0
Original Research: PNEUMONIA |

Antibiotic Timing and Diagnostic Uncertainty in Medicare Patients With Pneumonia*: Is it Reasonable to Expect All Patients to Receive Antibiotics Within 4 Hours? FREE TO VIEW

Mark L. Metersky, MD, FCCP; Thomas A. Sweeney, MD; Martin B. Getzow, MD; Farhan Siddiqui, MD; Wato Nsa, MD, PhD; Dale W. Bratzler, DO, MPH
Author and Funding Information

*From the Division of Pulmonary and Critical Care (Dr. Metersky) and the Department of Medicine (Dr. Siddiqui), University of Connecticut School of Medicine, Farmington, CT; the Department of Emergency Medicine (Dr. Sweeney), Christiana Care Health System, Newark, DE; Family Care Medical Center (Dr. Getzow), Chalfont, PA; Oklahoma Foundation for Medical Quality, Inc (Drs. Nsa and Bratzler), Oklahoma City, OK.

Correspondence to: Mark L. Metersky, MD, FCCP, Division of Pulmonary and Critical Care Medicine, University of Connecticut Health Center, 263 Farmington Ave, Farmington, CT 06030-1321; e-mail: Metersky@nso.uchc.edu



Chest. 2006;130(1):16-21. doi:10.1378/chest.130.1.16
Text Size: A A A
Published online

Background: Many organizations, including the Centers for Medicare & Medicaid Services, measure the percentage of patients hospitalized with pneumonia who receive antibiotics within 4 h of presentation. Because the diagnosis of pneumonia can be delayed in patients with an atypical presentation, there are concerns that attempts to achieve a performance target of 100% may encourage inappropriate antibiotic usage and the diversion of limited resources from seriously ill patients. This study was performed to determine how frequently Medicare patients with a hospital discharge diagnosis of pneumonia present in a manner that could potentially lead to diagnostic uncertainty and a resulting appropriate delay in antibiotic administration.

Methods: Randomly selected charts of hospitalized Medicare patients who have received diagnoses of pneumonia were reviewed independently by three reviewers to determine whether there was a potential reason for a delay of antibiotic administration other than quality of care. Antibiotic administration timing, patient demographic, and clinical characteristics were also abstracted.

Results: Nineteen of 86 patients (22%; 95% confidence interval, 13.7 to 32.2) presented in a manner that had the potential to result in delayed antibiotic treatment due to diagnostic uncertainty. Diagnostic uncertainty was significantly associated with the lack of rales, normal pulse oximetry findings, and lack of an infiltrate seen on the chest radiograph. There was a nonsignificant trend toward a longer time until antibiotic treatment in patients with diagnostic uncertainty.

Conclusions: Many Medicare patients in whom pneumonia has been diagnosed present in an atypical manner. Delivering antibiotic treatment within 4 h for all patients would necessitate the treatment of many patients before a firm diagnosis can be made.

The Centers for Medicare & Medicaid Services conducts a multipronged program to encourage improvements in quality of care for Medicare beneficiaries. The timing of antibiotic therapy for patients who have been admitted to the hospital with pneumonia has been an audited performance measure for pneumonia for many years,1as reports have demonstrated improved outcomes among patients who received antibiotics within 4 h of presentation.23

With the recent trend of using this and other performance measures as the basis for public reporting and pay-for-performance programs, there is increasing pressure for hospitals and physicians to drive their performance rates as high as possible.46 The only way to ensure not being outperformed by another institution in the competition for reimbursement (in many pay-for-performance programs) or for patients (in the case of publicly reported quality measures) is to achieve 100% adherence.4

It is well-known that many elderly patients with pneumonia present with atypical signs and symptoms.78 Thus, in order to achieve 100% performance, physicians may be compelled to administer antibiotics before a firm diagnosis can be established,46 a practice that has been described as “shoot first and ask questions later.”6 Inappropriate antibiotic use could induce increases in antibiotic resistance and the number of antibiotic-related adverse events, and may decrease the yield of subsequently required diagnostic tests.46 Another commonly voiced concern is that the increased focus on patients with pneumonia may delay care that should be given to other acutely ill patients.4,6,9

It has been suggested that one way of reducing the likelihood of these negative consequences is to determine the percentage of patients who present in a manner in which the treating physician could reasonably be expected to diagnose pneumonia and deliver antibiotic therapy within 4 h.45 Then, the performance goal could be set at that percentage, and any hospital reaching that goal could be reported as having achieved the goal, without reporting the actual percentage achieved.4 In this way, there would be no incentive for hospitals to strive for ever higher performance rates. This study was performed in order to determine how frequently Medicare patients with pneumonia present in a manner that is likely to allow diagnosis and antibiotic treatment within 4 h. A secondary aim of this study was to determine what patient characteristics predict a delay in the diagnosis of pneumonia or in delivering antibiotic therapy.

Randomly selected charts of Medicare patients with a hospital discharge diagnosis of pneumonia fulfilling the criteria for inclusion in Medicare pneumonia reporting were made available to the investigators. The criteria for inclusion in the Medicare reporting have been described in detail previously.10 A key inclusion criterion for the antibiotic timing measure was that pneumonia had to be among the diagnoses being considered at the time of hospital admission. A confirmatory chest radiograph was not required. Patients with documented antibiotic treatment within 24 h prior to presentation were excluded. A sample size of 80 to 100 patients was planned, due to investigator time and resource constraints, as this was an unfunded study. With this sample size, an analysis prior to commencing the study revealed an acceptable predicted 95% confidence interval based on an anticipated prevalence of diagnostic uncertainty of at least 15%.

Three clinicians from varied backgrounds were recruited to perform the chart reviews. One was a clinical academic pulmonologist at a suburban university hospital in Connecticut, one was an emergency department (ED) physician practicing at a large urban non-medical school-affiliated teaching hospital in Delaware, and one was a community-based family practitioner at a nonteaching hospital in Pennsylvania.

Each reviewer was supplied with any available prehospital records (ie, transfer notes and emergency services records) and ED records, including the results of diagnostic studies that would normally be available during the ED stay. These generally included blood chemistry measurements, CBC count, initial radiographic study findings, and occasionally a CT scan or urinalysis finding. Other than the official finding of the chest radiograph performed in the ED, no records that would have been unavailable to the ED physician were provided to the reviewers. In order to avoid bias related to knowledge of the actual timing of antibiotic administration, the timing of events occurring during the patient’s ED stay was redacted from the chart. Since patients who were admitted directly to the hospital ward are included in reporting for the Medicare performance indicators, they were also included in this study. For these patients, any data sent to the hospital with the patient, as well as the hospital admission history and physical examination and hospital admission laboratory results, were provided to the reviewers.

Each chart was reviewed independently by the reviewers, who were asked one key question, as follows: was there a potential reason for a delay of antibiotic administration other than quality of care? It was anticipated that such cases in which the answer was “yes” would generally be due to diagnostic uncertainty such that the diagnosis of pneumonia would likely not have been made soon enough to deliver antibiotics within 4 h. Inherent in this designation was acceptance of the premise that in the absence of diagnostic uncertainty or an unusual patient-related circumstance that prevented the timely delivery of antibiotics, a delay in antibiotic treatment represented a lapse in quality.

After a pilot study of 10 charts, it became clear that there were patients in whom a diagnosis of pneumonia was unlikely to have been made within 4 h, but in whom a need for therapy with antibiotics would have been clear within 4 h (ie, a patient who was more likely to have had a purulent exacerbation of chronic bronchitis [ie, COPD]). The reviewers felt that the ED physician could appropriately defer the choice of antibiotics to the admitting physician in such cases. This scenario was then allowed for in the reviewers’ responses. In patients who were critically ill with a probable infectious etiology, the reviewers felt that antibiotics should have been delivered within 4 h, even if the diagnosis of pneumonia was not clear. After the independent reviews, there were conference calls during which the reviewers attempted to achieve consensus on any cases for which there were discordant responses. The reasons for opinions were solicited and recorded.

We also attempted to determine what clinical factors were associated with the timing of antibiotic administration and diagnostic uncertainty. The chart abstraction for these analyses was performed independently of the above-described review process. These analyses were performed with the a priori understanding that we had limited statistical power. Therefore, antibiotic administration timing was treated as both a dichotomous variable (ie, > 4 h or < 4 h) and as a continuous variable. The initial chest radiograph reports were recorded as either suggesting pneumonia or not. The report of pneumonia, infiltrates, consolidation, opacity, or a similar synonym was accepted as representing pneumonia unless the conclusion made no mention of pneumonia and contained an alternate diagnosis. Summary statistics and univariate tests of association were performed. The χ2 test was used for categoric variables, and the Student t test was used for continuous variables. Statistical significance was accepted at p = 0.05. Analyses were conducted using a statistical software package (SAS, version 8.0; SAS Institute; Cary, NC). As a component of the Centers for Medicare & Medicaid Services quality-improvement activities, this study was exempt from institutional review board approval.

Eighty-seven charts were reviewed. Consensus among the three reviewers was achieved for 86 cases (99%). In one case, consensus could not be achieved as discordant interpretations of the chest radiograph in the medical record led to an inability to come to an agreement. This case was excluded from the study. The mean (± SD) age of the patients was 79 ± 9 years. Sixty-two patients (72%) were admitted to the hospital from home. Among the 86 included cases, there were 67 (78%) in which the reviewers’ consensus was that the presentation of pneumonia was clear enough that the delivery of antibiotics would not have been delayed unless there was a lapse in the quality of care. The interobserver agreement among all three reviewers for these cases was quite high, at 60 of 65 cases (92%).

In 19 of 86 cases (22%; 95% confidence interval, 13.7 to 32.2), the consensus was that there were factors that had the potential to delay antibiotic treatment, independent of quality of care. These were all due to diagnostic uncertainty such that a diagnosis of pneumonia would likely not have been made soon enough to allow antibiotic delivery within 4 h. The diagnostic uncertainty was caused by two distinct scenarios. In one scenario, the patient presented in an unusual fashion but was ultimately found to have clear evidence of pneumonia. Three examples include patients who presented with abdominal pain, focal neurologic deficits, and ventricular tachycardia. In the second scenario, the patient presented with clear pulmonary symptoms but there was difficulty in determining whether the cause was pneumonia or another etiology, such as heart failure or COPD exacerbation. Often, while pneumonia was listed by the ED physician as one of the final diagnoses, heart failure or exacerbation of COPD was also mentioned, and frequently was listed above pneumonia. In six of these cases, the reviewers felt that antibiotic treatment would have been appropriate for any alternative diagnosis (usually an acute exacerbation of COPD), but it would have been appropriate for the ED physician to defer the antibiotic choice to the admitting physician.

Among these 19 cases, there was initial interobserver disagreement for 14 cases (74%). There were three specific reasons for initial disagreement. In two cases, a reviewer overlooked a key piece of the data in the record, and once this was pointed out he changed his opinion. More often, the differences were due to diagnostic uncertainty. There were disagreements over how likely the diagnosis of pneumonia was for specific patients. In addition, there was not complete agreement on how likely the diagnosis of pneumonia should be before the ED physician should administer antibiotics. This arose most frequently in patients who clearly had respiratory symptoms but could have had pneumonia, heart failure, or an acute exacerbation of COPD. Although there is obviously no “correct” answer, the discussion of this issue can be paraphrased as follows: at what threshold should antibiotics be administered by the ED physician? Should antibiotics be administered only if the physician is nearly certain that the patient has pneumonia, or even if there is only 20% certainty?

Among the 86 patients included in the study, 8 (9%) were admitted directly to the hospital with a suspicion of pneumonia. Since there was no chance of diagnostic uncertainty and there are significant differences in how care is delivered to patients who are not admitted to the hospital through the ED, such patients were excluded from the analyses of the factors associated with diagnostic uncertainty and antibiotic timing. Table 1 demonstrates the demographic and clinical characteristics as well as the timing of antibiotic administration for the remaining 78 patients, stratified by whether or not there was diagnostic uncertainty. Acute mental status changes were almost twice as common in patients with diagnostic uncertainty, but this difference did not reach statistical significance (p = 0.12). The lack of rales, oxygen desaturation, or a chest radiograph suggesting pneumonia were all significantly more common in patients with diagnostic uncertainty. It is noteworthy that among these 78 patients, all with a hospital discharge diagnosis of pneumonia, 19 (24%) did not have a chest radiograph finding that was suggestive of pneumonia. Two of these patients also had a chest CT scan that did not demonstrate an infiltrate.

Twenty-one of the patients (27%) who had been admitted to the hospital through the ED received initial antibiotic therapy > 4 h after presentation. Thirty-eight percent of these patients had abdominal symptoms, while only 11% of those without delayed antibiotic administration had abdominal symptoms (p = 0.005). None of the other clinical factors listed in Table 1 were associated with the receipt of antibiotics after 4 h. Table 2 demonstrates the association between the mean time to antibiotic administration and patient characteristics. A significantly longer time until the receipt of antibiotics was associated with hospital admission from a facility other than home. Patients with abdominal symptoms received antibiotics a mean of 2 h and 23 min later than patients without such symptoms, but this difference fell short of achieving statistical significance (p = 0.07).

Among the 86 hospitalized Medicare patients with a hospital discharge diagnosis of pneumonia, 19 (22%) had a potential reason for a delay in the initial administration of antibiotics, all due to diagnostic uncertainty. These patients were significantly less likely to have rales, oxygen desaturation, or chest radiograph findings suggesting pneumonia. They were approximately twice as likely to have acute mental status changes, but this difference did not achieve statistical significance, probably due to insufficient statistical power.

These results have important implications for regulatory agencies and payers that use the timing of the initial administration of antibiotics as a pneumonia performance measure, with an implied target of 100% adherence. It may be inappropriate to expect hospitals to deliver antibiotics to all pneumonia patients within 4 h of presentation, as doing so would necessitate antibiotics being administered to many patients in whom the diagnosis of pneumonia would still be in doubt.

There is already concern that a target of 100% leads to inappropriate antibiotic use,46 which could decrease the yield of subsequent diagnostic tests in some patients and could stimulate the development of antibiotic resistance. Perhaps most importantly, efforts to achieve a target of 100% might result in the diversion of limited resources from other patients who might be more in need of prompt care.4,6,9

Our finding that patients without chest radiograph findings suggesting pneumonia received antibiotics almost as quickly as patients who had infiltrates seen on chest radiographs suggests that concerns regarding inappropriate antibiotic use may be well-founded. The presence of respiratory symptoms may prompt antibiotic usage, whether or not there is a confirmatory radiograph. It might be argued that these patients all received a hospital discharge diagnosis of pneumonia, so that antibiotic use was appropriate even if there was initial diagnostic uncertainty. However, studies employing CT scans demonstrate that only about 25% of patients with suspected pneumonia and a negative chest radiograph finding actually have pneumonia,11 so it is likely that many of the patients whose chest radiograph findings did not demonstrate the presence of infiltrates did not have pneumonia. It is virtually certain that the two patients without evidence of pneumonia on their CT scan did not have pneumonia.

Some limitations of our study must be acknowledged. Like all studies employing chart review, we were dependent on what was documented, so there may have been important factors that were not reflected in the chart that could have altered the reviewers’ findings. We did not find a significant difference in antibiotic timing between patients with diagnostic uncertainty and those without. This may have been in part due to our low sample size or the presence of other factors such as ED patient volume and staff/patient ratios that could affect the timing of the administration of antibiotics. This result could also suggest that the reviewers’ finding of diagnostic uncertainty was not relevant to what actually occurred when these patients were treated. However, this seems unlikely, given the lower frequency of physical examination and chest radiograph findings and frequent evidence of diagnostic uncertainty in the medical records of these patients. A more likely interpretation is that the timing of antibiotic administration performance measure prompts antibiotic treatment as soon as respiratory symptoms are elicited, often before a diagnosis of pneumonia is confirmed.

What change could mitigate the negative consequences associated with attempting to deliver antibiotics to 100% of pneumonia patients within 4 h? Our results suggest that the inclusion only of patients with a confirmatory chest radiograph finding would markedly lower the number of patients included in Medicare quality reporting who present in a manner that is likely to result in a delayed diagnosis of pneumonia. Even this change would not eliminate such patients, as 47% of them had a positive chest radiograph finding. We suggest that the implied performance goal be changed from 100%, to the percentage of patients who present in a manner that would not be likely to lead to a delay in diagnosis. Furthermore, for hospitals that achieve the goal, reporting only the achievement of the goal, and not the actual percentage, would prevent the negative consequences of competition between hospitals to achieve an unrealistically high performance. These changes would acknowledge the fact that the diagnosis of pneumonia is often imprecise and might lessen the criticism being directed at organizations employing the timing of antibiotic administration as a performance measure.6,9 Our study suggests that the appropriate goal may be approximately 75 to 80%, but this would need to be validated with a study having a larger sample size. The amount of hospital-to-hospital variation would also need to be determined.

In summary, we found that 22% of Medicare patients presented in a manner that was likely to result in a delayed diagnosis of pneumonia even in the setting of high-quality medical care. Several patient factors were associated with such a presentation. These results suggest that the performance target for the timing of antibiotic administration be set at an appropriate, evidence-based level.

Abbreviation: ED = emergency department

None of the authors have a conflict of interest to report. Drs. Bratzler and Nsa are employed by the Oklahoma Foundation for Medical Quality, Inc, which has a contract with the Centers for Medicare & Medicaid Services to perform quality improvement functions for Medicare beneficiaries. The analyses upon which this publication is based were performed under contract No. 500–02-OK-03, which is funded by the Centers for Medicare & Medicaid Services, an agency of the US Department of Health and Human Services. The content of this publication does not necessarily reflect the views of policies of the Department of Health and Human Services, nor does the mention of trade names, commercial products, or organizations imply endorsement by the US Government. The authors assume full responsibility for the accuracy and completeness of the ideas presented. PNEU-002-QISC-OK-1205.

Table Graphic Jump Location
Table 1. The Associations Among Diagnostic Uncertainty, Patient Characteristics, and Antibiotic Timing*
* 

Values are given as no. of patients (%), unless otherwise indicated.

Table Graphic Jump Location
Table 2. The Association Between the Mean Time to First Antibiotic Dose and Patient Characteristics Among Patients Admitted Through the ED
Jencks, SF, Cuerdon, T, Burwen, DR, et al (2000) Quality of medical care delivered to Medicare beneficiaries: a profile at state and national levels.JAMA284,1670-1676. [CrossRef] [PubMed]
 
Houck, PM, Bratzler, DW, Nsa, W, et al Timing of antibiotic administration and outcomes for Medicare patients hospitalized with community-acquired pneumonia.Arch Intern Med2004;164,637-644. [CrossRef] [PubMed]
 
Ziss, DR, Stowers, A, Feild, C Community-acquired pneumonia: compliance with Centers for Medicare and Medicaid Services, national guidelines, and factors associated with outcome.South Med J2003;96,949-959. [CrossRef] [PubMed]
 
Infectious Diseases Society of America.. Misuse of pneumonia guidelines raises concerns.IDSA News2006;15,1-16
 
Houck, PM, Bratzler, DW Administration of first hospital antibiotics for community-acquired pneumonia: does timeliness affect outcomes?Curr Opin Infect Dis2005;18,151-156. [CrossRef] [PubMed]
 
Walls RM, Resnick, JB. The CMS blood cultures for CAP program: the architects speak out. Available at: http://emergency-medicine.jwatch.org/cgi/content/full/2005/427/1. Accessed January 9, 2006.
 
Fein, AM, Feinsilver, SH, Niederman, MS Atypical manifestations of pneumonia in the elderly.Clin Chest Med1991;12,319-336. [PubMed]
 
Marrie, TJ Community-acquired pneumonia in the elderly.Clin Infect Dis2000;31,1066-1078. [CrossRef] [PubMed]
 
Walls, RM, Resnick, JB The Joint Commission on Accreditation of Healthcare Organizations and Centers for Medicare and Medicaid Services community-acquired pneumonia initiative: what went wrong?Ann Emerg Med2005;46,409-411. [CrossRef] [PubMed]
 
Metersky, ML, Ma, A, Bratzler, DW, et al Predicting bacteremia in patients with community-acquired pneumonia.Am J Respir Crit Care Med2004;169,342-347. [PubMed]
 
Syrjala, H, Broas, M, Suramo, I, et al High-resolution computed tomography for the diagnosis of community-acquired pneumonia.Clin Infect Dis1998;27,358-363. [CrossRef] [PubMed]
 

Figures

Tables

Table Graphic Jump Location
Table 1. The Associations Among Diagnostic Uncertainty, Patient Characteristics, and Antibiotic Timing*
* 

Values are given as no. of patients (%), unless otherwise indicated.

Table Graphic Jump Location
Table 2. The Association Between the Mean Time to First Antibiotic Dose and Patient Characteristics Among Patients Admitted Through the ED

References

Jencks, SF, Cuerdon, T, Burwen, DR, et al (2000) Quality of medical care delivered to Medicare beneficiaries: a profile at state and national levels.JAMA284,1670-1676. [CrossRef] [PubMed]
 
Houck, PM, Bratzler, DW, Nsa, W, et al Timing of antibiotic administration and outcomes for Medicare patients hospitalized with community-acquired pneumonia.Arch Intern Med2004;164,637-644. [CrossRef] [PubMed]
 
Ziss, DR, Stowers, A, Feild, C Community-acquired pneumonia: compliance with Centers for Medicare and Medicaid Services, national guidelines, and factors associated with outcome.South Med J2003;96,949-959. [CrossRef] [PubMed]
 
Infectious Diseases Society of America.. Misuse of pneumonia guidelines raises concerns.IDSA News2006;15,1-16
 
Houck, PM, Bratzler, DW Administration of first hospital antibiotics for community-acquired pneumonia: does timeliness affect outcomes?Curr Opin Infect Dis2005;18,151-156. [CrossRef] [PubMed]
 
Walls RM, Resnick, JB. The CMS blood cultures for CAP program: the architects speak out. Available at: http://emergency-medicine.jwatch.org/cgi/content/full/2005/427/1. Accessed January 9, 2006.
 
Fein, AM, Feinsilver, SH, Niederman, MS Atypical manifestations of pneumonia in the elderly.Clin Chest Med1991;12,319-336. [PubMed]
 
Marrie, TJ Community-acquired pneumonia in the elderly.Clin Infect Dis2000;31,1066-1078. [CrossRef] [PubMed]
 
Walls, RM, Resnick, JB The Joint Commission on Accreditation of Healthcare Organizations and Centers for Medicare and Medicaid Services community-acquired pneumonia initiative: what went wrong?Ann Emerg Med2005;46,409-411. [CrossRef] [PubMed]
 
Metersky, ML, Ma, A, Bratzler, DW, et al Predicting bacteremia in patients with community-acquired pneumonia.Am J Respir Crit Care Med2004;169,342-347. [PubMed]
 
Syrjala, H, Broas, M, Suramo, I, et al High-resolution computed tomography for the diagnosis of community-acquired pneumonia.Clin Infect Dis1998;27,358-363. [CrossRef] [PubMed]
 
NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging & repositioning the boxes below.

CHEST Journal Articles
PubMed Articles
Guidelines
Feverish illness in children: assessment and initial management in children younger than 5 years.
National Collaborating Centre for Women's and Children's Health | 8/28/2009
Blepharitis.
American Academy of Ophthalmology | 6/5/2009
  • CHEST Journal
    Print ISSN: 0012-3692
    Online ISSN: 1931-3543