PUE reporting description and case definition
National guidelines [1] require all inpatient and outpatient healthcare facilities to report cases meeting the PUE case definition. Clinicians should report cases to an expert consultation committee, which after review of clinical and laboratory data determines whether to report the case to the PUE surveillance system [1]. If a case is reported to the PUE system, the local center for disease control and prevention (CDC) will conduct a field investigation, collect respiratory specimens and send them to a national influenza surveillance network laboratory for testing of avian influenza viruses and, if associated with clusters of respiratory disease or relevant travel history, testing of Severe Acute Respiratory Syndrome Coronavirus (SARS-CoV) and Middle East Respiratory Syndrome Coronavirus (MERS-CoV).
A PUE case is defined as an illness of unknown etiology with 1) axillary temperature > 38 °C, 2) radiographic pneumonia, 3) low or normal leukocyte count or low lymphocyte count during the early stages of disease, and 4) no improvement or worsening symptoms after 3–5 days of antimicrobial treatment per clinical guidelines [1].
Evaluation sites
Participating hospitals were selected based on four criteria: if the facility 1) admitted at least 200 patients per month with a discharge diagnosis of pneumonia during February through May 2013–2015; 2) used an electronic Hospital Information System; 3) demonstrated willingness and capacity to collaborate with both national and local CDCs and 4) was located within one of China’s 21 of 31 provinces with previously identified H5N1 and/or H7N9 human cases.
We selected two tertiary hospitals in Anhui Province: the Second People’s Hospital of Fuyang City, a 1400-bed facility, which from February through May 2013–2015, admitted an average of 231 pneumonia patients per month, and Lu’an City People’s Hospital, a 2300-bed facility, which over the same time period admitted an average of 252 pneumonia patients per month. Fuyang Hospital, an infectious disease hospital, had experience treating human infections with avian influenza, while Lu’an Hospital, a general hospital, did not.
PUE case screening and enrollment
After reviewing the hospital information systems, the evaluation team developed a list of 56 admission diagnoses that captured the majority of acute respiratory infections (ARI) (Additional file 1).
Every day (including weekends) from February 29 through May 29, 2016, a designated, trained surveillance officer in each hospital 1) reviewed the hospital admission registry database to screen all admission diagnoses from the prior 3 days for diagnoses from the screening list (Additional file 1); 2) reviewed admission medical records with a matching diagnosis to identify and enroll patients with illnesses meeting the PUE case definition; 3) 2 days later, conducted a second medical record review for patients not enrolled during the first review to identify and enroll patients with illnesses newly meeting the case definition (for example, patients with no improvement or worsening symptoms after 3–5 days of antimicrobial treatment per clinical guidelines); and 4) 5 days later, conducted a third review of records for patients not enrolled during the first two reviews to enroll any remaining patients meeting the case definition. [Fig. 1].
Medical record review
For patients with illnesses meeting the PUE case definition, the surveillance officer used a standard questionnaire to collect information from the hospital information system related to demographics and, if available, epidemiological risk factors, including exposures to poultry, patients with similar symptoms, and travel history.
Case investigation
Surveillance officers followed the PUE surveillance protocol [1] to investigate enrolled PUE case-patients. Surveillance officers conducted face-to-face patient interviews using a standard questionnaire to collect the same information described in the medical records review section above to determine both accuracy and completeness of medical records.
Specimen collection and testing
Surveillance officers collected throat swabs from all identified PUE case-patients per the surveillance protocol [1]. Specimens were transported to the local CDC laboratory per standard procedures and tested for influenza viruses using real time reverse transcription polymerase chain reaction (rRT-PCR). If identified PUE case-patients were part of a cluster of epidemiologically-linked respiratory illnesses, specimens would also be tested for SARS-CoV and MERS-CoV. If a case-patient reported travel history to the Middle East, specimens would be tested for MERS-CoV.
Assessing reporting procedures
The PUE surveillance protocol describes a three-step procedure for reporting cases to the PUE system: 1) clinicians report identified PUE cases to their supervisor; if the supervisor concurs, the case is reported to the director; 2) the director determines whether to report the case to an expert consultation committee which usually includes specialists from the respiratory medicine department, the radiology department, and infection control; and 3) the expert committee determines whether to report the case to the PUE system. [Fig. 2].
In our evaluation, after patients with illnesses meeting the PUE case definition were enrolled, surveillance officers interviewed all clinicians who had primary medical responsibility for these patients. If the case were reported to the clinician’s supervisor, the surveillance officer also interviewed the supervisor and a representative member of the expert committee. [Fig. 3] Surveillance officers used a standard questionnaire to collect 1) demographic and occupational information about the clinician being interviewed and, when applicable, 2) demographic and occupational information about the senior clinicians who received the case report, and 3) reasons for reporting or not reporting PUE cases.
Data analysis
First, we described the number, proportion, ward distribution, and testing results of PUE cases identified by surveillance officers. Wilson score was used to calculate 95% confidence intervals (CI) for proportions. Second, we described and compared epidemiologically-linked exposures documented by clinicians in the medical records and used chi-square tests to compare differences in the frequency of exposure histories collected by surveillance officer interviews. Third, we described the number of PUE cases reported at each of the tiers of the three-step reporting process. Finally, we described factors associated with clinicians’ reporting or not reporting PUE cases.