis | 25 (16) | 33 (20) | 5 (15) | 63 (18) |
| Bacterial diagnosisa | 21 (13) | 48 (29) | 17 (51) | 86 (24) |
| Exacerbation of chronic condition | 4 (3) | 7 (4) | 0 (0) | 11 (3) |
| Viral diagnosisb | 99 (64) | 67 (37) | 7 (21) | 173 (49) |
| Symptom or other diagnoses | 6 (4) | 6 (4) | 2 (6) | 14 (4) |
| Total | 156 | 166 | 33 | 355 |
| Hospitalization | ||||
| Influenza and bacterial diagnosesa | 1 (3) | 3 (13) | 15 (16) | 19 (13) |
| Influenza diagnosis | 20 (67) | 8 (34) | 29 (30) | 57 (38) |
| Bacterial diagnosisa | 3 (10) | 5 (22) | 30 (31) | 38 (26) |
| Exacerbation of chronic condition | 1 (3) | 6 (26) | 12 (13) | 19 (13) |
| Viral diagnosisb | 1 (3) | 6 (26) | 12 (13) | 12 (8) |
| Symptom or other diagnoses | 0 (0) | 0 (0) | 4 (4) | 4 (3) |
| Total | 30 | 23 | 96 | 149 |
a Bacterial infections included the following diagnoses treated with antibiotics: bronchitis, chronic obstructive pulmonary disease exacerbation, otitis media, peritonitis, pharyngitis/strep throat/tonsillitis, pneumonia, sinusitis, and urinary tract infection.
b Because the hierarchical order began with influenza, viral illness excluded those with an influenza diagnosis or those with a diagnosis of viral illness in conjunction with a bacterial infection or exacerbation of chronic condition.
Pneumonia, bronchitis, and sinusitis were the most common bacterial diagnoses treated with antibiotics. Although chronic obstructive pulmonary disease was the most frequently diagnosed exacerbation, most were classified as bacterial infections because of a codiagnosis of bronchitis and treatment with antibiotics. In this hierarchical order, asthma exacerbation was the most frequent condition classified as an exacerbation. Overall, viral diagnoses were the most frequent category. Few (3%) had other diagnoses that included cough, chest pain, fever, neck pain, and syncope.
Among patients with high-risk conditions (Table 4), the hierarchical pattern of discharge diagnoses remained significantly different when stratified by age group (P < .001). For children, an influenza diagnosis with or without a bacterial diagnosis (45%) was most common, followed by viral diagnosis (33%). For adults aged 18–49 years, bacterial diagnosis alone (36%) was as common as influenza with or without a bacterial diagnosis (33%). For adults aged ≥50 years, influenza with or without bacterial diagnosis (42%) was most common, followed by bacterial diagnosis alone (34%).
Table 4.
Discharge Diagnosis Categories in Hierarchical Order Among All Patients With High-risk Medical Conditions and Laboratory-Confirmed Influenza, by Age Group
| Diagnosis(ies) | 0–17 y, No. (Column %) | 18–49 y, No. (Column %) | ≥50 y, No. (Column %) | Total, No. (Column %) |
|---|---|---|---|---|
| Influenza and bacterial diagnosesa | 2 (4) | 4 (5) | 14 (13) | 20 (8) |
| Influenza diagnosis | 19 (41) | 23 (27) | 31 (29) | 73 (31) |
| Bacterial diagnosisa | 5 (11) | 28 (33) | 35 (33) | 68 (29) |
| Exacerbation of chronic condition | 5 (11) | 12 (14) | 12 (11) | 29 (12) |
| Viral diagnosisb | 15 (33) | 15 (18) | 8 (8) | 38 (16) |
| Symptom or other diagnoses | 0 (0) | 2 (2) | 6 (5) | 8 (3) |
| Total | 46 | 84 | 106 | 236 |
a Bacterial infections included the following diagnoses treated with antibiotics: bronchitis, chronic obstructive pulmonary disease exacerbation, otitis media, peritonitis, pharyngitis/strep throat/tonsillitis, pneumonia, sinusitis, and urinary tract infection.
b Because the hierarchical order began with influenza, viral illness excluded those with an influenza diagnosis or those with a diagnosis of viral illness in conjunction with a bacterial infection or exacerbation of chronic condition.
Multivariate analysis of the entire study population identified factors that increased the likelihood a person with laboratory-confirmed influenza received a clinical influenza diagnosis. They included having a high-risk condition, being hospitalized, not having a bacterial diagnosis, having no or unknown insurance, being of Hispanic ethnicity or other or unknown race/ethnicity, and enrollment during the 2009–2010 influenza season (Table 5). Results were similar when the data from the atypical 2009–2010 pandemic influenza season were excluded (data not shown). Clinical suspicion of influenza could be indicated by ordering a rapid influenza test. Sensitivity analysis demonstrated similar results when ordering a rapid influenza test, irrespective of the results, was analyzed (data not shown).
Table 5.
Adjusted Odds Ratios for Influenza Diagnosis From a Multiple Logistic Regression Model Among All 504 Study-Confirmed, Influenza Virus–Positive Patients and From a Model of 236 With High-Risk Medical Conditions
| Characteristics | Adjusted OR (95% CI) |
|---|---|
| Overalla | |
| Race/ethnicity | |
| Non-Hispanic white | Reference |
| Black | 1.1 (.6–2.0) |
| Hispanic, other, or unknown | 2.7 (1.1–6.3) |
| Insurance | |
| Private insurance | 1.5 (.8–2.7) |
| No/unknown insurance | 3.8 (1.4–10.4) |
| Public insurance | Reference |
| Study year | |
| 2009–2010 | 3.1 (1.4–6.5) |
| 2010–2011 | 0.8 (.4–4.5) |
| 2011–2012 | 1.4 (.5–4.5) |
| 2012–2013 | Reference |
| Setting | |
| ED only | Reference |
| Inpatient | 5.4 (2.9–10.1) |
| High-risk condition | |
| Yes | 2.1 (1.2–3.8) |
| No | Reference |
| Bacterial diagnosis | |
| Yes | Reference |
| No | 3.7 (2.0–6.9) |
| High-risk subjectsb | |
| Setting | |
| ED only | Reference |
| Inpatient | 4.2 (2.0–8.9) |
| Bacterial diagnosis | |
| Yes | Reference |
| No | 3.7 (2.0–7.0) |
Abbreviations: CI, confidence interval; ED, emergency department; OR, odds ratio.
a Variables evaluated but not significantly associated with influenza diagnosis were age group, sex, timing in season, influenza vaccination status, and symptom days; these variables were not included in the model.
b Variables evaluated but not significantly associated with influenza diagnosis were age and interaction between age group and setting; age group was included in the model.
Among 236 patients with laboratory-confirmed influenza and high-risk conditions, 133 (56%; 95% CI, 50%–63%) received a clinical diagnosis of influenza. Similar to the overall study populations, setting and bacterial diagnosis remained significant (Table 1). Multivariate analysis of high-risk patients revealed that receipt of an influenza diagnosis was >3-fold more likely in the inpatient setting than in the ED setting and among patients without than among those with a bacterial diagnosis (Table 5).
DISCUSSION
Among persons with laboratory-confirmed influenza, 29% received a clinical diagnosis of influenza. For persons with high-risk conditions, this proportion increased to 56%. Multivariate analysis confirmed that clinicians were more likely to diagnose influenza among persons with high-risk conditions and among those who are sicker, as reflected by the increased diagnosis in the inpatient setting versus the ED setting. Although antiviral treatment is currently recommended for patients with influenza who are hospitalized or have high-risk conditions, many are not receiving a specific diagnosis of influenza or treated with antiviral medication. All patients, including those with high-risk conditions, are less likely to receive a diagnosis of influenza if they are discharged from the ED or receive a diagnosis of bacterial infection and an antibiotic prescription.
The clinical challenge is to identify all patients with influenza for whom an influenza diagnosis would alter the diagnostic evaluation or treatment plan. Many patients with laboratory-confirmed influenza are previously healthy and have had symptoms for ≥2 days. Diagnosing these patients, for whom treatment may not be indicated, with a non-specific viral illness may be sufficient. Although more diagnoses of influenza are made in the inpatient setting than in the ED setting, half of hospitalized patients with influenza are not receiving a diagnosis of or treatment for this disease, despite recommendation of antiviral treatment for such individuals [21, 22]. We found that 14% of patients enrolled and discharged from the ED with confirmed influenza were prescribed antivirals, which is similar to the 15% of outpatients with confirmed influenza being prescribed antivirals in the 2012–2013 US Vaccine Effectiveness Network study [11].
Although the association between influenza and bacterial infections is known [25–29], the proportion of patients with influenza virus infection who present with bacterial coinfection requiring treatment is not well defined. We found that 30% of patients with influenza had a diagnosis of bacterial disease treated with antibiotics and that, of these, only 18% received a diagnosis of influenza. Interestingly, the US Vaccine Effectiveness Network also found that 30% of patients with PCR-confirmed influenza in 1 of 5 ambulatory care centers were prescribed antibiotics and that, of these, 16% were prescribed antivirals [11]. These results are consistent with those from 2 inpatient studies. Dawood et al found that 28% of children hospitalized with laboratory-confirmed influenza had pneumonia [30]. Dao et al reported that 34% of adults hospitalized with laboratory-confirmed influenza had a consolidation or infiltrate on a chest radiograph [31].
Similar to Havers et al, we found that only 18% of patients with a bacterial diagnosis treated with antibiotics and with confirmed influenza received a diagnosis of influenza. It is possible that some patients would not have received a diagnosis of and treatment for a bacterial infection if influenza virus infection had been determined at the time of presentation [32]. Although most patients with acute bronchitis treated in the outpatient setting are not recommended to receive antibiotics [33], a recent study demonstrated that antibiotics are commonly prescribed [34]. These data could suggest that clinicians may be prone to premature closure (ie, suspension of differential diagnostic considerations once a bacterial diagnosis is made). Diagnostic errors among pediatric and adult patients are often multifactorial, with contributions from both system-based factors and cognitive factors [35, 36]. Cognitive factors include inadequate knowledge, data gathering, and information synthesis and verification. Inadequate knowledge is the least common cognitive error, whereas data gathering and synthesis are more common and often cluster with premature closure [37]. Interestingly, pediatricians self-reported that their most common diagnostic error was misdiagnosing a viral illness as a bacterial disease [36].
Our results are consistent with those from other published studies and highlight the clinical complexity of accurately diagnosing influenza among children and adults [38]. Clinical prediction rules can be helpful but are imperfect, reflecting that no symptom or set of symptoms has high specificity for influenza diagnosis [13, 16–18]. Our study demonstrates the broad range of symptoms among persons with influenza and demonstrates that variation occurs both across age groups and clinical settings. We, like others [7, 11, 18, 32], find that influenza is frequently not diagnosed or treated even among hospitalized and high-risk patients for whom antivirals are recommended [21, 22, 39].
We found that patients who are hospitalized and who have high-risk conditions are more likely to receive an influenza diagnosis, indicating that clinicians have a higher suspicion of influenza for this population. Multivariate analyses also found increased odds of an influenza diagnosis among patients with no insurance and among patients of Hispanic ethnicity or other or unknown race/ethnicity. The reasons for these findings are unclear but could relate to systematic differences in exposure, such as household size, the timing of presentation by symptom day or relative to the influenza peak, patient-reported symptoms, clinical suspicion of influenza, or patient request for a specific explanation for their symptoms.
This study has several strengths, including a novel approach to identifying predictors of influenza diagnosis among patients with confirmed influenza. The study population systematically enrolled persons of all ages presenting with fever or acute respiratory illnesses in inpatient and ED settings over 4 consecutive seasons. The study population all had molecular testing for influenza virus. Symptoms were determined by self-report or parent report at enrollment and not by chart review.
This study has several limitations. Patients who were and those who were not enrolled in the study may systematically differ. For example, patients not immunized against influenza might theoretically be less likely to participate in an influenza study. All patients were enrolled in one community, and results may vary geographically; however, our data on prescriptions of antivirals and antibiotics among patients with confirmed influenza were similar to those from the US Influenza Vaccine Effectiveness Network in 5 ambulatory care centers in 2012–2013 [11]. Other factors not measured may increase the likelihood of influenza diagnosis, such as media coverage and known influenza exposure. For example, national media coverage of the 2009–2010 pandemic likely contributed to a higher proportion of patients with laboratory-confirmed influenza receiving a diagnosis of influenza than in other years.
Fewer than half of patients with laboratory-confirmed influenza who have high-risk conditions or who are hospitalized receive a clinical diagnosis of influenza. A diagnosis of influenza would be important for all who would have their clinical treatment altered if the diagnosis of influenza was known. We found that the odds of an influenza diagnosis were >3-fold lower for all patients with a bacterial diagnosis, including those with high-risk conditions. Thus, during the influenza season, clinicians should consider whether persons with symptoms consistent with a bacterial infection could also have influenza and whether coinfection with influenza would alter the treatment recommendations. Future studies are needed to provide effective strategies to consistently identify and treat patients recommended to receive influenza antiviral medications according to the current recommendations, as well as the costs and benefits of this approach.
Notes
Acknowledgments. We thank all the patients and their families who participated in this study; all of the physicians and staff in the emergency departments at Forsyth Medical Center and Wake Forest Baptist Health, for making this study possible; Lauren Vannoy and Shannon Major at Wake Forest Baptist Health; Brian Coleman, James Hobbs, Wendy Hobbs, Eugenia Hutchinson, Mena Isnassuos, Brian Miller, Debra Norwood, Keisha Rodriguez, Robert Romanchuk, and Yvonne Whitley at Novant Clinical Research Institute, for facilitating enrollment or enrolling eligible patients; Lauren Vannoy, James Hobbs, and Brian Miller, for entering data into the database; Elizabeth Blakeney, for performing all viral cultures and reverse transcription–polymerase chain reaction analyses; and the anonymous reviewers, for enhancing the manuscript with their comments and suggestions.
Disclaimer. The views expressed in this article are solely those of the authors and do not necessarily represent the official views of the National Institutes of Health (NIH) or the US government.
Financial support. This work was supported by the NIH (grants R01 AI079226 and 5T35DK007400-34) and the Wachovia Research Fund.
Potential conflicts of interest. K. A. P. and T. R. P. have received research support from BD Diagnostics and MedImmune. All other authors report no potential conflicts.
All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
References
- 1. World Health Organization. Vaccines against influenza WHO position paper—November 2012. Wkly Epidemiol Rep 2012; 87:461–76. [PubMed] [Google Scholar]
- 2.Molinari NA, Ortega-Sanchez IR, Messonnier ML et al. The annual impact of seasonal influenza in the US: measuring disease burden and costs. Vaccine 2007; 25:5086–96. [DOI] [PubMed] [Google Scholar]
- 3.Fairbrother G, Cassedy A, Ortega-Sanchez IR et al. High costs of influenza: Direct medical costs of influenza disease in young children. Vaccine 2010; 28:4913–9. [DOI] [PubMed] [Google Scholar]
- 4.Ortega-Sanchez IR, Molinari NA, Fairbrother G et al. Indirect, out-of-pocket and medical costs from influenza-related illness in young children. Vaccine 2012; 30:4175–81. [DOI] [PubMed] [Google Scholar]
- 5.Bilcke J, Coenen S, Beutels P. Influenza-like-illness and clinically diagnosed flu: disease burden, costs and quality of life for patients seeking ambulatory care or no professional care at all. PLoS One 2014; 9:e102634. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Peters TR, Suerken CK, Snively BM et al. Influenza testing, diagnosis, and treatment in the emergency department in 2009–2010 and 2010–2011. Acad Emerg Med 2013; 20:786–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Poehling KA, Edwards KM, Griffin MR et al. The burden of influenza in young children, 2004–2009. Pediatrics 2013; 131:207–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Falsey AR, Walsh EE. Viral pneumonia in older adults. Clin Infect Dis 2006; 42:518–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Walsh EE, Cox C, Falsey AR. Clinical features of influenza A virus infection in older hospitalized persons. J Am Geriatr Soc 2002; 50:1498–503. [DOI] [PubMed] [Google Scholar]
- 10.Zhou H, Thompson WW, Viboud CG et al. Hospitalizations associated with influenza and respiratory syncytial virus in the United States, 1993–2008. Clin Infect Dis 2012; 54:1427–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Havers F, Thaker S, Clippard JR et al. Use of influenza antiviral agents by ambulatory care clinicians during the 2012–2013 influenza season. Clin Infect Dis 2014; 59:774–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Zambon MC, Stockton JD, Clewley JP, Fleming DM. Contribution of influenza and respiratory syncytial virus to community cases of influenza-like illness: an observational study. Lancet 2001; 358:1410–6. [DOI] [PubMed] [Google Scholar]
- 13.Woolpert T, Brodine S, Lemus H, Waalen J, Blair P, Faix D. Determination of clinical and demographic predictors of laboratory-confirmed influenza with subtype analysis. BMC Infect Dis 2012; 12:129. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Call SA, Vollenweider MA, Hornung CA, Simel DL, McKinney WP. Does this patient have influenza? JAMA 2005; 293:987–97. [DOI] [PubMed] [Google Scholar]
- 15.Jutel A, Banister E. “I was pretty sure I had the 'flu”: qualitative description of confirmed-influenza symptoms. Soc Sci Med 2013; 99:49–55. [DOI] [PubMed] [Google Scholar]
- 16.Ebell MH, Afonso AM, Gonzales R, Stein J, Genton B, Senn N. Development and validation of a clinical decision rule for the diagnosis of influenza. J Am Board Fam Med 2012; 25:55–62. [DOI] [PubMed] [Google Scholar]
- 17.Brim SN, Rudd RA, Funk RH, Callahan DB, Euler GL. Influenza vaccination coverage among children with asthma--United States, 2004–05 influenza season. MMWR Morb Mortal Wkly Rep 2007; 56(RR-9):193–6. [PubMed] [Google Scholar]
- 18.Mulpuru S, Roth VR, Lawrence N, Forster AJ. Influenza infection screening tools fail to accurately predict influenza status for patients during pandemic H1N1 influenza season. Can Respir J 2013; 20:e55–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Chartrand C, Leeflang MM, Minion J, Brewer T, Pai M. Accuracy of rapid influenza diagnostic tests: a meta-analysis. Ann Intern Med 2012; 156:500–11. [DOI] [PubMed] [Google Scholar]
- 20.Uyeki TM, Prasad R, Vukotich C et al. Low sensitivity of rapid diagnostic test for influenza. Clin Infect Dis 2009; 48:e89–92. [DOI] [PubMed] [Google Scholar]
- 21.Harper SA, Bradley JS, Englund JA et al. Seasonal influenza in adults and children--diagnosis, treatment, chemoprophylaxis, and institutional outbreak management: clinical practice guidelines of the Infectious Diseases Society of America. Clin Infect Dis 2009; 48:1003–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Fiore AE, Fry A, Shay D, Gubareva L, Bresee JS, Uyeki TM. Antiviral agents for the treatment and chemoprophylaxis of influenza --- recommendations of the Advisory Committee on Immunization Practices (ACIP). MMWR Recomm Rep 2011; 60(RR-1):1–24. [PubMed] [Google Scholar]
- 23.Fiore AE, Shay DK, Broder K et al. Prevention and control of seasonal influenza with vaccines: recommendations of the Advisory Committee on Immunization Practices (ACIP), 2009. MMWR Recomm Rep 2009; 58(RR-8):1–52. [PubMed] [Google Scholar]
- 24.Treanor JJ, Talbot HK, Ohmit SE et al. Effectiveness of seasonal influenza vaccines in the United States during a season with circulation of all three vaccine strains. Clin Infect Dis 2012; 55:951–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Morens DM, Taubenberger JK, Fauci AS. Predominant role of bacterial pneumonia as a cause of death in pandemic influenza: implications for pandemic influenza preparedness. J Infect Dis 2008; 198:962–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Brundage JF, Shanks GD. Deaths from bacterial pneumonia during 1918–19 influenza pandemic. Emerg Infect Dis 2008; 14:1193–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Falsey AR, Becker KL, Swinburne AJ et al. Bacterial complications of respiratory tract viral illness: a comprehensive evaluation. J Infect Dis 2013; 208:432–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Louie J, Jean C, Chen T-H et al. Bacterial coinfections in lung tissue specimens from fatal cases of 2009 pandemic influenza A (H1N1)—United States, May-August 2009. MMWR Morb Mortal Wkly Rep 2009; 58(RR-38):1071–4. [PubMed] [Google Scholar]
- 29.O'Brien L, Donegan N, Flaniken A et al. Severe coinfection with seasonal influenza A (H3N2) virus and Staphylococcus aureus—Maryland, February-March 2012. MMWR Morb Mortal Wkly Rep 2012; 61(RR-16):289–91. [PubMed] [Google Scholar]
- 30.Dawood FS, Chaves SS, Perez A et al. Complications and associated bacterial coinfections among children hospitalized with seasonal or pandemic influenza, United States, 2003–2010. J Infect Dis 2014; 209:686–94. [DOI] [PubMed] [Google Scholar]
- 31.Dao CN, Kamimoto L, Nowell M et al. Adult hospitalizations for laboratory-positive influenza during the 2005–2006 through 2007–2008 seasons in the United States. J Infect Dis 2010; 202:881–8. [DOI] [PubMed] [Google Scholar]
- 32.Martinot M, Heller R, Martin A et al. Contribution of systematic RT-PCR screening for influenza during the epidemic season. Med Mal Infect 2014; 44:123–7. [DOI] [PubMed] [Google Scholar]
- 33.Braman SS. Chronic cough due to acute bronchitis: ACCP evidence-based clinical practice guidelines. Chest 2006; 129:95S–103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Barnett ML, Linder JA. Antibiotic prescribing for adults with acute bronchitis in the United States, 1996–2010. JAMA 2014; 311:2020–2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Graber ML, Franklin N, Gordon R. Diagnostic error in internal medicine. Arch Intern Med 2005; 165:1493–9. [DOI] [PubMed] [Google Scholar]
- 36.Singh H, Thomas EJ, Wilson L et al. Errors of diagnosis in pediatric practice: a multisite survey. Pediatrics 2010; 126:70–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Arnold JC, Singh KK, Milder E et al. Human metapneumovirus associated with central nervous system infection in children. Pediatr Infect Dis J 2009; 28:1057–60. [DOI] [PubMed] [Google Scholar]
- 38.Redlberger-Fritz M, Hirk S, Buchinger D et al. Distinct differences in clinical manifestation and viral laboratory parameters between children and adults with influenza A(H1N1)pdm09 infection--a retrospective comparative analysis. J Med Virol 2014; 86:1048–55. [DOI] [PubMed] [Google Scholar]
- 39.Biggerstaff M, Jhung MA, Reed C, Fry AM, Balluz L, Finelli L. Influenza-like illness, the time to seek healthcare, and influenza antiviral receipt during the 2010–2011 influenza season-United States. J Infect Dis 2014; 210:535–44. [DOI] [PMC free article] [PubMed] [Google Scholar]