Clinical Predictors of Influenza in Children

ARTICLE Clinical Predictors of Influenza in Children

Marla J. Friedman, DO; Magdy W. Attia, MD

Background: It is difficult to diagnose influenza infec- Results: The mean±SD age of patients was 6.2±5.2 years; tion on clinical grounds alone. Available rapid diagnos- 50% were boys. Viral isolates included the following: in- tic tests have limited sensitivities. fluenza A, 45 patients (35%); influenza B, 13 (10%); other viruses, 10 (8%); negative results, 60 (47%). Demo- Objective: To develop a prediction model that identi- graphic and clinical findings were not significantly differ- fies children likely to have influenza infection. ent between the influenza A and influenza B groups. Cough (P=.003), headache (P=.04), and pharyngitis (P=.04) were Design: Prospective study. independently associated with influenza infection. This triad used as a prediction model for influenza infection had a Setting: Emergency department of a children’s hospital. sensitivity of 80% (95% confidence interval [CI], 69%- 91%); specificity, 78% (95% CI, 67%-89%); and likeli- Patients: All patients with a febrile respiratory illness hood ratio for a positive viral culture for influenza, 3.7 (95% during the influenza season of winter 2002 were eli- CI, 2.3-6.3). The posttest probability of this clinical defi- gible. A prospective sample of 128 children who were sus- nition is 77% (95% CI, 63%-91%). pected of having influenza infection based on predetermined criteria was enrolled. Each patient received a nasal Conclusion: The triad of cough, headache, and phar- wash for viral culture. yngitis is a predictor of influenza infection in children.

Main Outcome Measure: Clinical features that are most predictive of influenza infection in children. Arch Pediatr Adolesc Med. 2004;158:391-394

NFLUENZA IS A COMMON FEBRILE dren, but no characteristic symptom or illness with a significant impact symptom complex has been identified. Re- on the pediatric population. Dur- ported manifestations of influenza in chil- ing annual outbreaks, 15% to 20% dren include abrupt onset of high fever, of children are infected with in- coryza, cough, sore throat, vomiting, di- Ifluenza.1,2 Estimates of annual outpatient arrhea, abdominal pain, fatigue, head- visits attributable to influenza range from ache, and myalgias. The classic symp- 6 to 29 per 100 visits.3,4 School absentee- toms often associated with influenza in ism, parental work absenteeism, and sec- adults are not easily identified in chil- ondary illness among family members are dren. all significantly higher during influenza The gold standard for diagnosing in- season than throughout the rest of the win- fluenza infection is viral isolation by cul- 5 From the Department of ter. Infants and young children are hos- ture. However, viral culture results are not Pediatrics, Division of pitalized for influenza-associated ill- available in time to influence patient man- Emergency Medicine, nesses at rates comparable with those of agement. The rapid tests available to di- Alfred I. duPont Hospital for high-risk adults and elderly patients.3,6 Fur- agnose influenza are limited by their di- Children, Wilmington, Del; and thermore, influenza is commonly impli- agnostic abilities. A review of the studies Jefferson Medical College, cated as a cause of nosocomial infections performed on commercially available rapid Thomas Jefferson University, in pediatric inpatient units.7,8 tests reveals great variability in the sensi- Philadelphia, Pa. Dr Friedman is now with the Department of The nonspecific presentation of in- tivities (40%-100%), specificities (63%- Pediatrics, Division of fluenza infection makes it difficult to dis- 100%), positive predictive values (PPVs) Emergency Medicine, Miami tinguish from other febrile or respiratory (43%-100%), and negative predictive val- Children’s Hospital, illnesses. Many clinical features have been ues (56%-100%) of these tests. The higher Miami, Fla. associated with influenza illness in chil- sensitivities are often difficult to repro-

(REPRINTED) ARCH PEDIATR ADOLESC MED/ VOL 158, APR 2004 WWW.ARCHPEDIATRICS.COM 391

©2004 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/26/2021 duce in clinical practice, and false-negative results are This study was approved with consent waiver by the in- likely to occur.9 As predictive values are affected by dis- stitutional review board of our institution (Alfred I. duPont Hos- ease prevalence, these diagnostic tests are most helpful pital for Children). during periods of high influenza activity. However, it seems impractical and inefficient to test each patient with STUDY PERIOD suspected influenza infection, even during peak season. The beginning of influenza season was defined as the period The diagnostic abilities of these tests can be enhanced if following the detection of 2 consecutive isolates in one week patients with a high likelihood of influenza infection are of either influenza A or influenza B from viral culture in the selected for testing based on a set of clinical criteria. Dif- community. The season ended with the identification of the ferentiating influenza infection from infection caused by last of 2 consecutive isolates in one week. Patients were en- other respiratory pathogens is important for several rea- rolled from January 14, 2002, through March 29, 2002. sons. From a public health perspective, understanding the magnitude of illness attributable to influenza is cru- VIROLOGIC ANALYSIS cial for surveillance data across communities and states. Local and national surveillance is useful for predicting Viral culture specimens from nasal washes were inoculated onto human lung tissue (MRC-5) cells, rhesus monkey kid- health care resource needs (hospital beds, staffing, sup- ney cells, human embryonic kidney (HEK) cells, human epi- ply of diagnostic tests and antivirals needed) and for de- dermoid laryngeal carcinoma (Hep-2) cells, and human lung 10 tecting pandemics. Furthermore, children are consid- carcinoma (A-549) cells. These cells were incubated at ered an important factor in this process by introducing 34.5°C and observed daily for 28 days for cytopathic effect. infection into the home and transmitting it to adult and All viruses isolated by culture were confirmed by immuno- elderly patients.1 Timely diagnosis is imperative for the fluorescence staining. initiation of appropriate antiviral treatment and for the proper isolation of hospitalized patients. Both of these STATISTICAL ANALYSIS interventions may aid in reducing the overall disease impact. These many factors argue for enhancing the clini- The primary outcome measures were the clinical features that are most predictive of influenza infection in children. cal prediction of this disease. Interobserver agreement was evaluated for each of the clini- The purpose of this study was to identify a clinical cal examination findings. Two physicians performed separate prediction model for influenza infection in children. examinations and recorded their observations independently. Data was analyzed for agreement rates, and by ␬ statistics. METHODS Sample size was calculated to detect a 30% difference between assumed prevalence of influenza in tested patients (a 50-50 DESIGN AND DEFINITIONS chance, or 50%) and the hypothesis that a prediction model will possess a sensitivity that is approximately equal to rapid This prospective study was conducted during the 2002 influ- test (80%). After continuity correction, a sample size of 45 pa- enza season (January-March). All patients (birth through age tients in each of 2 groups representing influenza-positive and 17 years) who were seen in the emergency department of a sub- influenza-negative patients was estimated. urban tertiary pediatric center with a febrile respiratory illness Demographic and clinical findings of patients with influ- were eligible. A consecutive sample of children suspected of enza A and influenza B were compared with each other using ϫ having influenza infection were enrolled. Enrollment criteria 2 2 contingency tables, independent sample t test, or Mann- included fever and at least one of the following symptoms: co- Whitney U test, depending on their parametric distribution. A ryza, cough, headache, sore throat, or muscle aches. Fever was univariate analysis was then performed on all influenza pa- defined as an emergency department temperature higher than tients as a group compared with those patients who had no vi- 38°C or history of similar temperature within the previous 24 ral pathogen isolated. A binary logistic regression analysis us- hours. All enrolled patients received a nasal wash for viral cul- ing backwards stepwise elimination was performed to identify ture. At the time of evaluation, a standardized data collection variables independently associated with influenza infection. Fe- form was completed by the physician for each enrolled pa- ver greater than or equal to 39°C was entered as a dichoto- tient. Demographic information, duration of illness, day care mized variable. To simplify the analysis, the remainder of the attendance, and triage vital signs were recorded for each pa- vital signs were not included. All tests were 2-tailed and con- Յ tient. The presence or absence of 24 clinical features (12 his- sidered significant at P .05. Statistical analyses were per- torical features and 12 physical examination findings) was noted formed using SPSS version 11.5 (SPSS Inc, Chicago, Ill). on the same form. The historical features were recorded as di- chotomous variables as relayed by the caregiver at the time of evaluation. These features included fever, rhinorrhea, cough, RESULTS vomiting, diarrhea, decreased oral intake, decreased activity, headache, abdominal pain and/or nausea, sore throat, muscle DEMOGRAPHICS AND aches, and apnea. Pertinent physical examination findings con- VIROLOGIC TEST RESULTS sisted of coryza (rhinorrhea and/or nasal congestion), cough (observed during examination), wheezing, rales, retractions, During the study period, influenza tests were ordered on nasal flaring, rash (other than chronic dermatitis), conjuncti- 130 patients who were suspected of having influenza in- vitis (conjunctival injection and/or discharge), pharyngitis (ery- thema, swelling and/or exudate of the tonsils and/or phar- fection based on a predetermined set of criteria. Two pa- ynx), cervical adenopathy (enlarged, tender anterior cervical tients were excluded because they had no viral culture nodes without evidence of acute suppurative adenitis), otitis performed due to inadequate samples. Of the 128 eli- media (abnormal tympanic membrane appearance and de- gible patients, 45 (35%) had viral cultures positive for creased mobility), and abdominal tenderness. influenza A and 13 (10%) for influenza B. Ten patients

(REPRINTED) ARCH PEDIATR ADOLESC MED/ VOL 158, APR 2004 WWW.ARCHPEDIATRICS.COM 392

©2004 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/26/2021 Table 1. Demographics and Vital Signs of Patients Table 2. Univariate Analysis of Clinical Features With and Without Influenza* of Influenza-Positive Patients

Positive for Negative Frequency of Influenza Culture Findings P Reported Characteristic (n = 58) (n = 60) Value Finding Symptom, % OR (95% CI) Age, y 7 ± 5 6 ± 6 .22 Historical Sex, % male 50 50 Ͼ.99 Fever, Ն39°C 55 2.4 (1.2-5.0) Duration of symptoms, d 2.1 ± 1 2.6 ± 3 .21 Rhinorrhea 60 0.8 (0.4-1.7) Attended day care, % 41 40 Ͼ.99 Cough 83 2.2 (1.0-5.2) Temperature, °C 39.1 ± 1 38.7 ± 1 .06 Vomiting 36 1.6 (0.7-3.4) Temperature Ն39°C, % 57 37 .03 Diarrhea 10 0.8 (0.3-2.2) Heart rate, beats/min 138 ± 26 141 ± 32 .57 Decreased intake 57 1.0 (0.5-2.0) Respiratory rate, breaths/min 28 ± 8 21 ± 14 .03 Decreased activity 54 1.8 (0.8-3.6) Room air, % oxygen saturation 97 ± 2 96 ± 3 .10 Headache* 44 2.6 (1.2-5.8) Abdominal pain and/or nausea* 31 3.2 (1.2-8.2) *Values are expressed as mean ± SD unless otherwise noted. Sore throat 37 1.5 (0.7-3.2) Myalgia* 33 0.8 (0.4-1.8) Apnea† 0 0.5 (0.0-3.9) (8%) had other viral pathogens isolated (respiratory Clinical Coryza 57 1.6 (0.8-3.2) syncytial virus, 6; adenovirus, 3; rhinovirus, 1), and 60 Cough (observed) 33 1.3 (0.6-2.9) specimens (47%) were negative. Study patients had a Wheezing 5 0.3 (0.1-1.0) mean±SD age of 6.2±5.2 years and an equal sex distri- Rales 3 0.5 (0.1-2.4) bution (50% male). Retractions 0 0.1 (0.0-0.6) Nasal flaring 0 0.2 (0.0-1.2) CLINICAL CHARACTERISTICS OF INFLUENZA Rash 10 1.6 (0.3-8.1) Conjunctivitis 3 0.6 (0.1-3.7) Pharyngitis 35 2.0 (0.9-4.3) There was no significant difference between patients with Cervical adenopathy 14 1.7 (0.6-5.4) influenza A and influenza B in demographic or clinical Otitis media 10 1.0 (0.3-3.3) findings (PՆ.05 for all variables). Univariate analysis com- Abdominal tenderness 12 1.8 (0.5-6.0) paring all influenza patients with the viral-negative group detected a significant difference in their respiratory rates Abbreviations: CI, confidence interval; OR, odds ratio. *Symptoms were assessed in patients 3 years old or older (78 patients (28 vs 21 breaths/min; P=.03). The remainder of vital [66%]). signs was not significantly different. However, the influ- †Symptom was assessed in patients 1 year old or younger (15 patients enza patients were more likely to have temperature greater [12%]). than or equal to 39°C (57% vs 37%; P=.03) (Table 1). History of cough (odds ratio [OR], 2.2; 95% confidence Fifty-six pairs of observations were recorded. The interval [CI], 1.0-5.2), headache (OR, 2.6; 95% CI, 1.2- agreement rates of the variables in the model were as fol- 5.8), abdominal pain and/or nausea (OR, 3.2; 95% CI, lows: cough, 88% (␬=0.6); headache, 88% (␬=0.8); and 1.2-8.2), and clinical evidence of pharyngitis (OR, 2.0; pharyngitis, 76% (␬=0.5). 95% CI, 0.9-4.3) were more common in the influenza group (Table 2). COMMENT THE PREDICTIVE MODEL Influenza infection is a substantial cause of health care utilization and subsequent morbidity in the pediatric Backwards stepwise logistic regression analysis identi- population. The wide range of influenza-associated fied cough (adjusted OR, 7.2; 95% CI, 2-27; P=.003), symptoms in children often makes the clinical diagno- headache (adjusted OR, 4.3; 95% CI, 1-17; P=.04), and sis difficult. The diagnosis is further complicated by pharyngitis (adjusted OR, 3.9; 95% CI, 1-14; P=.04) as other respiratory viruses that cocirculate with influenza independent predictors of influenza infection in chil- and cause similar symptoms. Because influenza is po- dren. Hosmer-Lemeshow goodness-of-fit test results were tentially preventable by proper isolation and treatable not significant (␹2, 10.2; P=.25), which statistically sup- by antiviral agents, timely diagnosis is important for re- ports the robustness of this prediction model. The triad duction of disease spread and management of indi- of cough and headache by history and clinical finding of vidual cases. pharyngitis used as a prediction model for influenza had Traditionally, the laboratory diagnosis of influenza a sensitivity of 80% (95% CI, 69%-91%), a specificity of is made by isolating the virus in culture. However, this 78% (95% CI, 67%-89%), a PPV of 77% (95% CI, 61%- standard is impractical in clinical practice because of the 88%), a negative predictive value of 81% (95% CI, 70%- time required to obtain results. Rapid diagnostic tests are 92%), a likelihood ratio for positive viral culture find- available to aid in diagnosis. Their use is limited by their ings for influenza of 3.7 (95% CI, 2.3-6.3), and a likelihood variable availability in health care centers and their rela- ratio for negative culture findings of 0.26 (95% CI, tively low sensitivity compared with viral culture.10 In 0.14-0.44). The posttest probability of this clinical tool practice, most physicians rely on clinical symptoms to is 77% (95% CI, 63%-91%). diagnose influenza infection. However, studies compar-

(REPRINTED) ARCH PEDIATR ADOLESC MED/ VOL 158, APR 2004 WWW.ARCHPEDIATRICS.COM 393

©2004 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/26/2021 What This Study Adds season. The data were collected in a suburban tertiary care hospital, and the results may not be generalizable in all patient populations. Additional studies are needed Influenza is a frequent cause of outpatient visits, hospi- at different sites during another influenza season to vali- talizations, and nosocomial infections in children. It is date this model prospectively. often difficult to distinguish influenza from other fe- In conclusion, the clinical triad of cough, head- brile or respiratory illnesses on clinical grounds alone. The classic symptoms of influenza observed in older chil- ache, and pharyngitis accurately predicts influenza in- dren and adults are not easily identified in young chil- fection in febrile children during a community out- dren. The rapid diagnostic tests available to aid in the break. When influenza is circulating in the community, diagnosis of influenza are limited by their sensitivities. this pediatric prediction model can be used when con- Previous studies have identified clinical predictors of in- sidering management options for affected children. Fur- fluenza, but none have focused on infection in chil- thermore, this model is likely to improve the diagnostic dren. Our study found that cough, headache, and phar- abilities of rapid tests by identifying a subset of patients yngitis were significantly associated with influenza with a high likelihood of influenza infection to be tested. infection in children. The sensitivity of this pediatric model is comparable with that of rapid influenza tests, Accepted for publication December 11, 2003. and the 2 may be used in conjunction to improve clinical decision making. This study was presented in part at the Pediatric Aca- demic Societies Emergency Medicine Special Interest Group; May 3, 2003; Seattle, Wash; and at the Society for Aca- demic Emergency Medicine National Meeting; May 29, 2003; ing the accuracy of clinical diagnosis with laboratory- Boston, Mass. confirmed influenza reveal a wide range of PPVs Corresponding author and reprints: Marla J. Fried- (18%-87%).11-16 Several of these studies in the literature man, DO, Division of Emergency Medicine, Miami Chil- on adult patients include children in their methods; how- dren’s Hospital, 3100 SW 62nd Ave, Miami, FL 33155- ever, we are aware of no study that focuses exclusively 3009 (e-mail: [email protected]). on children. In our study, cough, headache, and pharyngitis were REFERENCES significantly associated with positive viral culture findings for influenza in children. The sensitivity of this model 1. Glezen WP. Emerging infections: pandemic influenza. Epidemiol Rev. 1996;18: is comparable with the rapid influenza tests (80%; 95% 64-76. CI, 69%-91%). Additionally, the likelihood ratio is 3.7 2. Neuzil KM, Zhu Y, Griffin MR, et al. Burden of interpandemic influenza in children younger than 5 years: a 25-year prospective study. J Infect Dis. 2002;185: (95% CI, 2.3-6.3). Likelihood ratios are a powerful clini- 147-152. cal tool that help clinicians determine how much more 3. Neuzil KM, Mellen BG, Wright PF, Mitchel EF, Griffin MR. The effect of influenza likely a test result is to occur in someone with the dis- on hospitalizations, outpatient visits, and courses of antibiotics in children. N Engl J Med. 2000;342:225-230. ease than in someone without the disease. The higher the 4. Glezen WP, Decker M, Joseph SW, Mercready RG Jr. Acute respiratory disease likelihood ratio, the better the test result is for ruling in associated with influenza epidemics in Houston, 1981-1983. J Infect Dis. 1987; a diagnosis. This information is used with the pretest prob- 155:1119-1126. 5. Neuzil KM, Hohlbein C, Zhu Y. Illness among schoolchildren during influenza sea- ability to estimate the posttest probability (or PPV) of dis- son. Arch Pediatr Adolesc Med. 2002;156:986-991. ease for that patient. Based on the gold standard of viral 6. Izurieta HS, Thompson WW, Kramarz P, et al. Influenza and the rates of hospi- talization for respiratory disease among infants and young children. N Engl J Med. culture, this model had a posttest probability of 77% (95% 2000;342:232-238. CI, 63%-91%). This is a significant improvement over our 7. Serwint JR, Miller RM. Why diagnose influenza infection in hospitalized pediat- physicians’ pretest probability of 45% using an inclu- ric patients? Pediatr Infect Dis J. 1993;12:200-204. 8. Hall CB. Nosocomial viral respiratory infections: perennial weeds on pediatric sive list of influenza-associated symptoms. wards. Am J Med. 1981;70:670-676. Adult studies have identified the complexes of fe- 9. Uyeki TM. Influenza diagnosis and treatment in children: a review of studies on ver and cough,11-13 or fever, cough, and abrupt onset,14,15 clinically useful tests and antiviral treatment for influenza. Pediatr Infect Dis J. 2003;22:164-177. as the best clinical predictors of influenza infection. Al- 10. Cheung M, Lieberman JM. Influenza: update on strategies for management. Con- though the difference in temperatures between our 2 temp Pediatr. 2002;19:82-94. 11. Boivin G, Hardy I, Tellier G, Maziade J. Predicting influenza infections during epi- groups of patients neared statistical significance demics with use of a clinical case definition. Clin Infect Dis. 2000;31:1166- (39.1°C in the influenza group vs 38.7°C in the negative 1169. group; P=.06), the difference is clinically negligible. 12. Zambon M, Hays J, Webster A, Newman R, Keene O. Diagnosis of influenza in the community: relationship of clinical diagnosis to confirmed virological, sero- This difference in the mean temperature could explain logic, or molecular detection of influenza. Arch Intern Med. 2001;161:2116- the higher respiratory rates in the influenza patients. 2122. Myalgias, which are classically associated with influ- 13. Monto AS, Gravenstein S, Elliott M, Colopy M, Schweinle J. Clinical signs and symptoms predicting influenza infection. Arch Intern Med. 2000;160:3243- enza infection, were not a clinical predictor in either 3247. our model or any of the quoted adult definitions. Nota- 14. Van Elden LJ, Van Essen GA, Boucher CA, et al. Clinical diagnosis of influenza virus infection: evaluation of diagnostic tools in general practice. Br J Gen Pract. bly, the triad of cough, headache, and pharyngitis is 2001;51:630-634. simple to elicit clinically, and each feature had a high 15. Govaert TM, Dinant GJ, Aretz K, Knottnerus JA. The predictive value of influenza agreement rate between observers. symptomatology in elderly people. Fam Pract. 1998;15:16-22. 16. Carrat F, Tachet A, Rouzioux C, Housset B, Valleron AJ. Evaluation of clinical case A recognized limitation of this study is that it was definitions of influenza: detailed investigation of patients during the 1995-1996 conducted in a single setting during a single influenza epidemic in France. Clin Infect Dis. 1999;28:283-290.

(REPRINTED) ARCH PEDIATR ADOLESC MED/ VOL 158, APR 2004 WWW.ARCHPEDIATRICS.COM 394