National Trends in Emergency Department Antibiotic Prescribing for Elders with Urinary Tract Infection, 1996–2005


  • Jeffrey M. Caterino MD,

    1. From the Departments of Emergency Medicine and Internal Medicine, The Ohio State University (JMC), Columbus, OH; the Duke University College of Medicine (SGW), Durham, NC; and the Department of Emergency Medicine, Massachusetts General Hospital (JAE, CAC), Boston, MA.
    Search for more papers by this author
  • Sarah Grace Weed,

    1. From the Departments of Emergency Medicine and Internal Medicine, The Ohio State University (JMC), Columbus, OH; the Duke University College of Medicine (SGW), Durham, NC; and the Department of Emergency Medicine, Massachusetts General Hospital (JAE, CAC), Boston, MA.
    Search for more papers by this author
  • Janice A. Espinola MPH,

    1. From the Departments of Emergency Medicine and Internal Medicine, The Ohio State University (JMC), Columbus, OH; the Duke University College of Medicine (SGW), Durham, NC; and the Department of Emergency Medicine, Massachusetts General Hospital (JAE, CAC), Boston, MA.
    Search for more papers by this author
  • Carlos A. Camargo, Jr MD, DrPH

    1. From the Departments of Emergency Medicine and Internal Medicine, The Ohio State University (JMC), Columbus, OH; the Duke University College of Medicine (SGW), Durham, NC; and the Department of Emergency Medicine, Massachusetts General Hospital (JAE, CAC), Boston, MA.
    Search for more papers by this author

  • Presented at the American College of Emergency Physicians Scientific Assembly, Chicago, IL, October 27–30, 2008.

  • This study was supported in part by a Dennis W. Jahnigen Career Development Award sponsored by the American Geriatrics Society, John A. Hartford Foundation, and Atlantic Philanthropies.

Address for correspondence and reprints: Jeffrey M. Caterino, MD; e-mail:


Objectives:  Given reported increases in antibiotic resistance among elders with urinary tract infection (UTI) and pyelonephritis, the authors identified national rates and trends in emergency department (ED) trimethoprim-sulfamethoxazole (TMP-SMX) and fluoroquinolone prescribing for older adults from 1996 to 2005.

Methods:  This was a retrospective analysis utilizing the ED component of the 1996–2005 National Hospital Ambulatory Medical Care Survey (NHAMCS). The authors included NHAMCS ED entries aged ≥18 years with a diagnosis of UTI or pyelonephritis; pregnancy was excluded. Records were divided into 18–64 years (“adults”) and ≥65 years (“elders”). Primary outcome measures were prescription of TMP-SMX monotherapy, fluoroquinolone monotherapy, and combination therapy with two or more antibiotics. Estimated visit totals and rates were calculated and trends analyzed.

Results:  From 1996 to 2005, there were 5 million elder ED visits for UTI or pyelonephritis. Approximately 9.4% (95% confidence interval [CI] = 7.9% to 11%) of elders received TMP-SMX monotherapy with rates decreasing over time (p-value for trend = 0.031). Overall, 35% (95% CI = 32% to 38%) of elders received fluoroquinolone monotherapy, which increased from 21% (95% CI = 14% to 27%) in 1996 to 45% (95% CI = 39% to 50%) in 2005 (p-value for trend < 0.001). Therapy with a fluoroquinolone plus a second antibiotic was used in only 4.2% (95% CI = 3.1% to 5.3%) of older patients.

Conclusions:  From 1996 to 2005, TMP-SMX monotherapy in elder ED patients decreased while fluoroquinolone therapy increased. The majority of older patients receiving fluoroquinolone therapy received a single agent. Given the continued prevalence of monotherapy for elder ED patients with UTI or pyelonephritis, antibiotic resistance patterns in these patients should be better characterized to ensure institution of appropriate empiric therapy.

In 2005, there were an estimated 16.7 million visits by patients ≥65 years of age to U.S. emergency departments (EDs), accounting for 14.5% of all visits.1 Among all ED patients, 4.6% received a genitourinary diagnosis and 1.8 million were diagnosed with urinary tract infection (UTI). This places UTI among the top 15 diagnoses annually given in the ED.1

Recent studies of the older adult population in a variety of settings, including the ED, have demonstrated a high prevalence of antibiotic resistant microorganisms among urinary tract pathogens.2–5 There is controversy over diagnosis of urinary infection in the absence of culture data, but emergency physicians (EPs) are usually required to choose empiric therapy without such information.6 Compared to younger adults, resistance rates are higher in elders.2,3 Elders are susceptible to infection with multidrug-resistant organisms, particularly in certain subpopulations, such as nursing home residents.2 Despite the potentially complex microbiologic spectrum of UTI and pyelonephritis in older ED patients, little data exist regarding current ED antimicrobial prescribing practices. Owing to increasing rates of trimethoprim-sulfamethoxazole (TMP-SMX) resistance, it has been suggested that TMP-SMX should not be used as empiric monotherapy in older patients.7–10 Rather, fluoroquinolones are the recommended agents for initial therapy. However, given increasing rates of resistance, even fluoroquinolone monotherapy may be insufficient in certain subgroups of elder patients.2

The objective of the present analysis was to identify national trends in ED antibiotic use for elders with both lower-tract UTI and upper-tract UTI (pyelonephritis) over the past decade. Specifically, we identified trends in TMP-SMX monotherapy and fluoroquinolone monotherapy. We then compared rates and trends between those aged 18 to 64 years and those 65 years and over. We hypothesized that there has been a downward trend in TMP-SMX prescribing in the elder population, accompanied by a concomitant increase in fluoroquinolone use for elders with UTIs. Additionally, we sought to identify emergence rates and trends of combination therapy, given the increasing resistance rates of antimicrobial resistance.


Survey Design

We performed a retrospective analysis of the ED component of the 1996–2005 National Hospital Ambulatory Medical Care Survey (NHAMCS) database. NHAMCS is a national probability survey of hospital outpatient and ED visits designed by the National Center for Health Statistics (NCHS) of the Centers for Disease Control and Prevention.11 The hospital’s institutional review board determined that the study did not require review as the use of deidentified registry data did not constitute human subjects research as defined in 45 CFR 46:102 (f).

Study Setting and Population

NHAMCS utilizes a four-stage probability sample based on visits to noninstitutional general and short-stay hospitals, excluding federal, military, and Veterans Administration (VA) hospitals, located in the 50 states and the District of Columbia.12,13 NHAMCS is conducted annually and covers geographic primary sampling units, hospitals within primary sampling units, EDs within hospitals, and patients within EDs. Trained hospital staff collected data during a randomly assigned 4-week data period for each of the sampled hospitals.14 Review of data collection was performed by a Bureau of Census field supervisor. Quality control checks were used. The nonresponse rate for most items was less than 5%, and error rates were less than 2% for items requiring medical coding.13 Diagnoses were coded by experienced personnel using the International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM). Included medications included those given in the ED, prescribed at discharge, or recommended by the EP. National estimates were obtained through use of a multistage estimation procedure with three basic components (inflation by reciprocals of the sampling selection probabilities, adjustment for nonresponse, and a population weighting ratio adjustment) and patient visit weights. A more detailed description of NHAMCS procedures is available for review at the NHAMCS Public Use Data Files Web page and in the technical notes section of each year’s NHAMCS Emergency Department Summary.15,16

For the present analysis, the study population included all ED visits by patients 18 years and older in the 1996–2005 NHAMCS database. Records were included if they had an ICD-9-CM code in any of the three diagnosis fields consistent with UTI or pyelonephritis (595.0; 595.89; 595.9; 590.xx; 599.0). Exclusion criteria included any patients with ICD-9-CM diagnosis codes related to pregnancy (630–677). Those aged 18–64 years were defined as “adults,” while those age 65 years or over were defined as “elders” or “older adults.”

Study Protocol and Measurements

We analyzed data by ED visit demographics, residency status (nursing home or other institution including rehabilitation facilities), and source of payment. Data regarding residency status were collected only during 2001 to 2005. Hospital data such as metropolitan statistical area (MSA) status of the hospital and U.S. region location (Northeast, Midwest, South, and West) were also examined. MSA and U.S. region categories represent standardized geographical divisions defined by the U.S. Bureau of the Census.

Our major outcomes of interest were rates of TMP-SMX and fluoroquinolone administration among ED visits with a UTI or pyelonephritis diagnosis. From 1996 to 2002, up to six medications were recorded on the survey per visit, while up to eight medications were recorded from 2003 to 2005.17,18 All medications are coded by published NCHS definitions that identify individual generic and trade names. In addition to individual medication codes, NHAMCS provides drug class codes based on the National Drug Code Directory for each medication. We identified ED patients administered TMP-SMX by including all medication codes for SMX/TMP combinations, including both generic and brand names (01017, 01163, 03321, 03430, 25338, 27835, 27840, 29843, 29888, 32433, 32438, and 35595). Fluoroquinolone use was identified by including the drug class code for quinolones/derivatives (0357). Ophthalmologic and otic preparations were excluded. Additional antibiotic use was identified by the presence of an additional drug class code for another antimicrobial agent. These codes included (0346, 0347, 0348, 0349, 0350, 0351, 0352, 0354, 0355, 0357, and 0358). Nitrofurantoin use was coded by presence of code 13105, 93090, or 18130. TMP-SMX and fluoroquinolone monotherapy was defined as administration of one of these agents plus the absence of a second drug class code for an additional antimicrobial agent. For diagnosis of UTI or pyelonephritis, numbers of lone diagnosis (no other diagnosis codes present) and primary diagnosis and presence of an additional ICD-9 diagnosis code for infection in addition to the UTI were calculated.

Data Analysis

We performed all analyses using STATA 10.0 (StataCorp, College Station, TX). To account for the complex four-stage sampling frame, all analyses were performed using survey design variables and appropriate survey commands in STATA. We determined point estimates and 95% confidence intervals (CIs) for visit characteristics overall and stratified by age (18–64 years vs. ≥65 years). Additionally, we calculated annual rates of antibiotic prescribing among visits by age. Estimates with standard errors >30% or based on less than 30 observations are considered unreliable and therefore are not reported. If necessary to obtain valid estimates, years were grouped by 2-year intervals. Trends were computed across years using weighted linear and logistic regression.


From 1996 to 2005, there were an estimated 786 million (95% CI = 718 to 853 million) all-cause visits to the ED by adults. Older adults accounted for approximately 158 million (95% CI = 144 to 172 million) or 20% of the total visits. The number of ED visits by elders was stable over this time period (p-value for trend = 0.23).

There were approximately 26.3 million visits for patients aged ≥18 years diagnosed with either UTI or pyelonephritis during the study period, including 7.5 million (95% CI = 6.7 to 8.4 million) elders. An estimated 4.8% (95% CI = 4.5% to 5.1%) of all elders were diagnosed with a UTI or pyelonephritis, compared to 3.0% (95% CI = 2.9% to 3.1%) of adults age 18–64 years. The number of UTI and pyelonephritis visits in older patients increased over the study period from 0.56 million (95% CI = 0.41 to 0.70 million) in 1996 to 0.86 million (95% CI = 0.67 to 1.1 million, p-value for trend < 0.001) in 2005. Characteristics of the study population stratified by age are shown in Table 1. Data on nursing home status were present only for 2001–2005. For these years, 23% of elders were documented as residing in a nursing home or other institution. Overall, admission rates were 8% (95% CI = 7% to 9%) for adults 18–64 years and 44% (95% CI = 41% to 47%) for elders.

Table 1. 
Characteristics of All Patients Diagnosed with a UTI in EDs, 1996–2005*
 Entire PopulationAge 18–64 YearsAge ≥65 Years
N (95% CI)(in Millions)% of Totaln (95% CI)(in Millions)% of Groupn (95% CI)(in Millions)% of Group
  1. CI = confidence interval; MSA = metropolitan statistical area; UTI = urinary tract infection.

  2. *Numbers may not equal 100 due to rounding.

  3. †Nursing home status is only available for 2001–2005 and includes other institutions. Percentages are based on those years.

  4. ‡CIs cannot be obtained for samples with fewer than 30 observations.

Overall26.3(23.9, 28.6)10018.7(17.1, 20.4) 7.5(6.7, 8.4) 
 Female21.6(19.6, 23.6)8216.4(14.9, 17.9)885.2(4.6, 5.9)69
 Male4.7(4.2, 5.2)182.4(2.1, 2.7)132.3(2.0, 2.6)31
 White 19.8(17.9, 21.8)7513.5(12.1, 14.8)726.4(5.6, 7.1)85
 African American5.7(4.9, 6.4)224.7(4.0, 5.3)251.0(0.8, 1.2)13
 Other0.8(0.6, 0.9)30.6(0.5, 0.7)30.2(0.1, 0.2)3
 Hispanic2.5(2.1, 3.0)102.2(1.8, 2.5)120.4(0.3, 0.4)5
 Northeast4.5(3.1, 5.4)183.1(2.6, 3.6)161.5(1.1, 1.9)20
 Midwest6.2(5.1, 7.3)244.3(3.6, 5.1)231.9(1.4, 2.3)25
 South10.5(8.8, 12.2)407.9(6.6, 9.1)422.6(2.1, 3.2)35
 West5.0(4.2, 5.9)193.5(2.9, 4.1)191.5(1.2, 1.8)20
 In MSA20.7(18.4, 23.0)7915.0(13.3, 16.7)805.7(4.9, 6.4)76
 Not in MSA5.6(3.8, 7.4)213.7(2.6, 4.9)201.9(1.2, 2.6)25
Nursing home resident†
 Yes1.2(0.9, 1.4)80.2(0.1, 0.2)21.0(0.8, 1.2)23
 No12.8(11.7, 13.9)869.9(9.0, 10.8)942.9(2.6, 3.2)68
 Unknown0.8(0.7, 1.0)60.5(0.4, 0.6)40.4(0.3, 0.5)9
Source of payment
 Private7.4(6.6, 8.2)286.5(5.8, 7.3)350.8(0.7, 1.0)11
 Public10.2(9.0, 11.3)54.6(4.1, 5.2)245.5(4.8, 6.2)73
 Self-pay3.7(3.2, 4.2)143.6(3.1, 4.1)190.1(0.0, 0.1)1
 Other/unknown1.9(1.6, 2.1)71.6(1.4, 1.9)80.2(0.2, 0.3)3
Seen by
 Staff physician19.6(17.8, 21.4)7514.2(12.9, 15.6)765.4(4.8, 6.0)72
 Other physician2.4(1.9, 3.0)91.4(1.0, 1.7)71.1(0.8, 1.3)15
 Resident/intern2.4(2.0, 2.8)91.7(1.4, 2.0)90.7(0.6, 0.9)9
 Physician’s assistant1.4(1.1, 1.7)51.1(0.9, 1.4)60.3(0.2, 0.4)4
 Nurse practitioner0.4(0.3, 0.5)20.3(0.2, 0.4)20.9(nc)‡12

In 39% of cases, UTI or pyelonephritis was the only diagnosis given in the NHAMCS record. This number was greater in younger (45%, 95% CI = 43% to 47%) than older (25%, 95% CI = 22% to 28%) adults. UTI or pyelonephritis was the primary diagnosis in 65% of patients, including 68% (95% CI = 66% to 70%) of younger and 58% (95% CI = 55% to 61%) of older adults. Overall, 8.5% (95% CI = 7.6% to 9.4%) of patients had another associated infectious diagnosis, and there was no difference between age groups.

TMP-SMX Monotherapy

Overall, 23% (95% CI = 21% to 25%) of adults 18–64 years old with UTI or pyelonephritis received TMP-SMX monotherapy, while only 9.4% (95% CI = 7.9% to 11%) of visits among elders did. Owing to small numbers of visits in the sample for elders (<30 per year), yearly estimates of TMP-SMX monotherapy could not be calculated. Figure 1 demonstrates the age-stratified percentages of TMP-SMX monotherapy in 2-year intervals. The estimated percentage of ED adults 18–64 years receiving TMP-SMX monotherapy decreased from 32% (95% CI = 28% to 36%) in 1996–1997 to 18% (95% CI = 13% to 22%) in 2004–2005 (p-value for trend < 0.001). Percentages of TMP-SMX monotherapy among elders decreased from 14% (95% CI = 9% to 19%) to 6% (95% CI = 4% to 9%; p-value for trend = 0.03).

Figure 1.

 Percentage receipt of TMP-SMX monotherapy from 1996 to 2005 in adults and elder ED patients. ED = emergency department; TMP-SMX = trimethoprim-sulfamethoxazole.

Fluoroquinolone Monotherapy

Overall, 26% (95% CI = 25% to 28%) of adults 18–64 years old received fluoroquinolone monotherapy, whereas 35% (95% CI = 32% to 38%) of older patients did. Prescribing of fluoroquinolone monotherapy increased in both groups over the study period (both p-value for trend < 0.001). Owing to small numbers of visits in the sample for elders (<30 per year) in two of the calendar years (1995, 1996), yearly estimates of fluoroquinolone monotherapy could not be calculated. Figure 2 demonstrates the age-stratified percentages of fluoroquinolone monotherapy in 2-year intervals. Rates of fluoroquinolone monotherapy in 2-year intervals increased among the adult 18- to 64-year-old population from 10% (95% CI = 7% to 13%) in 1996–1997 to 35% (95% CI = 30% to 39%) in 2004–2005 (p-value for trend < 0.001). In older adults, percentages in 2-year intervals of fluoroquinolone monotherapy increased from 21% (95% CI = 14% to 27%) to 45% (95% CI = 39% to 50%; p-value for trend < 0.001). Additionally, 39% of nursing home patients with a UTI or pyelonephritis were given fluoroquinolone monotherapy from 2001 to 2005.

Figure 2.

 Percentage receipt of fluoroquinolone monotherapy from 1996 to 2005 in adults and elder ED patients. ED = emergency department.

Combination and Other Antimicrobial Therapy

Trimethoprim-sulfamethoxazole was rarely used in combination with other antibiotics. Only 3.2% (95% CI = 2.5% to 3.8%) of adults and 1.1% (95% CI = 0.5% to 1.6%) of elders received TMP-SMX plus another antibiotic agent. Fluoroquinolones were similarly rarely used in combination with other antibiotics. A total of 5.0% (95% CI = 4.2% to 5.8%) of those 18–64 years and 4.2% (95% CI = 3.1% to 5.3%) of those ≥65 years received dual-antibiotic therapy including a fluoroquinolone over the study period. Both age groups exhibited a trend toward increasing therapy with a fluoroquinolone plus a second antibiotic (p-value for trend 0.01 and 0.03, respectively). Most estimates for individual years were considered unreliable using NHAMCS criteria due to large relative standard errors in the calculation. However, yearly rates for dual prescribing with fluoroquinolone were <7.5% for all years in both age groups.

Examining the monotherapy and combined therapy groups jointly, an estimated 31% (95% CI = 30% to 33%) of adults and 39% (95% CI = 36% to 42%) of elders received a fluoroquinolone either with or without another antibiotic. Visit rates for receiving any fluoroquinolone increased significantly for both adults, from 13% (95% CI = 9.3% to 17%) in 1996 to 44% (95% CI = 39% to 50%) in 2005 (p-value for trend < 0.001), and elder populations, from 26% (95% CI = 17% to 36%) in 1996 to 52% (95% CI = 44% to 61%) in 2005 (p-value for trend < 0.001; Figure 3).

Figure 3.

 Percentage receipt of any fluoroquinolone from 1996 to 2005 in adults and elder ED patients. ED = emergency department.

Administration of antimicrobials other than TMP-SMX or fluoroquinolones did not differ in general between adults and elders. Among the entire population, regardless of receipt of TMP-SMX or fluoroquinolones, rates were as follows. Penicillins were given to 4.4% (95% CI = 3.8% to 5.1%) of all UTI patients in the ED. Cephalosporins were given to 14.3% (95% CI = 12.9% to 15.6%), lincosamides or macrolides to 1.8% (95% CI = 1.4% to 2.2%), aminoglycosides to 1.3% (95% CI = 1.0% to 1.7%), and nitrofurantoin to 4.4% (95% CI = 3.7% to 5.0%). Nitrofurantoin was administered to 5.4% of adults ages 18–64 years (95% CI 4.6 to 6.3%) and 1.8% of elders (95% CI = 1.1% to 2.5%).


The objective of our analysis was to describe TMP-SMX and fluoroquinolone treatment patterns for older patients diagnosed with either upper or lower UTI in U.S. EDs from 1996 to 2005. We found that almost 5% of the over 15 million patients ≥65 years of age visiting U.S. EDs each year are diagnosed with a UTI or pyelonephritis, a greater percentage than that of other adults (3%). This corresponds to an estimated 7.5 million ED visits for elders with UTI or pyelonephritis over the study period. There are three main findings of the current study. First, between 1996 and 2005, rates of prescribing TMP-SMX for these conditions significantly decreased in both groups. Second, there has been a concomitant significant rise in fluoroquinolone prescribing in both age groups, with more elders receiving fluoroquinolones than younger adults. Third, use of more than one antibiotic in elder ED patients with UTI or pyelonephritis is uncommon.

The trends in ED antibiotic prescribing that we report are consistent with two previous studies using NHAMCS data in ambulatory patients.19,20 Huang and Stafford19 analyzed the non-ED portion of NHAMCS from 1989 to 1998. They reported an overall decline in TMP-SMX prescribing to women from 48% to 28% and an increase in fluoroquinolone prescribing from 19% to 29%. Patients <45 years of age were less likely to receive fluoroquinolones. In the second study, Kallen et al. examined women with UTIs using NHAMCS data from 2000 to 2002.20 ED visits accounted for 18% of the study’s population. Most patients received fluoroquinolones in this period (44%), and patients ≥30 years of age were more likely to get a fluoroquinolone. Our overall rate of TMP-SMX monotherapy is somewhat lower than that of these studies, likely due to the more complicated nature of ED patients. Our rates of fluoroquinolone prescribing were similar to these previous studies. The overall monotherapy rates of approximately 26% for young adults and 35% for elders are similar to the 29%–44% range found in the ambulatory patient studies.

Choice of an appropriate antibiotic regimen in older patients is particularly important because of the high burden of morbidity and mortality that accompanies UTI in this population. In one single-center chart review study, over 70% of elders diagnosed with UTI in the ED were admitted (compared to an overall elder admission rate of 50%) with an average length of stay of 5.4 days, 6% mortality, and 13% intensive care unit admission rate.21 This is consistent with the admission rate of 44% found among our elder NHAMCS cohort. Bacteremic UTI was associated with 33% mortality in one study of admitted older patients.22

Selection of antimicrobial therapy is based on several factors. These include physician preference/familiarity/comfort, patient allergies, expected resistance patterns, side effects, cost, and ease of administration. Studies seem to indicate that concerns over resistance are not the only factor driving the shift to fluoroquinolones, as higher rates of fluoroquinolone use do not directly track with higher resistance either temporally or geographically.20 As a result, additional factors may be causing the shift away from TMP-SMX use, including rising familiarity and comfort with fluoroquinolone use between 1996 and 2005.20 Among the side effects of fluoroquinolones are delirium, QTc prolongation, tendon rupture, and interaction with warfarin.23,24 However, fluoroquinolones may be better tolerated, are often administered once daily, and become more cost-effective as rates of TMP-SMX resistance rise.25 Despite their side effect profile, fluoroquinolones are reasonable choices for UTI and pyelonephritis, particularly if high rates of resistance are expected.2

Complicating empiric antibiotic choice in elders is their risk of drug-resistant organisms, which is greater than in younger adults.3 In 2003–2004, rates of TMP-SMX resistance in outpatient samples were 15.2% in those aged 15–49 years and 22%–24% in those 50 years and over. Fluoroquinolone resistance was <5% in patients aged 15–49 years, but it increased to 11.4% to 13.9% in patients 50 years and over. Elder ED patients are at even greater risk of antibiotic resistance than outpatients, as risk factors for drug resistance are particularly common. These factors include residence in a long-term care facility, presence of an indwelling catheter, structural genitourinary abnormality, recent antibiotic use, and baseline decreased functional status.2,4,5,26–28 Given these risk factors, ED rates of resistance may be higher than those found in other ambulatory populations.

Data on uropathogen resistance patterns specific to the ED elder population are limited. Wright et al.2 reported on ED urine culture data collected in the mid-1990s. Resistance was much greater in elder ED patients than in healthy young females and men <50 years. Forty-five percent of older adults had a multidrug-resistant organism, 46% had isolates resistant to TMP-SMX, and 11% had an organism resistant to fluoroquinolones. Elder patient characteristics had a strong relationship with resistance rates. For example, 61% of residents of long-term care facilities had multidrug-resistant organisms, 68% TMP-SMX resistance, and 41% fluoroquinolone resistance. Healthy elders, in contrast, had 0% fluoroquinolone resistance.2

Given the paucity of recent data, current antibiotic resistance patterns in elder ED patients with UTI or pyelonephritis are unclear. Considering the large number of risk factors for resistance found among ED patients, we would expect resistance rates to be at least as high as in outpatient cohorts. We would also expect rates to be at least as high as data obtained in the mid-1990s. Current ED antibiotic prescribing patterns appear to take these factors into account given the demonstrated decline in TMP-SMX and rise in fluoroquinolone use. However, in our study, ED patients were unlikely to receive combination antimicrobial therapy. Less than 5% of older adults received a second antibiotic in addition to a fluoroquinolone and less than 2% received a second antibiotic in addition to TMP-SMX. In light of the resistance patterns previously discussed, there is a concern that fluoroquinolone monotherapy may not be sufficient in some elder patients. In years with data available, about 23% of elders diagnosed with UTI or pyelonephritis were from nursing homes. These are patients at risk of fluoroquinolone and TMP-SMX resistance who may benefit from combination drug therapy.

These prescribing patterns indicate several areas for concern and future study. Given that the majority of elder patients receiving fluoroquinolones are receiving single drug therapy, there is a need for further study to clarify resistance patterns and risk factors for multidrug-resistant urinary pathogens in older ED patients. An important factor in the shift away from TMP-SMX in older adults may be the recognition that TMP-SMX resistance is increasing. However, the small amount of data available suggest a further shift may be needed.2 Fluoroquinolone resistance rates in elders are higher than in young adults. The increasing fluoroquinolone resistance in nursing home patients in particular may not be readily apparent to ED physicians, as they affect only a subgroup of the elder population.2 Antimicrobial resistance rates are increased by multiple risk factors (such as nursing home status, indwelling catheters, recent antibiotic use), and these are becoming more common in ED elders.26 Given these increases in risk factors, fully understanding their contributions to the patterns of antimicrobial resistance seen in the ED is necessary to ensure adequate therapy. For critically ill patients, the importance of empiric therapy that is effective against causative organisms is widely recognized.29 In sepsis, failure to administer an antibiotic with activity against the causative organism increases morbidity and mortality.30–32 Elders with UTI represent a potentially severely ill cohort with high admission rates.21 Fifteen percent of patients admitted from the ED with UTI have bacteremia, and mortality in older bacteremic UTI patients is 33%.22,33 Knowledge of likely pathogens and susceptibility patterns has been shown to improve the accuracy of empiric prescribing.29,32,34–38 Given the prevalence of monotherapy in this population, studies of antimicrobial resistance in older ED patients with UTI should be undertaken to fully explore overall resistance patterns, specific risk factors for resistance, and situations where dual antibiotic therapy may be indicated empirically.


We found that a substantial number of visits with a UTI or pyelonephritis diagnosis did not receive either TMP-SMX or a fluoroquinolone. Our estimates, however, are consistent with other NHAMCS studies.19,20 We chose to concentrate on these two antibiotic classes as they are generally the two most commonly prescribed antibiotics for UTI or pyelonephritis in ED patients. The three studies of UTI using NHAMCS data, including ours, have found that approximately 25% to 30% of patients did not have any antibiotic charted in the database. This is likely due to failure to capture all antibiotics for which prescriptions were written. If some prescriptions were missed, we would expect that this would increase the percentage of TMP-SMX monotherapy as well as fluoroquinolone use. NHAMCS in the time period studied inconsistently documented prescriptions written to be filled after the ED visit.

NHAMCS does not provide data on urine culture results. As a result, we were unable to identify the presence of positive urinary cultures and instead relied on clinical diagnosis by EPs. Difficulty with interpreting urine dipstick results due to poor sensitivity and specificity has been described.6 Also, no consensus exists on the appropriate colony count for consideration of an infection in elders.39 Also complicating diagnosis is the problem of asymptomatic bacteriuria, which is present in a large number of elder patients, and should not be treated with antimicrobial therapy.40 As a result, vague symptoms may not be due to active infection even in the presence of bacteriuria. These factors complicate diagnosis of true infection in the ED due to lack of an available criterion standard test at the time of ED diagnosis.

There is also debate over the significance of in vitro urinary pathogen resistance to TMP-SMX and fluoroquinolones.2,20 Some have argued that resistance is overestimated due to the preferential collection of cultures from high-risk patients.20 Others have noted that high urinary concentrations may allow drugs to overcome in vitro resistance.2 However, several authors have noted decreased success rates in patients with resistant UTIs who were treated with these antibiotics.2,41–43 For example, only 50% of infections caused by bacteria resistant to TMP-SMX were cleared after TMP-SMX treatment in one study.43 Additional clarification of the true effect of in vitro resistance on clinical outcome is warranted to help guide antibiotic prescribing.

Finally, we were unable to account for individual patient factors in choosing one antibiotic over others and thus are unable to determine the appropriateness of specific therapeutic decisions or risk factors for drug resistance. However, the persistence of fluoroquinolone monotherapy we have found among elders indicates that further study is warranted to identify if such appropriate reasons exist. We also based our analysis in real-world ED practice where the majority of antibiotic decisions are made empirically.


Over 7 million elder patients were diagnosed with UTI or pyelonephritis in U.S. EDs between 1996 and 2005. The rate of TMP-SMX monotherapy prescribed to these older adults decreased over the study period, while the rate of fluoroquinolone therapy increased. The majority of elder patients receiving fluoroquinolone therapy receive a single agent. Previous reports have indicated high levels of fluoroquinolone resistance in certain elder subpopulations. As a result, further study is warranted to characterize the appropriateness of fluoroquinolone monotherapy in older ED patients.