Abstract
Keywords
Introduction
Obesity is now recognized by the World Health Organization as a global pandemic spanning all socioeconomic classes with a projected 700 million affected by 2015. An alarming one-third of Americans are obese, with 1 out of 20 meeting criteria for morbid obesity (body mass index [BMI] ≥40 kg/m2). 1 From a public health standpoint, this raises serious concerns about associated morbidity and mortality, including increased risk of infection in this population. Not only has obesity been shown to be a predisposing factor in acquiring infections, but it also results in worse clinical outcomes than in patients with normal BMI. 2 Despite the increasing prevalence of obesity, there are few guidelines for antimicrobial dosing in obese patients and poor adherence to those already in place. In a recent retrospective study in the emergency department setting, emergency physicians adhered to their institution’s weight-based antibiotic guidelines for cefepime, cefazolin, and ciprofloxacin in 8.0%, 3.0%, and 1.2%, respectively. 3
Several studies have addressed antimicrobial dosing considerations in obese patients. Although the absorption of drugs does not seem to be significantly modified in obesity, the processes of volume of distribution (Vd), protein binding, metabolism, and antibacterial clearance are significantly altered. 4 According to Food and Drug Administration regulations, pharmaceutical companies are required to demonstrate average population effectiveness for new antimicrobial agents including children, the elderly, and those with renal and hepatic impairment. However, effectiveness data for those of higher than normal BMI are not required leading to underrepresentation of the obese population. 5 For primary care providers, this is of particular concern, as they serve on the frontline in administering antibiotics for most outpatient infections. Without appropriate data and guidelines to direct antibiotic treatment in the growing obese population, inadequate dosing is likely contributing to treatment failure, unnecessary escalation to broader spectrum antibiotics, and selection of resistant pathogens in obese patients.
Cellulitis is one of the most common bacterial infections seen in the primary care setting, with an incidence that has increased from 8.6 million to 14.2 million from 1997 to 2005. 6 From a recent retrospective analysis, both morbid obesity and inadequate empiric antibiotic therapy were independent risk factors for treatment failure in hospitalized patients with cellulitis (odds ratio 4.10, P = .02; and odds ratio 9.25, P < .01, respectively). 7 However, little is known about the relationship between specific BMI categories and weight with treatment failure in patients with skin and soft tissue infections. There is a need for more BMI-specific recommendations regarding antimicrobial dosing for patients >120 kg and normal organ function; such recommendations might include modified dosing for beta-lactams, trimethoprim-sulfamethoxazole, and many intravenous antibiotics. Many recommendations may be based on the available pharmacokinetic properties of the drug and the frequency and severity of known toxicities rather than published research studies.
For adult patients diagnosed with cellulitis in the primary care setting, we hypothesized that an increased BMI would correlate with adverse outcomes (specifically, prolonged antibiotic course, escalation of antibiotic choice, or hospitalization for intravenous administration of antibiotics). Since other factors such as the diagnosis of diabetes and tobacco use may be confounding issues, these conditions will specifically be controlled for in this study. With a recent (2014) institutional guideline suggesting increased doses for antibiotics in patients weighing more than 120 kg, we used weight in a secondary analysis.
Methods
This was a retrospective study involving primary care patients within a Mayo Clinic setting located in the Midwest. All patients who had a primary diagnosis of cellulitis from June 2008 through June 2013 were identified using ICD-9 (International Classification of Diseases, ninth revision) codes. After approval by the organization’s institutional review board, adult patients were screened and only those having given prior research authorization were included. Data were collected using both automated and manual processes. Demographic data recorded included age, gender, and race. Clinical data abstracted included height, weight, BMI, recent tobacco use, comorbid diagnosis of diabetes, antibiotic prescribed/ treatment methods used, and dosing. Any cellulitis case in which the patient was directly admitted to the hospital for treatment after being initially diagnosed was excluded from the analysis, as the focus of the study was outpatient management of cellulitis. Medical records with an obvious error in documentation (usually BMI) were also screened and removed.
Much of the data was retrieved using automated queries. However, because of limitations of the program, the data regarding each provider’s method or choice of treatment, including medication type, dose, and duration were manually abstracted by a family medicine resident and clinical research coordinator. The medical charts were also manually reviewed to determine smoking status within the previous 12 months.
We compared a cohort of patients with BMIs at 40 kg/m2 and higher with a cohort of age-and gender-matched controls having BMIs in the range of 18.5 to 25 kg/m2 at the time of their cellulitis diagnosis. The dependent variable was treatment failure as defined by (a) hospital admission for intravenous administration of antibiotics and (b) prolonged therapy (defined as an extended course of the initial antibiotic or a different antibiotic after initial course was completed.
Student’s t test was used for statistical analysis of the continuous variable. Categorical data were analyzed with chi-square testing. Multiple logistic regression modeling for the clinical outcome of an adverse outcome of either hospitalization or prolonged treatment within 30 days of the diagnosis of cellulitis as an outpatient was performed while retaining all independent variables studied. Statistical significance was set at P < .05. Calculations were performed on MedCalc software (www.medcalc.org, version 14.10.2).
Results
Of the 637 patients in the study cohort, 393 (61.7%) were of a BMI ≥40.0 kg/m2 and 244 (38.3%) were of normal BMI. Between the 2 groups of patients, those with a normal BMI were younger, more likely white women without diabetes and weigh less than 120 kg (Table 1). There was no difference between the 2 groups of patients with regard to tobacco use, which ranged from 16.5% to 19.1%. Overall, adverse outcomes were not statistically significant between the groups, with almost 1 in 5 patients experiencing either a prolonged therapy or hospitalization for their cellulitis. While hospitalizations were approximately 1 in 20 from each group; approximately 20% (range 17.9% to 19.6%) required prolonged or altered therapy.
Demographics and Clinical Factors by Body Mass Index (BMI; Normal vs ≥40 kg/m2) of Primary Care Patients Diagnosed With Cellulitis as an Outpatient, by Variable.
With subgroup analysis of the BMI ≥40 kg/m2 group, the patients with a BMI of 40 to 50 kg/m2 had similar adverse events compared with patients with a normal BMI; while the patients with a BMI ≥50 kg/m2 were noted to have almost twice the rate of adverse outcomes as the normal BMI group and the subgroup of patients with a BMI of ≥40 to <50 kg/m2 (Figure 1). These 71 patients with “super obesity” were almost all more than a weight of 120 kg (98.6%) and had a much higher incidence of diabetes when compared with the patients with normal BMI (40.8% vs 4.9%, P < .001; Table 2). Adverse outcomes were noted in 33.8% of these patients compared with 17.9% of the normal BMI patients. There were no demographic statistically significant differences noted between these 2 groups of patients for age, gender, race, or tobacco use.
Percentage of adverse events (either hospitalization or prolonged therapy) for outpatient patients diagnosed with cellulitis, by body mass index (BMI) category.
Demographics and Clinical Factors by Body Mass Index (BMI; Normal vs ≥50 kg/m2) of Primary Care Patients Diagnosed With Cellulitis as an Outpatient, by Variable.
Using a logistic regression analysis for presence of an adverse outcome within 30 days of the diagnosis of outpatient cellulitis in the cohorts of the 71 patients with BMI ≥50 kg/m2 and 244 patients with a normal BMI (N = 315) while retaining all the demographic variables, a BMI of ≥50 kg/m2 was associated with an odds ratio of 2.440 (95% CI, 1.260-4.724; P = .008) and weight >120 kg was associated with an odds ratio of 2.246 (95% CI, 1.154-4.369; P = .017; Table 3). Age, gender, race, the diagnosis of diabetes mellitus, or the use of tobacco were not associated with a worsening outcome of cellulitis, when controlling for all other variables for either model.
Odds Ratios for Adverse (Either Prolonged Treatment or Hospitalization) Event Within 30 Days of Diagnosis of Cellulitis (for Normal Body Mass Index [BMI] Patients vs BMI ≥50 kg/m2) by BMI or Weight, by Variable (N = 315).
Discussion
In this study, we determined rates of treatment failure in patients initially diagnosed with cellulitis in the outpatient setting with a particular emphasis on specific ranges of obesity, including morbid obesity (BMI 40-50 kg/m2) and super obesity (BMI ≥50 kg/m2). We were surprised to see minimal differences in the rates of prolonged antibiotic therapy and hospitalizations between normal BMI and the BMI 40 to 50 kg/m2 cohort; however, the rates of adverse outcomes in the BMI ≥50 kg/m2 cohort was quite marked and independent from other studied variables including age, gender, comorbid diabetes, and tobacco use. It was observed that most patients within this latter group had a weight of >120 kg, suggesting a weight at which concerns for inadequate antibiotic therapy should be triggered. This study adds to the knowledge base of outpatient treatment for cellulitis in patients with obesity, by controlling for the potential confounding factors of diabetes mellitus and tobacco use. Also, the significant change in adverse events at BMIs ≥50 kg/m2 vs BMI of 40 to 50 kg/m2 was intriguing, as prior studies have not subcategorized BMIs >40 kg/m2.
With these findings in mind, current prevalence data have shown that the highest BMI groups are increasing at the fastest rates. From a period of 2000 to 2005, self-reported BMI higher than 40 kg/m2 increased by 50% with BMI >50 kg/m2 increasing by 75%, approximately 3 times the rate of moderate obesity. 8 Based on obesity trends in Pennsylvania school children from 2007 to 2011, Lohrmann et al 9 project the prevalence of overweight, obesity, and extreme high obesity of Pennsylvania school children to be 16%, 6.6%, and 23% by 2031, lending additional evidence that morbid and severe obesity are surpassing general trends in moderate obesity. As clinically severe obesity continues to become a significant portion of the nation’s weight distribution, we will undoubtedly see more clinically relevant ramifications especially in terms of drug dosing.
From a pharmacologic standpoint, differences in drug metabolism among obese individuals have drawn significant attention over the past decade and are well described in the literature, although clinical evidence has lagged behind. In general, oral absorption may play a minor role in decreased drug concentrations due to delayed gastric emptying commonly seen in obesity. Volume of distribution (Vd), however, is significantly altered in obesity compared with normal-weight individuals. In terms of lipophilic antibiotics (eg, fluoroquinolones and macrolides), Vd is typically increased in obesity due to increases in adipose tissue. Additionally, Vd in hydrophilic antibiotics (eg, beta-lactams and aminoglycosides) is also increased with water comprising 30% of adipose tissue thus increasing lean body weight. Less is known about hepatic metabolism, although fatty infiltration of the liver and increased activity of the cytochrome P450 pathway has been implicated in altered metabolism. Glomerular filtration rate has been shown to be increased in obese individuals if otherwise healthy. Those with associated comorbidities such as hypertension and diabetes can have significantly impaired renal function posing risk for toxicity with increased antibiotic dosing.4,10,11
Reaching target antimicrobial concentrations in obese individuals has proven to be complex and certainly warrants additional clinical investigation. However, a few general considerations have been recognized which are reflected in our institutional guidelines for antimicrobial dosing in obesity. In general, penicillins and cephalosporins are suggested to be used at the higher end of suggested dosage range because of the relatively low rate of toxicity and adverse effects. 11 In a recent study looking at cefazolin concentrations in adipose tissue at surgical sites, cefazolin concentrations were shown to be inversely proportional to BMI (r = −0.67, P < .001) with a significant proportion of obese and extremely obese not achieving minimal inhibitory concentrations for Gram-negative rod coverage in adipose samples at 20% and 33.3%, respectively. 12 In a commonly cited study, Forse et al 13 showed that increasing cefazolin from 1 to 2 g resulted in 75% to 100% increase in tissue concentrations and resulted in a significant reduction in surgical site infections from 16.5% to 5.6%. Rather than using the higher end of dosage range for cephalosporins, our institutional guidelines as of July 2014, simply recommend doubling the dose in patients weighing greater than 120 kg with no known organ dysfunction.
Our findings support the recent update in our institutional guidelines directing providers to increase specific antibiotics such as penicillins and cephalosporins in patients weighing greater than 120 kg. Our data were collected from clinical episodes spanning from 2008 to 2012, predating these guidelines, with the vast majority of the patients included receiving cephalexin at standard dosing of 500 mg orally 3 to 4 times per day. It would be interesting to compare future rates of treatment failure with the new guidelines in place as well as rates of adherence among providers. Our data assessed risk of treatment failure in the outpatient setting and cannot be generalized to the inpatient setting. Another study showed weights of just 100 kg and BMI of >40 kg/m2 significantly associated with clinical failure in hospitalized patients with cellulitis. 7 We too may have had similar findings if we assessed clinical failure once patients reached the inpatient setting. These findings are also applicable to the dosing of antibiotics for prophylaxis of recurrent cellulitis. The PATCH I trial found that BMI >33 kg/m2, lymphedema, and multiple previous episodes were independent risk factors for failure of penicillin prophylaxis and that the increased prophylaxis failure rate in obese patients may be linked to underdosing of penicillin for BMI. 14
There were several limitations to this study with the most obvious being our reliance on the clinical diagnosis of cellulitis. Misdiagnosis of noninfectious conditions like stasis dermatitis as cellulitis is a common error, occurring as often as 28% in a recent study performed at University of California Los Angeles Medical Center. 15 This was minimized by excluding cases where there was a high degree of diagnostic uncertainty, but again we relied on accurate and detailed charting to do this. In an attempt to keep our study generalizable, we did not account for comorbid chronic medical conditions (chronic obstructive pulmonary disease, chronic kidney disease, cancer, etc),preexisting stasis dermatitis, chronic lymphedema, or immunocompromised status which may have an impact on cellulitis outcomes and could be considered for a future analysis. We also may have gleaned useful information from recording the mechanism of infection, as bites and trauma can predispose to infection of deeper structures that are more difficult to treat. There were also challenges in determining what defined prolonged treatment. If patients were found to have an allergic reaction to the medication initially prescribed, another medication was often prescribed in its place, thus extending the duration by several doses and days. This example was not coded as a prolonged therapeutic treatment, however, not all medication changes were as well documented leading to possible misinterpretation.
In addition, not all BMI categories were evaluated in this study. It was demonstrated in our study that there was no difference in the adverse outcomes noted between patients with a normal BMI and those of a BMI of 40 to 50 kg/m2 and that the difference in outcomes was only associated with those of a BMI >50 kg/m2 compared with a normal BMI. Our study was designed to compare those that would be eligible to require antibiotic dosing changes (BMI >40 kg/m2) and those who did not (BMI 18.5-25.0 kg/m2). Also, we did not determine the retrospective control of the patient’s diabetes at the index by determining the most recent hemoglobin A1c. Diabetes control would be best determined by an A1c at index date or within two months prior and we felt this would be best determined by a potential prospective analysis of cellulitis in diabetic patients. Retrospective data may or may not have been available, and would have diminished the potential number of diabetic patients.
Conclusions
Patients with BMI >50 kg/m2 and weight >120 kg have increased rates of antibiotic treatment failure, even when controlling for comorbid diabetes or tobacco use. Additionally, both adults and children with BMI >50 kg/m2 are growing faster than any other subset of obese patients. This highlights the need for increased provider awareness of dosing requirements in this population as well as continued research to support current guidelines and promote additional recommendations for oral antibiotics beyond beta-lactams.
Footnotes
Authors’ Note
Departmental resources were used for this study. All authors were involved in writing the article and had final approval of the submitted and published versions.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Department of Family Medicine, Mayo Clinic, Rochester, Minnesota.