Optimizing Antibiotic Stewardship in Nursing Homes: A Narrative Review and Recommendations for Improvement

The most common indication for initiation of antibiotics in nursing homes is UTI [14, 36, 92, 93]. Nevertheless, most residents treated for UTI do not manifest symptoms that are localized to the urinary tract [56]. Rather, studies suggest that a substantial number of these events represent unnecessary treatment of asymptomatic bacteriuria [58, 94]. Vague symptoms that have no clear relationship to the urinary tract, such as malaise or falls, are often cited as a reason to test resident urine samples [56]. Unfortunately, urine diagnostic tests—including the rapid urine reagent strip (dipstick), urinalysis and urine culture—fail to discriminate between UTI and asymptomatic bacteriuria [95‐97]. Nonetheless, there is compelling evidence that positive urine test results play an inordinate role in the decision to initiate an antibiotic [94]. For example, 60–90 % of the antibiotics prescribed for UTI in nursing homes are initiated only after urine cultures results are available [56, 98]. Moreover, withholding urine culture results, unless specifically requested by the provider, substantially reduces the frequency of antibiotic starts in hospitals without adversely affecting patient outcomes [99].

On the basis of these observations, it is reasonable to hypothesize that antibiotic prescribing for UTI can be reduced through interventions that target over-testing of resident urine samples. The first large-scale study to test this hypothesis employed criterion-based pathways to limit urinary testing and antibiotic prescribing to residents with fever and/or localizing urinary symptoms [80]. This study, performed in 24 nursing homes in the USA and Canada, demonstrated a reduction in antibiotic prescribing for UTI (1.17 versus 1.59 antibiotic starts per 1000 resident-days), but the effects waned over time, and there was a non-significant impact on orders for urine cultures (2.03 versus 2.48 cultures per 1000 resident-days), suggesting issues with intervention sustainability and fidelity [100]. Nevertheless, a subsequent study in a single Veterans Affairs (VA) nursing home demonstrated an impressive reduction in urine cultures (1.5 versus 3.7 cultures per 1000 resident-days), treatment of asymptomatic bacteriuria (0.6 versus 1.7 courses per 1000 resident-days) and overall days of antibiotic therapy (117 versus 168 days per 1000 resident-days) following implementation of the same intervention [82]. Similarly, treatment of asymptomatic bacteriuria was reduced from 52 to 10 % (P = 0.001) following implementation of a cognitive intervention [101] focused on limiting urine cultures in residents with guideline discordant symptoms (e.g. foul-smelling urine) [87]. Finally, significant reductions in urine cultures [incidence rate ratio (IRR) 0.73; 95 % confidence interval (CI) 0.66 to 0.79] and antibiotic therapy for UTI (IRR 0.67; 95 % CI 0.59–0.76) were achieved following implementation of a quality improvement collaborative focused on improving urine testing in 17 Massachusetts nursing homes [86].

Despite the availability of guidelines [102], there is evidence that assessments of residents with possible infection are suboptimal in many nursing homes [18, 34, 65]. Even when the assessment is thorough, the quality of the communication between facility nursing staff and providers, which has been previously linked to use of antipsychotics [103] and avoidable hospitalizations [73, 104], may be suboptimal. While the linkage between communication and antibiotic prescribing is less clear [42], interventions structured around standardizing resident assessments and interdisciplinary communication have led to reductions in antibiotic use in nursing homes [84, 85]. A medical provider and nurse-based intervention—which included training on antibiotic stewardship, a simple form that facilitated standardization of communication with providers and an informational brochure for families and residents—led to lower overall antibiotic prescribing in comparison with control facilities (IRR 0.86; 95 % CI 0.79–0.95) [85]. Similarly, antibiotic prescribing in Texas nursing homes was 33 % lower in nursing homes that implemented a standardized communication and UTI decision-support form with high fidelity in comparison with control nursing homes and facilities that implemented the form with low fidelity [84].

Education can be an effective behavioural intervention in situations where knowledge deficiencies drive observed practice patterns [105]. The available studies suggest that educational interventions have a modest impact on antibiotic use in nursing homes. The number of antibiotic starts dropped by 26 % and days of antibiotic therapy were reduced by 30 % following implementation of a case-based educational intervention focused on treatment of commonly encountered infections in a Chicago long-term care facility [81]. The impact of educational interventions when delivered across multiple facility settings has been less impressive. A cluster-randomized study of an educational intervention to improve adherence to a prescribing guideline in Canadian nursing homes demonstrated improvements in provider guideline adherence rates, but this difference did not remain significant during long-term follow-up [89]. Similarly, a cluster-randomized study of an educational intervention to reduce fluoroquinolone utilization in Swedish nursing homes did not demonstrate a significant impact on this particular prescribing pattern, although the proportion of infections treated with antibiotics was lower in intervention nursing homes compared with control facilities (difference-in-differences of proportions −0.124; 95 % CI −0.228 to −0.019) [83]. Interestingly, the effectiveness of educational interventions may rely on simultaneous delivery of content to facility nursing staff and prescribing providers. A cluster-randomized trial of an educational intervention to improve antibiotic prescribing for pneumonia in New York nursing homes demonstrated a significant improvement in adherence to prescribing guidelines (from 50 to 82 %) in facilities where education was targeted at both types of clinical staff [88]. In contrast, guideline adherence remained essentially unchanged (69 versus 65 %) in nursing homes where education was targeted solely at prescribing providers [88].

Post-prescribing review and de-escalation interventions (Q3–5 in Fig. 1) have had a profound impact on antibiotic utilization in hospitals [106]. Not surprisingly, there is evidence that de-escalation interventions have the potential for similar impact in nursing homes. Fluoroquinolones are among the most commonly prescribed antibiotics in US nursing homes [14]; however, studies have shown that providers rarely switch to agents with a more narrow spectrum (e.g. trimethoprim–sulfamethoxazole) even when culture results indicate a fully susceptible pathogen [23]. Moreover, a large observational study of nursing homes in Ontario demonstrated that the majority of antibiotic prescriptions exceeded durations of therapy sufficient for treatment of the most commonly encountered infections [17]. A follow-up study by these investigators confirmed that only a minority of prescribers (22 %) employed short-course therapy accepted in most guidelines [107]. Convincing providers who typically prescribed antibiotics for longer durations to prescribe antibiotics in a similar manner would reduce the total number of antibiotic days in Canadian nursing homes by 17 % [107]. A recently published study of 100 prescribing events in a VA nursing home demonstrated that 43 % of 1351 antibiotic days associated with these events could have been safely eliminated either by stopping the antibiotic because it was unnecessary or by shortening the duration of therapy [108].

Despite these opportunities, antibiotic de-escalation interventions remain poorly studied in the nursing home setting. A cluster-randomized study of a multi-component intervention, which included a routine post-prescribing review of empirically initiated antibiotics in 30 nursing homes in London, demonstrated a 5 % reduction in antibiotic consumption in intervention facilities compared with control facilities despite low adherence to the post-review procedures (26 %) [109]. Another study examined the impact of an infectious disease consultative service, which involved remote consultation services during daytime hours, as well as once-weekly in-person resident assessments, on antibiotic prescribing and resident outcomes in a 160-bed VA nursing home [110]. Significant declines in systemic antibiotic use (−30 %; P < 0.001) and facility rates of C. difficile (slope −0.03; P = 0.04) were observed over the 18-month intervention period in this study.

Changing nursing home culture requires commitment at an organizational level. Consequently, the first step is to obtain buy-in from the facility leadership, including the director of nursing and the medical director, to make antibiotic stewardship a facility quality improvement focus. Arguments that nursing home leaders find persuasive include (1) a need to satisfy regulatory requirements focused on appropriate use of medications; (2) federal mandates to demonstrate meaningful organizational quality assurance and performance improvement (QAPI) [113]; (3) emerging federal policies to promote and ultimately require antibiotic stewardship activities across all health care settings [6]; (4) organizational costs of treating antibiotic-resistant infections and CDI [25, 114]; and (5) how certain antibiotic stewardship interventions, particularly those focused around enhancing interdisciplinary communication, can generate corollary benefits in other nursing home processes and outcomes (e.g. enhanced management of resident changes in condition).

Antibiotic prescribing in nursing homes is a complex problem and involves a number of factors [115]. Facilities dedicated to improving the quality of antibiotic prescribing should assemble a multidisciplinary team that, at a minimum, involves the facility medical director, the director of nursing, an infection control practitioner and, if available, the facility pharmacist [116]. Initial meetings should focus on identification of outcome measures that will be targeted by improvement efforts, resources that will be needed to achieve these objectives, and duties and responsibilities of team members and expectations of nursing home staff and prescribing providers as they relate to antibiotic use in the facility. Seeking input from individuals with expertise in antibiotic stewardship and infectious diseases in the region can be particularly helpful at this stage. This team should plan to meet at least monthly when first getting started.

Provider orders for antibiotics should include clear documentation of the drug, dose, duration and indication. This latter requirement is critical for tracking antibiotic utilization (see below). Policies should be developed that discourage the use of prophylactic antibiotics for prevention of UTI and routine urine dipstick testing when evaluating a resident experiencing a change in condition. The antibiotic stewardship team should also consider developing policies and guidelines that limit the use of fluoroquinolone antibiotics, given their strong relationship to C. difficile [117] and high rates of resistance among urinary pathogens in many facilities [91, 118, 119]. More detailed guidelines that specify the choice and duration of antibiotic therapy for commonly encountered syndromes can be very effective in improving prescribing quality [81] but may require substantial effort to develop.

A capability to measure and track process and outcomes is a fundamental characteristic of successful quality improvement [120]. Most nursing homes routinely track commonly encountered infections [52, 121]. Adapting this process to monitor, at a minimum, UTI events and CDI should not prove too difficult for most facilities. We recommend that UTI be tracked in two ways: (1) treated UTI events (i.e. all antibiotic starts regardless of whether the event was a true infection); and (2) surveillance-defined UTI events (i.e. only those events that satisfy published criteria) [121, 122]. These measures can then be tracked in a number of ways. Point prevalence measurements (e.g. measuring the numbers of UTI events in the months before and after an intervention) are relatively easy to perform. Tracking outcomes using rates (e.g. UTI events per month/resident-days per month × 1000) is a more rigorous approach, which may allow a facility to more accurately assess the impact of multiple interventions over time. CDI can be tracked in a similar manner using the National Healthcare Surveillance Network (NHSN) criteria, which consider the rate of non-duplicative positive tests per 10,000 resident-days [123]. Nursing homes may also consider including residents transferred to their facility with C. difficile in that rate [124]. Extending monitoring to other types of infections, including those involving the respiratory tract and soft tissues, should be considered, as these are the second and third most common indications for initiation of antibiotic therapy in the nursing home setting [14, 125]. Finally, facilities should discuss developing a facility-specific antibiogram [119] in consultation with their contract laboratory. The antibiogram is a tool that facilities can use to identify common resistance patterns (e.g. high levels of fluoroquinolone resistance), which may influence providers’ empirical antibiotic decisions [91]. The small number of cultures collected in nursing homes may create instability in observed patterns of resistance and should be interpreted in consultation with individuals with infectious disease expertise.

Nursing homes embarking on quality improvement in antibiotic prescribing should seek to track antibiotic use in some manner. Logistics, as well as the types of improvement interventions that will be pursued, should guide the measure(s) selected. Most nursing homes employ tracking methods to identify residents experiencing a change in condition, including those residents who are currently receiving antibiotics [126]. Consequently, information on antibiotic starts is readily available and can be tracked at predefined time periods by the individual responsible for infection surveillance in the facility. Cross-sectional assessments of antibiotic use involve counting the number of resident on antibiotics during a given day or week. These assessments are fairly simple and do not require a huge amount of effort, but they may not be as sensitive to change as measures that are tracked more regularly. Tracking antibiotic starts per month (like other continuously monitored outcomes, this measure should be converted to a rate using resident-days as a denominator) is a more involved process, which allows facilities to trend and visualize their data. These data can be further stratified by indication (antibiotic starts for UTI, respiratory tract infection, skin and soft tissue infection), particularly if the indications for the antibiotic are routinely included on prescriber orders (see above). Antibiotic start data are ideal for tracking the effects of interventions focused on reducing unnecessary antibiotic use (e.g. reducing urine cultures) but may not be very informative for interventions that focus on reducing the duration of antibiotic therapy (e.g. guidelines or de-escalation). Facilities seeking to implement interventions along these lines should seek to track antibiotic utilization using either a days of therapy (DOT) measure [127] or a defined daily dose (DDD) measure [16]. The DOT utilization measure is derived by adding up every day on which a resident receives a unique antibiotic. By this method, receipt of multiple antibiotics has additive effects on the measure (e.g. a resident who receives 5 days of ceftriaxone and 5 days of azithromycin would generate 10 DOT even if both antibiotics are administered on the same day). In contrast, the DDD utilization measure is based on overall consumption of a particular antibiotic adjusted for typical daily dosing in an adult patient with intact renal function. By this approach, the total number of grams of each antibiotic used (or purchased) are summed over the period of interest and divided by the World Health Organization’s assigned DDD value [128]. Manual collection of these types of data poses a challenge. Consequently, working with the facility pharmacist to automate this process through queries of the pharmacy information system is recommended.

Education is a foundational activity of any ASP and can be used to target all three domains influencing prescribing decisions in nursing homes (Fig. 2—patient and family factors, facility and staff factors, health care provider factors). Educational content should cover the importance of stewardship, plans for implementation of specific stewardship activities and the responsibilities of clinical staff in achieving stewardship goals. Education should be targeted and tailored to nursing assistants, nursing staff, providers, residents and families. While not focused on antibiotic prescribing practices, a recently published study performed in 12 Michigan nursing homes demonstrated that a multimodal educational intervention focused on common infection prevention practices and care processes associated with indwelling urinary catheters and feeding tubes was associated with significant reductions in rates of MRSA acquisition [absolute risk reduction (ARR) 0.78; 95 % CI 0.65–0.95] and catheter-associated UTI (ARR 0.69; 95 % CI 0.49–0.99) [129]. Resident and family education, when combined with staff and provider education, as well as interventions to enhance interdisciplinary communication, has also recently been shown to be associated with reductions in antibiotic use in nursing homes [85]. Studies such as these demonstrate that educational interventions can be powerful tools for changing behaviours but likely need to target multiple individuals [88] and be delivered via a number of modalities—including in-service training sessions, newsletters, pocket guides, posters and brochures—in order to be maximally effective.

Once a facility has established its basic policies, outcome and process measures and has developed basic educational activities, the stewardship team should use the collected data to set new goals and select more active strategies in order to achieve these goals. As noted above, antibiotic decision-making in nursing homes is heavily influenced by diagnostic test results, and we believe that facilities will see the biggest return on investment through a focus on implementing interventions to reduce orders for urine cultures. Urine dipstick and urinalysis tests, despite being reliable methods for ruling out bacteriuria [130, 131], are often used to justify orders for urine cultures. Mild abnormalities on urinalysis, such as positive nitrite or leukocyte esterase readings on dipsticks, or moderate numbers of white blood cells on microscopy, are common in asymptomatic bacteriuria [96, 97], so neither test should be used as a screen for UTI. Consequently, nursing homes should instead develop and use protocols that restrict all urine testing to residents with a high probability of having a UTI (Fig. 3). These protocols should be operationalized not only through education of providers but, given their influence on provider decision-making [42, 43], also through engagement of nursing staff, who should be empowered to discourage providers from ordering diagnostic urine tests in the absence of specific evidence-based criteria. Tracking the frequency of urine cultures and the number of treated UTI events that do not satisfy surveillance definitions [121] provides targets that a facility can follow in order to assess the impact of the intervention. Other approaches based on improving interdisciplinary communication through tools that standardize documentation and sharing of important patient information, as well as development of a process to ensure consistent post-prescribing review and de-escalation, may represent a greater challenge but are worth considering, particularly if success in reducing the frequency of urine cultures has been achieved.