Routine Chest Films are Cost-Effective


Many radiographs are obtained in ICU patients without specific clinical indications. A significant percentage of these "routine" radiographs will demonstrate unexpected findings, some of which will result in a change of therapy. Brainsky and colleagues prospectively examined the potential impact on length-of-stay (LOS) and costs of not identifying those abnormalities, which were actually seen on routine radiographs in a MICU patient population.

Chest radiographs obtained between 6 and 9 a.m., ordered at least eight hours in advance and labeled "routine," were included in the study. Clinicians, blinded to the purpose of the study, were interviewed daily regarding radiographic findings and resultant changes in therapy. The impact on LOS and mortality was estimated by surveying 214 critical care practitioners in Philadelphia regarding the significance of radiographic findings and the effects of not knowing the identified abnormalities. One-hundred eighteen individuals (55%) responded to the survey, which consisted of case scenarios developed to represent the abnormalities identified in the clinical study. Expert opinions for each abnormality (severity and expected increase in LOS) were averaged, and summary information was used to project the impact on LOS. Radiography costs were estimated based on actual supplies and personnel time but did not include professional fees for film interpretation. ICU costs were based on an estimated daily cost of $1000. Cost effectiveness was calculated as the costs for the entire "routine" set of films compared to the cost savings from treating "positive" findings.

Data were collected on three randomly selected days of the week, to include weekends as well as weekdays. Patients were followed throughout their ICU stay. The study group consisted of 80 randomly selected patients in whom 221 routine chest radiographs over 312 patient days were evaluated. Nineteen patients (24% of the group) had no routine radiographs. There were no differences in age (52 ± 18 years), race (53% non-white), sex (males = 55%), or mortality (23%) between those with and without routine radiographs.

There were 80 new findings in 72 radiographs in 33 patients; thus, 33% of the routine films demonstrated at least one unanticipated observation. Of these changes, 80% were changes in previously known conditions, and 20% were completely new. Important findings were present in 44 routine radiographs (20%). Twenty actions were prompted in 15 patients from 18 routine films. Thirteen films lead to actions that were predicted to avoid increases in LOS by more than one day, and five prompted actions predicted to have a negligible effect on LOS. The average avoided increase in LOS by actions prompted routine films in the study group was 2.14 ± 1.7 days. The total predicted savings was 38.5 days or $38,500 in ICU costs. The cost using the highest cost estimate for all the routine films was $16,796, or $933 per finding that prompted a change in therapy. The average net saving for obtaining routine radiographs in the entire group was $271 per patient. The worst-case scenario was projected, and even with the highest cost of radiographs, lowest projected LOS was avoided and assumed non-helpful radiographs in all of the patients who had no routine films, the mean net savings was $133 per admission and $40 per routine radiograph. (Brainsky A, et al. Crit Care Med 1997;25:801-805.)


This paper presents a unique way of analyzing the cost effectiveness of routine chest radiographs in ICU patients. The finding that about 20% of routine films result in unexpected findings and a change of therapy is consistent with other reports. The novel contribution of Brainsky et al is the estimate of the extension of LOS that would have occurred by not having obtained the radiograph and made the therapeutic intervention. The method used was to ask a group of critical care physicians to guess how long the patient might stay in the ICU without the radiographic diagnosis and the prompted change of therapy. At least 35 experts responded to each possible scenario, and their estimates were averaged.

This group consensus process is also the major weakness of the study—its results are based entirely on opinion of the impact of therapeutic interventions, not on actual data. Despite this concern, the conclusions of Brainsky et al are supported by the sensitivity analysis, which indicated that if even 0.9 days were avoided by each intervention, cost-savings would be had by ordering daily, routine chest radiographs in all MICU patients. An even higher cost-benefit ratio can be obtained if films are requested based on clinical indicators. As many as 75% of these films will demonstrate abnormalities. The impact on LOS of treatment decisions triggered by these radiographic diagnoses has not been assessed.

The most common interventions were increasing diuretics for worsening congestive heart failure and withholding diuretics with improved congestion. Missing from the paper were the predicted changes in LOS for specific radiographic diagnoses and interventions. Evaluation of the credibility of this paper depends on these estimates. It would be helpful to see a table of these projections.