Radiology Resource Utilization During an H1N1 Influenza Outbreak

Introduction 

From its identification in the spring of 2009 to October 11, 2009, influenza A (H1N1) has been responsible for nearly 400,000 laboratory confirmed infections worldwide and more than 4,700 deaths [1]. Roughly 40,000 deaths are attributed to seasonal influenza annually in the United States alone [2], and converging lines of evidence suggest considerably higher incidence and case fatality rates from H1N1 than the usual seasonal flu [3, 4]. At press time the rate of influenza infection in the United States is well above the expected seasonal average, and the overwhelming majority of the recovered isolates are of the H1N1 type [5]. As a result, there is considerable anxiety regarding the capacity of health care organizations to cope with the potential effects, both medical and societal, of the threatened pandemic.

Many institutions have developed or are developing graded and detailed contingency plans to respond to the various severities of outbreak, to maintain the delivery of critical health care services while responding to increased demands due to H1N1 infection, and to cope with a reduction in available health care personnel to deliver care (because of illness among them). Our own university has an institution-wide plan covering all of our campuses, the student health service, and our University Health System. Within the Department of Radiology, academic faculty members who traditionally practice in subspecialities have been canvassed to determine their areas of best practice outside their specialties for reassignment to areas that could become short because of illness or be overwhelmed by volume. Radiology technologists who practice in specialized modalities, such as CT, MR, and ultrasound, have been refamiliarized with operating general radiology equipment. Radiology residents may be asked to assist in care delivery usually performed by personnel such as radiology nursing staff members or be reassigned to other clinical services in the health system, to aid in delivering care in specially set up flu clinics or on inpatient services. Although ramping up services for advanced ventilation and even extracorporeal life support may seem obvious, our experience during the early summer 2009 H1N1 outbreak indicates that because of the centrality of imaging in modern medical practice, preparing radiology departments with the most appropriate resources may be just as important. In this report, we describe the utilization of radiology services during the summer 2009 H1N1 outbreak, in the hope of providing information useful to others for pandemic planning.

Materials and Methods 

We retrospectively reviewed a population of 222 patients with flulike symptoms and confirmed influenza A (by direct fluorescent antibody testing or viral culture) who presented to the University of Michigan Health System via outpatient clinics, the emergency room, or transfer from outside hospitals between May 1, 2009, and July 18, 2009. Of these 222 patients, 31 had laboratory evidence of swine-origin influenza A virus (H1N1) on the basis of respiratory specimens tested with real-time reverse transcription polymerase chain reaction at the Michigan Department of Community Health or Centers for Disease Control and Prevention. The remaining subjects were presumed to have swine-origin influenza A virus (H1N1) because no other viruses were circulating in the community with any frequency during that time period.

A computerized search of our hospital-based medical record system was performed to determine the frequency and type of imaging studies used during their care between May 1, 2009, and August 10, 2009. The patient population included 14 patients treated in the intensive care unit (ICU) with mechanical ventilation, defined as either high-frequency oscillatory or bilevel ventilation (group 1), and the remaining 52 patients primarily treated as outpatients or briefly as inpatients without mechanical ventilation (group 2). All but one patient admitted to the ICU were transferred from other hospitals and had initial imaging at other facilities before arriving at our health system; for this cohort of patients, the resource utilization was determined only for the period of their hospital stays. For the one patient in group 1 who presented to our emergency department, the imaging studies include the emergency department visit and the entire ICU and inpatient stay. For group 2, the radiology examinations were included for a period ranging from 10 days before to 30 days after the diagnosis of influenza A.

This study was approved by the University of Michigan Institutional Review Board with a waiver of informed consent.

Results 

Of the 222 patients, 66 (30%) underwent radiologic imaging, which included chest radiography in all patients. These 66 patients were diverse in demographics, consistent with the broad swath of the population at risk during a pandemic (Table 1). Although the average age of the 66 patients was 26.7 years, there was a significant difference in age between the ICU patients (mean age, 43.5 years) and the non-ICU patients (mean age, 22.1 years) (P < .001; t test). Patients ranged in age from 9 months to 60 years. Whereas the non-ICU patients had a nearly equal distribution of men and women, 79% of the ICU population (11 of 14) were male. The average length of stay for the ICU population was 22 days (range, 9-63 days). Of the 52 non-ICU patients, 25 (48%) had brief periods of admission, with an average length of stay of 4 days; the average length of hospital stay for all of group 2 was only 2 days. Of note, the disposition of the ICU population reflects the nature of the quaternary care facility at which this study was performed. For example, only 29% percent (4 of 14) of the ICU patients were transferred from our ICU to our inpatient floors for ongoing management; 5 of the remaining patients died, and 5 patients were transferred back to their hospitals of origin. Presumably, further imaging studies were performed at these facilities, and these were not captured in this review.

Table 1. Patient demographics

		ICU	Non-ICU
	Number	14	52
	Average Age (in years)	44±11	22±16
	Male:Female	11:3	25:27
	Number and Percent of Patients Admitted	14(100%)	25(48%)
	Average Length of Hospital Stay	25±17	2.2±2.7
	In Hospital Mortality	5	0
	Total Number of Diagnostic Chest Radiographic Exams	469	71
	Average Number of Diagnostic Chest Radiographic Examinations per Patient	33.5	1.4
	Total Number of Thoracic CT Examinations	15	6⁎
	Average Number of Thoracic CT Examinations per Patient	1.1	0.1
	Total Number of Radiology Examinations	654	105

The ICU group received advanced mechanical ventilation while the non-ICU group did not. Average age and length of stay is reported with one standard deviation. The “Diagnostic” category of limited chest radiographic exams excludes the 21 studies performed in ICU patients for PICC placement only.

⁎ Of the 6 CT scans performed, one CT was not directly related to the influenza illness but was performed during the same period for follow-up of known comorbidities.

Although group 1 consisted of only 14 patients, this ICU population had extensive imaging resource utilization. They underwent a total of 654 imaging studies, including 469 single-view chest radiographic studies (mean, 33.5 per patient), 15 thoracic CT examinations (mean, 1.1 per patient), 59 other CT examinations (mean, 4.2 per patient), 71 abdominal plain radiographic series (mean, 5.1 per patient), and 39 other studies. Although the average ICU patient received 33.5 chest radiographic studies and 1.1 thoracic CT examinations, this was highly variable within any patient. For example, the number of chest radiographic examinations ranged from 14 to 99 across the 14 patients. In terms of CT, 9 of 14 patients (64%) underwent thoracic CT examinations, with 5 patients undergoing 1 thoracic CT, 3 patients undergoing 2 CT examinations, and 1 patient undergoing 4 CT examinations.

Group 2, the non-ICU population of 52 patients, underwent a total of 105 imaging studies (Figure 1), including 71 chest radiographic examinations (mean, 1.4 per patient), 6 thoracic CT examinations (mean, 0.1 per patient), and 28 other examinations. The latter included a variety of modalities, and most were related to medical issues other than the primary influenza infection, such as neurologic problems and hip infections. In terms of thoracic imaging, 39 of 52 (75%) received only 1 chest radiographic examination, 7 of 52 (13%) received 2 chest radiographic examinations, and 6 of 52 (11%) received ≥3 examinations. Six patients each had one thoracic CT examination; review of the indications for these demonstrated that 5 examinations were related to influenza symptoms. One thoracic CT examination was not directly related to the influenza illness but was performed during the same period for follow-up of known comorbidities (myeloma and lung nodules). Thus, thoracic CT was performed for the influenza illness in 5 of the 52 group 2 patients (9.6%).

View Large Image Download to PowerPoint

Fig 1. Radiology examinations in an H1N1-infected population. The average number of imaging examinations per patient performed on the study population are plotted for the two subgroups of patients who did (ICU) and did not (non ICU) need mechanical ventilation. These are shown in aggregate (total) as well as divided into single-view and two-view chest studies (diagnostic chest x-ray exams), chest CT examinations with and without contrast (chest CT exams), single and multiple views of the abdomen (abdominal plain film[s]), oblique chest radiographs for line placement (PICC placement), and all other CT, MRI, ultrasound, fluoroscopic, and interventional procedures (other). Error bars reflect one standard deviation. PICC = peripherally inserted central catheter.

The temporal distribution of the utilization of imaging resources (see Figure 2) by the ICU patients showed marked temporal variability, with relatively few studies performed in May and a precipitous decline in the number of studies performed in July compared with an impressive peak in mid-June, when a handful of patients accounted for nearly 100 studies a week. In contrast, the group 2 population showed a subtle peak in early June but was reasonably well distributed over the study period.

View Large Image Download to PowerPoint

Fig 2. Temporal pattern of thoracic imaging resource utilization. The number of thoracic imaging studies performed during each of the nine 10-day blocks of the 3.5-month study period are shown for the patients in aggregate (all) and separated into those who did (ICU) and did not (non ICU) require mechanical ventilation. Note that the last patients were enrolled on July 18, 2009, and thus examinations performed on patients diagnosed after this date are not included.

Discussion 

This review of the imaging performed on patients with H1N1 reveals both reassuring and troubling information. Of the 222 patients with documented H1N1 infection seen within our health system, only 30% (66 of 222) received any kind of related diagnostic radiology examination, with all 66 receiving at least one chest radiograph. Resource use was highest in the ICU population, with a mean of 33.5 chest radiographs, 5.1 abdominal radiographs, 1.1 thoracic CT examinations, and 4.2 nonthoracic CT examinations. The non-ICU patients who did not require any kind of mechanical ventilation had an average of less than 1.4 chest radiographic studies and 0.1 thoracic CT examinations. When the majority of these patients were in house at the same time during mid-June, they were together frequently receiving more than 10 radiology examinations a day in the university hospital or our children's hospital.

The potential for overwhelming study volumes is further suggested by our analysis of the temporal patterns of study distribution. Although radiology systems may be able to absorb steady increases in average study volume, sudden increases in peak volumes have the potential to overwhelm any reserve capacity. Radiology utilization was not uniform but dramatically asymmetric, with more than half of all ICU studies performed in the short time frame of 3 to 4 weeks. These peaks in demand during a large-scale pandemic are also likely to correspond to periods of maximal absenteeism among radiology technologists and radiologists, which may result in added stress to radiology providers. Another unfortunate, if less acute, finding from our temporal data is the fact that H1N1-related studies may persist for long after the pandemic peak has passed as a result of critically ill patients' lingering in the ICU for weeks or months after their initial infection.

The magnitude of extrathoracic studies performed on patients with H1N1 infection, specifically the ICU patients, was unexpected. Although teasing out the extent of cause and effect is impossible, there is no doubt that ICU patients often have comorbidities and often develop complications outside of the primarily diseased organ system. The ICU population in this study, for example, had a mean of 1.1 thoracic CT examinations performed but an additional 4.2 CT examinations involving the head, abdomen, pelvis, or extremities; an average of 5.1 abdominal radiographic studies; and a bevy of additional ultrasounds, fluoroscopic studies, and interventional procedures performed under imaging guidance. Our comparatively small study did not reveal any obvious trends in the types of extrathoracic studies that were performed, but we anticipate that as our experience with this infection increases, we will also have to address demand for currently unexpected studies. A recent paper, for example, suggests that H1N1 patients are at increased risk for pulmonary emboli [6], and thus we can safely predict that the number of ventilation-perfusion and pulmonary embolism CT studies is likely to increase along with chest radiographic examinations during an H1N1 pandemic.

Our results provide a useful representation of the radiology resource demands during a small-scale H1N1 outbreak. Despite the high anxiety relating to H1N1, not everyone infected with the virus sought treatment. Thus, the 222 patients almost certainly represent an underestimate of the number of infected individuals, and thus our 6% ICU fraction is hopefully overly pessimistic and biased by the quaternary care nature of our institution. It is also noteworthy that our estimate does not include the large population who may have been suspected of having influenza on the basis of their clinical presentation but who did not have lab confirmation of the same. This patient population may also have utilized radiology resources. Unfortunately, radiology requisition forms are notoriously lacking in clinical information, and thus we have no way of knowing whether any of the thousands of emergency department and outpatient diagnostic examinations conducted last summer were related to clinician or patient concerns of H1N1. Given the arguably even greater current anxiety, it is quite possible that radiology systems will be fielding greater than the usual number of examinations as a result of H1N1 but in patients who do not have the disease. Given the impact of the economic downturn in our region, it is also not possible to determine what impact the summer H1N1 outbreak had on our utilization of imaging in general.

As for the future, we simply do not know how bad the pandemic is going to be. As of this writing, the Centers for Disease Control and Prevention is reporting: widespread flu activity, primarily H1N1, in 48 states; that physician visits for influenza-like illness have steeply increased; that hospitalizations for confirmed H1N1 are rising, particularly in children and with a frequency higher than expected during a usual flu season; and that the proportion of deaths attributed to influenza and pneumonia is also greater than expected [5]. The literature on the expected infection rate and fraction of severe infections is frustratingly and inevitably speculative. Although we do not wish to be trite, these analyses largely boil down to a declaration that if H1N1 does infect a high fraction of the population or results in a higher frequency of serious illness, there will be many patients requiring ICU-level care, quite possible many more patients than there are currently ICU beds or other advanced support systems such as extracorporeal life support and oscillating ventilators. Such equipment was in high demand at our institution over the summer, and its shortage had a ripple effect on other services, such as pediatric cardiac surgery, which requires extracorporeal life support circuits for many complex surgical repairs. This study suggests that radiologists, radiology technologists, and imaging equipment should also be added to the list of resources in high demand if the more dire predictions are realized. Although certain limitations are hard (e.g., the number of trained radiologists available), others are softer, and our results encourage the immediate development of plans to utilize limited resources in the most effective way possible. Our considerations have suggested several broad areas of response that merit evaluation.