Pilot study of an online hospital antibiotic use tracking and reporting system

Abstract

Background

Antimicrobial use (AMU) varies widely among hospitals, suggesting a need to better monitor usage and evaluate the effectiveness of antimicrobial stewardship programs (ASPs). Our objective was to assess the feasibility of implementing an online voluntary hospital antibiotic use tracking and reporting system.

Methods

An online survey was sent to ASP clinicians representing hospitals across Ontario. Hospitals that tracked total hospital-wide inpatient antibiotic use in 2017 were asked to submit either days of therapy (DOT) or defined daily doses (DDD), along with separate inpatient days (PD), which were used as the denominator. Respondents who indicated no hospital-wide AMU tracking were asked to describe the barriers to its use. Antibiotic use was displayed on a public website for consenting hospitals.

Results

Of 201 eligible hospitals, 66 (33%) provided AMU data representing 10,634 of 25,208 (43%) eligible inpatient beds in the province. DOT and DDD data were provided by 36 hospitals, each. Weighted average antibiotic use was highest in acute teaching hospitals (513 DOT/1,000 PD, 709 DDD/1,000 PD) and lowest in complex continuing care and rehabilitation facilities (158 DOT/1,000 PD, 159 DDD/1,000 PD). Barriers cited for providing hospital-wide AMU data include lack of time and resources to collect and evaluate AMU data and technological limitations preventing data collection.

Conclusion

Integrating hospital AMU tracking and reporting as part of a voluntary initiative is feasible, with relatively broad participation. Short of a legislative mandate for participation, opportunities still exist to increase representation, including provision of guidance and technical support to help hospitals track and share AMU.

Rising rates of antimicrobial resistance are threatening the management of infectious diseases and contributing to increased risk with procedures that depend on antibiotic use, such as transplantation, chemotherapy, and surgery (1). Antimicrobial use (AMU) hastens antimicrobial resistance, at both the patient and the population level (2). As much as one-third of antibiotic prescriptions in hospitals are either unnecessary or inappropriate, providing a considerable opportunity to improve usage and mitigate harm for patients. Total antibiotic use varies widely among hospitals, even when accounting for aggregate hospital and patient characteristics (3), suggesting a need for standardized approaches to address overuse.

Antimicrobial stewardship can be described as

coordinated interventions designed to improve and measure the appropriate use of antimicrobial agents in terms of drug selection, dosing, duration of treatment and route of administration. (4 p323).

To determine the impact of local antimicrobial stewardship program (ASP) interventions, metrics such as days of therapy (DOTs) and defined daily doses (DDDs) standardized to patient days (PDs) are favoured as proxy indicators for antibiotic appropriateness, with a desired decrease associated with antimicrobial stewardship efforts. Although DDDs are the most widely used measure of AMU, DOTs offer an improved estimate of antibiotic exposure, particularly in populations with non-standard doses due to renal function, weight, and age (5). Technological limitations often prevent the measurement of DOTs in all facilities; hence, the current landscape includes a mix of both metrics (6).

Although tracking AMU over time enables hospitals to optimize antimicrobial stewardship activities locally, from a broader public health perspective access to AMU data at a regional level is needed to inform efforts in planning, evaluating, and strengthening provincial ASP initiatives. Because there is currently no central repository or comprehensive systematic effort to collect, synthesize, or report hospital-wide AMU data in Ontario, Public Health Ontario (PHO) had an opportunity to assess the feasibility of a voluntary, publicly available hospital AMU reporting mechanism as part of a wider, established provincial initiative that seeks to advance hospital ASPs.

Methods

Setting

This project was carried out in Ontario, Canada’s most populous province, led by PHO, an agency with a mandate to provide scientific and technical expertise to improve public health in the province.

Online survey

The Ontario ASP Landscape Survey, developed by PHO in 2016, is a voluntary survey of antimicrobial stewardship clinicians about ASPs at their institutions and was intended to obtain detailed information about program structure and strategy implementation in Ontario hospitals (4). On the basis of an overall response rate of 74% to the initial survey and stakeholder feedback, PHO expanded the survey questionnaire in 2018 to include additional questions about AMU measurement to determine whether this existing survey initiative would be a feasible mechanism to collect and report AMU. Respondents were asked to provide information about how AMU was measured at a hospital corporation level, and those who tracked total hospital-wide inpatient AMU for 2017 were then asked to submit these data at a hospital site level, either monthly or quarterly in DOTs or DDDs with separate inpatient day denominators for each eligible site within their corporation. World Health Organization Anatomic Therapeutic Classification J01 antibiotics (systemic antibacterial agents) were included (7). Respondents who indicated lack of hospital-wide AMU tracking were asked to describe the key barriers. Survey questions pertaining to AMU are available in the Appendix.

Survey distribution

The 2018 Ontario ASP Landscape Survey was available for 4 weeks (September 25–October 24, 2018) and was administered using an in-house online platform (Surveys@PHO). Targeted email distribution lists were used, and the survey was addressed to the clinician most responsible for antimicrobial stewardship in each organization. Contact information was available from a database of stakeholder contacts, a provincial email network, and a previous provincial ASP survey. Follow-up via email, telephone reminders, or both was provided to non-responders 2–3 weeks after survey launch. There were no monetary incentives to participate.

Ethics

This project was submitted to Ethics at PHO and was deemed to be outside the scope of Ethics review.

Eligibility

All hospitals in Ontario were eligible to complete the survey except for those that only provide mental health or ambulatory services.

Online reporting tool

The Ontario ASP Comparison Tool (https://www.public healthontario.ca/en/health-topics/antimicrobial-stewardship/asp-comparison-tool) is an online, publicly available interactive solution that was developed to enable users to compare characteristics of hospital ASPs on the basis of the results of the Ontario ASP Landscape Survey. With the addition of AMU data to the survey in 2018, this tool was expanded to include this new information. Survey respondents had the option of authorizing PHO to share all or part of their survey responses, including their AMU data, through the Ontario ASP Comparison Tool. Participating hospital names were displayed in the tool along with their quarterly and annual AMU data. The interactive tool allows stratification by geographical region and hospital type and inter-hospital comparison as well as comparison with the weighted average for the region, hospital type, or total for the province.

Analysis

Descriptive statistics for the AMU component of the survey were calculated using Microsoft Excel 2013 Version 14 (Microsoft Corp., Redmond, WA). Results are presented descriptively with median; inter-quartile ranges; and ranges across all hospitals, hospital types as defined by the Ontario Hospital Association (8), and health regions within Ontario. Seasonal variability, the measure of the proportional difference in use between the quarter with the highest use and the quarter with the lowest use, was measured overall and for each hospital type (9). Inpatient hospital bed numbers were based on information from the Ministry of Health and Long-Term Care. Barriers to collecting AMU data were presented descriptively.

Results

Antimicrobial use measurement at the corporation level

The overall response rate to the 2018 Ontario ASP Landscape Survey was 55% (70/127 hospital corporation respondents). Of those with or in the process of implementing a formal ASP (68/70; 97%), 75% (52/70) of all responding corporations indicated that AMU was tracked in the 2017 calendar year. Most corporations are tracking unit-specific AMU (77%; 40/52) or are focused on high-risk or targeted antimicrobials such as carbapenems and anti-pseudomonal agents (60%; 31/52). Other metrics such as drug-class specific use were tracked in 23% (12/52) of hospital corporations during 2017.

Hospital representation

Of 127 eligible corporations, there were 201 eligible hospital sites. Of these eligible hospital sites, 66 (33%) provided AMU data representing 42% of inpatient beds in the province (10,634/25,208 eligible beds). Regional representation varied widely from AMU capture for 89% of inpatient beds in the Central West health region to none in the North Simcoe Muskoka health region. Overall participation was highest in the regions around Niagara, Hamilton, and Toronto and lower in more rural regions in northern parts of the province. Representation was highest for large community hospitals (51%) and lowest for small community hospitals (18%; Figure 1). Almost all (92%; 61/66) authorized PHO to post their organization’s AMU using the Ontario ASP Comparison Tool.

Figure 1:

Representation of antibiotic use data provided by Ontario hospitals

CCC = complex continuing care; AMU = antimicrobial use

Use data

DOT data were provided by 36 hospitals, and DDD data were provided by 36 hospitals, with 6 hospitals providing both DOT and DDD data. Reported annual AMU varied widely across hospitals for both DOTs (133–728 DOTs/1,000 PDs) and DDDs (81–1,609 DDDs/1,000 PDs). The weighted average antibiotic use was highest in acute teaching hospitals (513 DOTs/1,000 PDs, 709 DDDs/1,000 PDs) and lowest in complex continuing care (CCC) and rehabilitation facilities (158 DOTs/1,000 PDs, 159 DDDs/1,000 PDs; Table 1). Variability between individual hospitals is shown in Figures 2 and 3.

Table 1:

Antibiotic use per 1,000 inpatient days by hospital type

	CCC & rehabilitation		Small community		Large community		Acute teaching
Metric	DOT (n = 2)	DDD (n = 6)	DOT (n = 3)	DDD (n = 4)	DOT (n = 30)	DDD (n = 19)	DOT (n = 1)	DDD (n = 7)
Weighted average	158	159	445	520	412	458	513	709
Median	193	163	483	525	448	425	513	614
IQR	164–223	139–217	341–593	306–699	350–540	404–515	513–513	484–886
Range	133–253	81–349	200–702	76–795	130–728	220–634	513–513	451–1,609
Seasonal variation, %*	63	26	4	43	8	10	6	5

For the 2017 calendar year in participating Ontario hospitals.

* Seasonal variation is calculated by measuring the percentage of difference between the quarter with highest use and the quarter with the lowest use

CCC = complex continuing care; DOT = days of therapy; DDD = defined daily dose; IQR = inter-quartile range

Figure 2:

Ontario hospital wide antibiotic use variability by hospital type in 2017: DDD

DDD = defined daily doses; PDs = patient days; CCC = complex continuing care

Figure 3:

Ontario hospital wide antibiotic use variability by hospital type in 2017: DOT

DOT = days of therapy; PDs = patient days; CCC = complex continuing care

Seasonality

Seasonal variability (difference between lowest use period and highest use period) was 6% overall, with use highest in the first quarter (January–March) and lowest in the third quarter (July–September). The highest degree of seasonal variability was seen in CCC and rehabilitation facilities (63% and 26% for DOTs and DDDs, respectively), whereas the lowest seasonal variability was noted in acute care facilities (6% and 5% for DOTs and DDDs, respectively; Table 1).

Barriers to participation

The 18 (25%) hospitals that indicated that they did not track hospital-wide AMU for 2017 cited technological limitations that prevented data collection or extraction from information systems (n = 10) or lack of time or resources to collect and evaluate AMU data (n = 7). They also reported that interventions were often focused on specific units or for specific classes, precluding the need to evaluate hospital-wide antibiotic use data (n = 5). A small number of hospitals (5/66; 8%) submitted AMU data to PHO but did not provide authorization for PHO to share their data using the Ontario ASP Comparison Tool.

Discussion

This voluntary survey on first attempt succeeded in obtaining AMU data for 33% of eligible Ontario hospitals and 42% of eligible inpatient beds, providing substantial progress in filling a vital gap in inpatient AMU data in the province. The survey also identified key barriers to address for future iterations to expand toward true population-based data.

AMU surveillance is well established in certain jurisdictions such as the European nations (10), but there are opportunities for improvement in North America where health care delivery is more decentralized. The US Centers for Disease Control and Prevention’s National Healthcare Safety Network launched the electronic collection of AMU data in 2012 (11). This voluntary approach allows for electronic reporting in a standardized format and as of 2017 had 300 hospitals participating across the United States (12), approximately 5% of the country’s facilities. In Canada, although multiple surveillance initiatives encompass some aspects of AMU, there remain significant gaps in tracking and reporting hospital AMU at the provincial level. For example, detailed longitudinal antimicrobial resistance and some AMU data are captured by the Canadian Nosocomial Infection Surveillance Program, but it is based on sentinel hospitals that make up less than 5% of sites across the country and is therefore not representative (13). AMU data included in the Canadian Integrated Program form Antimicrobial Resistance Surveillance and the Canadian Hospital Resistance Surveillance System are based on proprietary drug purchasing data and are not widely accessible and not presented at the hospital level. In Ontario, the Critical Care Information System collects DOT for antibacterials and antifungals and represents 100% of eligible facilities because of the mandatory nature of reporting, but this reporting is limited to levels 2 and 3 intensive care units. There is, hence, a need for more complete representation of AMU in hospitals, ideally facilitated by mandatory reporting, to better understand the challenges and evaluate the impact of antibiotic stewardship interventions from a provincial perspective.

This initial attempt to integrate AMU reporting as part of an established voluntary province-wide initiative supporting hospital ASPs resulted in the participation of 31% of eligible hospitals representing 42% of eligible inpatient beds in Ontario, relatively large coverage compared with previous attempts to voluntarily collect these data. Descriptive data from our survey reveals widespread variation in antibiotic use across hospitals, even when stratifying by hospital type. This is consistent with a study of 2014 antibiotic purchasing data in Ontario, showing seven-fold variation in hospital-wide usage and persistent variation after adjusting for hospital and aggregate patient characteristics (3).

Recognizing that participation may not be immediately feasible for all hospitals, we asked sites that could not provide hospital-wide AMU metrics to identify challenges to participating. Of particular importance were the inter-related concepts of lack of time and resources to collect and provide these metrics. Small community hospitals and CCC and rehabilitation hospitals had the lowest representation in providing hospital-wide AMU. Funding to support dedicated time for antimicrobial stewardship activities varies across hospitals. In Ontario, small community and CCC and rehabilitation hospitals are less likely to have designated funding or resources for their ASPs (6), so it is not surprising that these hospitals were also least likely to provide AMU metrics. Additional guidance to smaller hospitals on collecting and monitoring these AMU data may also encourage participation, allowing for improved representation. Technology limitations were also cited as an important factor preventing the acquisition of these data. Some respondents indicated that their hospital electronic health record or pharmacy software did not have the capability to provide AMU reports or that they were not sure how to obtain these reports even if they were available. In addition, many hospitals had concentrated AMU monitoring solely on specific inpatient units where antimicrobial stewardship activities are focussed and as a result were not evaluating hospital-wide usage. In the future, AMU surveillance efforts should focus on improvement in technology and supporting the skills necessary to acquire and report hospital-wide AMU data. The initial workload to establish AMU tracking in hospitals would be expected to decrease as efforts shift to maintaining AMU measurement. So this approach is expected to be sustainable over time.

This voluntary survey approach to hospital-wide AMU tracking and reporting has some limitations. Because AMU is captured at the hospital level and represents the total of all antibiotics used, it lacks the granularity necessary for evaluating usage in specific areas of the hospital and specific classes of antimicrobials, and it does not necessarily correlate with broad-spectrum antibiotic use in the facility (14). However, these high-level data provide the basis for inter-hospital comparison and encourage hospitals to monitor AMU across the facility in addition to a more focused look at individual patient care units or targeted drugs. The sample size was low for some hospital types, particularly small community hospitals and acute teaching hospitals measuring DOTs, which reduced the generalizability of estimates of total usage and seasonal variability in these hospital types. AMU is not adjusted for hospital or patient characteristics such as length of stay, proportion of ICU admissions, and case mix, which are known to affect AMU. Future efforts should focus on improved methods of stratification and risk adjustment based on patient and facility characteristics to allow for better comparison across facilities and identify benchmarks for optimal usage. An even mix of DOTs and DDDs was provided by participating hospitals, and there are known challenges in using these metrics to predict each other (15). Hence, they tend to be reported separately, which reduces the ability to perform inter-hospital comparison (16). Antibiotic use is a proxy for appropriateness, but it does not indicate explicitly whether antibiotics are used optimally; for those reasons, other methodologies such as point prevalence studies may be more ideal for assessing appropriateness of antibiotic use (17,18). In fact, a recent study by Lee et al. found that point prevalence studies may have predictive value for other antibiotic utilization metrics such as DDDs (18). Evaluation of AMU trends over time (19) is an opportunity to make this dataset more robust and allow facilities to identify site-specific targets for use.

Despite these limitations, this study highlights key considerations for improving AMU tracking and reporting at a regional level, including understanding and addressing barriers to participation, improving the ease of collection and submission of data, and performing risk adjustment to better understand variability in usage. In addition, to our knowledge, the Ontario ASP Comparison Tool is the first online, publicly available interactive solution that allows stakeholders to interact with hospital AMU data in the context of antibiotic stewardship practices with advanced functionality that allows users to search, filter, compare, and graphically visualize the information.

Conclusion

Integrating hospital AMU tracking and reporting as part of a voluntary provincial ASP initiative is a feasible approach with relatively broad representation. Short of a legislative mandate for participation, opportunities still exist to increase representation, including provision of guidance and technical support to help hospitals track and share AMU, with the aim that these initial investments in workload will lead to increased uptake and a sustainable approach to monitoring AMU across the province.

Appendix: Antimicrobial Use Data

Antimicrobial use (AMU) data help track antibiotic exposure and the impact of ASP strategies. Defined daily dose (DDD) and days of therapy (DOT) are the most commonly used metrics in ASP efforts.

1. Drug utilization data

   1. Did your corporation track antibiotic use (i.e., DDD or DOT) in the 2017 calendar year (January to December 2017)?

      • ☐ Yes

      • ☐ No; specify reason: [freetext]

   2. In which area(s) has your corporation tracked antibiotic use in the last 2017 calendar year?

      • ☐ Unit-specific

      • ☐ “High risk” or “targeted” antimicrobial use

      • ☐ Other: [freetext]

   3. Did your corporation track TOTAL hospital-wide antimicrobial use (either DDD and/or DOT) for the 2017 calendar year (January to December 2017)?

      • ☐ Yes

      • ☐ No; please specify any reasons for lack of hospital-wide antimicrobial use tracking: [freetext] .

2. 2. Please enter total hospital-wide antibiotic use data (WHO ATC J01—all systemic antibacterials), in DDD or DOT, including total inpatient patient days (PDs). If you are a multi-site corporation, please enter hospital-wide antibiotic use data for EACH site in separate rows.

   Enter either monthly (a) or quarterly (b) data for the 2017 calendar year. For any questions, please email asp@oahpp.ca

   1. Please enter antibiotic use data tracked monthly here:

      (one hospital per row, fill in as many as you have)

      #	Hospital site	AMU (specify DDD or DOT)	Jan AMU	Jan PDs	Feb AMU	Feb PDs	Mar AMU	Mar PDs	Apr AMU	Apr PDs	May AMU	May PDs	June AMU
      1
      2
      3
      4
      5
      6
      7

      June PDs	Jul AMU	Jul PDs	Aug AMU	Aug PDs	Sep AMU	Sep PDs	Oct AMU	Oct PDs	Nov AMU	Nov PDs	Dec AMU	Dec PDs

   2. Please enter antibiotic use data tracked quarterly _ here:

      (one hospital per row, fill in as many as you have)

      #	Hospital site	AMU (Specify DDD or DOT)	Jan–Mar AMU	Jan–Mar PDs	Apr–Jun AMU	Apr–Jun PDs	Jul–Sep AMU	Jul–Sep PDs	Oct–Dec AMU	Oct–Dec PDs
      1
      2
      3
      4
      5
      6
      7

Competing Interests:

The authors have nothing to disclose.

Ethics Approval:

This project was submitted to Ethics at Public Health Ontario and was deemed to be outside the scope of ethics review.

Informed Consent:

N/A

Registry and the Registration No. of the Study/Trial:

N/A

Animal Studies:

N/A

Funding:

No funding was received for this article.

Peer Review:

This article has been peer reviewed.