Management by Primary Care Clinicians of Patients Suspected of Having Community-Acquired Methicillin-Resistant Staphylococcus Aureus Infections
Purulent skin and soft tissue infections (SSTIs) requiring medical attention are often managed in primary care, although in the United States they account for only a small percentage of total primary care office visits. According to Medical Expenditure Panel Survey (MEPS) data from 2005, this condition presents in less than 0.5 percent of outpatient visits, accounting for slightly over 2.5 million visits by just under 900,000 people annually.1 Staphylococcus aureus is the most common pathogen causing this condition.
SSTI management is becoming more complex. The prevalence of SSTIs caused by community-acquired methicillin-resistant Staphylococcus aureus (CA-MRSA) has been increasing rapidly, including in otherwise healthy individuals.2 CA-MRSA is a significant public health concern as it has the potential to develop quickly into an invasive skin infection and cause other life-threatening complications.3-5 Like other SSTIs, most CA-MRSA infections are managed initially on an outpatient basis in primary care settings. Therefore, it is critical that primary care clinicians recognize and appropriately treat potential CA-MRSA infections. Current evidence suggests that treatments that take into account the possibility of CA-MRSA are not the norm. For instance, the MEPS 2005 data indicates that when an antibiotic was prescribed for an abscess, it covered MRSA in only 17 percent of the cases.1 Cellulitis, an SSTI without purulence (not an abscess), is typically caused by Staphylococcus aureus or streptococcus.
In response to this growing public health problem, the Centers for Disease Control and Prevention (CDC) convened an expert panel and published recommendations and a clinical algorithm/flowsheet for outpatient management of CA-MRSA. The guideline for management of CA-MRSA involves the following, alone or in combination: incision and drainage (I&D), culture of the purulent material, and use of systemic antibiotics. Although the CDC guidelines are being widely disseminated, their feasibility and therefore uptake in busy primary care settings are unknown. Our project in this task order advanced our understanding of CA-MRSA by developing real-world strategies consistent with the CDC guidelines and by testing these strategies in busy primary care settings.
The purpose of this study was to test interventions specifically designed for two health networks to optimize treatment for SSTIs consistent with the CDC CA-MRSA guidelines. This study had three specific aims:
- Describe the baseline incidence and clinical practice for SSTIs in three large health systems: two private health systems and a large integrated urban community health center.
- Design two intervention strategies and associated outcome measures consistent with the CDC CA-MRSA guidelines.
- Conduct a clinical trial evaluating the effect of the intervention strategies on the outcomes in primary care clinics within two health care systems.
Previous reports described the complete findings for aims 1 and 2, which we only summarize in this report. We fully describe findings from aim 3 in this report.
For the intervention phase of the project, we worked with two health networks, Medical Clinic of North Texas (MCNT) and Wilmington Health Associates (WHA). (Denver Health and Hospital Association [DHHA] only participated during the pre-intervention phase of the project.)
MCNT is a large group practice in north Dallas that includes 35 practice locations and approximately 120 clinicians, the majority of whom are in primary care and are family physicians, general internists, and general pediatricians. The system also includes one obstetrics-gynecology practice and one rheumatology practice. The system cares for approximately 150,000 people per year. All providers are on the same, linked electronic medical record system, NextGen.
WHA is a multispecialty group practice in Wilmington, North Carolina, with about 100 clinicians, of whom approximately 35 are in primary care. Primary care clinicians include family physicians, general internists, and general pediatricians. The entire group uses a single electronic medical record, A4 Healthmatics.
DHHA, a fully integrated safety net urban health system, includes a federally qualified community health center component consisting of 20 primary care and school-based clinics that receive more than 320,000 annual visits. The DHHA system implemented outpatient guidelines for CA-MRSA in August 2004, well before the CDC guidelines and flowsheet were published. The DHHA guidelines are very similar to the CDC guidelines. Therefore, the DHHA system provided an opportunity to seek "best practices" approaches to guideline implementation. Neither the WHA nor MCNT systems had implemented any system-wide approach to CA-MRSA.
We conducted a clinical trial for 6 to 7 months in 16 clinics within the WHA and MCNT systems. We worked with each organization's existing internal quality improvement (QI) team to conduct ongoing practice implementation of the new SSTI guidelines. We conducted the trial using a before-after study design, which includes 12 months of historical data review of care prior to the initial start date for all participating clinics. Treatment of SSTIs during the 6- to 7-month intervention period was compared to 12 months of historical data. The prolonged historical timeframe helped us discern if there had been any secular trends in the SSTI outcome measures prior to the implementation. These trends were adjusted for in the quantitative analysis. All data were de-identified.
Data for this project came from two main sources: de-identified electronic health record data and QI data made available to the research team. CINA extracted standardized, electronic chart audit data from participating WHA and MCNT sites according to International Classification of Diseases, 9th Revision (ICD-9) codes for SSTIs (680.x, 681.x, 682.x). A randomly generated number linked the de-identified patient-level data provided to the research team. CINA, which has a Health Insurance Portability and Accountability Act business agreement with all organizations, managed the data collection. WHA and MCNT conducted manual chart audits that were used as a supplement to the electronic chart audit.
Quality improvement data were made available from patient followups conducted within and by each network and from case report evaluations that providers completed on select cases. On behalf of the health systems, the investigators agreed to follow up by phone with providers to learn about the management of specific cases.
Primary Outcomes (Electronic Chart Audit Data)
Descriptive statistics, including frequencies and percentages, were used to characterize the historical and intervention data. For the primary outcomes, bivariate Chi-square, Student's t-tests, and Fisher's exact tests were conducted to compare pre-intervention and intervention electronic chart audit data. Generalized estimating equations (GEE) with exchangeable variance-covariance structures were used to model the odds of a patient receiving a culture, a drainage procedure, an antibiotic, and a MRSA antibiotic while accounting for correlations due to clustering of patients within providers. Independent variables included in the models were gender, child or adult status, the presence of diabetes, a history of previous skin infection, the health care system through which services were received, and the specialties of the providers who delivered the services.
A longitudinal growth model was used to determine whether evidence existed of the increasing use of antibiotics as well as MRSA antibiotics across the population of patients served during the November 2007 through October 2008 historical control period. The longitudinal analysis revealed that this was indeed the case for patients with a 681.x or 682.x diagnosis but not for a 680.x diagnosis. The longitudinal growth model predicted that the percent of 681.x and 682.x patients who were prescribed antibiotics by a provider increased by 1 percent per month (p=0.0008). In addition, the percent of patients prescribed MRSA antibiotics increased by 0.5 percent per month (p=.0480). In both cases, the increases were significant. To hold the impact of the historical data constant, a piecewise GEE model was used to determine whether the odds of antibiotic and MRSA antibiotic use was greater during the intervention period.
Patient Follow-Up Data
For the patient follow-up data, bivariate chi-square and Fisher's exact tests were conducted to compare pre-intervention and intervention outcomes on several important measures, including hospitalization, emergency department (ED) or urgent care visits, medication adherence, whether the infection resolved, and time to resolution. Linear regression was used to model approximate duration of infection, and logistic regression was used to model the odds of an unplanned care event (combined ED and urgent care visits and hospitalization).
Time to resolution was based on patient response when asked how long ago did the infection clear (this was not asked to patients who said the infection was still present). The answer was selected by the patient from several options and was converted to days (e.g., "about a week ago" converted to 7 days, "more than a week ago" estimated to be 11 days). From the interval between the initial SSTI clinical visit and the patient follow-up response, we subtracted the number of days since the infection cleared to determine time to resolution. Our plan was to have the patient follow-up contacts occur between 15 and 31 days after the initial SSTI visit. However, due to logistical difficulties, some patients, including most at WHA, had very long intervals (up to 6 months) between the SSTI visit and the patient follow-up contact (see the Limitations section for further discussion). In these cases with a long interval, it is possible there was a recall bias in interpreting patient responses, and this bias would be especially likely when the patient was asked how long ago did the infection clear. In addition, because the longest response possible was "more than a week ago," time to resolution could not be accurately determined when there was a long interval. Therefore, in calculating time to resolution, cases were excluded when the interval between the initial SSTI and the follow-up patient contact was greater than 31 days. Cases were also excluded when data elements required to calculate time to resolution were missing. The other patient follow-up data collected on the cases with long intervals were included in the results.
Provider Quality Improvement Case Reports
The case reports were summarized with basic descriptive statistics (e.g., frequencies and percents). Follow-up interviews were treated as qualitative data and were analyzed iteratively beginning with an initial review of the data for major themes, stratified by cellulitis and abscess cases. The initial themes were shared with the study team for feedback and comments. The interview data then were analyzed fully, arriving at a final set of themes for further review and verification by the team.
Manual Chart Audits
For the patient follow-up data, univariate analyses were conducted to compare the pre-intervention and intervention periods.
Aim 1. Describe the baseline incidence and clinical practice for SSTIs in three large health systems: two private health systems and a large integrated urban community health center.
There were 92 manual chart audits in the three systems. A majority of abscesses in each system and overall (57.6 percent) were classified as 680.x. Rarely was cellulitis coded as 680.x. Overall, antibiotics were started in almost two-thirds of cases, but of these, less than 40 percent were agents that covered MRSA. DHHA had higher procedure rates and higher utilization of antibiotics (including MRSA coverage), but a higher culture rate was not found (only increased antibiotic usage was significant). Most of the abscesses were small and solitary (Table 1).
Table 1. Manual chart audit data (pre-intervention baseline only)
|System (primary care only)||WHA||MCNT||DHHA||Total|
|Months/years of data collection||10/30/06-10/16/08||4/10/07-06/1709||1/0/08-7/07/09||—||Sig.|
|Number of abscess cases||34||38||20||92|
|Classified 680||21( 61.7%)||20 (52.6%)||12 (60.0%)||53 (57.6%)|
|Classified 681,682||13 (38.2%)||18 (47.4%)||8 (40%)||39 (42.4%)|
|Culture obtained, n/%||10 (30.3%)||17 (44.7%)||5(25.0%)||32 (34.8%)|
|% of cultures obtained that are Staphylococcus aureus||7 (70.0%)||14 (82.4%)||4(80.0%)||25 (78.1%)|
|% of Staphylococcus aureus that are MRSA||1 (14.3%)||2 (14.3%)||3 (75.0%)||6 (24.0%)||*|
|Antibiotic started, n/%||12 (35.3%)||29(76.3%)||19 (95.0%)||60 (65.2%)||*|
|% of antibiotic starts that are MRSA antibiotics||4(33.3%)||8(27.6%)||11(57.9%)||23 (38.3%)|
|Culture obtained and MRSA antibiotics started, n/%||2 (5.9%)||3 (7.9%)||4 (20.0%)||9 (9.8%)|
|Procedures done, n/%||6 (18.2%)||9 (23.7%)||6 (30.0%)||21(23.8%)|
|% of procedures where antibiotics started||1 (16.7%)||6 (66.7%)||6 (100.0%)||13 (61.9%)||*|
|Adults (18 + years), n/%||17 (50.0%)||20 (47.4%)||16 (80.0%)||53 (57.6%)|
|% of adults with procedures done||2 (12.5%)||4 (20.0%)||5(31.3%)||11 (20.8%)|
|Children (<18 years), n/%||17 (50.0%)||18 (52.6%)||4 (20.0%)||39 (42.4%)|
|% of children with procedures done||4 (23.5 %)||5 (27.8% )||1 (25.0% )||10 (25.6%)|
|Abscess characteristics||Number of cases with more than one lesion||7 (20.6%)||8 (21.1%)||4 (20.0%)|
|Lesion size (cm) when known, mean(std)||0.9 cm (1.5)||2.0 cm (1.7)||1.2 cm (1.2)||1.4 cm
|% lesions with spontaneous drainage||0 (0%)||13 (26.5% )||4(16.7%)||17 (15.2)||*|
|% referred||0 (0%)||0 (0%)||1 (5.0%)||1 (1.1%)|
|% with documentation of follow-up plans||30 (88.2%)||34(89.5%)||3 (15.0%)||76(82.6%)||*|
|% subsequent unplanned visits outside of clinic||0 (0%)||0 (0%)||1 (5.0%)||1 (1.1%)|
|Number of cellulitis without an abscess||5||8||14||27|
|Classified 680||0 (0%)||1 (12.5%)||0||1 (3.7%)|
|Classified 681,682||5 (100%)||7 (87.5%)||14||26 (96.3%)|
|Antibiotic started, n/%||2 (40%)||5 (62.5%)||14 (100%)||21 (77.8%)||*|
|% of antibiotic starts that are MRSA antibiotics||1 (50%)||2 (40%)||4 (28.6%)||7 (33.3%)|
Aim 2. Design two intervention strategies and associated outcome measures consistent with the CDC CA-MRSA guidelines.
Based on the results from Aim 2 focus group results, the intervention was developed to specifically deal with: (1) time constraints, (2) forgetting to do a culture and/or not having the culturette readily available, and (3) provider concerns about performing I&Ds. Therefore, we developed ideas for a prepackaged kit with point-of-care information for patients and providers to be ready and on hand. To remind providers to do a culture, culturettes would be readily available with the kits.
We worked with the QI contracts from WHA and MCNT and developed the following intervention strategies to be implemented in each system (Table 2). The study team took the findings from the focus groups and worked closely with the key contacts from MCNT and WHA to solve feasibility issues and refine the specific intervention strategies. A feasibility pilot test of the intervention was conducted.
Table 2. Intervention components in each health system
|Type of Intervention||MCNT||WHA|
|Provider information (at the point-of-care)||X||X|
Key: MCNT = Medical Clinic of North Texas, WHA = Wilmington Health Associates. HER=electronic health record.
Continuing medical education (CME) presentations on SSTIs and CA-MRSA were given in both MCNT and WHA. For MCNT, Dr. Connie Price, an infectious disease consultant for the project, conducted an Internet-based CME conference for providers. For WHA, CME was provided during an in-person meeting by an in-house infectious disease specialist and an in-house surgeon. Overall, the intervention in WHA and MCNT were very similar; the CME and EHR differences between the two systems were minor, therefore the data from the two systems were combined for analysis.
Aim 3. Conduct a clinical trial evaluating the effect of the intervention strategies on the outcomes in primary care clinics within two health care systems.
The electronic chart audit from MCNT and WHA resulted in a total of 4,518 SSTI cases during this study, including 3,112 during the pre-intervention period and 1,406 cases during the intervention. The lower number of cases during the intervention is largely because the intervention occurred over 6 to 7 months, whereas the pre-intervention period was 12 months. The demographics of the cases are presented below (Table 3). At MCNT, the proportion of children and adolescents fell in the intervention, and there were fewer children and adolescents compared with WHA. In the two systems combined, the mean age rose slightly but significantly during the intervention period compared with the pre-intervention period. Family physicians and general internists each saw about one-third of the cases, with the remainder split among pediatricians and midlevels. Due to an increase in MCNT, there were more patients with diabetes in the intervention period, whereas patients with previous SSTIs were more common in the intervention period due to an increase in WHA.
Table 3. Characteristics of all SSTI cases (680.x � 682.x) in both health systems
|Pre||Intervention||p Value*||Pre||Intervention||p Value||Pre||Intervention||p Value|
|Total number of cases (n)||1870||643||-||1242||763||-||3112||1406||-|
|IM & Peds (n)||81||27||-||0||0||-||81||27||-|
|IM & Peds (%)||4.33||4.2||0.8864||0||0||-||2.6||1.92||0.1644|
|Previous case SSTI (n)||321||114||-||123||150||-||444||264||-|
|Previous case SSTI (%)||17.17||17.73||0.7445||9.9||19.66||<.0001||14.27||18.78||0.0001|
Key: MCNT = Medical Clinic of North Texas, WHA = Wilmington Health Associates, FM = family medicine, IM = internal medicine, Peds = pediatrics, SSTI = skin and soft tissue infection
*p values only calculated for proportions or mean values.