Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Trends and impact of antimicrobial resistance on older inpatients with urinary tract infections (UTIs): A national retrospective observational study

  • Hoa Q. Nguyen ,

    Contributed equally to this work with: Hoa Q. Nguyen, Nga T. Q. Nguyen

    Roles Conceptualization, Formal analysis, Methodology, Writing – original draft, Writing – review & editing

    Affiliation School of Pharmacy, Queen's University Belfast, Belfast, United Kingdom

  • Nga T. Q. Nguyen ,

    Contributed equally to this work with: Hoa Q. Nguyen, Nga T. Q. Nguyen

    Roles Conceptualization, Formal analysis, Methodology, Software, Visualization, Writing – original draft, Writing – review & editing

    Affiliation Centre for Public Health, School of Medicine, Dentistry and Biomedical Sciences, Queen's University Belfast, Belfast, United Kingdom

  • Carmel M. Hughes ,

    Roles Supervision, Writing – review & editing

    ‡ These authors also contributed equally to this work.

    Affiliation School of Pharmacy, Queen's University Belfast, Belfast, United Kingdom

  • Ciaran O’Neill

    Roles Conceptualization, Methodology, Software, Supervision, Writing – review & editing

    ‡ These authors also contributed equally to this work.

    Affiliation Centre for Public Health, School of Medicine, Dentistry and Biomedical Sciences, Queen's University Belfast, Belfast, United Kingdom

Trends and impact of antimicrobial resistance on older inpatients with urinary tract infections (UTIs): A national retrospective observational study

  • Hoa Q. Nguyen, 
  • Nga T. Q. Nguyen, 
  • Carmel M. Hughes, 
  • Ciaran O’Neill


Urinary tract infections (UTIs) are one of the most common infections in older people and are associated with increased morbidity and mortality. UTIs are also associated with increased risk of antimicrobial resistance (AR). This study examined changes in AR among older inpatients with a primary diagnosis of UTIs in the United States over an 8-year period and the impact of AR on clinical outcomes and hospital costs. Data were obtained from the longitudinal hospital HCUP-NIS database from 2009 to 2016 for inpatient episodes that involved those aged 65+ years. The ICD-9 and ICD-10 codes were used to identify episodes with a primary diagnosis of UTIs, comorbidities, AR status and age-adjusted Deyo-Charlson comorbidity index (ACCI) for the patient concerned. Weighted multivariable regression was used to examine the impact of AR on all-cause inpatient mortality, discharge destination, length of stay and hospital expenditures, adjusted for socio-demographic and clinical covariates. The proportion of admissions with AR increased, from 3.64% in 2009 to 6.88% in 2016 (p<0.001), with distinct patterns for different types of resistance. The likelihood of AR was higher in admissions with high ACCI scores and admissions to hospitals in urban areas. Admissions with AR were more likely to be discharged to healthcare facilities (e.g. care homes) compared to routine discharge (OR 1.81; 95%CI, 1.75–1.86), had increased length of stay (1.12 days; 95%CI, 1.06–1.18) and hospital costs (1259 USD; 95%CI, 1178–1340). Resistance due to MRSA was specifically associated with increased hospital mortality (OR 1.33; 95%CI, 1.15–1.53). Our findings suggest that the prevalence of AR has increased among older inpatients with UTIs in the USA. The study highlights the impact of AR among older inpatients with a primary diagnosis of UTIs on clinical outcomes and hospital costs. These relationships and their implications for the care homes to which patients are frequently discharged warrant further research.


Antimicrobial resistance (AR) is recognized as a global threat to health and healthcare systems. AR could lead to negative impacts on patient outcomes such as increased length of hospital stay, functional decline, increased healthcare expenditure and all-cause mortality [1]. In the United States of America (USA), it is calculated that approximately two million people suffer from infections which are resistant to antimicrobials each year and at least 23,000 die of these conditions [2]. By 2050, it is estimated that AR will trigger 10 million deaths worldwide every year [3]. In addition, AR has a massive economic impact; in the USA, for example it is estimated to cost approximately $2.2 billion annually and adds $1,383 for every patient treated [4]. The challenge presented by AR has also increased over time with the appearance of new resistant strains [2,5].

Older people are more susceptible to AR due to physiological changes and comorbidity [6]. This population has potentially higher risks of developing AR-related illness due to higher exposure to infection from hospital and institutional settings [7,8]. Long-term care facilities for older people may also increase exposure to AR because they have been recognized as reservoirs of resistant bacteria and have high rates of antimicrobial prescribing [911]. In the USA, the percentage of older people admitted to hospital with resistant infections increased by 48.8% from 1997 to 2006 [12]. The population of older persons is expected to increase globally to 1.4 billion by 2030 [13]; therefore, AR may become a key consideration in the management of older people’s care in the future.

Urinary tract infections (UTIs) are one of the most common infections in older people and are associated with increased morbidity and mortality [7,8]. Difficulty in diagnosis of UTIs among older people may lead to complications such as bacteremia if untreated [14] or Clostridium difficile infection if treated inappropriately [15]. High rates of AR have also been recorded amongst patients with UTIs [1618]. This may contribute to poor outcomes of treating older patients with UTIs.

There is a paucity of evidence in respect of the impact of AR on the treatment and outcomes of older patients with UTIs in an inpatient setting. This study aimed to examine changes in AR over time and its impact on clinical outcomes and hospital costs among older people hospitalized with UTIs in the USA.

Materials and methods

Data source

We conducted a retrospective observational study using the National (Nationwide) Inpatient Sample—Healthcare Cost and Utilization Project (HCUP-NIS), Agency for Healthcare Research and Quality (AHRQ). The HCUP-NIS is the largest available all-payer inpatient healthcare administrative dataset in the USA corresponding to an approximately 20 percent stratified sample of all discharges from community hospitals but excluding rehabilitation and long-term acute care hospitals [19]. Each record includes patient demographic details, diagnoses, procedures and other information associated with a single hospital admission. Data in relation to medication use, including antimicrobials, and microbiological results were not available; however, administrative datasets without such data have the potential to explore the impact of AR [12,20]. Due to the de-identified and publicly-available nature of HCUP-NIS data, this study was considered exempt by the School Research Ethics Committee/Institutional Review Board at Queen’s University Belfast.

Case definition

We obtained HCUP-NIS data from 2009 to 2016. The study cohort was restricted to episodes of care that involved those aged 65 years and over, the age commonly used to define older persons in the USA [21]. The International Classification of Diseases 9th Revision and 10th Revision (ICD-9 and ICD-10) primary procedure and diagnostic codes were used to identify admissions with a primary diagnosis of UTIs. Following coding practice in the USA, ICD-9 codes were used for data before the fourth quarter of 2015, and ICD-10 codes were used for data from the fourth quarter of 2015 until the end of 2016. Cases were identified as admissions with a primary diagnosis of UTIs based on the definition of AHRQ Prevention Quality Indicator 12 for Urinary Tract Infections [22]. This definition comprises urinary tract infection, acute cystitis, cystitis, acute pyelonephritis, renal/perirenal abscess, pyeloureteritis cystica, pyelonephritis or pyonephrosis not specified as acute or chronic, and infection of kidney; it excludes cases with any codes for kidney/urinary tract disorder or transferred from another health care facility.

Infections due to Methicillin-resistant Staphylococcus aureus (MRSA) have been associated with high rates of mortality and prolonged hospital stay [23], and MRSA is the only resistant organism for which specific codes have been created by the ICD system since October 2008 [24]. Therefore, we decided to report resistance due to MRSA separately from other types of AR. Episodes with all types of AR, including beta-lactam resistance (BR), resistance due to MRSA, multidrug resistance (MR), quinolone resistance (QR), were identified using ICD codes (See S1 Appendix). We were also particularly interested in BR, resistance due to MRSA, MR, and QR as these types of resistance can develop in patients with UTIs [18] and have a major impact on patient outcomes [3,5]. AR was defined as having any ICD-9 V09.XX codes or diagnosis codes for MRSA (038.12, 041.12, 482.42), and equivalent ICD-10 codes. The ICD-9 codes were also used to define BR (V09.0, V09.1), QR (V09.5X), and MR (V09.81, V09.91), along with equivalent ICD-10 codes. An age-adjusted Deyo-Charlson Comorbidity Index (ACCI) was established using ICD codes to identify any of the following comorbidities for each admission: congestive heart failure, chronic pulmonary disease, cerebrovascular disease, diabetes mellitus with or without chronic complications, dementia, myocardial infarction, rheumatic disease, peripheral vascular disease, mild, moderate or severe liver disease, peptic ulcer disease, renal disease, hemiplegia or paraplegia, and HIV/AIDS. We used the Deyo-Charlson Comorbidity Index as this provided details on ICD-9 codes to define comorbidities, and applied age-adjusted version of Charlson Comorbidity Index as older patient data were used in this study [25,26]. Episodes with comorbidities were then weighted using the Deyo-Charlson algorithm and ACCI score calculated [25,26].

Study variables

HCUP-NIS provided a range of socio-demographic variables comprising insurance type (Medicare/Medicaid/Private/Others), ethnicity, sex (male/female), age at admission, median household income for patients’ ZIP code and location (broadly in terms of rurality). In order to control for hospital characteristics, we used the unique HCUP hospital number to link the core data to the hospital data. These variables comprised a hospital's ownership/control category (government/private owned), location (rural, urban), and its teaching status (non-teaching, teaching). Clinical variables included the comorbidity index ACCI (as calculated above), mode of presentation (elective/non-elective) and total number of procedures performed in the hospital. We also identified cases with sepsis, a severe complication of UTIs, and C. difficile infection, a common complication of long-term use of broad-spectrum antibiotics potentially due to UTIs treatment, using ICD codes (see S1 Appendix for more details). These variables were chosen because they were either reported to be associated with mortality [14,27] or found to be clinically relevant in the literature [2831].

Study outcomes

We first described the characteristics of admissions with UTIs, proportion of admissions with AR in general and with specific types of resistance (BR, resistance due to MRSA, MR, and QR). The four main outcomes associated with AR, comprising all-cause inpatient mortality, discharge destination, length of stay in the hospital, and hospital incurred costs, were measured to estimate the impact of AR [1]. Discharge destination included routine discharge (i.e. discharge to home, self-care, and court/law enforcement), and discharge to healthcare facilities (including long-term care facilities or care homes, short-term hospitals, home healthcare and other rehabilitation centers), excluding those who died at the hospital. The hospital-specific all-payer inpatient ‘cost-to-charge’ ratio tool provided by AHRQ-HCUP was used to convert discharges to hospital costs. Hospital costs were then adjusted for inflation using the personal consumption expenditure health component price index based on its ability to capture information on expenditures by all payers [32,33]. This is based on the assumption that average all-payer reimbursements or expenditures could be a proxy for underlying resource costs, a common practice used in literature [32].

Statistical analyses

Sample characteristics were described using median and interquartile range (IQR) for continuous variables with skewed distribution, mean and standard deviation (SD) for other continuous variables and proportions for categorical variables. In order to generate national estimates using HCUP-NIS data that span multiple years, all analyses were weighted by trend weight for data years prior to 2012 and by the discharge-level weight for data from 2012 and later as recommended by the AHRQ [19] unless specified otherwise. This practice reflects the major change in sampling method of HCUP-NIS in 2012 (changing from a 20% stratified sample of hospitals to a 20% national patient-level sample, with non-representative sampling across hospitals) [19].

Trends in antimicrobial resistance.

The incidence of UTIs as a primary diagnosis and proportion of admissions with AR, BR, resistance due to MRSA, MR, or QR were described on a monthly basis and weighted to provide national estimates. Multivariable logistic regression was used to examine factors affecting the likelihood of having AR, BR, MR, and QR. Explanatory variables used in these analyses comprised socio-demographic variables (insurance type, ethnicity, sex, age at admission, median household income, location, and year), and clinical variables (comorbidity score ACCI, hospital characteristics, mode of presentation) as suggested by the literature [2831].

Impact of antimicrobial resistance.

We estimated the impact of AR by examining the four outcomes: all-cause inpatient mortality, discharge to another healthcare facility, length of stay in the hospital, and hospital incurred costs. We examined factors associated with the likelihood of inpatient mortality and discharge to a healthcare facility among those admitted with a primary diagnosis of UTIs using weighted multivariable logistic regression. Models were adjusted for socio-demographic variables (insurance type, ethnicity, gender, age at admission, median household income, location), and clinical variables (comorbidity score ACCI, hospital teaching status, hospital location, hospital ownership, mode of presentation, number of procedures performed, C. difficile infection and sepsis status). Year was added as a predictor to control for variability over time in all regression models.

Similarly, based on the nature of count data and evidence of over-dispersion of length of stay in our data, we fitted weighted negative binomial regression models to examine the impact of AR, BR, resistance due to MRSA, MR, and QR on length of stay. Generalized linear models (GLM) were used to accommodate the continuous, positive and skewed nature of hospital cost data in the cost analysis. Akaike information criterion (AIC) and Bayesian information criterion (BIC) were used to assess the fit of the GLM model [34]. The link function and distribution family were jointly chosen using AIC and BIC while running a series of GLM models. Marginal effect analyses were used to estimate the incremental hospital costs attributable to AR, BR, resistance due to MRSA, MR, and QR.

Sensitivity analyses

The UTI definition of AHRQ Prevention Quality Indicator 12 for Urinary Tract Infections includes acute pyelonephritis, renal/perirenal abscess, pyeloureteritis cystica, pyelonephritis or pyonephrosis not specified as acute or chronic, and infection of kidney [22]. Such conditions have been recognized to result in serious complications or treatment failure [35]. Therefore, we conducted a set of sensitivity analyses with a refined sample without ICD codes for acute pyelonephritis, renal/perirenal abscess, pyeloureteritis cystica, pyelonephritis or pyonephrosis not specified as acute or chronic, and infection of kidney.

All analyses and data manipulations were performed using STATA version 15 (StataCorp LLC, College Station, TX).


Characteristics of the cohort

Over the 8-year period, a total of 546,305 eligible admissions with a primary diagnosis of UTIs were included in the study. Characteristics of episodes with primary UTIs are presented in Table 1. Most of the cohort was white female, with more White American and fewer females in the AR group and resistance due to MRSA group. Medicare was the predominant type of insurance, accounting for approximately 92% in all groups, while more people in the non-AR group had private insurance. Admissions with AR were more likely to be in the highest income group, those who were slightly younger, had more comorbidities, increased length of stay and higher hospital costs. Among subgroups of AR, resistance due to MRSA group had the highest cost while those with MR were the youngest and most likely to have private insurance.

Table 1. Characteristics of the pooled cohort from 2009–2016.

Trends in antimicrobial resistance

Fig 1 describes the total number of admissions with a primary diagnosis of UTIs and the proportion of cases with any type of AR by month. All data were weighted by HCUP-NIS weights to reflect the national estimates over the 8-year period. In the pooled data, while UTIs admission rate fluctuated over this period with a seasonal peak in the summer months, there was also an upward trend in AR with a gradual upward trend in MR and QR, a sharp increase in BR in 2015 and a downward trend in resistance due to MRSA. The proportion of AR among inpatient episodes those aged 65+ with a primary diagnosis of UTIs increased from 3.64% in 2008 to 6.88% in 2016 (p<0.001).

Fig 1. Trends in primary UTIs admission and antimicrobial resistance from 2009–2016.

Factors associated with the likelihood of AR, BR, MR and QR

As shown in Fig 2, factors associated with higher likelihood of having AR included higher ACCI score (OR 1.06, 95%CI, 1.06–1.07) and admissions to hospitals in urban areas (OR 1.07, 95%CI, 1.00–1.15). On the other hand, factors associated with lower likelihood of having AR included being Black American (OR 0.93, 95%CI, 0.89–0.98), being Hispanic (OR 0.94, 95%CI, 0.89–0.99), having private insurance (OR 0.76, 95%CI, 0.67–0.87), being female (OR 0.69, 95%CI, 0.67–0.71), having low income, and admission to a private (OR 0.93, 95%CI, 0.89–0.97) or teaching hospital (OR 0.91, 95%CI, 0.89–0.95) and increased age (OR 0.99, 95%CI, 0.99–0.99). A comparable pattern was observed among those with BR or MR that factors were associated with increased likelihood of having BR or MR included being Asians, Hispanics or other races, higher income, and higher ACCI. Notably, a decreased likelihood of having resistance due to MRSA was found among those who were Black, Hispanic Americans and other ethnicities, female, had lower ACCI, and private insurance. S2 Appendix provides further details on the likelihood of having different types of AR.

Fig 2. Factors associated with AR, BR, resistance due to MRSA, MR and QR.

Impact of antimicrobial resistance

All-cause inpatient mortality.

Over the 8-year period, the all-cause inpatient mortality rate was 1.42% for UTI admissions with associated AR, compared to 1.09% in the non-AR group (p < 0.001). In the multivariable logistic model, we found non-significant difference in the odds of death between the two groups during the observed inpatient episode (see Table 2). However, when examining specific types of AR, (BR, resistance due to MRSA, MR, and QR), we found a higher risk of in-patient death among those with resistance due to MRSA (OR 1.33, 95%CI, 1.15–1.53), but lower risk among those with BR (OR 0.48, 95%CI, 0.28–0.80), and MR (OR 0.67, 95%CI, 0.51–0.89). Other factors that were predictive of mortality were having C. difficile infection, having sepsis, being Black American, older age and higher ACCI. Moreover, having any types of insurance (Medicaid/Medicare/private insurance), being admitted to private hospital or hospital in an urban area reduced risk of inpatient mortality. Risk of inpatient mortality also fell over time.

Discharge destination.

Among survivors, 22.8% of patients with AR were routinely discharged, while 75.4% were transferred to other healthcare facilities (including long-term care facilities, short-term hospitals, home healthcare or other rehabilitation centers), compared to 35.8% and 62.8% in the non-AR group, respectively. Multivariable regression showed that admissions with any types of AR were more likely to be discharged to a healthcare facility than those without resistance. Among different types of AR, admissions with resistance due to MRSA showed the highest odds of being discharged to a health facility, which was over two times higher than those without resistance (OR 2. 21, 95%CI, 2. 10–2. 33) (see Table 2).

Length of stay in the hospital.

Compared to those without AR, the unadjusted additional length of stay for admissions with AR was 1. 77 days (p<0.001). After being adjusted for other covariates, admissions with AR were 1. 12 days (95%CI, 1. 06–1. 18) longer in hospital compared to non-AR. The predicted length of stay of those with BR, resistance due to MRSA, and MR were 0.74, 1.36, 0.83 days longer than those without AR, respectively (see Table 2). There was no significant difference in length of stay between admissions with QR and those without AR.

Hospital incurred costs.

Hospital discharges were converted to costs using cost-to-charge ratio provided by the HCUP and were inflated to the 2016 value. The unadjusted costs associated with hospitalization with AR were 2730 USD (95%CI, 2596–2864) higher than non-AR group (p<0.001). After estimating a series of GLM models, the AIC and BIC statistics supported the use of log link function and distribution family gamma. In the multivariable regression, admissions with AR, on average, consumed 1259 USD (95%:1178–1340) more than those without AR, though distinct patterns were observed in different types of AR (see Table 2).

Sensitivity analyses

Results of the sensitivity analyses on a refined sample without upper UTIs diagnosis confirmed our findings (see S3 Appendix). The general patterns of factors associated with increased/decreased likelihood of having AR, BR, resistance due to MRSA, MR, and QR remained comparable to the main analysis. Regarding impact of antimicrobial resistance, having AR was associated with increased likelihood of being discharged to a healthcare facility (OR 1.74, 95%CI, 1.69–1.80), increased hospital stays by 1.11 days (95%CI, 1.05–1.18), and increased hospital costs by 1236 USD (95%CI, 1152–1320). Resistance due to MRSA remained the costliest group with 1.35 days longer hospital stay, 2.12 times more likely to have healthcare discharge, and 1.29 times more likely to die in hospital.


While the prevalence of AR has been shown to increase over time [5], especially in hospital admissions with infections [12], to our knowledge, this is the first study to explore trends in the impact of AR on older inpatients with UTIs. Our findings suggest that AR among older inpatients with UTIs increased over time in the USA from 2009 to 2016. Socio-demographic factors associated with the likelihood of resistance were gender, age, median household income, type of insurance, ethnicity, ACCI score, and type of admitting hospital. Patients with AR were associated with an increased likelihood of discharge to other healthcare facilities, increased length of stay, hospital costs and all-cause inpatient mortality, although distinct patterns were evident between specific types of AR.

Within our study, the number of admissions with UTIs did not significantly change during the 8-year period of observation; however, there was a distinct seasonal pattern. Summer peaks among episodes with UTIs has been reported in a previous study [36]. Episodes with recorded AR did, however, increase substantially with distinct patterns for BR, QR, and MR except for resistance due to MRSA. The slight decrease over time in percentage of resistance due to MRSA has been also reported in literature [37,38]. However, this trend in our study may only be representative of inpatients aged from 65 with UTIs, and ICD codes for MRSA have been reported to have less sensitivity in detecting cases with MRSA [39]. The sharp rise of AR and BR from October 2015 may be partly due to introduction of the ICD-10 codes for infection with extended spectrum beta-lactamase resistance, which had not been recorded separately in the ICD- 9 codes. Although antimicrobial stewardship programs have been widely promoted since 2007 [40], AR prevalence seems to have risen continuously regardless, which suggests better and more intensive approaches should be implemented.

Risk factors associated with the likelihood of AR among inpatients with UTIs in our study are consistent with findings in previous studies. High ACCI scores indicated multiple comorbidities which are considered as a risk factor for antimicrobial resistance [31,41]. Male patients with UTIs were more likely to develop resistance compared to female patients [30,42]; this is perhaps unsurprising as older men are likely to develop UTI complications [14]. Additionally, we found that although male admissions were younger (mean age 79.8 ±7.8) compared to female (mean age 81.3±7.8) (p<0.001), they had higher ACCI scores (mean ACCI 5.9±2.4) compared to female (mean ACCI 5.6±2.1) (p<0.001). This could partly explain the lower risk of AR resistance among female in our cohort. Regarding ethnicities, we found that being Black or Hispanic is associated with lower odds of having AR compared to Whites. Compared to White Americans (mean 5.7±2.2), Black American had higher ACCI score (mean 6.0±2.3), and Hispanic Americans had lower ACCI score (mean 5.6±2.2) (p<0.001). While one may suspect higher ACCI among White admissions could contribute to their observed higher likelihood of AR, we suppose that there may be further unobserved factors such as the use of antimicrobials among this subgroup. In fact, White persons have been reported to be prescribed higher amounts of antimicrobials than other ethnicities including Black and Hispanic persons [28] which may help to explain the higher rates of AR among this group. The increased likelihood of BR and MR among Asian and Hispanic persons could be due to the fact that they were carriers of resistance organisms from Asian and Latin American countries where high rates of AR and antibiotic consumption have been reported [43].

We found that increased age was associated with a decreased likelihood of AR, BR and MR; nevertheless, the AR and MR group had higher ACCI and lower age compared to the non-AR group (Table 1). This could suggest that inpatients with AR or MR who had high ACCI may not have lived as long as those without AR. Admissions to hospitals in urban areas were also associated with an increased likelihood of having AR. This was perhaps due to a high density of people and hospitals in urban areas which may accelerate transmission of AR bacteria [44]. We also found that having health insurance (including private insurance) or having lower income was associated with lower risk of AR. A previous study also reported an increased AR-associated infection risk among patients without health insurance [12]. Additionally, although individual data of outpatient healthcare expenditure was not available, populations with high income may be more likely to pay for private healthcare, which has been associated with antimicrobial resistance [29,45], perhaps again related to easier access to antimicrobials. One might argue that people with high income would be likely to have private insurance, thus the AR pattern should be consistent between these groups. However, in HCUP-NIS, income-related data were constructed based on median household income for patients’ ZIP code; a degree of caution is, therefore, warranted in the interpretation of this result given the potential for ecological fallacy.

With regard to the burden of AR, episodes with recorded resistance due to MRSA were associated with an increased likelihood in inpatient mortality and an increase in length of hospital stay, which is consistent with findings of previous studies [46,47]. Interestingly, episodes with BR and MR were associated with lower risks of hospital death. However, we found episodes with BR and MR were 1.39 and 1.57 times more likely to be discharged to healthcare facilities (including long-term care facilities or care homes) compared to those without resistance. It should also be noted that ACCI scores in the BR and MR groups were both significantly higher than those in the non-AR group (Table 1). Although there was no clear reason to explain a reduction in inpatient observed mortality among inpatients with BR or MR, it is possible that patients with BR or MR were severely ill so perhaps the decision was made to discharge them to other long-term care facilities for end-of life care [48]. Higher number of admissions with AR sent to healthcare facilities at discharge may well contribute to the reservoirs of antimicrobial resistance in this setting [10,11] an issue that may be particularly concerning given the risk this presents to others. Our findings were also similar to those of previous studies which reported that AR infection increased length of hospital stay and hospital expenditure [46,47,49]. The addition to hospital costs associated with AR in our study was lower than those reported previously which was presented for specific resistant strains such as MRSA or Extended spectrum beta-lactamase resistance-producing Escherichia coli [46,47]. The full cost of infections should ideally include those associated with subsequent care, where we did not factor in the higher likelihood of transfer to another care facility among those with AR. The data available to us does not permit this follow-up however. We note though the recommendations made elsewhere in respect of the better estimation of economic impact [46,50]. Given these recommendations, we adopted a robust approach to utilize a representative sample of the population of interest as well as choosing GLM models to accommodate the distinct nature of costs data while taking into account potential confounding factors. These approaches represented our endeavor to increase the reliability of the estimated costs while noting the limitations imposed by the data.

The current study used a population-based data which represented 20% of national inpatient data in the USA. By pooling data in 8 years, the large number of episodes supported the study power. However, our study had several limitations. Chronological age (e.g. 65 years and over) is commonly used as cut-off to define older populations but its usefulness as an indicator for physiological and functional status of the older population has been disputed [51]. Frailty, a condition of functional decline among older people, has been reported to be associated with adverse inpatient outcomes, including hospital mortality [52,53]. We were unable to identify older inpatients with frailty using ICD codes for this administrative dataset. However, as proportions of frailty increase with age in the US older population [54], we used an age-adjusted version of CCI score as a proxy of health status and frailty of our cohort, and subsequently adjusted all models for this variable. We noted the development of an ICD-10 code for sarcopenia, but as this did not become available until September 2016, we were unable to exploit it in our analyses. We concur, however, that in future releases of HCUP data, the use of this code to distinguish between older persons in terms of frailty may benefit this research area.

Moreover, HCUP-NIS data are collected mainly for administrative purposes; thus, important clinical data such as medication use or microbiological results were not available. The cross-sectional nature of this study also limited our ability to make further suggestions regarding our findings. In addition, primary and secondary diagnoses and antimicrobial resistance were identified using ICD codes which have been reported with potentially missing cases [24]. Nevertheless, we endeavored to exploit available information in the data which could be surrogate measures for clinical data such as complications, types of resistance or discharge information. Medication use data during hospital admission may not be necessary to explore AR as patients may have developed resistance due to transmission or previous antimicrobial use. In order to reduce bias from ICD codes, we applied the definition of UTIs using ICD codes by AHRQ and used all codes that clearly defined resistant infections in ICD-9 and ICD-10 [22,24]. As noted, we cannot observe what happens to patients after discharge and cannot therefore state confidently as to longer term mortality or healthcare cost implications of AR. However, even with these caveats, our findings indicate a considerable impact of this issue on clinical and economic outcomes.


Our findings suggest that AR is increasing among older inpatients with UTIs. We also found distinct patterns in the relationship between specific types of AR and the likelihood of all-cause inpatient mortality, hospital discharge destination, length of stay, and hospital costs. These relationships and their implications for the care homes to which patients are likely discharged warrant further research.

Supporting information

S1 Appendix. ICD codes used to identify antimicrobial resistance.


S2 Appendix. Factors associated with the likelihood of antimicrobial resistance.



We are thankful to Luke Barry for supporting us access to HCUP-NIS data. We thank the reviewers for their thoughtful comments and efforts towards improving our manuscript.


  1. 1. Friedman ND, Temkin E, Carmeli Y. The negative impact of antibiotic resistance. Clin Microbiol Infect. 2016;22(5):416–22. pmid:26706614
  2. 2. Centers for Disease Control and Prevention (CDC). Antibiotic/antimicrobial resistance [Internet]. [cited 2019 Jun 30].
  3. 3. de Kraker MEA, Stewardson AJ, Harbarth S. Will 10 Million People Die a Year due to Antimicrobial Resistance by 2050? PLOS Med. 2016;13(11):e1002184. pmid:27898664
  4. 4. Thorpe KE, Joski P, Johnston KJ. Antibiotic-Resistant Infection Treatment Costs Have Doubled Since 2002, Now Exceeding $2 Billion Annually. Health Aff. 2018;37(4):662–9.
  5. 5. Harbarth S, Balkhy HH, Goossens H, Jarlier V, Kluytmans J, Laxminarayan R, et al. Antimicrobial resistance: one world, one fight! Antimicrob Resist Infect Control. 2015;4(1):49.
  6. 6. Smith P, Bennett G, Bradley S, Drinka P, Lautenbach E, Marx J, et al. SHEA/APIC Guideline: Infection Prevention and Control in the Long-Term Care Facility. Infect Control Hosp Epidemiol. 2008;29(9):785–814. pmid:18767983
  7. 7. Detweiler K, Mayers D, Fletcher SG. Bacteruria and Urinary Tract Infections in the Elderly. Urol Clin North Am. 2015;42(4):561–8. pmid:26475952
  8. 8. Mouton CP, Bazaldua O V, Pierce B, Espino DV. Common infections in older adults. Am Fam Physician. 2001;63(2):257–68. pmid:11201692
  9. 9. Peron EP, Hirsch AA, Jury LA, Jump RLP, Donskey CJ. Another Setting for Stewardship: High Rate of Unnecessary Antimicrobial Use in a Veterans Affairs Long-Term Care Facility. J Am Geriatr Soc. 2013;61(2):289–90. pmid:23405923
  10. 10. van Buul LW, van der Steen JT, Veenhuizen RB, Achterberg WP, Schellevis FG, Essink RTGM, et al. Antibiotic Use and Resistance in Long Term Care Facilities. J Am Med Dir Assoc. 2012;13(6):568.e1–568.e13.
  11. 11. Esposito S, Leone S, Noviello S, Ianniello F, Fiore M. Antibiotic resistance in long-term care facilities. New Microbiol. 2007;30(3):326–31. pmid:17802920
  12. 12. Mainous AG, Diaz VA, Matheson EM, Gregorie SH, Hueston WJ. Trends in Hospitalizations with Antibiotic-Resistant Infections: U.S., 1997–2006. Public Health Rep. 2011;126(3):354–60. pmid:21553664
  13. 13. United Nations—Department of Economic and Social Affairs—Population Division. World Population Ageing [Internet]. 2015 [cited 2019 Jun 30].
  14. 14. Gharbi M, Drysdale JH, Lishman H, Goudie R, Molokhia M, Johnson AP, et al. Antibiotic management of urinary tract infection in elderly patients in primary care and its association with bloodstream infections and all cause mortality: population based cohort study. BMJ. 2019;364:l525. pmid:30814048
  15. 15. Gopal Rao G, Patel M. Urinary tract infection in hospitalized elderly patients in the United Kingdom: the importance of making an accurate diagnosis in the post broad-spectrum antibiotic era. J Antimicrob Chemother. 2008;63(1):5–6. pmid:19022779
  16. 16. Lob SH, Nicolle LE, Hoban DJ, Kazmierczak KM, Badal RE, Sahm DF. Susceptibility patterns and ESBL rates of Escherichia coli from urinary tract infections in Canada and the United States, SMART 2010–2014. Diagn Microbiol Infect Dis. 2016;85(4):459–65. pmid:27306116
  17. 17. Frazee BW, Trivedi T, Montgomery M, Petrovic DF, Yamaji R, Riley L. Emergency Department Urinary Tract Infections Caused by Extended-Spectrum β-Lactamase–Producing Enterobacteriaceae: Many Patients Have No Identifiable Risk Factor and Discordant Empiric Therapy Is Common. Ann Emerg Med. 2018;72(4):449–56. pmid:29980462
  18. 18. Fagan M, Lindbæk M, Grude N, Reiso H, Romøren M, Skaare D, et al. Antibiotic resistance patterns of bacteria causing urinary tract infections in the elderly living in nursing homes versus the elderly living at home: an observational study. BMC Geriatr. 2015;15(1):98.
  19. 19. Agency for Healthcare Research and Quality. HCUP National (Nationwide) Inpatient Sample (NIS). Healthcare Cost and Utilization Project (HCUP) [Internet]. [cited 2019 Jun 30].
  20. 20. Schultz L, Lowe TJ, Srinivasan A, Neilson D, Pugliese G. Economic Impact of Redundant Antimicrobial Therapy in US Hospitals. Infect Control Hosp Epidemiol. 2014;35(10):1229–35. pmid:25203175
  21. 21. Ricardo Rodrigues, Manfred Huber, Giovanni Lamura. Facts and Figures on Healthy Ageing and Long-term Care [Internet]. Vienna; 2012 [cited 2019 Jun 30].
  22. 22. Agency for Healthcare Research and Quality. AHRQ Quality Indicators ICD-9-CM and ICD-10-CM/ PCS Specification Enhanced Version 5.0 Prevention Quality Indicators #12 Urinary Tract Infection Admission Rate [Internet]. 2015 [cited 2019 Jun 30].
  23. 23. de Kraker MEA, Davey PG, Grundmann H, group B study. Mortality and hospital stay associated with resistant Staphylococcus aureus and Escherichia coli bacteremia: estimating the burden of antibiotic resistance in Europe. PLoS Med. 2011/10/11. 2011;8(10):e1001104–e1001104. pmid:22022233
  24. 24. Burnham JP, Kwon JH, Babcock HM, Olsen MA, Kollef MH. ICD-9-CM Coding for Multidrug Resistant Infection Correlates Poorly With Microbiologically Confirmed Multidrug Resistant Infection. Infect Control Hosp Epidemiol. 2017;38(11):1381–3. pmid:28870271
  25. 25. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis. 1987;40(5):373–83. pmid:3558716
  26. 26. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45(6):613–9. pmid:1607900
  27. 27. Abou Chakra CN, Pepin J, Sirard S, Valiquette L. Risk Factors for Recurrence, Complications and Mortality in Clostridium difficile Infection: A Systematic Review. PLoS One. 2014;9(6):e98400. pmid:24897375
  28. 28. Olesen SW, Grad YH. Racial/ethnic disparities in antimicrobial drug use, United States, 2014–2015. Emerg Infect Dis. 2018;24(11):2126–8. pmid:30334733
  29. 29. Collignon P, Beggs JJ, Walsh TR, Gandra S, Laxminarayan R. Anthropological and socioeconomic factors contributing to global antimicrobial resistance: a univariate and multivariable analysis. Lancet Planet Heal. 2018;2(9):e398–405.
  30. 30. Colodner R, Rock W, Chazan B, Keller N, Guy N, Sakran W, et al. Risk factors for the development of extended-spectrum beta-lactamase-producing bacteria in nonhospitalized patients. Eur J Clin Microbiol Infect Dis. 2004;23(3):163–7. pmid:14986159
  31. 31. Laudisio A, Marinosci F, Gemma A, Bartoli IR, Montenegro N, Incalzi RA. The Burden of Comorbidity Is Associated with Antibiotic Resistance Among Institutionalized Elderly with Urinary Infection: A Retrospective Cohort Study in a Single Italian Nursing Home Between 2009 and 2014. Microb Drug Resist. 2016;23(4):500–6. pmid:27525808
  32. 32. Dunn A, Grosse SD, Zuvekas SH. Adjusting Health Expenditures for Inflation: A Review of Measures for Health Services Research in the United States. Health Serv Res. 2016;53(1):175–96. pmid:27873305
  33. 33. FRED Economic Data. Personal consumption expenditures: Services: Health care (chain-type price index) [Internet]. 2018 [cited 2019 Jun 30].
  34. 34. Deb P, Norton EC, Manning WG. Health econometrics using Stata. 1st ed. Stata Press; 2017.
  35. 35. Lichtenberger P, Hooton TM. Complicated urinary tract infections. Curr Infect Dis Rep. 2008;10(6):499–504. pmid:18945392
  36. 36. Simmering JE, Tang F, Cavanaugh JE, Polgreen LA, Polgreen PM. The Increase in Hospitalizations for Urinary Tract Infections and the Associated Costs in the United States, 1998–2011. Open forum Infect Dis. 2017;4(1):ofw281. pmid:28480273
  37. 37. Klein EY, Mojica N, Jiang W, Cosgrove SE, Septimus E, Morgan DJ, et al. Trends in Methicillin-Resistant Staphylococcus aureus Hospitalizations in the United States, 2010–2014. Clin Infect Dis. 2017;65(11):1921–3. pmid:29020322
  38. 38. Kourtis AP, Hatfield K, Baggs J, Mu Y, See I, Epson E, et al. Epidemiology and Recent Trends in Methicillin-Resistant and in Methicillin-Susceptible Staphylococcus aureus Bloodstream Infections—United States. MMWR Morb Mortal Wkly Rep. 2019;68:214–219. pmid:30845118
  39. 39. Schweizer ML, Eber MR, Laxminarayan R, Furuno JP, Popovich KJ, Hota B, et al. Validity of ICD-9-CM coding for identifying incident methicillin-resistant Staphylococcus aureus (MRSA) infections: is MRSA infection coded as a chronic disease? Infect Control Hosp Epidemiol. 2011;32(2):148–54. pmid:21460469
  40. 40. Dyar OJ, Huttner B, Schouten J, Pulcini C. What is antimicrobial stewardship? Clin Microbiol Infect. 2017;23(11):793–8. pmid:28882725
  41. 41. Nouvenne A, Ticinesi A, Lauretani F, Maggio M, Lippi G, Guida L, et al. Comorbidities and Disease Severity as Risk Factors for Carbapenem-Resistant Klebsiella pneumoniae Colonization: Report of an Experience in an Internal Medicine Unit. PLoS One. 2014;9(10):e110001. pmid:25335100
  42. 42. Sahuquillo-Arce JM, Selva M, Perpiñán H, Gobernado M, Armero C, López-Quílez A, et al. Antimicrobial resistance in more than 100,000 Escherichia coli isolates according to culture site and patient age, gender, and location. Antimicrob Agents Chemother. 2011;55(3):1222–8. pmid:21220537
  43. 43. Versporten A, Zarb P, Caniaux I, Gros M-F, Drapier N, Miller M, et al. Antimicrobial consumption and resistance in adult hospital inpatients in 53 countries: results of an internet-based global point prevalence survey. Lancet Glob Heal. 2018;6(6):e619–29.
  44. 44. Smith DL, Levin SA, Laxminarayan R. Strategic interactions in multi-institutional epidemics of antibiotic resistance. Proc Natl Acad Sci U S A. 2005;102(8):3153–8. pmid:15677330
  45. 45. Collignon P, Athukorala P-C, Senanayake S, Khan F. Antimicrobial resistance: the major contribution of poor governance and corruption to this growing problem. PLoS One. 2015;10(3):e0116746–e0116746. pmid:25786027
  46. 46. Naylor NR, Atun R, Zhu N, Kulasabanathan K, Silva S, Chatterjee A, et al. Estimating the burden of antimicrobial resistance: a systematic literature review. Antimicrob Resist Infect Control. 2018;7:58. pmid:29713465
  47. 47. Cosgrove SE. The Relationship between Antimicrobial Resistance and Patient Outcomes: Mortality, Length of Hospital Stay, and Health Care Costs. Clin Infect Dis. 2006;42(Suppl 2):S82–9.
  48. 48. Oxenham D, Finucane A, Arnold E, Russell P. Delivering preference for place of death in a specialist palliative care setting. BMJ Qual Improv Reports. 2013;2(1):u201375.w897.
  49. 49. Mauldin PD, Salgado CD, Hansen IS, Durup DT, Bosso JA. Attributable Hospital Cost and Length of Stay Associated with Health Care-Associated Infections Caused by Antibiotic-Resistant Gram-Negative Bacteria. Antimicrob Agents Chemother. 2009;54(1):109–15. pmid:19841152
  50. 50. Gandra S, Barter DM, Laxminarayan R. Economic burden of antibiotic resistance: how much do we really know? Clin Microbiol Infect. 2014;20(10):973–80. pmid:25273968
  51. 51. Lowsky DJ, Olshansky SJ, Bhattacharya J, Goldman DP. Heterogeneity in healthy aging. J Gerontol A Biol Sci Med Sci. 2013/11/17. 2014;69(6):640–9. pmid:24249734
  52. 52. Basic D, Shanley C. Frailty in an Older Inpatient Population: Using the Clinical Frailty Scale to Predict Patient Outcomes. J Aging Health. 2014;27(4):670–85. pmid:25414168
  53. 53. Gilardi F, Scarcella P, Proietti MG, Capobianco G, Rocco G, Capanna A, et al. Frailty as a predictor of mortality and hospital services use in older adults: a cluster analysis in a cohort study. Eur J Public Health. 2018;28(5):842–6. pmid:29590362
  54. 54. Bandeen-Roche K, Seplaki CL, Huang J, Buta B, Kalyani RR, Varadhan R, et al. Frailty in Older Adults: A Nationally Representative Profile in the United States. J Gerontol A Biol Sci Med Sci. 2015/08/21. 2015;70(11):1427–34. pmid:26297656