Impact of the Joint Commission Pneumonia Core Measure on Antibiotic Use and Selection for Community-Acquired Pneumonia in the Emergency Room

Abstract

Background

Prior to 2012, The Joint Commission (TJC) pneumonia core measure (PN-5) required antibiotic administration for suspected community-acquired pneumonia (CAP) within 6 hours of arrival to the emergency room (ER). In 2012, TJC issued PN-6 requiring antibiotic administration within 24 hours of presentation. Though PN-6 was anticipated to reduce overuse and inappropriate antibiotic use and improve appropriate antibiotic selection, the impact of PN-5 and PN-6 on optimizing care for CAP in the ER remains unknown.

Objective

To investigate the impact of TJC pneumonia core measures on antibiotic use in the ER for suspected CAP.

Methods

In this single-center study, medical records of patients 18 years old and older diagnosed with CAP in the ER during 2011 (PN-5) and 2012 (PN-6) and admitted for 1 day or longer were reviewed. Exclusion criteria included criteria for health care–associated pneumonia. Comparisons between groups were performed using descriptive statistics and contingency table analysis with chi-square or Fisher exact tests for categorical variables and t tests for continuous variables. Statistical analyses were performed using Microsoft Excel 2010 and SAS version 9.4.

Results

Antibiotic use was comparable between PN-5 and PN-6. Approximately half of patients in each group received an appropriate empiric CAP regimen (52% vs 54%; P = .807). Among inappropriate regimens, the most common reason was use of a beta-lactam alone (69% vs 83%; P = .26). More patients had an ultimate diagnosis of CAP with PN-6 (78% vs 86%; P = .3).

Conclusion

Changes in pneumonia core measure requirements did not have a significant impact on appropriate antibiotic use in the ER.

Over 5 million cases of community-acquired pneumonia (CAP) are diagnosed in the United States annually.^1–3 Approximately one-fourth of these patients require hospital admission at an average cost of $5,424 per patient and over $10 billion annually overall.^2,4,5 Of those patients requiring inpatient treatment, approximately 600,000 are Medicare recipients. ⁶ CAP accounts for approximately 5% of emergency room (ER) visits, and patients who require hospitalization have an estimated mortality rate of 6% to 12%.^1,4,6

The Joint Commission (TJC), the accreditation body for the Centers for Medicare & Medicaid Services (CMS), evaluates hospitals' compliance with core measures on varying disease states that are linked to financial reimbursement for Medicare and Medicaid patients.^7–9 Additional bonuses and pay-for-performances are also linked to such quality measure performances.^7–8 In 2004, TJC instituted a core measure directive to address pneumonia that listed several explicit requirements for the management of pneumonia in the ER. ⁸ One of the most controversial standards was the requirement to administer antibiotics within 4 hours of ER arrival to all patients with a working diagnosis of CAP (PN-5). ⁸ This requirement was based on 2 controversial studies that suggested an increase in mortality with delayed antibiotic administration.^1,4,8,10 One concern with the initial 4-hour administration requirement was that antibiotics would be used inappropriately to ensure compliance with this core measure and subsequently with Medicare reimbursement. ⁸ Most of these concerns were expressed through individual feedback or published editorials and review articles.^{1,4,6–7,10–14} The concerns do not appear to be unfounded. In a study by Nicks et al, 55% of ER physicians reported prescribing antibiotics to patients they did not believe had pneumonia in order to comply with TJC standards. ⁶ Furthermore, 42% of those who admitted to this practice said they did so at least 3 times per month. ⁶ Another study by Fee et al found that almost 25% of patients who met TJC pneumonia core measure inclusion criteria and had CAP included in the ER differential diagnosis did not have a final ER diagnosis of pneumonia. ⁷ The authors also highlighted that the majority of ER diagnoses are symptom based, which limits the certainty of diagnosis. ⁷ Other clinical conditions, such as heart failure, may present with similar respiratory symptoms but without infectious etiology. Anecdotal reports have been published that state that patients with respiratory symptoms were generically treated with antibiotics upon presentation to comply with the time requirements of the pneumonia core measure with only minimal amounts of patient-and clinical-specific information being applied to the choice of that initial antibiotic agent.^11–13 Any potential overuse and misuse use of available antimicrobial agents can complicate the already challenging situation of antibiotic resistance patterns.

In 2007, the Infectious Disease Society of America and the American Thoracic Society jointly published updated CAP guidelines. ¹⁵ The new guidelines did not suggest as stringent a requirement for the timing of antibiotics; they only recommended that antibiotics be initiated while the patient was still in the ER. ¹⁵ The authors of the guidelines elected not to recommend a specific time window for initial antibiotics, because they noted that administering antibiotics within the first 4 to 8 hours of presentation did not provide an increased survival benefit nor shorten time to clinical stability in most studies. ¹⁵ However, for logistical purposes, they did feel that if pneumonia were considered likely, the first dose should be administered while still in the ER to prevent excessive delays in therapy. ¹⁵

Following the publication of the updated CAP guidelines, TJC revised their pneumonia core measure (PN-5) so that the antibiotic administration requirement was extended from 4 to 6 hours. ⁸ TJC also allowed for an exclusionary provision for patients for whom there is “diagnostic uncertainty” regarding pneumonia diagnosis. ⁸ In January 2012, PN-5 was retired and replaced with the pneumonia core measure PN-6 (including PN-6, PN-6a, and PN-6b), which requires that patients with CAP receive initial antibiotics with 24 hours of presentation and that the selection of such antibiotics be consistent with current guidelines according to the patient care area within the hospital to which the patient is admitted. ⁸ To our knowledge, no study has been conducted to evaluate the relationship between antibiotic administration compliance with PN-5 or PN-6 in association with final or primary hospital diagnosis since the change in the pneumonia core measure's time to first antibiotic requirement was instituted. ⁸ Though the revision of the pneumonia core measure (PN-6) was anticipated to reduce the overuse and inappropriate use of antibiotics in patients without infectious causes of their respiratory symptoms as well as improve the appropriateness of antibiotic selection for patients with CAP, the impact of PN-5 or PN-6 on optimizing care for CAP in the ER remains unknown.

Objective

The objective of this single-center study was to investigate the impact of TJC pneumonia core measures on antibiotic use in the ER in patients presenting with suspected CAP. The primary outcome was the appropriateness of the initial antibiotic agent(s), including regimen and dosing, in the ER for patients presenting with suspected CAP. Secondary outcomes were final or primary diagnosis at time of hospital discharge, appropriateness of antibiotic usage in the ER in relation to hospital admission unit (intensive care unit [ICU] or non-ICU unit) and age (≥65 years or ≤64 years) for patients with suspected CAP before and after the implementation of the PN-6 pneumonia core measure.

Methods

A retrospective chart review was conducted of patients with suspected CAP admitted to DCH Regional Medical Center, Tuscaloosa, Alabama, from the ER between January 1, 2011 to July 31, 2011 (PN-5 pneumonia core measure) and January 1, 2012 to July 31, 2012 (PN-6 pneumonia core measure). To meet inclusion criteria, patients had to be age 18 years or older, have a documented diagnosis in the ER of CAP, and be admitted for an inpatient stay of at least 1 day. Patients were excluded if they met 1 or more criteria for hospital-acquired or health care–associated pneumonia (HCAP) including hospitalization for 2 days or more in the preceding 90 days; residence in a nursing home, rehabilitation center, or other extended care facility; or the receipt of home infusion therapy, outpatient intravenous antibiotics, chemotherapy, wound care, or hemodialysis within the 30 days preceding presentation to the ER. Institutional review board approvals were granted, and informed consent was waived due to the retrospective design of the study. Data collected included gender, age, antibiotic allergy information, length of stay, residence prior to admission, patient care area patient admitted to (ICU or non-ICU), pertinent medical history, antibiotic use within previous 30 days, antimicrobial use in the ER, use of inpatient antibiotics, supporting objective evidence for pneumonia, and final inpatient diagnosis. Antibiotic use in the ER was evaluated for appropriateness using the Infectious Diseases Society of America/American Thoracic Society Consensus Guidelines on the Management of Community-Acquired Pneumonia in Adults and the NAF-Endorsed Voluntary Consensus Standards for Hospital Care Pneumonia Measure Information Form version 4.0.^8,15 To achieve at least 80% power, at least 50 patients had to be evaluated per group. Basic descriptive statistics were used to evaluate continuous variables, and proportions and frequencies were used for categorical variables. Comparisons between the 2 groups were performed using contingency table analysis with a chi-square test or Fisher exact test for categorical variables and a Student t test for continuous variables. The a priori level of significance was set at .05. Statistical analyses were performed using Microsoft Excel 2010 and SAS version 9.4 (SAS, Inc., Cary, NC).

Results

There were 245 patients admitted to DCH Regional Medical Center after being diagnosed with and treated for CAP in the ER between January 1, 2011 to July 31, 2011 (PN-5) and January 1, 2012 to July 31, 2012 (PN-6). Almost half of those patients were excluded from review for meeting the qualifications for HCAP diagnosis and 18 patients were excluded due to age. The remaining 110 patients were divided similarly between the PN-5 group (n = 60) and PN-6 group (n = 50). There were no significant differences in baseline characteristics between the 2 groups ( Table 1 ). The mean age was similar in both groups (63 vs 57 years), and most participants were female. Although 25 documented antibiotic allergies were noted, no patient received an antibiotic for which he or she had a known allergy. Most patients were admitted to a non-ICU unit.

Table 1.

Demographics and baseline characteristics

	2011 (PN-5) (n = 60)	2012 (PN-6) (n = 50)
Mean age, years	63	57
Female, n (%)	39 (65)	28 (56)
Length of hospital stay, mean days	7.46	7.4
Admission source, n (%)
Community	58 (97)	49 (98)
Outside hospital a	2 (3)	1 (2)
Admission unit, n (%)
ICU	12 (20)	13 (26)
Non-ICU	48 (80)	37 (74)
Past medical history, n (%)
Arrhythmia	3 (5)	0 (0)
Asthma	14 (23)	14 (28)
Atrial fibrillation	8 (13)	4 (8)
Cancer	8 (13)	10 (20)
COPD	20 (33)	15 (30)
Heart failure	13 (22)	8 (16)
Immunosuppression b	3 (5)	0 (0)
Lung disease – other c	6 (10)	3 (6)
Thrombosis	3 (5)	2 (4)
Tuberculosis	2 (3)	3 (6)
Tobacco use, n (%)	21 (35)	17 (34)
Use of oral antibiotics within 90 days prior to admission, n (%)	3 (5)	7 (14)

Note: COPD = chronic obstructive pulmonary disease; PN-5 = pneumonia core measure 5; PN=6 = pneumonia core measure 6.

^aOnly patients transferred from outside hospitals meeting the timing definitions for community acquired pneumonia were included.

^bImmunosuppression defined as any chronic medical condition and/or chronic medication use that suppresses the immune system.

^cAny pulmonary disorder other than asthma, COPD, tuberculosis, cancer, or pulmonary arterial hypertension. No patient had a history of pulmonary arterial hypertension.

Antibiotic selection to treat suspected CAP in the ER was similar under the PN-5 and PN-6 pneumonia core measures. The majority of patients were treated with azithromycin (12 vs 15; P = .225), ceftriaxone (37 vs 39; P = .06), levofloxacin (25 vs 16; P = .3), or a combination these antibiotics. The only significant change noted in the use of specific antibiotic dosing was an increase in the use of ceftriaxone 1 g from 50% of patients under PN-5 to 65% of patients under PN-6 (P = .019). Approximately half of patients in each group (52% vs 54%; P = .807) received an appropriate regimen to treat suspected CAP per guidelines and dosing standards ( Table 2 ). The most common reason for an inappropriate regimen was the failure to utilize a macrolide, doxycycline, or levofloxacin within the regimen as directed by the guidelines, a practice that increased after the implementation of PN-6 (69% vs 83%; P = .26). There were no significant differences in the appropriate use of individual antibiotics and dosing under PN-5 compared to PN-6 with the exception of levofloxacin 500 mg IV (0% vs 8%; P = .04). There were nonsignificant improvements in the appropriate use of azithromycin and ceftriaxone individually, but not levofloxacin. However, overall use of levofloxacin also declined after PN-6 implementation, and therefore the decrease in appropriate levofloxacin use could be a reflection of an overall decline in drug utilization. Twelve patients in each group received broad-spectrum antibiotics for Pseudomonas aeruginosa, methicillin-resistant Staphylococcus aureus, or anaerobic coverage; none of these patients had risk factors for any of these pathogens.

Table 2.

Antibiotic use for community-acquired pneumonia

	2011 (PN-5) (n = 60)	2012 (PN-6) (n = 50)	P value
Appropriate regimen in ER, n (%)	31 (52)	27 (54)	.807
Correct utilization in ER of:
Azithromycin a	92% (11/12 patients)	100% (15/15 patients)	.255
Ceftriaxone	43% (16/37 patients)	51% (20/39 patients)	.483
Levofloxacin b	76% (19/25 patients)	81% (13/16 patients)	.692
Continued antibiotic use as an inpatient for CAP, n (%)	54 (90)	48 (96)	.227
Mean length of inpatient antibiotics for CAP, days	5.72	6.33	.38

Note: CAP = community-acquired pneumonia; ER = emergency room; PN-5 = pneumonia core measure 5; PN-6 = pneumonia core measure 6.

^aAzithromycin was the only macrolide prescribed for CAP during the study period.

^bLevofloxacin is the institution's formulary fluoroquinolone. No other fluoroquinolone was prescribed for CAP during the study period.

The percentage of patients diagnosed and treated for suspected CAP in the ER who ultimately had pneumonia versus a noninfectious diagnosis with respiratory signs or symptoms increased after the implementation of PN-6 (86%) compared to PN-5 (78%) (P = .3) ( Table 3 ). Almost all patients continued antibiotics during their inpatient stay (90% vs 96%; P = .227), with a mean length of stay exceeding 5 days in both groups (5.72 vs 6.33; P = .38). There were no significant differences noted in type of noninfectious final diagnosis between the 2 groups.

Table 3.

Final diagnosis at discharge

	2011 (PN-5) (n = 60)	2012 (PN-6) (n = 50)	P value
CAP, overall, n (%)	47 (73)	43 (86)	.3
CAP, single diagnosis, n	35	32	.544
CAP, with another diagnosis, n	12	11	.979
Other final diagnoses, a n	24	17	.517
Asthma b	3	1	.624
Cancer	0	2 c	.204
COPD	8	7	.919
GERD	1 d	0	1
Heart failure d	2	2	1
Medication overdose	1 e	0	1
Pulmonary edema	0	1 d	.4545
Viral infection	4 c	2 c	.687
Other f	6 e	4 c	.753

Note: CAP = community-acquired pneumonia; COPD = chronic obstructive pulmonary disease; GERD = gastroesophageal reflux disease; PN-5 = pneumonia core measure 5; PN-6 = pneumonia core measure 6.

^aSome patients had more than 1 additional noninfectious final diagnosis.

^b67% of asthma diagnoses in 2011 were documented along with a CAP diagnosis.

^c1 diagnosis documented along with a CAP diagnosis.

^dAll diagnoses were documented along with a CAP diagnosis.

^eSingular diagnosis; no concurrent CAP final diagnosis documented.

^fOther final diagnosis documented included: acute gastroenteritis, respiratory failure (noninfectious), severe back pain, bacteremia, pneumothorax, and abscess.

An analysis based on age and admission unit was conducted to determine whether differences within those subgroups existed for patients admitted under PN-5 versus PN-6 or between each other. There were no differences between any groups for receipt of an appropriate regimen. Patients admitted to the ICU in 2011 (PN-5) were administered significantly more vancomycin doses in the ER (33% vs 0%; P = .04). On the other hand, patients admitted to a non-ICU unit in 2012 (PN-6) were administered more ceftriaxone doses in the ER than in 2011 (70% vs 48%; P = .04). Geriatric patients received more azithromycin in 2012 than 2011 (53% vs 16%; P = .006), but no differences in antibiotic selection were seen in younger patients between the 2 core measures. There were no differences noted in antibiotic selection based on overall unit admission or age. The total number of patients with a final CAP diagnosis at discharge did not differ between the 2 groups when they were grouped by admission unit or age. However, among patients admitted to a non-ICU unit, more had a final CAP diagnosis in addition to another final diagnosis documented at discharge in 2011 (79%), whereas the majority of patients admitted in 2012 had a singular final diagnosis of CAP (87%) (P <.001). No other differences were seen between the 2 groups based on admission unit or age.

Discussion and Conclusion

Our study demonstrates that the change in TJC pneumonia core measure from one that focused on antibiotic timing to antibiotic selection had no impact on overall antibiotic selection in the ER for patients with suspected CAP. The purpose of modifying the pneumonia core measure to PN-6 was to improve antibiotic selection. However we found only a 2% improvement in appropriate antibiotic regimen use, including drug selection and dosing, after the change to PN-6. The pneumonia core measure failure to improve appropriate initial CAP antibiotic use is discouraging but not alarming. It has been shown that initial treatment failure in CAP is correlated with significantly higher mortality, longer hospital stays, and higher hospital costs. ¹⁶ On the other hand, most of the rational for classifying an initial regimen as inappropriate in our study was due to lack of atypical coverage. Recent studies present conflicting conclusions about whether beta-lactam/macrolide or fluoroquinolone therapy results in better outcomes than a beta-lactam alone.^17–20 Several studies suggest that beta-lactam therapy alone, as was common in our study, is at least equivocal to the inclusion of atypical coverage.^18,20 Even among some of the studies that conclude that the beta-lactam/macrolide combination was superior to beta-lactam monotherapy, it is suggested that the benefits may be as much from the immunomodulating effect of the macrolides as from the antimicrobial activity. ¹⁸ There was only 1 death in our study that was not considered to be caused by CAP. Although we did not specifically evaluate clinical outcomes such as mortality, our findings seem to corroborate those of studies such as Postma et al, which indicate that beta-lactam therapy is non-inferior to combination therapy. ²⁰ If the patients who received beta-lactam therapy in our study were to be reclassified as receiving an appropriate initial regimen, the amount of appropriate initial empiric therapy would be significantly higher in both the PN-5 group (52% vs 85%; P <.001) and the PN-6 group (54% vs 92%; P <.001). However, the difference between the 2 groups would remain insignificant (P = .26), thus reiterating the absence of influence of the change in the pneumonia core measure on CAP antibiotic prescribing in the ER.

Ceftriaxone, the most commonly used beta-lactam in our study, is the most prescribed antibiotic for pneumonia in the United States and is administered to 20% of all ER patients. ²¹ Approximately 69% of the patients we evaluated received ceftriaxone in the ER. Concerns about the potential overuse of antibiotics in the ER to maintain compliance with the pneumonia core measures prompted our evaluation of whether the practice of prescribing antibiotics in the ER to patients without an active infection would decline after the antibiotic timing was extended from 6 to 24 hours. The increased time allotment did appear to have a positive impact. The amount of patients inappropriately diagnosed with and treated for CAP decreased from 22% with PN-5 to 14% with PN-6, therefore subsequently reducing the number of patients who would have been inappropriately treated for a nonexistent infection. Of the non-CAP final diagnoses, only 1, chronic obstructive pulmonary disease (COPD), had the potential to warrant antibiotic therapy.

The finding of patients admitted with a diagnosis of CAP whose discharge diagnosis was not CAP is not unique to our institution. Studies have reported an increase in these types of patients since the pneumonia core measure was first introduced.^3,21 The suggestion was that patients were given a premature CAP diagnosis to avoid noncompliance with core measure's antibiotic timing requirement.^3,21 A survey by Nicks et al found that 55% of emergency physicians reported prescribing antibiotics to patients they did not believe had CAP to comply with the core measure. ²¹ Although the decrease we found was not statistically significant, it is a trend in the right direction for antibiotic stewardship. We noted a high rate of antibiotic continuation during hospitalization. Over 10% of patients in each group who were continued on antibiotics during their inpatient stay had a final diagnosis that was not ultimately CAP. The length of therapy in these patients was inappropriate for patients without an infection and was significantly longer with the PN-6 group (4.5 vs 8 days; P = .04). Fluoroquinolones and third-generation cephalosporins, both standard therapies for CAP, promote extended-spectrum beta-lactamase (ESBL)-mediated resistance. ²² Incorrectly classifying patients with respiratory symptoms as having CAP increases the potential for mismanagement and development of pathogens for which there are few effective treatment options.

We had anticipated reviewing many more patients. Initially, 245 patients were identified with an ICD-9 code for CAP within the 2 study periods, but 55% of them (n = 135) met criteria for HCAP. This appears to be consistent with other studies. Micek et al conducted a retrospective study of prescribing for CAP in the ER over an 8-year period and found that almost 50% of patients treated for CAP in the ER had a risk factor for HCAP; they identified only 259 patients with actual CAP. ¹⁶ Although we did not evaluate the HCAP patients' antibiotic regimens, it is concerning that such a high percentage of patients with HCAP were incorrectly classified as CAP and potentially inappropriately managed. It has been shown that patients with HCAP risk factors are significantly less likely to receive appropriate antibiotic treatment compared to those without HCAP risk factors and that suboptimal antibiotic selection in these patients leads to untoward outcomes. ¹⁶

Our study has several limitations. First it was a small, single-center study. Although power was met, a larger, multicenter study would be able to consider differences in prescribing patterns that may vary by locality. Second, we focused on prescribing antibiotics in the ER only. The timing of antibiotic receipt was extended from 6 hours to 24 hours under PN-6. It is possible that the patients whose antibiotic regimens were considered inappropriate due to use of monotherapy where combination therapy was recommended by the guidelines may have received the appropriate additional antimicrobial within that 24-hour window. We focused on ER prescribing as it is the most likely area where initial, empiric antibiotic therapy for a diagnosed infection will be ordered; this is in accordance with the guideline recommendations, which are primarily for empiric treatment. Although advances in antimicrobial testing have been made, rapid and accurate diagnostic methods are not currently available in most institutions. As such, the microbiology etiology of CAP is unknown at time of diagnosis and therefore initial treatment is primarily empiric. ¹⁸ Our study excluded patients discharged from the ER. This was done to coincide with the intent of the core measures. As a result, a full evaluation of antibiotic prescribing patterns during the 2 study periods may not have been captured. Additionally, the number of patients with HCAP risk factors or with non-infectious causes of symptoms inappropriately treated for CAP may have been understated.

TJC removed pneumonia from the required core measures effective January 2015. ⁹ Although the burden of pneumonia makes it an appropriate target for quality and performance measurements, use of these metrics did not appear to have a strong impact on CAP treatment in the ER in our study. The types of antibiotics and regimens used did not differ between the 2 versions of the core measure. Although just fewer than half of regimens complied with guidelines, the majority were at least evidence based. Because the choice of antimicrobial therapy did not improve after the change to a more selection-based core measure, it is uncertain what benefit the PN-6 core measure had on patient outcomes over previous versions. We saw an improvement in appropriate CAP diagnoses in ER after the change to PN-6, but any benefit derived may have been related more to undoing the harm caused by the early pneumonia core measure, which led to increased incorrect CAP diagnoses in the ER, than to any true improvement on CAP recognition and management. It is notable that regulatory bodies and payers implement programs to improve care and patient outcomes. However, it is important that the extent of unintended consequences be considered and measured against the potential benefit of any implemented standard. The removal of CAP as a core measure seems to be appropriate at this time. If TJC should decide to include pneumonia as a core measure in the future, the chosen measurement should be well determined to improve quality in CAP diagnosis and management across the continuum of care in the hospital, without putting most of the burden within a single domain.