Treatment of bacterial skin infections in ED observation units: factors influencing prescribing practice

a b s t r a c t

Objective: The Infectious Disease Society of America publishes Evidence-based guidelines for the treatment of skin and soft tissue infections. How closely physicians follow these guidelines is unknown, particularly in the emergency department observation unit (EDOU) where increasing numbers of patients are treatment for these infections. Our objectives were to describe (1) the antibiotic treatment patterns EDOU patients, (2) physicians’ adherence to the IDSA guidelines, and (3) factors that influence physician’s prescribing practices.

Methods: This prospective cohort enrolled adult patients discharged from an EDOU at an academic medical center after treatment for a skin or soft tissue infection. Information was collected from chart review and patient inter- view pertaining to the patient’s sociodemographic characteristics, presenting illness, and antibiotic Treatment regimens. Treatment regimens were compared with national guidelines.

Results: The study included 193 patients of which only 43% were treated according to IDSA guidelines, 42% were overtreated, and 15% were undertreated. Women were more likely to be undertreated (relative risk, 1.58; 95% confidence interval, 1.21-2.06), whereas patients 50 years and older were at risk for overtreatment (relative risk, 1.44; 95% confidence interval, 1.03-2.02). Women also received shorter courses of antibiotic therapy with an average of 9.6 days of treatment compared with 10.6 days for men.

Conclusions: Physician Antibiotic prescribing practices demonstrated poor adherence to IDSA guidelines and were influenced by the patient’s age and sex. Standardized antibiotic protocols for treatment of skin and soft tissue in- fections to IDSA guidelines in the EDOU would minimize physician bias.

Introduction

Failure of oral antibiotic treatment for Skin and soft tissue infections has led to the need for intravenous therapy that is often deliv- ered in emergency department observation units (EDOUs) [1-3]. As

? Meetings: None.

?? Grant: This study was designed and carried out at the University of Massachusetts

Medical School and was supported by an intradepartmental grant through the Department of Emergency Medicine. VB is supported by an National Institutes of Health award 1R15AI112985-01A1 and PLH by 5K24AT003683-09.

? Conflicts of interest: None of the authors listed have any conflict of interest.

?? Author contributions: JPH, EWB, and PLH conceived the study, designed the trial, and

obtained research funding. JPH, GW, and AF supervised the conduct of the trial and data collection. JPH, GW, and AF recruited participating centers and patients and managed the data, including collection and quality control. JPH and VB provided statistical advice on study design and analyzed the data; JPH drafted the manuscript, and all authors contribut- ed substantially to its revision. JPH takes responsibility for the paper as a whole.

* Corresponding author at: 55 Lake Avenue North, Worcester, MA 01655. Tel.: +1 508 450 8688; fax: +1508 421 1490.

E-mail address: [email protected] (J.P. Haran).

treatment locations that focus on the brief treatment of medical prob- lems of limited complexity, EDOUs are increasingly the location in which Intravenous antibiotic therapy for resistant SSTI is administered [4]. Guidelines from the Infectious Disease Society of America (IDSA) recommend antibiotic coverage for abscesses and intravenous antibi- otics for cellulitis, only in the presence of a systemic inflammatory re- sponse, if the patient is severely immunocompromised, or failing outpatient treatment [5]. Adherence to IDSA guidelines is important given the increasing resistance to many commonly used antimicrobial agents caused by widespread antibiotic use. Unfortunately, physician discretion in selecting antibiotic regimens leads to a variability in emer- gency department (ED) treatment approaches to common bacterial in- fections [6]. Given the risks of bacterial resistance due to inappropriate antibiotic use, physician practices in prescribing antibiotics in the EDOU are unknown and deserve attention [4,7]. The objectives of this cohort study, therefore, were to (1) describe the prevalence of various antibiotic prescribing practices for patients with SSTIs, (2) compare these management practices with national guidelines, and (3) identify factors that might influence physician’s prescribing practices.

http://dx.doi.org/10.1016/j.ajem.2015.08.035

Methods

Study design

We conducted a single-center, prospective cohort study. The hospital’s institutional review board approved the study (IRB docket no. H00001871).

Study setting and population

We identified adult ED patients retrospectively and prospectively enrolled and confirmed patient data into this cohort study after dis- charge from an EDOU at a large Urban academic ED between January and December 2013. Patients were eligible for participation if they were discharged home after an EDOU stay and either received antibi- otics for an SSTI in the EDOU or upon discharge. Skin and soft tissue infec- tion was defined as clinician diagnosis of a bacterially caused abscess or cellulitis upon discharge or a diagnosis of abscess or cellulitis that was treated with antibiotics. We identified potential subjects using EDOU census logs at the end of each month whom we contacted by telephone 4 weeks after the discharge date. Eligible patients were English speak- ing, 18 years of age or older, treated for an SSTI, and had a working cell or landline telephone. We analyzed separately those patients who were admitted directly to the hospital after failing EDOU treatment. We excluded patients if they were unable to cooperate with the ques- tionnaire or recall events surrounding their care. To minimize recall bias, we also excluded patients whom we were unable to contact within 8 weeks of discharge from EDOU.

Data collection

Research assistants trained in chart review and patient data extrac- tion by the authors administered a standardized survey to consenting and enrolled patients with an SSTI over the telephone. We obtained data related to the patient’s antibiotic compliance, health care visits/ hospitalizations, and any other complications since initial EDOU dis- charge. We obtained medical histories, allergies, and antibiotic treat- ment regimens that we confirmed through chart review by reviewers blinded to the Study objectives. We calculated the Charlson Comorbidity Index (CCI) to characterize patient’s medical comorbidities [8,9]. In ad- dition, we obtained information pertaining to initial ED presentation, EDOU hospital course, and antibiotic treatments used from the medical

records in a blinded manner. We reviewed the medical records of pa- tients admitted to the hospital after failing EDOU therapy to obtain medical histories, EDOU visit details, and information about their subse- quent hospital stay. We captured and extracted study data using RED- Cap electronic data capture tools [10].

Infectious Disease Society of America classification

We determined 2 treatment categories of nonpurulent and purulent SSTIs and then classified the disease as mild, moderate, or severe using the IDSA guidelines comparing observed antibiotic treatment regimens with recommended national guidelines (Table 1) [5]. Each patient re- ceived a score of 1 for mild, 2 for moderate, and 3 for severe. Indepen- dently, we ranked the observed antibiotic treatment regimen also as mild, moderate, or severe based on IDSA recommendations. We com- pared the 2 scores with each other to determine if the observed antibi- otic regimen matched the anticipated treatment class. We performed this in a blinder manner by 2 independent ED clinicians. Where there was disagreement, a third adjudicator served as a tie breaker in the final determination. We categorized patients into expected treatment when the scores equaled each other, undertreated if the observed score was lower than the anticipated score, and overtreated when the observed score was higher than the anticipated score.

Outcomes

The primary study outcome was both undertreatment and over- treatment group classification after EDOU therapy for an SSTI. Second- ary outcomes were failure of EDOU treatment defined as the patient being admitted to an inpatient ward from the EDOU, hospitalization up to 1 week after EDOU discharge for the same infection, completion of antibiotics as planned, and development of antibiotic-associated diar- rhea defined as 3 or more loose stools per day for 2 or more consecutive days up to 30 days after EDOU discharge [11,12]. Patients who were ad- mitted to the hospital from the EDOU after failing therapy were ana- lyzed for sociodemographic characteristics, presenting illness, and antibiotic treatment regimens. We defined failure of EDOU therapy as a continuation of SSTI symptoms in which additional antibiotic treat- ment was necessary to resolve the infection within 2 weeks of initial visit. This was identified as repeat emergency department visits, prima- ry care or specialist visits, or hospitalizations. We counted only visits that resulted in a change in antibiotic therapy. For example, well visits

Table 1

Practice guidelines for SSTIs adapted from the 2014 IDSA update

Type	Nonpurulent			Purulent
Class	Severe	Moderate	Mild	Severe	Moderate	Mild
Signs and symptoms	Malignancy on chemotherapy Neutropenia Severe cell-mediated immunodeficiency Immersion injuries	Systemic signs of infection	No SIRS No AMS Hemodynamically stable	Failed initial antibiotic treatment Impaired host defenses SIRS with hypotension	SIRS	No SIRS

- - Animal bites
  - Penetrating trauma
  - Evidence of MRSA infection/ colonization elsewhere
  - Injection drug use
  - SIRS

Treatments Intravenous Rx

- - Combination Rx^a

Intravenous Rx

Single class^a

Oral Rx

- Penicillin VK or
- Cephalosporin or
- Dicloxacillin or
- Clindamycin

I&D Empiric Rx

Vancomycin
Clindamycin
Other^a

I&D Empiric Rx

- TMP/SMX or
- Doxycycline or
- Dicloxacillin or
- Cephalexin

I&D alone

Abbreviations: SIRS, systemic inflammatory response syndrome; Rx, antibiotic; AMS, altered mental status; I&D, incision and drainage; MRSA, methicillin-resistant Staphylococcus aureus; TMP/SMX, trimethoprim/sulfamethoxazole; VK, V potassium.

^a Please refer to Stevens et al [5] for complete list of antibiotic choices.

or visits for rechecking abscess wounds were not considered as a treat- ment failure. We also asked patients if they completed their antibiotic prescription, were hospitalized for the same infection, and if they had any diarrheal symptoms after discharge from the EDOU.

Data analysis

We used ?² tests to compare study patient characteristics among the 3 classification groups in categories of single variables and analysis of variance for continuous variables. To confirm these results, we per- formed proportional logistic regression with ordered factor response as response variable [13]. We considered as our ordered response vari- able the study outcome (undertreated N expected N overtreated), such that, depending on how we ordered the variables, this approach allows determining the risk of being undertreated with respect to being ex- pected or overtreated (when combined) per unitary change in each co- variate when all the others are kept constant. To do this, we used the polr command from the R package [14]. As logistic regression with polr returns odd ratios, we used the function orsk to convert odd ratios into Relative risk estimates [15].

Based on our previous work [6] where we observed a 10% sex differ- ence in the undertreated group and using a desired power of 0.8 with an ? of .05, we arrived at a sample size of 188 patients. Given the EDOU vol- ume from the preceding year, we estimated that a 12-month enroll- ment period would be sufficient to obtain this sample size.

Results

Characteristics of study subjects

During the 1-year study period, 284 patients treated in the EDOU for an SSTI were discharged to their homes; 58 patients (17.0%) were ad- mitted to the hospital after failing EDOU therapy. Because of a lack of contact information, 53 patients were ineligible for enrollment. Among the remaining 231 patients, 24 (10.4%) declined to participate in the study, 7 (3.0%) could not communicate in English, 6 (2.6%) fell out of the 4-week window for contact, and 1 (0.4%) was unable to recall events surrounding their EDOU stay. The final study consisted of 193 patients.

Patients directly admitted to the hospital from the EDOU after failing antibiotic therapy had a mean age of 49.0 +- 17.8 years old, and 57% were women. From this cohort, 29.6% of patients were treated in com- pliance with IDSA guidelines, whereas 66.7% were overtreated and 3.7% were undertreated. This admitted cohort did not differ from the pa- tients discharged home by either age, sex, ethnicity, CCI score, or infec- tion type or location. The prevalence of intravenous drug abuse was greater in patients admitted to the hospital (17.2%) than those discharged home (3.1%). This resulted in patients with an intravenous drug abuse history being 3 times more likely to be admitted to the hospital from the EDOU than discharged home (RR, 3.07; 95% confi- dence interval [CI], 1.95-4.85). In addition, there were a greater percent- age of abscesses in the Admitted group than in the group discharged home (22.4% vs 15.0%, respectively). Of the patients admitted, 33.3% underwent surgical drainage of infectious material after EDOU antibiot- ic treatment failure.

The final study population was primarily white non-Hispanic with a mean age of 47 +- 16.7 years and a similar distribution of men and women. Most patients had no medical comorbidities (CCI of 0) with 21.8% of patients taking antibiotics before the EDOU admission (ie, fail- ing Outpatient therapy). Most patients were treated for cellulitis (85.0%). The most common EDOU antibiotics prescribed were vancomy- cin followed by first-generation cephalosporins, clindamycin, and peni- cillin inhibitor combination antibiotics. Approximately two-thirds of patients were treated with a single antibiotic class within the EDOU.

Comparison to national guidelines

To investigate how closely the observed EDOU patient treatment regimens correlated with published national guidelines, we compared their observED treatments with the IDSA 2014 published guideline clas- sifications. Accordingly, 43.3% of patients were treated in compliance with IDSA guidelines, 41.8% were overtreated (ie, given antibiotic treat- ments consistent with a higher IDSA class than their own observed treatment), and 15.0% were undertreated.

Factors influencing prescribing

We observed differences in various patient characteristics among the overtreated, undertreated, or expected groups (Table 2). Patients in the overtreated group were older with most of them being men. Older patients (>= 50 years old) were 48% more likely to be overtreated than their younger counterparts (RR, 1.44; 95% CI, 1.03-2.02; RR regres- sion, 1.39; 95% CI, 1.00-1.90; Figure, panel A). In addition, patients with abscesses were more likely to be overtreated (RR regression, 3.33; 95% CI, 2.20-5.96; Figure, panel B). Conversely, patients in the undertreated group were younger with more than 2 of 3 patients being female. Thus, women were 58% more likely to be undertreated in comparison to men (RR, 1.58; 95% CI, 1.21-2.06; RR regression, 1.47; 95% CI, 1.07-2.05;

Figure, panel C). Moreover, women comprised 80% of patients in the group that was undertreated by 2 class categories (ie, they should have been treated in the severe category and instead received antibiotic treatment consistent with the mild category). Women also received a lower number of antibiotic classes with 1.6 classes prescribed (95% CI, 1.5-1.7), as compared to 1.9 classes prescribed (95% CI, 1.7-2.1) for men. In addition, women received shorter courses of antibiotic therapy with an average of 9.6 days of treatment (95% CI, 9.0-10.3) compared to

10.6 days of treatment for men (95% CI, 9.7-11.4). In addition, patients with previous history of antibiotics are more likely to be undertreated (RR regression, 3.64; 95% CI, 2.36-5.66; Figure, panel D). Among the in- fection types, abscesses were overtreated most of the time (Table 3). Table 4 lists the antibiotics used in each group. Vancomycin followed by clindamycin and first-generation cephalosporins were the classes of antibiotics used the most with most patients receiving more than 1 class of antibiotic during the course of their treatment.

Patient outcomes

Upon follow-up, most patients (87.1%) completed the antibiotics prescribed from the EDOU. Major reasons for stopping the home antibi- otic prescription were gastrointestinal upset followed by resolving in- fection and stoppage based on another clinician’s advice. Diarrhea was reported in 26.4% of all patients. Patients reported a failure of antibiotic therapy after EDOU discharge only 6.7% of the time with 38.5% of these

Table 2

Characteristics of study patients

Demographics Undertreated Matched Overtreated

	n= 41		n= 72		n= 80
Age (SD)	44.7 (18.1)	45.1 (14.2)		50.1 (17.8)
Female	29 (70.7)	36 (50.0)		32 (40.0)
White	34 (82.9)	59 (81.9)		62 (77.5)
Hispanic	5 (12.2)	6 (8.3)		11 (13.8)
African American	0 (0.0)	3 (4.2)		3 (3.8)
Asian	0 (0.0)	2 (2.8)		1 (1.3)
Medical history CCI 0 (SD)	70.7 (0.46)	61.1 (0.49)		58.8 (0.50)
CCI 1 (SD)	12.2 (0.33)	19.4 (0.40)		23.8 (0.43)
CCI >=2 (SD)	17.1 (0.38)	19.4 (0.40)		17.5 (0.38)
Current Abx	18 (43.9)	23 (31.9)		1 (1.3)

Data are presented as percentage unless otherwise indicated. Abbreviation: Abx, antibiotics.

20 20

15 15

Gender

Counts

10 F 10

5 5

Abscess No Yes

0 0

Age

Counts

<50 10

>=50

Antibotic No Yes

0 0

Under-treated

Matched

Over-treated

Under-treated

Matched

Over-treated

Figure. Proportional Logistic Regression Model Results Broken by Treatment Class.

patients subsequently requiring hospital admission to the cure the in- fection. We observed no differences in these secondary outcomes among patients in the overtreated, undertreated, or expected groups.

Study strengths and limitations

This study is the first, to our knowledge, to combine SSTI Treatment outcomes reported from EDOU antibiotic therapy while assessing phy- sician compliance with IDSA guidelines. It is, however, limited by reporting outcomes from a single site. We did not investigate other fac- tors that may influence physician prescribing practice, such as proximal streaking with cellulitis; however, these factors are not included in the guideline Decision-making algorithms. We did, however, have com- pleteness of data concerning systemic symptoms such as fever and the other factors necessary to make a decision using the guidelines obtained from both the patient and through the medical record. Finally, the

failure rate of EDOU treatment was lower than expected and did not permit analyzing associations of Failure rates to treatment categories. There is a lack in patient-centered outcomes of treatment failures be- tween IDSA-compliant and noncompliant treatment groups, which needs to be better explored in future studies to link IDSA guideline com- pliance to outcomes.

Discussion

Emergency department observation unit treatment of bacterial in- fections poorly follow IDSA guidelines and conformed with these

Table 4

Antibiotics

Table 3

Infection characteristics

Type Undertreated Matched Overtreated n = 41 n = 72 n = 80

Cellulitis 37 (90.2) 68 (94.4) 59 (73.8)

Abscess 4 (9.8) 4 (5.6) 21 (26.3)

				Quinolone	1 (2.4)	0 (0.0)	2 (2.5)
Location				Sulfonamide	0 (0.0)	6 (8.3)	3 (3.8)
Face	11 (26.8)	8 (11.1)	8 (10.0)	Doxycycline	1 (2.4)	1 (1.4)	1 (1.3)
Hand	11 (26.8)	7 (9.7)	14 (17.5)	Nitrofuantoin	0 (0.0)	0 (0.0)	0 (0.0)
Foot	2 (4.9)	9 (12.5)	11 (13.8)	Metronidazole	0 (0.0)	1 (1.4)	0 (0.0)
Trunk	4 (9.8)	7 (9.7)	6 (7.5)	No. of classes (SD)	1.4 (0.59)	2.1 (1.1)	2.1 (0.90)

Undertreated Matched Overtreated

	n= 41		n= 72		n= 80
First-generation cephalosporin	4 (9.8)		16 (22.2)		17 (21.3)
Third-generation cephalosporin	5 (12.2)		4 (5.6)		5 (6.3)
Vancomycin	0 (0.0)		30 (41.7)		46 (57.5)
Clindamycin	9 (22.0)		9 (12.5)		18 (22.5)
Macrolide	0 (0.0)		1 (1.4)		0 (0.0)
Penicillin	0 (0.0)	1 (1.4)		0 (0.0)
Penicillin/I	11 (26.8)	3 (4.2)		4 (5.0)

Data are presented as percentage unless otherwise indicated.

Penicillin/I indicates penicillin inhibitor combination; the total for frequency prescribed columns is greater than 100% as patients received multiple antibiotics.

Extremity	11 (26.8)	34 (47.2)	33 (41.3)
Buttocks	2 (4.9)	6 (8.3)	7 (8.8)

guidelines less than half the time. Furthermore, we observed striking disparities in treatment regimen based on sex: women were 58% more likely to be undertreated, whereas adults older than 50 years were 48% more likely to be overtreated. In addition, women received antibiotic regimens that used a lower total number of antibiotic classes used and they were treated for a shorter period despite having the same Infection types and severity as men.

The misuse and overuse of antibiotics continue to be a widespread public health and quality concern in the United States. Bacterial skin in- fections are 1 infection type with clear treatment guidelines [5] that has risen to become one of the most common reasons for seeking Emergent care with a tripling of ED visits between 1993 and 2005 [2]. Antibiotic treatment of community-associated methicillin-resistant Staphylococ- cus aureus continues to rise despite national guidelines advising against their use [16]. Given the increasing rates of antibiotic resistance in com- mon bacterial Community-acquired infections [17], the overuse of anti- biotics and the increasing incidence and severity of Clostridium Difficile infections (CDIs) [18] are important public health issues. We found that Intravenous use of vancomycin and clindamycin in the EDOU were the most common reasons for patients being overtreated when an oral or narrow spectrum antibiotic would have been a more appro- priate choice. The need not only for standardized evidence-based ap- propriate antibiotic Treatment protocols but also for evaluation of how well physicians adhere to these protocols is critical.

There is growing recognition of the public health importance of Sex disparities in the treatment of common medical conditions including acute myocardial infarction [19], stroke [20], and sepsis [21,22]. These disparities not only exist for women but also for minorities and the el- derly [23,24]. Differences between men and women in terms of antibi- otic prescribing practices in patients with SSTIs has not been previously explored. Despite the fact that SSTIs are common in the el- derly, the medically accepted treatment regimens for this group of pa- tients should not differ from their younger counterparts [25,26]. Per IDSA guidelines, these treatments should also remain consistent across age and sex. Older age is not a risk factor for EDOU treatment failure for SSTI [27]; however, advanced age and antibiotic exposure are the strongest known risk factors for CDI [28,29]. Therefore, more aggressive antibiotic treatment of older patients may impart more harm without providing any additional benefit. The older adults in our study received more aggressive antibiotic therapy with Intravenous antibiotics where an oral antibiotic was indicated. Intravenous antibiotic use is associated with higher rates of Antibiotic-associated diarrhea [6]. We identified 1 case of CDI after EDOU treatment in a 54-year-old woman that was, ac- cording to national guidelines, overtreated for leg cellulitis.

This study contributes to the growing body of literature that demon- strates sex disparities in medicine [22,30]. Our study showed that women were less aggressively treated for an SSTI than men. Women were undertreated not only by the IDSA guideline antibiotic choice but also in the duration of therapy. This poor adherence to SSTI guide- lines with physician prescribing patterns being influenced by factors not tied to clinical outcomes speaks to the need for strict protocol- driven ED and EDOU treatment algorithms. This study was not powered to detect differences among the groups with respect to treatment failure after EDOU discharge; however, it is not surprising that patient who did fail were older adults with higher medical comorbidity scores. Addition research will further define how undertreating patients contributes to worse outcomes.

We also identified patient characteristics that herald the likely fail- ure EDOU therapy. For example, intravenous drug abusers were 3 times more likely to fail EDOU therapy and require subsequent hospital admission for further management. The other major factor associated with treatment failure was infection type; 1 in 3 abscesses managed in the EDOU required subsequent admission and surgical intervention. We, therefore, recommend that hospital admission be considered in any patient with a history of intravenous drug abuse; furthermore, abscesses may deserve incision and drainage in conjunction with

antibiotic therapy. Importantly, we observed that older adults faired just as well as younger during the course of therapy within the EDOU making this an important alternative treatment location compared to hospital admission.

Conclusions

Based on our findings, we recommend that EDOUs and EDs stan- dardize antibiotic therapy to better align with national guidelines. Physician-driven Treatment decisions led to undertreatment of women and overtreatment of older adults. The EDOU is a good alternative to hospitalization for the treatment of SSTIs, especially among older adults; however, antibiotic treatments should follow IDSA guidelines, which would hopefully lead to a reduction in bacterial resistance.

References

AJEM