Rotjanapan P, Dosa D, Thomas KS. Potentially Inappropriate Treatment of Urinary Tract Infections in Two Rhode Island Nursing Homes. Arch Intern Med. 2011;171(5):438–443. doi:10.1001/archinternmed.2011.13
The aim of this study was to determine the appropriateness of antibiotic initiation, selection, and duration of therapy among patients in nursing homes with results of a urinalysis showing urinary tract infection.
A retrospective chart review was conducted on patients of 2 nursing homes during a 6-month period (June 1-November 30, 2008). If a urinalysis had been ordered, the case was evaluated for the appropriateness of antibiotic initiation based on the McGeer criteria. For patients receiving antibiotics, the appropriateness of the initial selection, dosing schedule, and duration of treatment were assessed using patient-specific information and Infectious Diseases Society of American criteria. Patients' records were also reviewed for information on the development of Clostridium difficile colitis.
A total of 519 records were reviewed for documentation of a urinalysis; 132 patients, with a total of 172 case patients (ie, urinalysis showing infection) met inclusion criteria. Antibiotic treatment was initiated in 96 of the 172 case patients (56%); 146 case patients (85%) did not meet the McGeer criteria, yet antibiotic treatment was initiated in 70 of these (41%). Furthermore, 69 case patients (72%) received an inappropriate drug based on Infectious Diseases Society of American criteria, 44 case patients (46%) received inappropriate drug dosing based on creatinine clearance, and 64 case patients (67%) received treatment for longer than recommended. Patients who did not meet the McGeer criteria but received antibiotic therapy were 8.5 (95% confidence interval, 1.7-42.2) times more likely to develop C difficile within 3 months of treatment.
Opportunities exist to improve provider practice related to the appropriate treatment of urinary tract infections in the nursing home.
Urinary tract infections (UTIs) are the most common bacterial infections in older patients residing in nursing homes (NHs).1- 4 Fifty percent of female and 40% of male NH residents have been reported to have a UTI during their admission. Urinary tract infections are also the most common reason for antimicrobial prescribing in NHs and are responsible for initiation of 20% to 60% of systemic antimicrobial courses among the residents.1,5
The majority of NH residents with pyuria or bacteriuria documented on urinalysis (UA) do not manifest symptoms of an infection.6 Current consensus guidelines7- 10 from organizations such as the Infectious Diseases Society of America (IDSA) and the American Medical Directors Association suggest that asymptomatic UTIs should not be treated, taking into account the high percentage of recolonization after treatment; the potential complications of therapy, including the increased incidence of Clostridium difficile ; and the risks of precipitating antibiotic resistance. In 1991, McGeer et al11 published guidelines to identify when antibiotic therapy should be initiated in patients with results of a UA showing infection (positive UA). Additional guidelines have been sanctioned by the IDSA that guide the appropriateness of initial antibiotic selection and the duration of treatment.12
Despite these guidelines, telephone calls to physicians from NH staff reporting asymptomatic bacteriuria are commonplace, and physicians often prescribe antibiotic treatment pending urine culture results. To evaluate how antibiotics are used in UTIs, we conducted a retrospective review of all residents of 2 Rhode Island NHs during a 6-month period. Each case in which a positive UA was identified was evaluated to determine the appropriateness of antibiotic initiation using the McGeer criteria. The appropriateness of initial antibiotic selection and the duration of therapy were reviewed using IDSA criteria. Finally, the number of adverse events resulting from antibiotic initiation in this cohort was documented.
The protocol used in the study was reviewed by the Institutional Review Board at Rhode Island Hospital. In addition, permission was solicited from the administrators of the 2 NHs in Rhode Island. Participating research sites included a 120-bed facility in Providence, Rhode Island, and a 150-bed facility in Pawtucket, Rhode Island. Inclusion criteria for the study population consisted of any patient admitted to a NH between June 1, 2008, and November 30, 2008, who had a positive UA while residing in the facility. A positive urinalysis was defined as more than 3 erythrocytes per high-power field, more than 3 leukocytes in men or more than 5 leukocytes in women per high-power field, more than 2 renal tubular cells per high-power field, or more than 10 bacteria per high-power field.13
Exclusion criteria included patients with indwelling Foley catheters, patients receiving only comfort measures, patients with restrictions on antibiotic use stated in their care plans, and those with suspected upper UTI (pyelonephritis) on presentation. Charts were reviewed for all patients who met study criteria. If a patient had multiple UAs, each positive episode was treated as a separate event as long as the event occurred after a course of antibiotics had been completed. The chart audit included reviewing relevant demographic information, comorbid conditions, physicians' notes, nursing notes, laboratory evaluations, telephone order sheets, and medication sheets. Background characteristics of the patient population, including cognition and functional status, were obtained from the patient's most recent Minimum Data Set before the UA using the Cognitive Performance Scales and Activities of Daily Living scale, respectively.
Charts were reviewed for the appropriateness of antibiotic initiation using established criteria for surveillance published by McGeer et al.11 For patients in a NH without an indwelling catheter, 3 of the following criteria must be met to identify UTI and necessitate treatment: (1) a temperature of 38°C or higher; (2) new or increased burning sensation on urination, increased frequency of urination, or urgency of urination; (3) new flank or suprapubic pain or tenderness; (4) change in character of urine; and (5) worsening of mental or functional status.11 Antibiotic initiation was deemed inappropriate if McGeer criteria were not met.
Before beginning the analysis, a review of each facility's resistance patterns was conducted to ensure that Escherichia coli resistance to trimethoprim-sulfamethoxazole (TMP-SMZ) did not exceed 20% at either facility, as recommended by the IDSA.8 Finding no special resistance patterns, charts were reviewed to determine the initial antibiotic selection based on existing recommendations for uncomplicated UTIs sanctioned by the IDSA.12 Empiric treatment of UTIs was considered appropriate if:
TMP-SMZ was chosen as an empiric treatment, providing there was no documented contraindication to its use (eg, allergy);
in cases in which a contraindication to TMP-SMZ was identified, a fluoroquinolone was selected as an alternative therapy unless a contraindication to its use was identified; and
in cases in which a contraindication to both TMP-SMZ and a fluoroquinolone was identified, a third-generation cephalosporin or similarly broad-spectrum agent (eg, piperacillin sodium and tazobactam sodium or aztreonam) was selected as an alternative therapy in patients who required empiric treatment.
In cases in which the organism and sensitivities were known before antibiotic initiation, the initial antibiotic selection was deemed appropriate if:
the antibiotic with the narrowest effective spectrum was selected; in cases in which a patient had an allergy to the indicated drug, the next most narrow-spectrum antibiotic was deemed appropriate; and
nitrofurantoin was used and the calculated creatinine clearance (CrCl) was more than 60 mL/min.
Table 1 contains the appropriate dosages used by investigators for antibiotics routinely prescribed for treatment of uncomplicated acute bacterial cystitis in patients with normal or impaired renal function based on recommendations in the literature.12,14,15 Each patient's calculated CrCl, in milliliters per minute, was determined using the Cockcroft and Gault equation16 and the Jelliffe equation.17 For patients with normal renal function, defined as CrCl more than 60 mL/min, consistency with these guidelines was evaluated. If the dosing deviated from the Table 1 guidelines, it was deemed inappropriate. Consistency with guidelines was similarly determined for patients with impaired renal function.
As per IDSA guidelines, duration of treatment for acute bacterial cystitis (lower UTI) was deemed appropriate in women if antibiotics were administered for 3 days, except for nitrofurantoin therapy, in which the recommended duration of treatment was 7 days.8,12 For men, duration of treatment was deemed appropriate if antibiotics were administered for at least 7 to 14 days.18
The unit of analysis was each case patient. Descriptive statistics presented in Table 2 were calculated using SAS (version 9.2; SAS Institute Inc, Cary, North Carolina) stratified by facility.19 Logistic regression analysis was used to examine patient characteristics (age, sex, and Cognitive Performance Scales and Activities of Daily Living Scale scores) associated with the initiation of antibiotics, the inappropriate initiation of antibiotics, and the association of developing C difficile colitis within 3 months of antibiotic initiation when controlling for other patient characteristics.
The medical records of 519 patients were screened for eligibility. Of those records, 150, with a total of 188 UAs, were initially enrolled. Twelve patients were excluded due to the presence of indwelling Foley catheters at the time of UA and 4 patients with suspected acute pyelonephritis on presentation were also excluded. In total, 132 patient charts were reviewed, with a total of 172 UAs (ie, case patients). Of the 30 patients in whom repeat UAs were suspicious for UTIs, the median duration between events was 40.6 days (range, 7-150 days).
Table 2 shows the patient characteristics of the study population from the 2 NHs. Women constituted 78% of all case patients, and mean age was 83 years (range, 65-99 years). Approximately 86% of the case patients studied were white, 2% were black, and 12% were Hispanic. Eighty-nine urine specimens (52%) were obtained from patients with intact, borderline intact, or mildly impaired cognitive status (Cognitive Performance Scale score, 0-2); 60 urine specimens (35%) from patients with moderate to moderately severe impairment (Cognitive Performance Scale score, 3-4); and 23 urine specimens (13%) from those with severe impairment to very severe impairment (Cognitive Performance Scale score, >4). Most patients in the study sample (77%) needed extensive assistance based on their Activities of Daily Living Scale scores (score, 3-4), and 7% were totally dependent or almost totally dependent (score, 5-6).
Comorbid conditions included hypertension (75% of all case patients), dementia (55%), coronary artery disease (18%), diabetes mellitus (17%), and cerebrovascular accident (9%). A total of 36 case patients (21%) received antimicrobial therapy within 1 month before urine collection. The characteristics of the patients did not differ significantly between the 2 NHs except on cognitive and disability measures (Table 2). Patients in NH No. 2 were more cognitively impaired and functionally dependent compared with patients in NH No. 1.
Antibiotic therapy was initiated in 96 case patients. Women were 2.4 times more likely to receive antibiotics than were men (95% confidence interval, 1.1-5.2), controlling for other patient characteristics. Twenty-six case patients met the McGeer criteria for appropriate antibiotic initiation based on chart review; 100% of those case patients received antimicrobial therapy. A total of 146 case patients (85%) did not meet the McGeer criteria; of those, antibiotic therapy was used in 70 (41%). Among the 96 case patients who were given antibiotics, only 73 (76%) eventually had positive results of urine cultures. None of the tested patient characteristics (sex, race, cognitive status, functional status, and diagnosis) was correlated with inappropriate initiation of antibiotics on logistic regression.
Twenty-seven case patients received empiric treatment for suspected UTI before urine culture and sensitivity results were available. Among those who were treated empirically before the culture and sensitivity information was available, 18 (67%) were started on a fluoroquinolone, 5 (19%) on a sulfonamide, 2 (7%) on nitrofurantoin, and 2 (7%) on other medications. Based on the IDSA criteria specified in the “Guidelines for Determining the Appropriateness of Initial Antibiotic Selection” subsection of the “Methods” section, 15 case patients (56%) were given the wrong antibiotic—all received fluoroquinolones despite not having an allergy to sulfa.
Forty-four of the 96 case patients (46%) received inappropriate dosages by the criteria established (Table 3); 20 dosages were too high. Of those 44 case patients, 13 with normal renal function received levofloxacin, 500 mg, daily and 5 received ciprofloxacin hydrochloride, 500 mg, twice daily to treat uncomplicated UTI. Two case patients who had impaired renal function (CrCl <30 mL/min) received 1 tablet of TMP-SMZ, 160 mg to 800 mg, twice daily.
An additional 12 case patients received antibiotics despite contraindications. Ten patients who had impaired renal function (CrCl ≤60 mL/min) received nitrofurantoin and 2 received moxifloxacin hydrochloride, a medication that is not recommended for the treatment of UTI. Finally, 12 case patients without renal impairment received inadequate dosages of TMP-SMZ, penicillin, a cephalosporin, or an aminoglycoside.
Among the 96 case patients who received antibiotics, the duration of treatment varied, depending mainly on the severity of infection, with a range of 3 to 14 days. The mean duration of treatment was 7.8 days for lower (uncomplicated) UTI. Duration of aminoglycoside treatment was the longest (mean, 8.6 days), followed by nitrofurantoin (8.6 days); none of those case patients had upper UTI. Based on IDSA criteria, 64 case patients (67%) received antibiotics for longer than the recommended course.
None of the 76 case patients who did not receive antibiotics for asymptomatic bacteriuria experienced adverse consequences. There were also no deaths or hospitalizations attributed to worsening infection or sepsis during the 3-month follow-up period. Among those who received antibiotic therapy, 11 case patients (12%) developed C difficile colitis within 3 weeks of treatment; none of those case patients met McGeer criteria for initiation of antibiotics. Patients who were inappropriately given antibiotics were 8 times more likely to develop C difficile colitis within 3 months (odds ratio, 8.5; 95% confidence interval, 1.7-42.2) compared with the rest of the NH population. Every patient who developed C difficile recovered after metronidazole or vancomycin therapy.
We conducted a retrospective audit of the appropriateness of UTI treatment in 2 large NHs in Rhode Island for a 6-month period to determine how frequently errors occurred with antibiotic initiation, antibiotic selection, and duration of therapy compared with published guidelines. When compared with published criteria, 41% of case patients received antibiotic treatment despite never meeting McGeer criteria. This percentage is higher than the findings by Loeb et al,20 who reported that 30% of patients with asymptomatic bacteriuria received antibiotics. This inappropriate treatment was not without consequence because patients who received antibiotics without meeting McGeer criteria were 8 times more likely to develop C difficile in logistic regression than the NH population. Conversely, our study found that there were no complications (eg, hospitalizations, sepsis) among any of the case patients who did not meet McGeer criteria and were not given antibiotics. These results corroborate the findings of other investigators21,22 who have found no effect of asymptomatic bacteriuria on morbidity and mortality and lend credence to the idea of potentially using the McGeer criteria as a treatment guide.
In addition to the inappropriate initiation of antibiotics, our study noted several other critical prescribing errors involving this drug class. Among case patients started on antibiotics, 56% received an inappropriate drug based on suggested therapy for asymptomatic cystitis. An additional 46% received inappropriate dosages based on their calculated CrCl, and 67% received antibiotics for longer than indicated. It is noteworthy that the mean duration of antibiotic treatment in our study population was generally twice as long as what is recommended for uncomplicated UTIs.
Placed into context, these results spotlight the aggressive and potentially haphazard use of antibiotics in the NH environment and raise the question of patient safety issues. In a well-publicized report from the Institute of Medicine entitled To Err is Human,23 NHs were identified as the most common site for adverse drug events, with more than 800 000 estimated prescription-related errors occurring annually. Studies24 show that many of these errors are antibiotic related, contributing to significant morbidity and mortality annually. Beyond the potential for medical errors, the increased prevalence of antibiotic-resistant organisms resulting from overuse of antibiotics can result in 5 times higher mortality rates and more frequent hospitalizations than similar infections with nonresistant organisms.25,26 Greater understanding of why NH physicians prescribe antibiotics while waiting for urine culture and sensitivity results is required.
Somewhat unexpectedly, when evaluating patient-level correlates of inappropriate initiation of antibiotics, we did not identify any trends toward increased use of these medications with increasing levels of cognitive impairment. Previous research27 has suggested that patients with cognitive impairment frequently receive antibiotics during their last months of life. There was also no observed increase in inappropriate antibiotic use based on functional impairment, sex, race, or a documented diagnosis of congestive heart failure, diabetes, cerebrovascular disease, or dementia.
Nevertheless, we did identify a trend toward initiation of antibiotics among women. It is plausible to speculate that physicians might take more care in starting antibiotic therapy in men, in whom the duration of treatment is longer than in women; 3 days of treatment for women might seem reasonable while awaiting culture and sensitivity results. Further research is needed to identify whether this holds true in other studies and to determine whether NH characteristics or provider characteristics play a role in inappropriate initiation of antibiotics.
We believe that the results of this study provide important information regarding how antibiotics are used for patients in NHs who develop UTIs; however, there are several limitations to our analyses. Given the retrospective nature of this study, the review of antibiotic initiation was limited to the documentation in the chart. It was not possible to recreate the patient-specific concerns that a provider needed to consider when making decisions about antibiotic initiation. Second, some question the validity of guidelines, such as the McGeer criteria.28 Surveys of NH physicians have also found that more than one-third are unaware of the McGeer criteria and only 55% use them in clinical practice.29 Study is needed to understand why the criteria are not used, particularly in light of the findings discussed in this study.
Finally, in this study, we assumed that all the case patients had uncomplicated UTIs. Although we excluded patients with catheters, the formal definition of a complicated UTI includes those occurring in a patient with other structural or functional abnormalities. Given that the recommended treatment duration of complicated UTIs is longer (10-14 days),30 it is possible that some of the 67% of case patients who received antibiotics longer than the recommended course might have been appropriately treated for their level of infection.
There are clear opportunities to improve the care of patients in NHs who have UTIs. This study suggests that nearly 41% of case patients in 2 NHs received antibiotics despite not meeting guidelines for their initiation. A number of these patients eventually developed complications from antibiotic use, including an 8-fold increase in the rate of C difficile. Case patients who did not meet McGeer criteria did not develop complications from failure to treat. Other routine errors among this population included starting the wrong antibiotic, receiving dosages at higher rates than recommended, and exceeding the recommended duration of antibiotic treatment for UTIs.
Correspondence: David Dosa, MD, MPH, Center for Gerontology and Health Care Research, Warren Alpert Medical School, Brown University, 121 S Main St, Floor S-6, Providence, RI 02912 (firstname.lastname@example.org).
Accepted for Publication: August 23, 2010.
Author Contributions:Study concept and design: Rotjanapan and Dosa. Acquisition of data: Rotjanapan. Analysis and interpretation of data: Rotjanapan, Dosa, and Thomas. Drafting of the manuscript: Rotjanapan, Dosa, and Thomas. Critical revision of the manuscript for important intellectual content: Rotjanapan, Dosa, and Thomas. Statistical analysis: Dosa and Thomas. Obtained funding: Dosa. Administrative, technical, and material support: Rotjanapan. Study supervision: Dosa.
Financial Disclosure: None reported.
Funding/Support: This study was supported by the Hartford Foundation.
Additional Contributions: We thank Vincent Mor, PhD, and Richard Besdine, MD, for their critical review of the manuscript.