Audit of antimicrobial stewardship in medical inpatients in Waikato, New Zealand 2021

With the creation of Te Aka Whai Ora – Māori Health Authority and Te Whatu Ora – Health New Zealand, a coordinated national plan for antimicrobial stewardship (AMS) to reduce antimicrobial resistance (AMR) is highly relevant.^1,2 Upsurges in AMR remain a critical risk to global health and economic development, with global antibiotic use increasing by 65% from 2000 to 2015.³

In a recent systematic analysis, AMR was associated with 4.95 million deaths (3.62 to 6.57) in 2019, making AMR the third leading cause of death in the 2019 Global Burden of Disease report, after ischaemic heart disease and stroke.⁴ The World Health Organization has denoted AMR as one of the top 10 global public health threats.⁵

While antimicrobials are essential to protect human health, they can be used inappropriately and excessively; conversely, relative under-prescribing of antimicrobials occurs for Māori and Pacific peoples in New Zealand.^6,7 In the most recent OECD comparison, New Zealand had the fourth highest level of antibiotic prescribing,⁸ with more than 50% of use classed as inappropriate.⁹ Although most inpatient settings in New Zealand have antimicrobial prescribing policies, adherence with these policies is not known.¹⁰

The December 2021 report from the Prime Minister’s Chief Science Advisor strongly recommends AMS in all sectors to combat AMR, with an equity focus and Māori and Pacific engagement.² A key AMS strategy is prospective audit and feedback after antimicrobial prescriptions, recommended by international guidelines.¹¹ This provides an educational benefit to clinicians while maintaining prescriber autonomy. Audit data can identify areas requiring improvements, although the process is typically labour intensive and relies on the availability of antimicrobial specialists. Therefore, we selected standards for antimicrobial audit which can be easily replicated for quality improvement initiatives.

Methods

Setting

We audited medical inpatients at two hospitals in the mid-North Island of New Zealand: Waikato Hospital in Hamilton, a 673-bed tertiary care hospital, and Thames Hospital, a 52-bed rural secondary care hospital. Neither hospital had electronic prescribing or a formulary restriction programme. Antimicrobial advice from the Waikato Hospital Infectious Diseases (ID) department needs to be actively requested and there is no coordinated surveillance of antimicrobial use.

Inclusion criteria

Inpatients aged 15 years and above were eligible if they received at least one dose of an IV antibiotic in the six general medicine, stroke or respiratory wards at Waikato Hospital or the Thames inpatient unit, which is the single inpatient ward at Thames Hospital. Patients were included if their first IV antibiotic was prescribed in the national medication chart between Sunday 7 November and Friday 3 December 2021. Patients receiving only oral antibiotics were not included.

We excluded patients where IV antibiotics were not commenced in one of the medical wards or the emergency department to reflect medical inpatient prescribing and minimise auditing of prophylactic antibiotics.

Audit process

At Waikato Hospital, the clinical informatics pharmacist generated daily electronic lists of patients receiving IV antibiotics from medDispense® machines (TouchPoint Medical) located in the six medical wards. At Thames Hospital, all new inpatient notes were checked for IV antibiotic prescriptions.

Eleven auditors (medical students, house officers, registrars, pharmacists and consultants) reviewed paper-based national medication charts, clinical notes and electronic laboratory records on weekdays on the wards. Individual clinicians were not aware of the audit, to minimise the Hawthorne effect. Acknowledging the limitations,¹² ethnicity data were collected from the hospital patient management database and categorised as Māori and non-Māori. Multiple ethnicities were managed using prioritisation. Information on colonisation with multidrug-resistant organisms was taken from alerts on the electronic record.

Auditors reviewed what happened within the first 24 hours (defined as the end of the post-acute ward-round for new admissions) and at 48 to 72 hours, to document if antibiotics had been rationalised according to microbiology or changed to the oral route if appropriate. Our aim was to have two separate prospective reviews for all patients. Due to time limitations or patients admitted on weekends, data were collected prospectively at the 24-hour and 48-to-72-hour time points in 93 of 205 patients (45.4%) by the same auditor. For 112 of 205 patients (54.6%), data were collected at 48 to 72 hours, and information for the first 24 hours was retrospectively collected at that review. All patients had complete data for both time points.

Single data entry was standardised using a pre-coded Microsoft Form™ on smartphone browsers (available via Appendix 1 and online: https://forms.office.com/r/8fKeiKGAbf). Data were stored on an online, secure server on Microsoft Teams™.

Audit standards

We selected 10 audit standards shown in Table 1, defined from Waikato Hospital’s antimicrobial prescribing policy (Appendix 2, version 01, issued 23 June 2020) and adult antimicrobial guide on the MicroGuide™ app (version 4.22, November 2021, Horizon Strategic Partners Ltd. Leeds, UK): https://viewer.microguide.global/WDHB. Aspirational audit targets of 100% were chosen by the ID department, after applying inclusion and exclusion criteria to make this as practical as possible.

View Table 1–4, Figure 1.

We referenced MicroGuide™ to categorise documented indications and define the recommended empirical antibiotic regimens. The most senior clinicians’ documented diagnoses within the first 24 hours were matched to MicroGuide™ categories. To focus on antibiotic choice, dose optimisation was not audited. Gentamicin use was not obligatory in sepsis of unknown source, as ceftriaxone monotherapy was acceptable. Patients without neutropenia were categorised as having non-neutropenic sepsis if the word “sepsis” or “urosepsis” was documented in the clinical impression. Diagnostic accuracy and infection severity scores were not verified, to audit against real world practice. Uncertain entries were clarified by an ID physician.

Local approval for the audit and reporting of results was obtained from the Waikato audit and research unit (registration number 4289P). Data interpretation was reviewed by a senior Māori researcher and the local Māori research review committee in line with the CONSIDER statement for strengthening reporting of health research involving Indigenous peoples.¹³

The sample size was determined by a 4-week auditing period. Based on medDispense® data, we estimated at least 280 patients would be commenced on IV antibiotics in this period. A non-stratified random sample size of 500 would be required (125 Māori, 375 non-Māori) for 80% power to detect a difference of 10% between Māori and non-Māori with the Chi-squared test. Given this number was not feasible, ethnicity groups were not compared directly. Proportions were presented for categorical data and compared using Two-Sample tests of proportions, with a confidence level of 95%. Normally distributed continuous data were presented as means with standard deviations (SDs) and compared using Two-Sample t-Tests, with a confidence level of 95%. We analysed data using Microsoft Excel™ and STATA™ software (StataCorp. 2019. Stata Statistical Software: Release 16.1 College Station, TX: StataCorp LLC).

Results

Baseline characteristics

There were 728 discharges from the selected wards during the audit period, 578 from Waikato Hospital and 150 from Thames. IV antibiotics were dispensed to 262 of 578 Waikato Hospital patients (45.3%). We excluded 87 of these 262 patients (33.2%): those starting antibiotics outside the audit period or the selected wards, patients whose notes were unavailable and patients with missing data. A remaining 175 of 262 patients were audited (66.8%). Adding 30 patients from Thames Hospital, this totalled 205 audited patients.

Compared to 20.3% of people aged 15 and older in the Waikato Region identifying as Māori in the 2018 Census,¹⁴ Māori comprised 182 of all 728 discharges (25.0%) from the seven wards during the audit period (p = 0.002), and 52 of 205 audited patients (25.4%, p=0.074), acknowledging likely undercount of Māori.¹² Ethnicity differences in methicillin-resistant Staphylococcus aureus (MRSA) colonisation and ward location are shown in Table 2.

The mean age of Māori was 61.1 years (SD 16.1) compared to non-Māori at 71.2 years (SD 18.2), a mean difference of 10 years (p=0.001). Waikato data from the 2018 Census for people aged 15 and older showed a mean age of 39.0 years (SD 17.2) for Māori and 47.9 years (SD 19.5) for non-Māori, with a similar mean difference of 9 years.¹⁴ Figure 1 demonstrates the age distributions.

Antimicrobial use

The clinician-documented indications for antibiotics are shown in Table 3, most commonly respiratory, genitourinary, skin and soft tissue infections in 137 of 205 patients (66.8%). The indication was not documented in 12 of 205 cases (5.9%). Sepsis was documented in 55 of 205 patients (26.8%), with 8 of 55 recorded as Māori (14.5%). Only one patient had documented COVID-19 infection. Of the 205 audited patients, the most common initial antibiotics were ceftriaxone and amoxicillin/clavulanate, together comprising 149 of 205 prescriptions (72.7%).

Audit standards

The primary outcomes are shown in Table 4.

Diagnostic stewardship

1: Blood cultures were taken prior to IV antibiotics for 117 of the 205 audited patients (57.1%) and for 57 of 86 patients with essential diagnoses (66.3%): non-neutropenic sepsis (52), neutropenic sepsis (3), meningitis (6), endocarditis (1), septic arthritis (4), pyelonephritis (11) and urinary tract infection receiving IV antibiotics (9). Blood cultures were taken prior to antibiotics for 13 of 19 patients with sepsis of unknown source (68.4%). Our audit was not designed to compare Māori and non-Māori outcomes and the difference in outcomes for this standard may be due to chance, particularly with a low proportion of Māori patients documented as having sepsis (14.5%).

2: Urine culture was taken before IV antibiotics for 27 of 34 (79.4%) patients with pyelonephritis (11), urinary tract infection receiving IV antibiotics (9) or urinary sepsis/“urosepsis” (14). Urine culture prior to IV antibiotics is not mandatory in sepsis guidelines,¹⁵ as this can cause unnecessary delays. Therefore, we only applied this standard to sepsis with suspected urinary tract origin, and not to other sources of sepsis.

3: CSF was sampled before IV antibiotics, or up to 4 hours after, for three of six patients with meningitis suspected initially (50.0%). None of the patients had bacterial meningitis on follow-up.

Antimicrobial stewardship—indication and antimicrobial choice

4: An indication was written in the notes for 193 of 205 patients (94.1%). In contrast, an indication was written in the medication chart in 23 of 205 patients (11.2%).

5: A planned duration or a review date was present in the notes for 87 of 205 patients (42.4%) and in the medication chart in 25 of 205 patients (12.2%).

Only three of 205 (1.5%) patients had the indication and duration documented in both the notes and medication chart.

6: A relevant MicroGuide™ page was available in 167 of 205 patients (81.5%), with 91 of these 167 patients having antibiotic choices consistent with MicroGuide™ (54.5%). We excluded 38 of 205 patients (18.5%): guideline not available (21), ID specialist advice was given (2), significant antibiotic allergy or intolerances (3), known causative microbiology within the prior 7 days (4), already failing the recommended antibiotic (6) and MRSA/ESBL carriage not covered by the guideline (2). Ceftriaxone use was consistent with MicroGuide™ in 28 of 80 patients (35.0%) and amoxicillin/clavulanate in 46 of 69 patients (66.7%).

7: Piperacillin/tazobactam or a carbapenem were administered to 13 of 205 patients (6.3%) and discussed with ID in only two of 13 patients (15.4%). Piperacillin/tazobactam was prescribed for nine patients: four were consistent with MicroGuide™, three had no relevant MicroGuide™ page available and two were not consistent with MicroGuide™. A carbapenem was administered empirically to four patients: one was discussed with ID and the other three were colonised by multidrug-resistant organisms.

Antimicrobial review—duration, IV-oral switch and de-escalation

8: Antibiotics were reviewed within 48 to 72 hours after the start date for 186 of 205 patients (90.7%). Antibiotics were stopped at this point for 31 of these 186 patients (16.7%).

9: IV-oral SWITCH criteria were met for 140 of 205 patients (68.3%). A switch to oral antibiotics occurred for 101 of 140 patients (72.1%) and antibiotics were stopped for 23 of 140 (16.4%), totalling 124 of 140 patients who met the audit standard (88.6%).

10: Microbiology results were available to target antibiotic therapy for 132 of 205 patients (64.4%). Antibiotics were targeted in 122 of these 132 patients (92.4%) at the 48-to-72-hour review.

Discussion

Our audit identified specific areas for AMS quality improvement initiatives. The ID and microbiology departments currently do not engage in regular planned stewardship rounds. A business case for increased AMS resources to enable this activity has been submitted. To complete the audit cycle, the Waikato AMS programme plans to support medical and surgical teams to undertake quarterly antimicrobial prescribing audits, to measure improvements from planned AMS interventions outlined below.

Diagnostic stewardship

Microbiological sampling standards were met on 69.0% of occasions. For comparison, 24.0% of patients prescribed antibiotics in an AMS study in Vietnam had microbiological sampling. This occurred before starting antibiotics for 34.8% of those patients.¹⁶ Microbiological testing sensitivity reduces rapidly after commencing IV antibiotics.^17,18 When sampling is delayed, opportunities for antimicrobial optimisation may be lost. Auditing the timing of microbiological sampling in relation to antibiotics for specific diagnoses has not been widely reported and is not measured by the Australasian National Antibiotic Prescribing Survey (NAPS).

We are in the process of updating MicroGuide™ to reflect our local laboratory guidance on optimising blood and urine cultures.¹⁹ For adults, we now advise taking two sets of blood cultures from a single venepuncture site, with 8–10mL of blood per bottle. Single-site sampling for the first two blood culture sets is compatible with updated Duke-ISCVID endocarditis criteria.²⁰ In future audits we would document the number and type of blood culture bottles taken before antibiotics. Online surveys, educational campaigns and audits around improving microbiological sampling for phlebotomy, nursing and medical colleagues are planned.

Antimicrobial stewardship—indication and antimicrobial choice

Documenting indications and duration for antimicrobial prescriptions is strongly recommended by the US Centers for Disease Control and Prevention²¹ and the UK National Institute for Health and Care Excellence.²² This facilitates AMS audit and is included in our antimicrobial prescribing policy. Benefits include error prevention, enhanced communication, patient empowerment and promoting responsible antimicrobial prescribing.²³ Our documentation results were similar to Canterbury NAPS data, where the indication was documented in 73.5% of prescriptions, and a review or stop date in 30.2%.²⁴ One factor may be that the New Zealand national medication chart does not have a mandatory space for documenting indications and duration. Subsequent to this audit, an antimicrobial sticker was designed to place on the national medication chart with areas to document indication and review date. This has been implemented in the intensive care unit. Electronic prescribing significantly improves the documentation of antimicrobial indication.²⁵ Until this is available, our AMS committee is working with pharmacy and nursing colleagues to empower them to remind prescribers to include indications and durations for antibiotic prescriptions.

MicroGuide™ adherence was 54.5% in our audit. Of concern, 65.0% of empirical ceftriaxone and 33.3% of amoxicillin/clavulanate prescribing was outside of guidelines. This may be due to familiarity with these antibiotics to cover for sepsis when there is clinical uncertainty, and the absence of formulary restriction for ceftriaxone. In the Canterbury NAPS, guideline adherence was 74%²⁴ and adherence to the Auckland SCRIPT app has rates from 9 to 50%.²⁶ Given only one of four carbapenem prescriptions were discussed with ID in our audit, we implemented a carbapenem restriction policy in January 2023 and are in the process of auditing this policy. Empirical prescribing of restricted antibiotics does not need immediate ID approval when consistent with MicroGuide™. However, discussion with ID within 48 to 72 hours allows for dose optimisation, defined durations, targeted prescribing based on microbiology results and facilitation of outpatient IV antibiotics if required.

To improve documentation in the medication chart and familiarity with MicroGuide™, our local AMS committee is introducing an antimicrobial prescribing journey initiative. This is an educational campaign outlining antimicrobial prescribing for a patient from admission until discharge following the antimicrobial prescribing policy. It incorporates elements from other local campaigns, including sepsis tools and IV-oral SWITCH. Interventions include visual aids, posters and education sessions with prescribers, pharmacists and nursing staff. Utilising a straightforward infographic, it encourages holistic staff, patient and whānau engagement.

Antimicrobial review—duration, IV-oral switch and de-escalation

The results for these standards were around 90%. Only 64.4% of patients had microbiological results available to optimise antimicrobials at 48 to 72 hours, highlighting the importance of diagnostic stewardship to enhance AMS interventions. Our results were encouraging, as a study in Melbourne found IV-oral switch occurrence in only 57.0% of patients, despite a tightly regulated AMS programme.²⁷

Strengths of our audit include a range of infections over a representative 1-month period, urban and rural locations and reporting by ethnicity. The inclusion method ensured most patients on IV antibiotics in these wards were audited. Data were collected for the early 24-hour period and also for the 48-to-72-hour review, capturing the effect of initial diagnostic uncertainty on empirical antimicrobial prescribing. There was only one COVID-19 infection, minimising confounding by this condition.

Limitations include retrospective data collection in 54.6% of patients. Ethnicity was not self-identified.¹² Direct comparisons by ethnicity were limited by differences in baseline characteristics and the sample size; however, these data could help to plan for future audits with sufficient power. Results cannot be extrapolated to critical care or surgical specialties, particularly for perioperative IV antibiotic prophylaxis. To reduce complexity, we focused on IV antibiotic choice. In future audits we would include dose optimisation and oral antimicrobials. The 100% target for each standard was aspirational, as we did not want to choose arbitrary targets for these important standards of care. Retaining high targets specifically for Māori, Pacific and rural patients may help to address documented inequities.^28,29

Our audit adds to the narrative of AMS intervention in New Zealand. There is a need for increased use of equity-focused audit and feedback as an essential element of the New Zealand AMS strategy. We suggest small, focused AMS audits at frequent intervals, with Māori and Pacific patients included to allow for better understanding around inequities related to infectious diseases. As up to 95% of antibiotic consumption is in the community,^7,30 dedicated audits on community antibiotic use are also required, including in residential care facilities. We hope that our audit findings may contribute to the process of developing a strong, nation-wide AMS programme. We believe that increased ID and AMS resources are vital for success, as has been advocated by AMS colleagues across New Zealand.¹

‍

View Appendices.