University of Alberta



Measuring influenza and ILI – Some thoughts

Clinical syndrome of influenza

- Not specific

- May not be sensitive

- Evidence from high prevalence studies – good sens and spec but low NPV.

- More likely that prevalence is lower – Govarert et al. – bad PPV.

In any event, clinical diagnoses would require manual chart reviews – we have ICD data only.

- Codes identified by earlier study.

- Increase sensitivity for influenza

- Not specific for influenza

- Previous work on sens/spec of ICD codes for influenza

We are stuck good. But will get through this using the on vs off-season idea.

- Cover Serfling models, Farr’s insight, etc.

- Use studies showing that ICD codes can predict influenza epidemics.

Diagnosing influenza and influenza-like illness – in the clinic, and in administrative data.

Govaert TM, Dinant GJ, Aretz K, Knottnerus JA. The predictive value of influenza symptomatology in elderly people. Family Practice, 15(1): 1998.

- Objective – Determine the complex of symptoms which ahs the highest predictive value for the diagnosis of influenza.

- Cross-sectional diagnostic study – part of a prospective RCT

o 1992-1992 influenza season

o 34 GPs in 15 practices in the Netherlands.

o Patients aged 60 or older.

o Gold standard – serology.

o Patient followed for a vaccination RCT, data measured at time of presentation with ILI.

- Detecting ILI

o GP registration of symptoms of patients presenting with ILI.

o All participants also asked to fill out and return a questionnaire asking about possible influenza episodes.

o Cases ascertained by GP application of ICHPPC-2 criteria, or by researcher application of ACHPPC-2 criteria to patients questionaires.

o All participants had acute and convalescent-phase serology.

- Patients enrolled – 927 in vaccine arm, 911 in placebo arm.

- 121 cases of influenza serologically confirmed. 6.6% prevalence.

- GP reports – 48 cases of ILI, of which 17 were serologically confirmed.

- Participant questionnaires – 144 cases found, 35 serologically confirmed.

- Symptoms – Highest OR for coughing, fever, then acute onset, and malaise.

o Coughing sens = 66, spec = 76, PPV = 17%, NPV = 97%.

o Fever sens = 34, spec = 91, PPV = 21%, NPV = 95%.

o Acute onset sens = 51, spec = 81, PPV = 16%, NPV = 96%.

- Highest NPV – Absence of …

o Coughing 0.97.

o Fever, 0.95.

o Acute onset or malaise, 0.96 each.

o PPV quite low, though.

- Specific combinations of symptoms have a higher predictive value than the presence of all symptoms.

o Combination of fever, coughing, and acute onset – PPV = 30.3%.

o If this was used for only those patients who presented to medical attention, the PPV would be 44%.

- Highlights the fact that health-care based influenza studies examine a different set of patients – those presenting to medical attention.

- Differences from Monto et al. and Boivin et al. show the importance of underlying prevalence of disease for PPV and NPV. Uncertain if this explains differences in sensitivity and specificity.

Boivin G, Hardy I, Tellier G, Maziade J. Predicting influenza infections during epidemics with use of a clinical case definition. Clin Infect Dis, 31(5): 2000.

- Objective – Validate the use of a clinical case definition for the diagnosis of influenz infections by sentinel physicians. Identify the best clinical predictors of influenza infections.

- Cross-sectional diagnostic study

o 3 outpatient clinics in Quebec

o Patients presenting with a flu-like illness of /= 55 years – OR = 1.60

- Diagnostic value

o Fever alone, PPV = 77%, NPV = 49%, sens = 68%, spec = 60%.

o Cough was sensitive (93%) but not specific (20%).

o The best PPV (85%) was obtained by fever and cough with onset between 36-48 hours prior to presentation, but sens (50%) was low for this (spec = 81%).

o Fever and cough (without respect to time of onset relative to presentation) had PPV = 77%, NPV = 51%, sens = 63%, and spec = 68%.

- As with Zambon, a surprising large number of patients recruited were positive for influenza (2470 (66%) vs 1274 (34%)). Probably because patients were included if they already had fever, etc..

- I.e.: spectrum bias – “patients who enroll in RCTs assessing either treatment or prevention of influenza may not represent the population of persons [presenting to a primary care office.” (Call et al., JAMA, 2005).

- Interesting reference: “The precise origins for these illnesses are rarely identified. This is largely because for most respiratory viral diseases, establishment of the specific viral cause is neither necessary (i.e.: does not direct therapy) and thus is not cost-effective.” Ref = Long Hall Cunningham et al., Arch Fam Med, 1997.

- Limitations – Exclusions of certain patient populations.

- Limitations – Requirement of fever or feverishness for enrollment means …

o Findings may not generalize well to those without these symptoms

o The diagnostic specifications of the test may have been altered, i.e.: overestimation of sensitivity and PPV.

- Conclusion: Cough and fever are good predictors of infection among patients with influenza-like illness when influenza is present within the community. To maximize their predictive value, clinicians should keep apprised of influenza in the community. During the start of epidemics, when information is scanty and PPV may be lower, physicians might collect diagnostic specimens.

Zambon M, Hays J, Webster A, Newman R, Keene O. Diagnosis of influenza in the community: relationship of clinical diagnosis to confirmed virological, serologic, or molecular detection of influenza. Archives of Internal Medicine, 161(17): 2001.

- Objective – Examine the relationship between clinically diagnosed influenza and laboratory confirmation using different diagnostic test results.

- Cross-sectional diagnostic study

o Methods of detection

▪ RT-PCR

▪ Culture

▪ Serology

o Included patients had ILI detected during a period of intense prospective case ascertainment as recruitment for RCTs of inhaled anti-virals.

o Inclusion - patients initially seen within 48 hours of the onset of illness in the community when influenza was known to be circulating.

o Patients were recruited from multiple sites in Europe and North America during influenza epidemics.

- 1133 patients enrolled.

- Specimens for all three methods of detection available for 1033 patients (91%).

- 791 of 1033 (77%) patients were confirmed as influenza positive on one or more tests.

- This is much higher than community rates of influenza in ILI. Probably because patients were included if they already had fever, etc..

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, 358(9291): 2001

- Objective – Assess the contribution of influenza and RSV to cases of ILI in the community.

- Prospective series - descriptive

o 75 sentinel practices.

o All new episodes of ILI recorded, indexed, and reported weekly via routine clinical surveillance.

o Virologic survellance in a subset of GPs from 10 to 15 general practices – systematic NP swabs in patients with clinical case of ILI.

o Oct 1 to April 30 for the years of 1995-96, 1996-97, and 1997-98.

- Detection of influenza and RSV by PCR.

- NP swabs from 2226 cases of ILI = 7.5% of total consultations for ILI in the sentinel network.

o RSV responsible for 16-25% of ILI.

o Influenza responsible for 20% - 41% of ILI.

- RSV detected before influenza virus detections, and co-circulated with influenza virus detections, although peaks of detection did not always coincide.

- Influenza peaks coincide with peak of ILI.

- “When influenza is prevalent, there are excess consultations with ILI and acute bronchitis, and excess deaths from all causes. However, the frequency of acute bronchitis in Winter is always substantially higher than that ascribed to ILI, whether or not influenza is circulating, and we think that at least a proportion of cases of acute bronchitis might be due to RSV as well as influenza, although we have not virologically analyzed cases of acute bronchitis.”

Kelly H, Birch C. The causes and diagnosis of influenza-like illness. Aust Fam Physician, 33(5): 2004.

- Review of Australian ILI surveillance and causes of ILI.

- Use sentinel surveillance of ILI syndrome to enumerate viral etiologies of ILI. Important causes are influenza, RSV, adenovirus, picornavirus, parainfluenza virus, and, less frequently, coronaviruses and human metapneumovirus. Possibly rhinovirus for severe colds.

- Rates of isolates positive for influenza – 43% in 2002, 37% in 2003 – Victoria, Australia.

- Rates of isolates positive for other respiratory viral infections – 17% in 2002, 13% in 2003 – picornavirus most common – Victoria, Australia.

Ebel MH, White LL, Casault T. A systematic review of the history and physical examination to diagnose influenza. J Am Board Fam Pract, 17(1): 2004

- SR of diagnostic performance of signs and symptoms on history and physical exam (HPE) for influenza.

o Independent cohort studies and RCTs that were functionally equivalent included. Influenza A and B combined.

o Any reference laboratory test.

o Random effects model for pooling sensitivity and specificity.

o Positive and negative likelihood ratios calculated from random effects estimates.

o AUC calculated.

- 7 studies included.

- The HPE elements that best ruled in influenza when present were rigors (LR = 7.2, 1 study), fever presenting after >=3 days of illness onset (LR = 4.0, 1 study), and sweating (LR = 3.0, 1 study). The HPE elements best able to rule out influenza when absent were having systemic symptoms (LR = -0.36, 1 study), coughing (LR = -0.38, 4 studies), not being able to cope with daily activities (LR = -0.39, 1 study), and being confined to bed (LR = -0.50, 1 study).

- AUC could not be calculated for many variables with LR+ > 2.0 or LR- < 0.5.

- A bias toward lower estimates of sensitivity and specificity may have ben introduced by the fact that most studies only included patients with suspected influenza – e.g.: making fever an inclusion criteria will make it impossible for the variable to conribute to discriminating between patients with and without influenza – especially affects estimates that were part of the large Monto study.

- Also, many results come from a single study. Commonly studied symptoms had less diagnostic value – also higher heterogeneity.

- Individual signs and symptoms rarely include or exclude a disease. More successful strategy is to use several key symptoms to stratify patients into low-, moderate-, and high-risk groups. This study has identified 3 variables that help rule in influenza, and 4 symptoms that rule it out.

Ebell MH. Diagnosing and treating patients with suspected influenza. American Family Physician, 72(9): 2005

- EBM brief.

- Identifies two recent meta-analyses showing that fever, cough, rigors, and sweats increased the likelihood of influenza, but that these symptoms individually had a relatively poor predictive value. Also note that WBC < 4000 per cubic mm may increase likelihood of influenza.

- The two meta-analyses were Ebell et al., J Am Board Fam Pract., 2004; and Call et al., JAMA, 2005.

- Also discusses when to use rapid testing for influenza (only if the pre-test likelihood of influenza is less than 30%), and when to prescribe an anti-viral.

Call SA, Vollenweider MA, Hornung CA, Simel DL, McKinney WP. Does this patient have influenza? JAMA, 293(8): 2005

- 2003 – a year of sub-optimal antigenic match? A year of vaccine shortage.

- Also 2004-2005 – one of the manufacturers of TIV ceased providing vaccine to the US – available vaccine only half of projected for the US.

- CDC sentinel data – patient visits to a PCP office for ILI peak at 2.2% to 7.1% during influenza season (reference 4). Majority of samples in the 2003-2004 influenza season tested negative for influenza.

- Phyisician must be able to accurately estimate the probability of influenza as opposed to other infections to guide further diagnostic testing and anti-viral therapy.

- Objective – Identify clinical factors that may be valuable in distinguishing which patients with influenza-like illness have a higher probability of truly having influenza.

- Systematic review of MEDLINE from Jan 1966 – September 2004.

o Inclusion – Prospective cohort, RCT, or meta-analysis; gold standard = culture, 4-fold rise in serologic titres, PCR, or immunofluorescent antibody; quality graded A or B using the Rational Clinical Examination series scheme.

o Random effects pooled LR+, LR-, and DOR (diagnostic OR = LR+ / LR-, comparison of odds given the positive finding vs not having the finding).

- 6 studies (7105 patients) included – out of 10 meeting the inclusion criteria – unable to obtain primary data for 3 studies, and 1 study was a multiple report.

- “No single clinical finding consistently had a positive LR high enough to clinically ‘rule in’ influenza, nor did any single finding have a negative LR low enough to clinically ‘rule out’ influenza”.

- Good negative LR – Absence of fever, cough or nasal congestion were the only findings with a LR- < 0.5.

- Ineffective positive and negative LRs – Feverishness, myalgia, malaise, sore throat, and sneezing each had a positive and negative LR indistinguishable from 1.0.

- Inconsistent positive LRs – heterogeneity – E.g.: fever – 3.1 and 3.8 in two studies, 1.7, 1.1, and 1.8 in three studies.

- Inclusion = any age

o Fever and cough (LR = 1.9 [1.8, 2.1], 1 study) (Monto et al.)

o Good DOR

▪ Fever (DOR = 4.5 [1.8, 11.0] 3 studies)

▪ Cough (DOR = 2.8 [2.1, 3.7], 4 studies) were most useful single findings

▪ Fever and cough had intermediate value (DOR = 3.6 [3.1, 4.2], 1 study).

- Inclusion = aged >= 60

o “Reasonable” positive LR, i.e.: LR > 2.0

▪ Fever (LR = 3.8 [2.8, 5.0], 1 study)

▪ Malaise (LR = 2.6 [2.2, 3.1], 1 study)

▪ Chills (LR = 2.6 [2.0, 3.2], 1 study)

o Moderate positive LR, i.e.: LR > 5.0

▪ Fever and cough (LR = 5.0 [3.5, 6.9], 1 study) (Govaert et al.)

o Good DOR – for fever and cough, as single findings, much higher for those aged >= 60 years than for all ages.

▪ Fever (DOR = 5.2 [3.4, 7.9] 1 study)

▪ Cough (DOR = 3.4 [1.2, 9.7], 2 studies)

▪ Malaise also performed well in this group (DOR = 4.9 [3.3, 7.1], 1 study)

- Sensitivity and specificity ranged quite highly.

- Spectrum bias, “as patients who enroll in RCTs assessing either treatment or prevention of influenza may not represent the population of persons [presenting to a primary care office.” Spectrum bias a particular issue in study by Govaert et al., in that signs and symptoms were assessed in all the persons enrolling in the vaccine trial whether they had complaints of illness or not. The prevalence of laboratory confirmed influenza was 6.6% in Govaert et al.,, compared to the prevalence in other studies, which ranged from 8% to 67%.

- “No specific symptom or combination of symptoms is diagnostic of this common infection.”

- A combination of fever and cough during influenza season suggests a significantly increased likelihood of influenza among elderly individuals.

Hessen MT. In the clinic. Influenza. Annals of Internal Medicine, 151(9): 2009

- Symptoms of influenza frequently overlap those of other viral respiratory symptoms.

- Differentiable by higher temperatures, acute symptom onset, more severe diagnosis. Also, presence of fever and cough distinguishing; presence of influenza in the community raises index of suspicion.

- References to Zambon et al., Arch Intern Med, 2001; Monto et al., Arch Intern Med., 2000; and Boivin et al., Clin Infect Dis, 2000 for diagnostic performance of the influenza case definition (syndromic ILI).

- GI symptoms may suggest other diagnosis, although kids and H1N1 often present with GI symptoms.

- Complications of influenza references Connolly et al., BMJ, 1993.

Influenza as a cause of CAP

Johnstone J, Majumdar SR, Fox JD, Marrie TJ. Viral infection in adults hospitalized with community-acquired pneumonia: Prevalence, pathogens, and presentation. Chest, 134(6): 2008

- Objective – Describe pathogens, clinical presentation, and outcomes in consecutive adults admitted to hospital with CAP.

- CAP prospective cohort in Edmonton, consecutive adult patients meeting inclusion criteria presenting to Edmonton EDs, hospitalized for CAP.

- 300 patients enrolled, 193 with evaluable nasopharygeal (NP) specimens.

- Protocol violations (missed collection) in 39 patients.

o Those with non-evaluable NP specimens not different, except …

o Lower functional status (predisposes to viral infection)

o Higher rates of lobar pneumonia on CXR (suggestive of bacterial infection)

o Slightly shorter LOS.

- 193 patients with evaluable NP.

o Identification of respiratory pathogen in 75 patients

o “Unknown” in the other 118.

- Viral infection in 29 patients = 15% of 193 patients with evaluable NP specimens.

o Viral infection with influenza, hMPV, or RSV in 18 / 29 patients

o 62% of those with identifiable viral respiratory pathogen.

o Influenza A infection in 3 patients, influenza B in 4 patients.

- Most common bacteria – S. pneumoniae.

- Those presenting with a viral infection were older, more likely to have cardiac disease, more frail, and more likely to have a normal leukocyte count than those presenting with bacterial infection.

- Study notes that most older etiology studies have reported influenza infection in patients with pneumonia 4 to 19% of the time, followed by RSV (Falsey AR, Walsh EE. Viral pneumonia in older adults. Clin Infect Dis, 42: 2006).

- Influenza was the most common virus identified – however, hMPV occurred as commonly as influenza, and more frequently than RSV – this is a novel finding that has not been documented due to lack of testing for hMPV in the past.

- Influenza most common virus identified = 4% of patients (7 patients).

- Limitations: Presence of virus does not equal virus as a cause of pneumonia, merely potential cause. Unable to determine cause – would require lung parenchyma samples. Invasive! Missed investigations – high rate of unknown infections. Patients without evaluable NP swabs excluded. Potential for diagnostic inaccuracy.

- Study conclusion – one sixth of all included patients had a respiratory virus identified; alternatively, more than one third of those in which a respiratory pathogen was identified had a respiratory viral infection.

File TM, Marrie TJ. Burden of community-acquired pneumonia in North American adults. Postgraduate medicine, 122(2): 2010

- Review of evidence on epidemiology, outcomes, etiology, and costs of CAP from studies 1997 to current.

- Hospitalizations for infectious diseases increased from 1998 to 2005, and pneumonia in patients aged >= 50 years accounted for 35% of these hospitalizations.

- Etiology – cites Johnstone, Chest, 2008 – Only study cited that gives insight into the role of viruses as etiologies of CAP

- While viruses can cause CAP, “diagnostic studies to detect them are rarely performed”.

- CAP – outpatients – most commonly isolated microorganisms – may exclude viruses – S pneumoniae and H influenzae. S arueaus much less common, but physicians must be vigilant due to increasing reports of severe CAP due to MRSA.

Use of ICD codes to identify influenza

Quenel et al., International Journal of Epidemiology, 1994

- Evaluate the sensitivity, specificity, and positive predictive value of health service based indicators collected by an epidemiology network in the Paris region of France for detecting influenza epidemics.

- Gold standard = epidemic, defined as a week in which sentinel surveillance viral isolates >1% positive for influenza.

- Heath service indicators collected from a field epidemiologic network of 51 GPs and 25 paediatricians enrolled for this purpose. Indicators considered positive for an epidemic if weekly value is greater than the upper 95% confidence limit of average “non-epidemic weeks” (no regression for seasonal trends, etc.):

o Medical activity (number of home visits + consultations / week, percentage of GP visits declared ILI-related)

o Absenteeism

o Drug consumption

o Short-term admissions to hospitals

o Number of hospital deaths.

o Health services medical activity indicators

- GP activity due to influenza-like illness – sens = 69.04%, spec = 68.62%. GP overall activity had lower sensitivity and lower specificity. GP visits as an indicator and sens = 66.66% and spec = 71.56%. (VS gold standard.)

- Secondary analysis in which sens and spec to detect an epidemic, as opposed to epidemic week, was performed. An epidemc was defined as >2 consecutive weeks with indicator positive, and community presence of influenza defined by presence of influenza positive isolates. GP activity due to influenza-like illness had sens = 54.71%, spec = 94.11%, and PPV = 79.31%. Emergency visits had high sens and spec. Sick leave as well. Hospital fatality had 100% PPV.

Tsui FC, Wagner MM, Dato V, Chang CC. Value of ICD-9 coded chief complaints for detection of epidemics. Proc AMIA Symp: 2001

- Objective – Determine the diagnostic characteristics of ICD-codes for detecting influenza epidemics, compared with the use of a narrowing set of P&I codes.

- Methods

o Data from a single ED in Pittsburgh.

o Year = 1999

o Codes = data for input into a detection system

▪ Respiratory set (RS) and influenza set (iS) of ICD codes developed for content validity by general internists.

▪ P&I deaths = gold standard

o Detection system – Serfling models – Code frequencies above expected signaled an influenza epidemic.

- Three epidemics detected by gold standard.

- ICD codes were 100% sensitive for influenza epidemics. However, the positive predictive value was 50% for the RS and 25% for the IS.

- The study was unable to estimate specificity or negative predictive value – missing one box in the 2x2 table.

- The RS and IS sets were able to detect epidemics one week earlier than P&I deaths.

Keren R, Wheeler A, Coffin SE, Zaoutis T, Hondinka R, Heydon K. ICD-9 codes for identifying influenza hospitalizations in children. Emerging Infect Dis, 12(10): 2006

- Study of the sensitivity and PPV of influenza specific ICD-9 admission or discharge codes (487.0, 487.1, 487.8).

- Retrospective cohort of all patients 48 mg/ml.

o Two of these patients had C. pneumoniae or M. pneumoniae identified.

o Negative or low CRP in patients for whom a viral etiology was identified.

o CRP used because of lack of diagnostic tests for bacteria in acute bronchitis (e.g.: no sputum production).

- “When acute bronchitis is confirmed as of infective origin, viruses are still the commonest causative agents”.

Flemming et al., Communicable Disease and Public Health, 2000

- Unable to locate full paper, abstract only.

- Epidemic versus non-epidemic periods, 1989-1998, comparison of GP diagnoses for ILI, acute otitis media, acute bronchitis, and all respiratory infections. No information on acute bronchitis reported in abstract.

Macfarlane J, Holmes W, Gard P, Macfarlane R, Rose D, Weston V, Leinonen M, Saikku P, Myint S. Prospective study of the incidence, aetioogy and outcome of adult lower respiratory tract illness in the community. Thorax, 56(2): 2001.

- Objective – Investigate the incidence, causes, and outcomes of community acquired lower respiratory tract illness.

- LRTI taken here to be a more precise term for “acute bronchitis”, which is labeled with little knowledge of the true etiology of disease.

o Acute illness present for 21 days or less

o Cough = cardinal symptom

o At least one other lower respiratory tract symptom – sputum, dyspnea, wheeze, or chest discomfort

o No alternative explanation (e.g.: asthma, pharyngitis)

- Prospective case series

o 10 GPs from 2 practices reported all previously well adults consulting with acute LRTI.

o October 1997 to September 1998.

o Sputum samples, NP swabs, and blood for serology (acute and convalescent) obtained for viral and bacterial studies.

- LRTI recorded in 638 patients – LRTI rate = 54/1000 previously well adults per year.

- 316 patients investigated for etiology. Patients missed were not studied due to non-consenting patient, practice too busy, research nurse not available, etc..

- Patients studied more likely to have systemic symptoms, other LRT symptoms, and were more likely to be prescribed antibiotics.

- Pathogens identified in 173/316 (55%).

o Bacterial: 82/173 (47.4%).

o Viral: 51/173 (29.5%) – Influenza A most common (23/51) – 45% of viral LRTI.

- High prevalence of bacterial infection may be due to crossing of case definition with pneumonia. Also – there may be an issue with bacterial colonization not leading to infection.

- Influenza is one of the most common pathogens isolated in LRTI. If Strep pneumoniae isolations are ignored, influenza is the most common pathogen isolated in LRTI – acute bronchitis.

- The data provide crude evidence that antibiotic treatment is un-associated with decreased rates of return visits for bronchitis.

Connolly AM, Salmon RL, Lervy B, Williams DH. What are the complications of influenza and can they be prevented? Experience form the 1989 epidemic of H3N2 influenza A in general practice. BMJ, 306(6890): 2003

- Objective – Measure the incidence and risk of complications, identify the risk of hospitalization.

- Prospective cohort study

o 34 sentinel practices in Wales

o Seasonal influenza epidemic in 1989 (weeks 45-52)

o Cases of ILI identified by clinical definition.

o 1 control per case matched on age and sex – next patient occurring at least three places from the case on the practice register.

- Outcomes data – from MRR extraction.

- Analysis – Rates calculated, compared using the Mantel-Haenszel version of the chi-squared test.

- 395 cases obtained. 352 cases included.

- Low rate of hospital admission – Only 2 cases admitted – 9.1/100000

- Bronchitis (OR = 12.8 [5.2, 31.8]) and pneumonia (OR = 9.0 [1.1, 71.0]) significantly commoner in cases than controls. Bronchitis was the most common complication of influenza (rate = 190.1/1000 cases, 95% CI [148.5, 231.7]).

- Bronchiolitis, otitis media, and depression NS.

- Other complications (e.g.: pericarditis, Reye’s syndrome, death) did not occur.

Louie JK, Hacker JK, Gonzales R, Mark J, Maselli JH, Yagi S, Drew WL. Charactieriation of viral agents causing acute respiratory infection in a San Francisco University Medical Center Clinic during the influenza season. Clinical Infectious Diseases, 41(6): 2005.

- Objective – Characterize the spectrum of viral agents cause acute respiratory infection.

- Prospective series of previously health adults presenting to medical attention

o January to March, 2002.

o Consecutive adults presenting to UCSF acute ambulatory care clinic or ED with symptoms of an acute respiratory infection.

o ARI defined as new illness within the past 3 weeks with cough, sinus pain, congestion, sore throat, or fever.

o Exclusion of certain noninfectious, self-reported conditions.

o NP specimens obtained at enrollment.

o Culture and PCR for viral etiologies, either sufficient for diagnosis – bacterial etiologies ignored because these are not thought to be common etiologies of acute bronchitis in adults without underlying lung disease.

- 408 patients evaluated, 289 enrolled, 281 with clinical specimens, 266 with specimens suitable for testing.

- Most common cause of URTI was influenza (24%).

- Non-specific URTI in 54% (142/266).

- Bronchitis in 16% (42/266).

o 17/42 had viral pathogen identified

▪ Most common agent was picornavirus (10 patients, 24%)

▪ 9 of the picornavirus isolates was rhinovirus.

▪ Next most common was influenza in 6 (14%) patients.

o 25/42 had no pathogen identified.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download