Management by Primary Care Clinicians of Patients Suspected of Having Community-Acquired Methicillin-Resistant Staphylococcus Aureus Infections

Patient Follow-Up Data

Results (continued)

Antibiotics and MRSA Antibiotics, Accounting for Temporal Changes

Temporal changes did not account for the increase in antibiotic use or MRSA antibiotics during the intervention in 680.x cases, as seen in Tables 11 and 13. Among patients with a 681.x or 682.x diagnosis, the piecewise GEE model indicated that the intervention generated an increase in antibiotic and MRSA-antibiotic prescriptions that exceeded the natural rate of change that occurred over the period of the study. However, this increase was not statistically significant (p=0.0539 for antibiotics and p=0.1220 for MRSA antibiotics), and thus we cannot rule out that the increase in antibiotics and MRSA antibiotics weren't due to temporal trends. There was a 12 percent per month increase in the odds of receiving an antibiotic and a 13 percent per month increase in the odds of receiving a MRSA antibiotic during the intervention period. These increases were only 5 percent and 4 percent, respectively, during the historical period (Table 16).

Other findings from the model for patients with a 681.x or 682.x diagnosis were quite revealing. For example, the odds of a patient receiving an antibiotic and an antibiotic to treat MRSA were higher for patients from Texas clinics than for patients from North Carolina clinics. The odds that a patient seen by a pediatrician received a MRSA antibiotic were higher than the odds that patients seen by a family medicine physician received such treatment.

Return to Contents

Assessment of Hawthorne Effect

The Hawthorne Effect is often discussed in practice-based research networks research but rarely measured. In this project, approximately 25 percent of clinicians participated in the evaluation of cases via the QI component of the project, which also included follow-up interviews. The other 75 percent of clinicians were "controls" for those clinicians who were "observed." We found no Hawthorne Effects, including in the prescribing of antibiotics or the selection of antibiotics that cover MRSA, which are two outcome measures in which a Hawthorne Effect might be expected (Tables 17 and 18).

Table 17. Hawthorne Effect, purulent (680.x) intervention cases only

Outcome "Observed" Clinicians n=14 "Control" Clinicians n=77 p value*
1. Number of cases 21 127 -
2. Cases with antibiotics prescribed (n) 9 60 -
3. Rate of antibiotic prescribing (%) (row 2/row 1) 42.86 47.24 0.7089
4. Cases with antibiotics that cover MRSA (n) 3 29 -
5. MRSA coverage rate (%) (row 4/row 3) 33.33 48.33 0.4001
6. Cases with procedures performed (n) 1 6 -
7. Procedure rate (%) (row 6/row 1) 4.76 4.72 0.994
8. Cases with cultures obtained (n) 2 19 -
9. Culture rate (%) (row 8/row1) 9.52 14.96 0.5083

*p values only calculated for proportions or mean values.

Table 18. Hawthorne Effect, 681.x and 682.x intervention cases only

Outcome "Observed" Clinicians n=14 "Control" Clinicians n=77 p value*
1. Number of cases 250 1008 -
2. Cases with antibiotics prescribed (n) 105 465 -
3. Rate of antibiotic prescribing (%) (row 2/row 1) 47.04 43.08 0.2402
4. Cases with antibiotics that cover MRSA (n) 42 180 -
5. MRSA coverage rate (%) (row 4/row 3) 40.00 38.71 0.8065

*p values only calculated for proportions or mean values.

Patient Follow-Up Data

Each health system conducted its own patient follow-up evaluations to gather information about patients who were treated for SSTIs in the participating clinics. Calls were made at least 10 days after the initial clinic visit. Patients were asked about outcomes, such as hospitalizations, ED or urgent care visits, status of the infection, and use of antibiotics. CINA provided de-identified data to the research team. The data collection methods were different: In MCNT, hybrid calls were made to patients (live caller asks patient to answer questions then questions were answered by keypad selections from an automated system with branching), and in WHA, a staff person contacted and completed all data collection live.

A total of 681 calls were completed in the two health systems (Table 19). There were greater than 13 times more calls for 681.x-682.x cases, compared to 680.x (only 48 total 680.x cases including both WHA and MCNT pre-intervention and intervention periods combined). This limited the analysis that could be done on 680.x cases. The characteristics of the patient samples did not change significantly between pre-intervention and intervention periods, except in MCNT where there were somewhat more men and adult cases (consistent with the electronic data, Table 3). The mean age did not significantly change.

Among the outcomes patients were asked about, there were no significant differences reported between the pre-intervention and intervention periods. Although unplanned care events were more common in MCNT, these events were infrequent across both groups with no significant change from pre-intervention to intervention. The proportion of cases that resolved at the time of the patient telephone contact did not change between the pre-intervention and intervention period, although there was a much higher proportion resolved in WHA compared to MCNT (Table 20). The number of cases with a 680.x diagnosis and time to resolution data was too small to discern any trends. Among 681-682.x cases, there were sufficient cases for MCNT only, and during the intervention, the time to resolution decreased slightly by 0.7 days, but this was not significant (p=0.4881). Because of the delays in the vast majority of WHA calls resulting in their exclusion from the time to resolution analysis, the sample size for time to resolution at WHA for 681-682.x cases is very small and difficult to interpret. The overwhelming majority of patients who were prescribed antibiotics said they picked them up, indicating good adherence. Second courses of antibiotics occurred in less than 10 percent of cases overall and were more common in MCNT. However, there was no change in the rate of this comparing the pre-intervention to the intervention period.

In the multivariate analysis of 680.x-682.x cases (but comprised almost entirely of 681.x-682.x because of the small n in 680.x), the intervention did not have an effect on unplanned care events (odds ratio [OR] 0.957, 95% confidence interval [CI] 0.44 to 2.08, p=.912), whereas having a previous case of MRSA increased the likelihood of unplanned care events (OR 2.45, 95% CI 1.17 to 5.13, p=.018). The multivariate analysis of time to resolution was limited by the dataset consisting of almost entirely 681.-682.x cases from MCNT. The intervention resulted in a 1.2-day decrease in the time to resolution, but this was not significant (p=.2426); there were no other significant predictors for time to resolution.

Return to Contents

Provider QI Case Reports

Each health system also conducted QI evaluations of selected SSTI cases to better understand the management of these infections by their providers. Health care providers who agreed to participate completed case reports using an online system through their respective health systems. The research team received de-identified case reports from CINA and reviewed them. The research team then conducted interviews to gather more detailed information about certain cases. Initially, clinicians with any type of SSTI case were interviewed, but after the study team determined that sufficient information had been obtained for the more common diagnosis of cellulitis, only cases of abscess or mixed abscess and cellulitis were interviewed.

Providers completed QI case reports on 111 SSTI cases (Table 21). Interviews were conducted for 42 of these cases, based on whether we determined that further information about SSTI management might be obtained (Table 22).

Table 21. Characteristics of patients seen with SSTI diagnosis in QI case reports

  MCNT WHA Total
Female 20 (37.7%) 34 (59.7%) 54 (49.1%)
Male 33 (62.3%) 23 (40.4%) 56 (47.4%)
Age < 18 6 (11.3%) 13 (22.8%) 19 (17.3%)
Age ≥18 47 (88.7%) 44 (77.2%) 91 (82.7%)
New abscess* 16 (30.2%) 12 (21.1%) 28 (25.5%)
New mixed abscess and cellulitis* 17 (32.1%) 16 (28.1%) 33 (30.0%)
New cellulitis without abscess* 18 (34.0%) 28 (49.1%) 44 (41.8%)
Unknown 2 (3.8%) 1 (1.8%) 3 (2.7%)

Key: MCNT = Medical Clinic of North Texas, WHA = Wilmington Health Associates.
* New SSTI indicates this was the first time that this patient had presented to a health care provider with this infection. Diagnoses are based on clinician's assessment, not on ICD-9 codes.

Table 22. Number of qualitative provider interview cases by infection type
n = 42

Type of SSTI MCNT WHA Total
Cellulitis 5 8 13
Abscess or mixed Abscess and Cellulitis 14 15 29

Key: SSTI = skin and soft tissue infection, MCNT = Medical Clinic of North Texas, WHA = Wilmington Health Associates.

The case reports and interviews allowed us to respond to some key questions about the initial management of SSTIs. Because the initial management of SSTIs differs for those that are abscesses versus those that are strictly cellulitis, we considered them separately.

Return to Contents

Management and Treatment of Abscesses

The CDC guidelines recommend I&D as the primary treatment for abscess and culturing all purulent discharge. The guidelines note that if an antibiotic is used in the treatment of abscesses, it should cover CA-MRSA. In light of these recommendations, the key questions for purulent skin infections were:

  1. What factors influence whether a provider elects to drain an abscess?
  2. What factors influence whether a provider packs an abscess after an I&D?
  3. In a patient with an abscess, what factors influence whether or not a culture is done?
  4. When antibiotics are used, what factors influence the decision whether or not to cover MRSA?
  5. How did providers use the intervention materials (I&D kits, patient education, and provider guidelines), and what were the barriers to implementation?

1. What factors influence whether a provider elects to drain an abscess?

Among the QI case reports, 66 were cases with abscesses. The primary care providers provided QI information for 29 of these cases. These providers reported either performing a drainage procedure or referring the patient for a drainage procedure in 65 percent of these patients. Large deep abscesses were more likely to be drained. Reasons for not draining an abscess included: Spontaneous drainage, patient self manipulation leading to drainage, small size, folliculitis, or being located in a sensitive position like the scrotum.

2. What factors influence whether a provider packs an abscess after an I&D?

Provider opinions about packing varied, with some reporting that they pack most abscesses and some almost never pack. The most common reason for packing was large abscess size. Providers who did not use packing reported not seeing treatment failures. Other reasons for not packing abscesses included: Not having or being unable to find packing material, insufficient time, lack of available follow-up appointments, packing requires a larger incision that makes a larger scar, a superficial abscess, or an abscess that was fully drained.

3. In a patient with an abscess, what factors influence whether or not a culture is done?

All providers interviewed reported that they generally obtain cultures when treating abscesses. Cultures were actually obtained in 67 percent of the cases discussed in the provider interviews. Reasons for not culturing included: No procedure was done, no pus was seen, or the culture, regardless of the results, would not change the treatment. No providers indicated that culture results for these cases changed the treatment.

4. When antibiotics are used, what factors influence the decision whether or not to cover MRSA?

Among the cases where an interview was conducted, providers reported using antibiotics for over 90 percent of patients with abscesses (Table 23). The majority of providers prescribed antibiotics for all abscesses when MRSA was suspected. This was considered to be the community standard of care, even if an abscess was drained completely. The most commonly used antibiotic was TMP/SMX (trimethoprim\sulfamethosazole), which was prescribed for 73 percent of patients (Table 24). Providers prescribed antibiotics that the CDC recommends to cover CA-MRSA in all patients with abscesses who received antibiotics. Many clinicians reported that essentially all or the majority of abscess cultures at their site grew MRSA. However, cultures grew MRSA in only 44 percent of interview cases.

Table 23. Antibiotics prescriptions by interviewed providers

Any Antibiotic 22 (92%)
No Antibiotic 2 (8%)

Table 24. Type of antibiotic prescribed

TMP/SMX 18 (81.8%)*
Doxycycline 3 (13.6%)
Clindamycin 1 (4.6%)

*TMP/SMX+ Rifampin 3 patients
TMP/SMX + Cephalexin 1
TMP/SMX + Ceftriaxone

5. How did providers use the intervention materials (I&D kits, patient education, and provider guideline), and what were the barriers to implementation?

By the end of the study, all providers were aware of the I&D kits, patient education, and provider guidelines. Providers who used the I&D kits thought they were very helpful. Reasons for not using the kits included: Provider habits and routines, alternate I&D equipment was readily available nearer to exam rooms, and not being able to find or access the kits quickly.

More providers reported using the provider guidelines or the patient education materials at the point of care than used the kits. The majority thought that the materials were useful. Most providers especially liked the patient education materials.

Initially, some clinics did not receive the kits, and a number of providers noted that they were not aware of the I&D kits, patient education materials, or provider guidelines. This resulted from difficulties educating providers about and implementing an intervention in widely dispersed clinics when the study team was not on site.

In the end, the intervention appeared to change patient management for some providers, and many providers agreed that the intervention was helpful, especially the written materials. This is particularly relevant because the written materials would be more easily generalizable to other setting than the I&D kits, due to cost and other factors.

Return to Contents

Management and Treatment of Cellulitis

As expected, antibiotics were used to treat cellulitis 100 percent of the time, based on the data from the QI case reports. The role of CA-MRSA in cellulitis without abscess or purulent drainage is less clear because cultures are rarely obtained. The CDC recommended covering Streptococcus and other suspected pathogens initially, providing close followup, and adding coverage for CA-MRSA if patients do not respond to initial treatments. We were interested in answering two key questions:

  1. What antibiotics do providers use when treating cellulitis?
  2. When do providers cover CA-MRSA when treating cellulitis?

1. What antibiotics do providers use when treating cellulitis?

The majority of the patients (58 percent) in this QI report sample received antibiotics that are recommended by the CDC for suspected CA-MRSA. Antibiotics covering Group A Streptococci were prescribed for 75 percent of patients.

2. When do providers cover CA-MRSA when treating cellulitis?

Eight of the 12 providers providing QI data were more likely to consider CA-MRSA as the cause of a cellulitis when the appearance of the cellulitis suggested CA-MRSA. Cellulitis from MRSA was described as "very red," inflamed, or draining pus and as having a rapid onset associated with fever or an ill-appearing patient. A "gut feeling" that the cellulitis was caused by MRSA or a cellulitis that looked like it might turn into an abscess also made providers more likely to treat a cellulitis patient with a CA-MRSA-covering antibiotic. Patients at higher risk for risk MRSA infection described by clinicians included recently hospitalized patients, frail older patients, nursing home residents, hospital workers, diabetics, and patients who played contact sports.

Return to Contents

Manual Chart Audit

Manual chart audits were conducted in each health system for comparison with CINA's automated data extraction (i.e., electronic chart audit). In addition to the manual chart audit conducted prior to the intervention period, a manual chart audit was conducted after the intervention period on 40 randomly selected abscess cases per health system for the 6- to 7-month intervention time period (with up to 100 cases total including a short audit of non-abscess SSTIs cases to verify non-abscess status). This chart audit was used validate the automated audit information and collect additional information not available electronically, such as key elements of the history (e.g., length of time from initial symptoms until presentation, whether the patient in contact sports program) and physical description of the abscess (location, size, and description of the SSTI).

A total of 160 cases were reviewed for the chart audit, including 40 pre-intervention in each system and 40 during the intervention in each system (Table 25). The procedure rate among abscesses was about 25 percent in the pre-intervention and intervention audits combined, which was three times higher than the procedure rate found in the electronic audit. This suggests that procedures were not being captured in the electronic audit. Similarly, the overall culture rate in the audit was 37 percent, 2.3 times higher than found in the electronically obtained data.

Table 25. Manual chart audit results, including pre-intervention and intervention periods

System (primary care only) WHA MCNT Total
Months/years of data collection Pre-Intervention Intervention Pre-Intervention Intervention Pre-Intervention Intervention
Number of abscess cases 34 39 38 40 72 79
  Classified 680 21
20 (52.6%) 20 (50.0%) 41
39 (49.4%)
  Classified 681,682 13 (38.2%) 20 (51.3%) 18 (47.4%) 19 (50.0%) 31
  Culture obtained, n/% 10 (30.3%) 13 (33.3%) 17 (44.7%) 16 (40.0%) 27
  % of Cultures obtained that are Staphylococcus aureus 7 (70.0%) 9 (69.2%) 14 (82.4%) 10 (62.5%) 21 (77.8%) 19 (65.5%)
  % of Staphylococcus aureus that are MRSA 1 (14.3%) 4
2 (14.3%) 7 (70.0%) 3 (14.3%) 11
  Procedures done, n/% 6 (18.2%) 7 (18.0%) 9 (23.7%) 15 (37.5%) 15
Abscess characteristics Number of cases with more than one lesion 7 (20.6%) 7 (18.0%) 8 (21.1%) 2 (21.1%) 15 (20.83%) 9
% Referred 0
4 (10.0%) 0
% with documentation of follow-up plans 30 (88.2%) 35 (89.7%) 34
25 (62.5%) 64 (88.9%) 60 (75.9%)
% Subsequent unplanned visits outside of clinic 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%)
Number of cellulitis without an abscess 5 0 8 0 13 0 (0%)
  Classified 680 0
1 (12.5%) 0
  Classified 681,682 5 (100%) 0
7 (87.5%) 0

Key: MCNT = Medical Clinic of North Texas, WHA = Wilmington Health Associates.

Return to Contents
Proceed to Next Section

Page last reviewed October 2014
Internet Citation: Patient Follow-Up Data. October 2014. Agency for Healthcare Research and Quality, Rockville, MD.