PEDIATRICS Vol. 107 No. 2 February 2001, pp. 241-248

Routine Chest Radiographs in Pediatric Intensive Care Units

Michael W. Quasney, PhD, MD^*, Denise M. Goodman, MD, Michael Billow, DO^§, Hsiaoting Chiu, PhD, Larry Easterling, MD, Lorry Frankel, MD^¶, David Habib, MD^#, Mary Heitschmidt, RN, MS^**, Stephan Kurachek, MD, Frank Moler, MD^§§, Vicki Montgomery, MD^||, Michele Moss, MD^¶¶, Sheila Murman, RN^##, Thomas Rice, MD^***, Barbara Richman, MD, and Samuel Tilden, MD^§§§

From the ^* Department of Pediatrics, Crippled Children's Foundation Research Center, Le Bonheur Children's Medical Center, University of Tennessee, Memphis, Tennessee;  Department of Pediatrics, Children's Memorial Hospital, Northwestern University, Evanston, Illinois; ^§ Department of Pediatric Critical Care, Children's Hospital Medical Center of Akron, Akron, Ohio;  Division of Pediatric Critical Care, Cook Children's Medical Center; Fort Worth, Texas; ^¶ Department of Pediatrics, Lucile Salter Packard Children's Hospital, Stanford University, Stanford, California; ^# Department of Pediatrics, Children's Hospital, Medical University of South Carolina, Charleston, South Carolina; ^** Department of Nursing, University of Chicago Children's Hospital, Chicago, Illinois;  Pediatric Pulmonary and Intensive Care, Children's Hospitals and Clinics of Minneapolis, Minneapolis, Minnesota; ^§§ Department of Pediatrics and Communicable Diseases, C. S. Mott Children's Hospital, University of Michigan, Ann Arbor, Michigan; ^|| Department of Pediatrics-Critical Care, Kosair Children's Hospital, Louisville, Kentucky; ^¶¶ Department of Pediatrics, Arkansas Children's Hospital, University of Arkansas for Medical Sciences, Little Rock, Arkansas; ^## Department of Nursing, Miami Children's Hospital, Miami, Florida; ^*** Department of Pediatrics, Children's Hospital of Wisconsin, Milwaukee, Wisconsin;  Departments of Anesthesiology and Pediatrics, Vanderbilt Children's Hospital, Vanderbilt University, Nashville, Tennessee; and the ^§§§ Department of Pediatrics, Children's Hospital of Alabama, University of Alabama-Birmingham, Birmingham, Alabama.

	ABSTRACT

Top Abstract Methods Results Discussion References

Objectives.  To determine whether interventions were performed based on portable routine morning chest x-rays (CXRs) in pediatric intensive care unit (PICU) patients and to identify patient subgroups for whom the routine CXR is most useful.

Design.  Prospective multiinstitutional study.

Setting.  PICUs of 15 tertiary care hospitals.

Patients.  PICU patients who received a routine morning CXR were included in the study.

Outcome Measures.  Recorded data included: weight, diagnosis, presence of active cardiopulmonary problems, length of stay, and number and type of devices. The number and types of interventions based on the interpretation of the CXR were recorded.

Results.  Five hundred twelve routine CXRs were evaluated. The majority of the routine chest radiographs were obtained on patients who were admitted for cardiovascular disease (195/512; 38%) or respiratory failure (186/512; 36%), and 465/512 of the routine CXRs (91%) were performed on patients with one or more devices. Two hundred thirty-one of the 512 routine CXRs (45%) resulted in 1 or more interventions. One hundred fifty-five of the 284 routine CXRs (55%) obtained in children 10 kg resulted in one or more interventions, compared with 61/152 (40%) and 15/76 (20%) of routine CXRs obtained in children 10 to 40 kg and 40 kg, respectively. The frequency of interventions increased from 19% in children with no devices to >50% in children with 2 or more devices. One or more interventions were performed in 27% of routine CXRs when no active cardiopulmonary problems were present, compared with 51% of routine CXRs when active cardiopulmonary problems were present. Diagnosis and length of intensive care unit stay at the time the routine CXR was obtained did not affect the percentage of CXRs that resulted in interventions.

Conclusions.  Routine CXRs are more likely to result in interventions in the smaller, critically ill child with one or more devices and if active cardiopulmonary problems are present.  Key words:  routine chest radiographs, pediatrics, intensive care unit, interventions.

Chest radiographs are commonly performed in the pediatric intensive care unit (PICU) setting. Many such radiographs are routinely performed without any specific clinical indication because they help to confirm the position of various devices such as endotracheal tubes, chest tubes, and central venous catheters. Indeed, the American College of Radiology and others suggest that daily chest radiographs are indicated on mechanically ventilated patients as well as those with acute cardiopulmonary problems.^1-4 The rationale for these recommendations is based on studies suggesting that the routine chest radiograph is likely to identify unexpected problems that have not yet manifested clinically.^5-8 Other studies suggest that there is little value in obtaining the routine chest radiograph in the intensive care unit (ICU) setting because clinical data can be used to obtain the information needed to manage the patient^9,10 and portable chest radiography is poor at identifying pathophysiologic manifestations of acute respiratory failure.¹¹ In the era of cost containment in the health care industry, there is economic pressure to reduce the number of tests that have little impact on patient management.

Most studies examining the efficacy and value of routine chest radiographs have been performed in adult ICUs. These studies have demonstrated a high frequency of unsuspected or abnormal findings on the chest radiographs obtained on patients in the ICU,^5-8 and most of these findings led to changes in the management of the patient.⁸ Brainsky and coworkers¹² found that 8% of routine chest radiographs in an adult ICU led to interventions and concluded that routine chest radiographs resulted in decreased overall costs. Others have concluded that the routine morning chest radiograph is diagnostically or clinically important^5,13-15 particularly in adult patients with pulmonary or complicated cardiac disorders⁸ or Swan-Ganz catheters.¹⁶ The usefulness of the routine chest radiograph in the smaller, critically ill child may be different than it is in adults for various anatomic and physiologic reasons. In the only study evaluating the efficacy of routine chest radiographs in a PICU, Hauser and colleagues¹⁷ demonstrated that significant findings requiring alterations in patient management could have been missed had the routine chest radiograph not been performed. They concluded that routine daily chest radiographs have a significant clinical impact on the management of critically ill children.

We evaluated the usefulness of routine morning chest radiographs in 15 PICUs by examining the frequency of interventions performed based on the results of the routine chest radiograph. We hypothesize that routine morning chest radiographs in smaller patients in PICUs would be more likely to result in interventions than in larger patients. We also analyzed various subgroups of critically ill children in an effort to determine those for whom the routine chest radiograph was more likely to result in an intervention.

	METHODS

Top Abstract Methods Results Discussion References

Fifteen PICUs participated in the study. The institutions represent PICUs that are university-affiliated as well as nonuniversity-affiliated. Admissions to the PICUs ranged from 50 to 120 children per month.

Children admitted to the PICU who had a routine chest radiograph performed were enrolled in the study. For the purposes of this study, a routine chest radiograph was defined as one that was ordered before midnight the previous night and performed the following morning. Chest radiographs were excluded if they were ordered before midnight the previous night for the purpose of following up specific therapies or processes. Each patient had only one chest radiograph evaluated for the study, and once the patient was enrolled in the study, no further chest radiographs of that patient were evaluated for the purposes of this study. All chest radiographs were performed using portable devices. Approval of the study was obtained from the institutional review boards at the participating institutions. Each institution randomly selected nonconsecutive days over a 6-month period on which to collect the data, and all patients in the ICUs who met the criteria for having a routine chest radiograph on that day were included in the evaluation. A standardized data collection form was used by all participating institutions to record the following: patient's weight; admission diagnosis; the presence of active cardiopulmonary problems; length of ICU stay at the time of the evaluation; primary service to which the patient was admitted; devices such as endotracheal tubes, central venous catheters, and chest tubes present at the time of the chest radiograph; whether the patient was mechanically ventilated; and adverse events that occurred during radiography. Active cardiopulmonary problems were defined as infiltrates, atelectasis, pneumonitis, pneumothorax, pleural effusions, reactive airway disease, congestive heart failure, pericardial effusion, and cyanotic heart disease. The interpretation of the routine chest radiographs and decisions regarding the need for interventions were made by the attending physicians and PICU fellows making the clinical decisions on the patients; for the purposes of this study, the interpretations were not confirmed by radiologists. Physicians were asked whether the result of the routine chest radiograph led to an intervention; if so, the intervention was recorded (see Table 4 for reported interventions). All interventions were recorded, and no attempt was made to judge the importance of an intervention. The collection of data was not performed by individuals involved in the care of the patient. The physicians interpreting the chest radiographs and deciding whether interventions were needed were blinded to the hypothesis of this study.

Statistical analysis was performed using ² analysis. P values for multiple comparisons are reported. Univariate logistic regression analysis was performed to obtain odds ratios between the independent variables weight group, presence of devices, and the presence of active cardiopulmonary problems and the outcome measure of interventions. The independent contribution of hypothesized factors affecting the probability of an intervention was evaluated using multiple logistic regressions. All hypothesized factors (listed above) were modeled simultaneously. Statistical analysis was performed using the statistical software package GB-Stat (New England Software, Inc, Greenwich, CT).

	RESULTS

Top Abstract Methods Results Discussion References

Descriptions of the Participating PICUs

Fifteen PICUs participated in this study (Table 1). All but 2 units had residents or pediatric critical care fellows as part of the decision-making team. The number of beds ranged from 12 to 30, whereas the median daily census was 7 to 21 patients. Two of the 15 units admitted medical and surgical (noncardiovascular) patients and 13 admitted medical, surgical, and postoperative cardiovascular surgery patients.

Mortalities in the individual PICUs ranged from 2.4% to 8.5%. Policies regarding routine chest radiographs included routine chest radiographs obtained on all intubated patients, all cardiovascular surgery patients, all postoperative or postinstrumented patients, and/or selected patients at the discretion of the clinical management team. No institution had a written policy governing the use of routine chest radiographs, and policies on routine chest radiographs reflected the current practices at the participating PICUs.

Descriptions of the Routine Chest Radiographs

Five hundred twelve chest radiographs met our definition of routine. These routine chest radiographs were evaluated at each of the participating institutions on nonconsecutive days from June to December 1997. Each institution contributed 30 to 50 evaluations. Weights of the patients for whom the routine chest radiographs were obtained ranged from 2 kg to 103 kg (median: 9 kg). The majority of the chest radiographs were obtained from patients who were 10 kg (284/512; 55%; Table 2). Almost 75% of the chest radiographs were obtained from patients with either cardiovascular disease or respiratory failure. The majority of the routine chest radiographs were obtained on mechanically ventilated patients (331/512; 65%). Of those routine chest radiographs that were obtained on patients 10 kg, 75% (213/284) were obtained on patients who were also mechanically ventilated. The length of ICU stay for patients at the time the chest radiograph was performed was 1 to 156 days (median: 4 days). Devices were present in 465 of the 512 routine chest radiographs evaluated (91%) and are listed in Table 3.

There were 3 adverse events recorded that occurred during the performance of the routine chest radiograph. These included the loss of peripheral intravenous catheters in 2 patients (1 patient in the 10-kg group, the other patient in the 10-40-kg weight group) and the dislodgment of an endotracheal tube that was replaced without difficulty (patient in the 10-kg group). This represents a <1% risk of adverse events in the performing of routine chest radiographs.

Interventions Performed as a Result of the Routine Chest Radiograph

Physicians were asked whether the interpretation of the routine chest radiograph led to an intervention. The results of the 512 routine chest radiographs led to 1 intervention in 152 cases, 2 interventions in 71 cases, and 3 interventions in 8 cases. Thus, of the 512 routine chest radiographs, 231 resulted in 1 or more interventions (45%). The types and number of interventions are listed in Table 4.

Effect of Diagnosis on Percentage of Chest X-Rays (CXRs) Resulting in Interventions

The percentage of routine chest radiographs that resulted in one or more interventions was compared among the various diagnostic groups (Table 2). Interventions were less likely to be performed as a result of the routine chest radiographs when the chest radiographs were obtained from children who had undergone noncardiovascular surgical procedures (P < .025). There was no difference in the percentage of routine chest radiographs that resulted in one or more interventions among the other diagnostic groups.

Effect of the Presence of One or More Active Cardiopulmonary Problems on Percentage of CXRs Resulting in Interventions

We compared the percentage of routine chest radiographs that resulted in one or more interventions if no active cardiopulmonary problems were present with the percentage of routine chest radiographs that resulted in one or more interventions if one or more active cardiopulmonary problems were present (Table 5). One or more interventions were performed as a result of 27% of the routine chest radiographs when no active cardiopulmonary problems were present. This percentage increased to 51% (P < .001) if active cardiopulmonary problems were present. Univariate logistic regression analysis demonstrated that the odds of an intervention being performed as a result of the chest radiograph increased 2.9 times (95% confidence interval [CI]: 1.8-4.6; corresponding to relative risk of 1.9) with the presence of an active cardiopulmonary problem.

Effect of Length of Stay on Percentage of CXRs Resulting in Interventions

The median length of stay in the ICU at the time of the routine chest radiograph for patients for whom the routine chest radiographs were evaluated was 4 days. There was no statistically significant difference in the percentage of routine chest radiographs that resulted in one or more interventions if the patients had a length of stay of 1 to 2 days (43%), compared with patients whose length of stay was 3 to 4 days (54%; P < .1), 5 to 10 days (44%; P < .9), or 11 days (43%; P < .9).

Effect of Weight on Percentage of CXRs Resulting in Interventions

Routine chest radiographs obtained on smaller patients (10 kg) were more likely to result in interventions than chest radiographs obtained on larger patients regardless of whether the patient was mechanically ventilated or not mechanically ventilated (Table 2). Of the chest radiographs obtained on patients  10 kg, 155/284 (55%) resulted in one or more interventions. This compared with 61/152 (40%; P < .005) and 15/76 (20%; P < .001) of the chest radiographs obtained on patients 10 to 40 kg and >40 kg, respectively. Univariate logistic regression analysis demonstrated that the odds of an intervention being performed as a result of the chest radiograph increased 4.9 times (95% CI: 2.7-9.0; corresponding to relative risk of 2.8) in children 10 kg compared with children >40 kg and 1.8 times (95% CI: 1.2-2.7; corresponding to relative risk of 2.0) compared with children 10 to 40 kg. The frequency of routine chest radiographs resulting in one or more interventions was significantly higher in patients who were 10 kg and mechanically ventilated (59%) and 10 to 40 kg (52%; P < .005) compared with patients >40 kg who were mechanically ventilated (23%; P < .001) (Table 2). The frequency of routine chest radiographs resulting in one or more interventions in patients who were 10 to 40 kg and mechanically ventilated was not different (56%; P < .7) from the smaller mechanically ventilated patient. Patients 10 kg who were not mechanically ventilated had a significantly higher frequency (29/71; 41%) of routine chest radiographs that resulted in one or more interventions than patients who were 10 to 40 kg (12/64; 19%; P < .01) or >40 kg (8/46; 17%; P < .01) and were not mechanically ventilated.

Effect of the Number of Devices on Percentage of CXRs Resulting in Interventions

Nearly 91% of the routine chest radiographs (465/512) evaluated for this study were performed on patients who had one or more of the devices. The frequency of interventions increased from 19% in routine chest radiographs obtained on patients with no devices to >50% in patients with more than 2 devices (P < .001; Table 5). Univariate logistic regression analysis demonstrated that the odds of an intervention being performed as a result of a chest radiograph increased 2.6 times (95% CI: 1.2-5.6; corresponding to a relative risk of 2.0) in children with 1 or 2 devices and 5.3 times (95% CI: 2.5-11.5; corresponding to a relative risk of 2.9) in children with more than 2 devices.

Effect of Patient Weight and Number of Devices on Percentage of CXRs Resulting in Interventions

As shown in Table 5, the percentage of routine chest radiographs resulting in one or more interventions is strongly associated with the number of devices. We determined whether this observation was consistent in each of the weight groups. Figure 1A demonstrates that the number of devices present significantly increased the percentage of chest radiographs resulting in one or more interventions in both the 10-kg (P < .001) and 10- to 40-kg weight groups (P < .025) but not in the percentage of chest radiographs resulting in one or more interventions in the >40-kg weight group (P < .5). In contrast, Fig 1B demonstrates that in patients who were mechanically ventilated the number of devices present had no significant effect on the percentage of chest radiographs resulting in one or more interventions in all patient weight groups (10 kg and 10-40 kg, P < .5; >40 kg, P < .75). However, there was a significant difference between the percentage of routine chest radiographs that resulted in one or more interventions between the mechanically ventilated patients >40 kg and smaller patients with 1 to 2 devices and between the mechanically ventilated patients >40 kg and smaller patients with >2 devices (P < .005). This is in sharp contrast in the group of patients who were not mechanically ventilated as demonstrated in Fig 1C. As the number of devices present increased, there was a significant increase in the percentage of chest radiographs resulting in one or more interventions in patients 10 kg (P < .025) but not in larger patients (P < .5). There was in addition a significantly greater percentage of routine chest radiographs that resulted in one or more interventions in the 10-kg weight group with 1 to 2 devices than either of the other 2 weight groups with a similar number of devices (P < .001). After inclusion of the variables in a multiple logistic regression model, all remain statistically significant independent predictors of whether an intervention was performed. Adjusted odds ratios for each of the independent variables were 3.2 (95% CI: 1.5-7.0) for the presence of devices, 2.3 (95% CI: 1.5-3.7) for the presence of active cardiopulmonary problems, and 2.1 (95% CI: 1.5-3.1) for weight 10 kg. Thus, the probability of an intervention being performed is 60% if all 3 variables are present, and 8.7% if none of the variables are present.

View larger version (34K): [in this window] [in a new window]   Fig. 1.   Effect of number of devices in patients in the various weight groups on percentage of routine chest radiographs that resulted in 1 or more interventions. The effect of the number of devices in patients in the various weight groups was assessed: dark gray boxes, 10 kg; medium gray boxes, 10 to 40 kg; light gray boxes, >40 kg. A, All patients; 10 kg with no devices compared with 1 to 2 devices or >2 devices (P < .001), 10 to 40 kg group with no devices compared with 1 to 2 devices (NS) or with >2 devices (P < .025), >40 kg with no devices compared with 1 to 2 devices (NS) or with >2 devices (NS). B, Mechanically ventilated patients; all weight groups with 1 to 2 devices compared with same weight group with >2 devices (NS), 10 kg or 10 to 40 kg group with 1 to 2 devices or >2 devices compared >40 kg group with same number of devices (P < .025). C, Not mechanically ventilated patients; 10 kg with no devices compared with 1 to 2 devices (P < .025) and 10 kg with no devices compared with >2 devices (NS). Ten to 40 kg and >40 kg with no devices compared with 1 to 2 devices or >2 devices (NS), 10 kg with 1 to 2 devices compared with 10 to 40 kg and >40 kg with 1 to 2 devices (P < .001). NS = not significant.

	DISCUSSION

Top Abstract Methods Results Discussion References

Routine diagnostic tests performed in the ICU setting should aid the clinician in the management of the patient, ideally, without the addition of high costs or morbidity to the patient. Routine chest radiographs have been suggested on mechanically ventilated adult patients as well as those with acute cardiopulmonary problems to help identify unsuspected problems that might require a change in patient management.^1-4,8 We have evaluated the usefulness of the routine chest radiograph in critically ill children in a prospective, multiinstitutional survey by asking physicians what, if any, interventions were performed as the result of the routine chest radiograph. Almost 45% of all routine chest radiographs evaluated in this study resulted in one or more interventions. We were also able to identify subgroups of patients in whom the routine chest radiograph was more likely to result in interventions. For example, >50% of routine chest radiographs performed on critically ill children 10 kg with one or more devices resulted in one or more interventions in our study, suggesting that routine chest radiographs have a significant impact on the management of this population of critically ill children. Routine chest radiographs performed on critically ill children >40 kg with 2 or fewer devices, in contrast, were less likely to result in interventions, suggesting that these routine chest radiographs had a smaller impact on the management of the child.

Our study differs from previous reports in several ways. First, most studies examining the usefulness of routine chest radiographs have involved the adult ICU population, which is significantly different from the PICU population. Second, our study examines the usual practice of how the routine chest radiograph is interpreted and used; that is, the physician team making the clinical decisions regarding the patient often interprets the chest radiograph early in the morning before or during rounds and makes decisions regarding interventions at that time. Confirmation by a radiologist often occurs after most of those decisions are made. Our study was not designed to examine whether interventions were performed based on the interpretation by the radiologist, rather whether the chest radiograph resulted in an intervention based on a decision made by the clinical management team. That is the current practice in the PICUs that participated in this study. Further analysis of our data is needed to determine whether there is congruence between the interpretation by the clinical decision-making team and the interpretation by the radiologist.

Although there are similarities in our study with a previous report examining routine chest radiographs in the PICU population, there are significant differences. In the study by Hauser et al,¹⁷ nearly 24% of chest radiographs resulted in interventions. However, it is difficult to compare the population in their study with the population in our study because detailed information regarding which patients in their study required interventions is lacking. Furthermore, patients had multiple chest radiographs evaluated for the study; that is, 353 chest radiographs from only 101 different patients from the same institution were evaluated, whereas 512 different patients from 15 different institutions were evaluated in our study. It is unclear whether the patients described by Hauser et al who had interventions came from a specific subgroup of the 101 patients. Furthermore, although they summarize that young, intubated, mechanically ventilated patients with indwelling central venous catheters benefited the most from routine chest radiographs, this is based on whether interval changes could be predicted, not whether the routine chest radiograph altered the management of the patient. Our study differs from their study in that the actual interventions performed because of the routine chest radiograph were analyzed rather than whether one could predict the findings on the chest radiograph.

One of the objectives of this study was to identify patient subgroups for whom the routine chest radiograph is most useful. In this manner, we could also identify subgroups for whom the routine chest radiograph was less likely to result in changes in management, thereby reducing costs by decreasing the use of routine chest radiographs. We did not attempt, however, to analyze cost-savings that would have occurred if routine chest radiographs had not been performed. We believe that the type of analysis performed in this study is important because it attempts to identify subgroups of patients in whom certain diagnostic tests need not be performed because of the low likelihood that the test will lead to a change in clinical management.

Based on our study alone, however, the conclusion that routine chest radiographs should not be performed on certain subgroups of patients must be viewed with caution. Less than 20% of the routine chest radiographs obtained in the >40 kg group with no devices led to an intervention; however, some of those interventions such as changes in diuretic therapy and mechanical ventilation could significantly affect patient outcome. Arguably, the use of other clinical data, such as fluid balance and arterial blood gases, might have obviated the need for the chest radiograph in making management decisions regarding diuretics or the ventilator. The patients in this group, however, still had clinically important decisions made based on the results of the routine chest radiograph. It is difficult to judge the level of interventions that would justify not performing routine chest radiographs. For some physicians, even a low level of interventions could potentially be very important in the care of critically ill children. The cost-savings of not performing routine chest radiographs could be substantial; however, without being able to reliably place a value on the interventions performed during this study, it is difficult to determine the additional costs of caring for patients in whom interventions because of the routine chest radiograph prevented bad outcomes. In addition, chest radiographs that did not lead to an intervention but eliminated a potential cause for a clinical finding are not deemed important by the type of analysis performed in this study. These chest radiographs can also be of great significance because they could lead to further clinical decisions. A randomized trial in which patients receive a daily chest radiograph or one less frequently would better evaluate issues of costs and impact on outcome. The need for routine chest radiographs in patients in whom there is a low frequency of interventions based on the radiograph, therefore, should continue to be evaluated carefully.

Some of the interventions that were performed based on the result of the routine chest radiograph might be considered minor and have little clinical importance. We did not attempt to judge how important the intervention was in the care of the patient. There are differing views on the importance of the various interventions that we report in this study and some would argue that even seemingly minor interventions such as an adjustment in a nasogastric tube or the institution of chest physiotherapy could be of significant clinical importance in the management of a critically ill child. A study designed to evaluate the importance of these various types of interventions may be warranted.

In addition, our study design identified interventions based on routine chest radiographs on only one day of the patient's hospital stay of which the median for the entire study population was 4 days. A cumulative risk of interventions based on the median length of stay increases the likelihood that a routine chest radiograph for a given patient would result in an intervention. If one examines the group of patients for whom an intervention was less likely, that is, the >40-kg patient with no devices, ~20% of the routine chest radiographs on any given day led to an intervention. The cumulative risk that an intervention would be performed would increase to ~50% for any given patient in this group over the course of their ICU stay.

This study also emphasizes some of the differences between pediatric and adult patients. For example, the ausculatory examination in small children often can be misleading. The routine chest radiograph can aid the clinician in identifying areas of atelectasis and infiltrates; indeed, changes in chest physiotherapy were guided by the routine chest radiograph 62 times, and this intervention was more likely in the child 10 kg than in larger children (Table 4). The distance between the vocal cords and carina in the child 10 kg is smaller making appropriate positioning of the endotracheal tube more difficult. Adjustments in the position of the endotracheal tube based on the result of the routine chest radiograph accounted for the greatest number of interventions in the smallest weight group of patients. Indeed, 75 of the total interventions were adjustments in devices with the majority (61/75) in the children 10 kg. The data also demonstrate that as the patients in our study approached adult size, the percentage of routine chest radiographs that led to one or more interventions is similar to that reported by others.^12,14,15 We also confirm the findings of others that there is a greater likelihood of interventions based on the routine chest radiograph in mechanically ventilated patients, although the differences in those patients >40 kg are not as large.¹⁵ The use of routine chest radiographs in the smaller critically ill child, therefore, aids in the overall assessment of these challenging patients but whether routine chest radiographs have an impact on patient outcomes remains to be determined.

There are several limitations to this type of analysis. First, there was an unequal number of chest radiographs evaluated by each institution. This number varied between 30 and 50; there is, thus, a selection bias introduced by the practice patterns of physicians from institutions that contributed a larger number of evaluations. We attempted to limit this selection bias by obtaining more evaluations from the smaller ICUs. Second, the design of the study may have led to changes in the practice patterns of the physicians managing the patients, thereby affecting the frequency of interventions. We attempted to minimize this effect by collecting data during different weeks when different teams of physicians may have been making the management decisions on the patients and attempting to keep the decision-making team blinded to the hypothesis of the study. Third, some of the decisions made by the bedside physician to intervene based on the results of the chest radiograph most likely also took into account other clinical data. For example, the decision to use diuretic therapy could also have included fluid balance in the previous 24 hours. We attempted to minimize this by the specific question that we asked (ie, did the result of the routine chest radiograph lead to an intervention?). Finally, this study did not examine whether the use of routine chest radiographs improves patient outcome or reduces costs. We believe that the best approach to examine these issues is a study in which patients are randomized to receive or to not receive a routine daily chest radiograph.

Despite its limitations, we believe that this study identifies populations of critically ill children in whom the routine chest radiograph has a significant impact in patient management and populations in whom the routine chest radiograph is less likely to have an impact on management decisions. Specifically, critically ill children who are mechanically ventilated and <40 kg are more likely to have interventions based on a routine chest radiograph than children >40 kg. Children who are not mechanically ventilated and 10 kg are more likely to have interventions based on a routine chest radiograph than children >10 kg. We hope that additional studies will identify other patient populations in whom the use of routine tests can be minimized to reduce health care costs while maintaining high quality care of critically ill children.

	ACKNOWLEDGMENTS

This project was supported in part by the Crippled Children's Foundation and the National Association of Children's Hospitals and Related Institutions.

	FOOTNOTES

Received for publication Jan 13, 2000; accepted Jul 20, 2000.

Reprint requests to (M.W.Q.) Division of Critical Care, Department of Pediatrics, Crippled Children's Foundation Research Center, Le Bonheur Children's Medical Center, 50 N Dunlap. Memphis, TN 38103. E-mail: mquasney{at}utmem.edu

	ABBREVIATIONS

PICU, pediatric intensive care unit; ICU, intensive care unit; CXR, chest x-rays; CI, confidence interval.

	REFERENCES

Top Abstract Methods Results Discussion References

1. American College of Radiology. American College of Radiology Standard for Performance of Pediatric, and Adult Bedside (Portable) Chest Radiography ACR Standards. American College of Radiology, Reston, VA; 1998:33
2. Goldenheim PD, Kazemi H Cardiopulmonary monitoring of critically ill patients. New Engl J Med 1984; 311:717-720 [Medline]
3. Horst HM, Fagan B, Beute GH Chest radiographs in surgical intensive care patients: a valuable "routine." Henry Ford Hosp Med J 1986; 34:84-86 [Medline]
4. Henschke CI, Yankelevitz DF, Wand A, Davis SD, Shiau M Accuracy and efficacy of chest radiography in the intensive care unit. Radiol Clin North Am 1996; 34:21-31 [Medline]
5. Greenbaum DM, Marschall KE The value of routine daily chest x-rays in intubated patients in the medical intensive care unit. Crit Care Med 1982; 10:29-30 [Medline]
6. Henschke CI, Pasternack GS, Schroeder S, Hart KK, Herman PG Bedside chest radiography: diagnostic efficacy. Radiology 1983; 149:23-26 [Abstract/Free Full Text]
7. Janower ML, Jennas-Nocera Z, Mukai J Utility and efficacy of portable chest radiographs. Am J Radiol 1984; 142:265-267 [Abstract/Free Full Text]
8. Strain DS, Kinasewitz GT, Vereen LE, George RB Value of routine daily chest x-rays in the medical intensive care unit. Crit Care Med 1985; 13:534-536 [Medline]
9. Bhagwanjee S, Muckart DJJ Routine daily chest radiography is not indicated for ventilated patients in a surgical ICU. Intensive Care Med 1996; 22:1335-1338 [Medline]
10. Silverstein DS, Livingston DH, Elcavage J, Kovar L, Kelly KM The utility of routine chest radiography in the surgical intensive care unit. J Trauma 1993; 35:43646
11. Liebman PR, Philips E, Weisel R, Ali J, Hechtman HB Limitations of portable radiography of the chest in patients with acute respiratory failure. Surg Gynecol Obstet 1978; 146:705-708 [Medline]
12. Brainsky A, Fletcher RH, Glick HA, Lanken PN, Williams SV, Kundel HL Routine portable chest radiographs in the medical intensive care unit: effects and costs. Crit Care Med 1997; 25:801-805 [CrossRef][Medline]
13. Bekemeyer WB, Crapo RO, Calhoon S, Cannon CY, Clayton PD Efficacy of chest radiography in a respiratory intensive care unit: a prospective study. Chest 1985; 88:691-696 [Abstract/Free Full Text]
14. Hall JB, White SR, Karrison T Efficacy of daily routine chest radiographs in intubated, mechanically ventilated patients. Crit Care Med 1991; 19:689-693 [Medline]
15. Marik PE, Janower ML The impact of routine chest radiography on ICU management decisions: an observational study. Am J Crit Care 1997; 6:95-98
16. Fong Y, Whalen GF, Hariri RJ, Barie PS Utility of routine chest radiographs in the surgical intensive care unit. Arch Surg 1995; 130:764-768 [Abstract/Free Full Text]
17. Hauser GJ, Pollack MM, Sivit CJ, Taylor GA, Bulas DI, Guion CJ Routine chest radiographs in pediatric intensive care: a prospective study. Pediatrics 1989; 83:465-470 [Abstract/Free Full Text]