A practical guide to implementing clinical audit

Abstract

Clinical audit is a quality improvement tool for evaluating and improving patient care and outcomes. This is achieved by systematically reviewing current practices against explicit criteria and measuring the impact of change(s) introduced to generate improvement. The clinical audit process can be described by “Plan,” “Do,” “Study,” “Act” phases that comprise an audit cycle. The phases are moved through in turn to attempt quality improvement. Clinical audits are widely used in human medicine at both local (individual clinic or hospital) and national (to achieve nationwide improvements in care) levels. Substantial and sustained improvements in patient care have been attributed to the use of clinical audits. Clinical audits have been described in the veterinary literature since the 1990s, but their adoption does not appear widespread. This paper is intended as a practical, “how to” guide to applying clinical audit in veterinary practice.

Introduction

Clinical audit is a quality improvement tool that is used to monitor, assess, and improve the quality of care in human and veterinary medicine (1–7). The procedure involves measuring an outcome or process and comparing this to current evidence or best practice, then implementing changes to improve the quality of care (8,9). In human medicine, clinical audit is considered a standard of care (2,9,10). For example, The National Hip Fracture Database (UK) has allowed institutions to audit the care they provide to hip fracture patients according to national standards (2). Through clinical audit, 1 facility was successful in shortening the median time to surgery and discharged more patients to rehabilitation units following a hip fracture (2). In veterinary medicine, the Royal College of Veterinary Surgeons (RCVS) includes clinical audit in its Practice Standards Scheme, in which approximately 65% of UK practices participate (11). Performing clinical audit is a requirement for practices accredited as Veterinary Hospitals and encouraged for those accredited as General Practice and Emergency Service Clinics (11). However, as audits are not frequently published (there is no requirement for publication in the Practice Standards Scheme), it is unclear to what extent clinical audits are being conducted (5). At the time of writing, the American Veterinary Medical Association (AVMA), and the Canadian Veterinary Medical Association (CVMA) do not appear to provide resources on clinical audit. The aim of this paper is to provide a practical “how-to” guide on performing clinical audit. This guide is supplemented by a fictitious example of a clinical audit (Appendix, “Antimicrobial use in bacterial cystitis”). Key principles are illustrated with examples, including one based on a clinical audit performed on the subject of postoperative hypothermia in dogs in a large referral hospital (4).

Appendix.

Antimicrobial use in bacterial cystitis.

A small animal clinic is looking into performing a clinical audit. The group begins by answering the 3 preparatory questions. “What are we trying to accomplish?” After reading the Canadian Veterinary Medical Association recommendations on judicious use of antimicrobials, the veterinary team would like to improve their diagnosis and treatment of bacterial cystitis in dogs (1). The team agrees to follow the International Society for Companion Animal Infectious Diseases (ISCAID) guidelines (2). “How will we know that a change is an improvement?” Based on their clinical experience, the team identifies the following points from the ISCAID guidelines as areas that would qualify as improvement (2): Performing a urine culture on all cases of sporadic cystitis. Treat sporadic cystitis cases for 3 to 5 days. “What changes can we make that will result in improvement?” An initial short audit cycle was conducted over a week and identified that very few sporadic cystitis cases were being cultured and that treatment duration was usually 7 to 10 days. Upon further discussion with the staff there was no clear definition for what was considered sporadic cystitis and therefore what length of treatment was required. The staff created a checklist to aid in identifying cases of sporadic cystitis which included the following i) less than 3 episodes of cystitis in the preceding 12 months; ii) healthy non-pregnant females or neutered males; iii) no known urinary tract anatomical or functional abnormalities or co-morbidities. This information was then brought to the Plan stage.

Plan The team’s purpose statement was to improve antimicrobial use following the diagnosis and treatment of bacterial cystitis by: Culturing all urine samples from dogs with clinical signs consistent with bacterial cystitis. Limiting the treatment period to 3 to 5 days. Collecting data for items a. and b. over a 2-month period To help achieve these goals the RVTs were in charge of filling the prescriptions and confirming the duration of treatment.

Do One RVT was allocated 2 h a week to collect the data. This involved searching the computer database for “cystitis,” verifying the duration of antimicrobial treatment and determining whether a urine culture was performed.

Study After the 2-month period, the team reviewed the data. A bar graph was used to compare the number of cystitis cases against those cultured (Figure 4a). Following discussion, the team realized that discrepancies in performing cultures were largely explained by different levels of comfort with collecting urine via ultrasound-guided cystocentesis. A run chart was plotted to monitor treatment duration. The team was happy to see a trend towards treating between 3 to 5 days except for 2 astronomical points (Figure 4b). On further investigation they realized that this case was from a locum who was not made aware of the practice policy.

Act To improve urine culturing, the clinic elected to support a continuing education day on basic ultrasonography. Once the CE course was completed, a new clinical audit cycle was performed.

References

¹Canadian Veterinary Medical Association [homepage on the Internet] c2020 Veterinary oversight of antimicrobial use — A pan-Canadian framework of professional standards for veterinarians. Available from: https://www.canadianveterinarians.net/documents/pan-canadian-framework Last accessed December 10, 2020.

²Weese JS, Blondeau J, Boothe D, et al. International Society for Companion Animal Infectious Diseases (ISCAID) guidelines for the diagnosis and management of bacterial urinary tract infections in dogs and cats. Vet J 2019;247:8–25.

Materials and methods

The process of achieving sustained improvement in a complex system, such as the provision of veterinary care, may not be straightforward (12,13). This is because of the complexity of the system and the habits of the workforce. A successful change in practice is reflected in a measurable improvement that can be maintained (7–9,13,14). Common to most existing descriptions of clinical audit is the concept that an audit is cyclical, repeated at intervals to track performance. Multiple cycles may be necessary to achieve the desired changes, particularly in complex processes or where the key activities resulting in change are not obvious. While several audit cycle models exist, one which has been widely used in the domains of both business and medicine, is the “Model for Improvement” (9,13). The original model consists of 3 preparatory questions linked to a Plan-Do-Study-Act (PDSA) cycle (Figure 1). “Plan” is identifying the process(es) and outcome(s) to be studied and how data will be collected, “Do” is collecting the data, “Study” is interpreting the collected data, and “Act” brings the current audit cycle to completion by reviewing the results and planning for ongoing improvement. The following is an elaboration of each component of the model.

Figure 1.

The audit cycle, showing the 4 necessary steps of a complete audit cycle, from planning an audit (“Plan”), to the collection and analysis of data (“Do” and “Study”) and identifying steps leading towards improvement (“Act”). The cycle is intended to be repeated with the goal of continuous improvement. Adapted from Langley et al (13).

Preparatory questions

The 3 preparatory questions of the Model for Improvement precede the first step, the “Plan” stage of the PDSA cycle and improve the likelihood of conducting a successful audit (13). The questions are: 1. “What are we trying to accomplish?” 2. “How will we know that a change is an improvement?” 3. “What changes can we make that will result in an improvement?”

1. “What are we trying to accomplish?”

Identify the aim of the clinical audit. The selected topic may be determined by relevance to an individual practitioner or clinic, with common topics often related to patient/staff safety, clinical practice or finances (Table 1). When choosing a topic, it is essential that staff understand the rationale for the desired improvement, as this will build buy-in and improve compliance with introduced changes. Ideally, the selected topic has an associated set of explicit criteria/guidelines based on current evidence, against which performance can be compared [e.g., the use of physiologic monitors during cardiopulmonary resuscitation (15)]. Where guidelines exist, they should be written in a way that identifies measurable criteria that have been shown to improve the quality of care (16,17). Unfortunately, in veterinary medicine the evidence-base for outcome criteria or the existence of guidelines is often limited, resulting in a reliance on a combination of evidence and expert opinion, such as seen in consensus guidelines (15,18). Where neither evidence-based criteria nor consensus guidelines exist, criteria may be derived from observation of achievements in comparable settings: this is known as benchmarking. For example, the RCVS Knowledge website has a cumulative online database reporting on the outcomes of neutering procedures (spays and neuters in dogs, cats, and rabbits; currently reports outcomes from over 39 000 procedures) (19). These outcome data can be used as a benchmark for performance and to potentially identify areas for clinical audit/investigation. For example, the current data show that approximately 1% of cat castrations and 10% of dog castrations have a postoperative complication requiring medical treatment (19). So, if a practice was experiencing a higher incidence of complications requiring medical treatment with their canine neuters, such as 15%, this suggests an area for improvement and a percentage goal to achieve.

Table 1.

Sample subjects for clinical audit.

Subject	Potential measurement	Associated criteria/guidelines?	Potential target
Staff safety
Bite wounds	Cat bites that penetrate skin	None	Zero cat bites
Patients escaping	Patients escaping kennel, patients loose on walks	None	Zero kennel or walk escapes
Finances
Appointment scheduling	Client wait time beyond scheduled appointment time	None	Within 5 min
Emergencies	Returning calls for after hour emergencies	None	Triaging calls within 30 min
Clinical practice
Pre-operative antimicrobials	Antimicrobial administration time and surgical incision time	Within 30 to 60 min of surgery (25)	Within 60 min
Diabetes mellitus management	Appropriate dietary management	American Anima Hospital Association diabetes guidelines (26)	Achieving and maintaining ideal body mass

Using our hypothermia example, the response to “What are we trying to accomplish?” would be a reduction in incidence of postoperative hypothermia in dogs. A quick literature search in human and veterinary medicine reveals multiple adverse effects of perioperative hypothermia in humans and a high incidence of postoperative hypothermia in dogs and cats (4,20–23). The latter provides indirect support for the perception that the clinic has a high incidence of post-operative hypothermia (an example of benchmarking). The reported incidence of hypothermia in the literature can be used as a performance target, with the goal of providing a similar or better level of care. It is important with this first question that all staff understand the importance of the task identified for improvement. By taking the time to explain the potential adverse effects of hypothermia on patients, staff are encouraged to support the audit and comply with implemented changes. For example, kennel staff members may be more vigilant in monitoring patients’ post-operative temperature if they understand the potential consequences of hypothermia.

2. “How will we know that a change is an improvement?”

It is often easy to introduce a change, but it may not be straightforward, particularly in complex systems, to know if that change has a positive effect. Some form of data collection confirms that a change results in an improvement (and supports future assessments of performance and improves support from staff ). The data collected should provide information relevant to the outcome of interest. In its simplest form, the outcome itself can be measured (e.g., incidence of postoperative hypothermia). However, in complex systems or to better understand weaknesses in a system, it is often helpful to measure processes within a system (e.g., How many cases had active warming applied? Was rectal temperature recorded during surgery?). It is often tempting to focus on outcome measures as these reflect success (or failure). However, focusing on an outcome can be frustrating when it is unclear which processes resulted in an improved outcome. Additionally, if an instituted change does not result in improvement and data collection was limited to an outcome, it is impossible to know if the underlying processes were performed properly or if the right processes were altered, or both [e.g., if the change is to provide active warming during all surgeries and the incidence of hypothermia (outcome measurement) remains unchanged, the cause(s) of failure to generate improvement is unknown]. Similarly, if improvement occurs, the assumption is that this resulted entirely from the introduced change, unless a measure of process(es) is included to ensure that the change was implemented as planned or that other processes were affected/unaffected as predicted (e.g., the incidence of hypothermia drops when active warmers are made available, but perhaps the timing of active warmer use also has an important effect on the incidence of hypothermia). Therefore, including process measurements in an audit provides richer information than focusing exclusively on outcome.

Process measurements relate to specific activities (actions or steps) associated with outcomes. Documenting processes is often more sensitive than measuring outcomes (8,24). Process measurements may be the only feasible option when the outcome of interest is delayed (e.g., time to death or disease development) or the outcome occurs infrequently (e.g., anesthetic death). However, it is not always evident which processes influence outcome and to what extent (not all change in a system will result in improvement) (13). Ideally, process measurement selection would focus on those with the greatest influence on outcome and amenability to measurement.

Returning to the hypothermia example, focusing on the outcome alone (hypothermia at recovery) could miss processes with a direct and important effect on outcome. These could include, when/if rectal temperatures are recorded in the perioperative period, the use of active warmers (timing of application, temperature setting, availability of warmers), and procedure duration.

3. “What changes can we make that will result in an improvement?”

In veterinary medicine, evidence linking specific activities to beneficial outcomes can be sparse. Methods to help identify items for change (a “needs assessment”) may include conducting initial surveys, receiving feedback from relevant parties, and collecting initial data on performance. In complex systems in which numerous outcomes could be improved, an initial survey of users may help identify which outcomes to initially pursue (13). For example, with the goal of improved client service, a survey of clients may identify that delays in scheduled appointments are less important than segregated waiting areas for dogs and cats. This information can then be discussed with staff to explore if, and how, suggested changes can be introduced (e.g., feasibility of a screen to provide a visual barrier between cats and dogs), followed by an audit cycle to assess the impact of that change (survey of clients, record of pet density in segregated areas, feedback from staff on effectiveness). When considering a clinical audit, seeking contributions from relevant parties can identify items that may contribute to, or obstruct, the success of an audit (e.g., achieving an outcome of reducing medication errors is more likely to be successful if the number of different concentration preparations is minimized). Performing a short audit cycle, with limited scope, can enable rapid feedback and provide information on the impact of the tested changes. This can be valuable when it is unclear if a change is likely to be successful or there is a financial commitment associated with a change in practice. The resulting information could then be used in a larger scale clinical audit.

In the hypothermia example, it was important to verify the perception that the incidence of postoperative hypothermia was high. An initial, short period of data collection (over 1 wk) showed that 88% of dogs were hypothermic at extubation and that post-operative temperature recording, and warming were inconsistent. Discussing the findings with the registered veterinary technicians (RVTs) revealed that there was no consistent definition of hypothermia applied in the clinic, there was no standard practice for frequency of temperature measurements, and the availability of forced air warmers was limited (4). This process of acting on the perception that hypothermia was occurring frequently by collecting data to document its incidence and discussing the findings, was effectively a short audit cycle, the outcome of which was to confirm that a problem (hypothermia) existed, and to identify barriers and opportunities for improvement. These findings were incorporated into the “Plan” stage of the larger, more comprehensive, PDSA cycle.

PDSA cycle

Step 1: “Plan” (preparing for the clinical audit)

The planning stage is critical to the success of the audit cycle and considers: identification of the process(es)/outcome(s) to be studied based upon responses to the 3 preparatory questions (above), identification of the Who, What, When, and Where of the cycle, and preparation for data collection (“Do” phase) (8,13). The 3 preparatory questions should result in a statement of the aim/purpose of the audit cycle (1). An example of a purpose statement could be: “To improve the management of postoperative hypothermia in dogs presenting for non-urgent procedures.” However, such a statement is vague because no specific target is identified. Should the target be to achieve 0% hypothermia after surgery? This is ideal, but probably unrealistic. A more realistic target might be, “90% of dogs returning to normothermia within 4 h after surgery.” This latter statement recognizes that maintaining normothermia during surgery is difficult, but that management of hypothermia can be improved.

Once the purpose statement has been created, the groups involved and resources required are identified (7). This allows an appraisal of whether the selected outcome can be realistically manipulated and (hopefully) improved (14). For the hypothermia example, it is important to ensure that there are sufficient resources (staff and equipment) to support the goal. Similarly, it is important that key parties are invested in the project: presenting the current situation (including the current incidence of hypothermia), the aim of the audit cycle, the reasons why reducing the incidence of hypothermia is important, the planned changes and contributions of staff, are more likely to result in support and eventual success (5). Absence of staff involvement is a common source of failure or of resistance to bring about change (8,27). Similarly, a key aspect of successful audit is creating an environment in which the goal is improvement rather than placing blame on specific groups or individuals (8). Finally, an initial timeframe for data collection (the “Do” phase) can be set out. This is determined based upon local constraints but should set a minimum time to have a chance of identifying an improvement (if one were to occur; this could naturally be determined by caseload, for example) and be short enough that lack of success is not unnecessarily prolonged. Visualizing and interpreting data as they are collected (see Step 3: “Study”) can identify when change has occurred and a suitable time to stop data collection.

From this “Plan” stage, the following was described for the hypothermia clinical audit:

1. Purpose statement: “to have 90% of dogs return to normothermia within 4 h after surgery.”

2. The results of the short data collection period (identifying an 88% incidence of postoperative hypothermia) were presented to the practice, with attendance by the majority of veterinarians, registered veterinary technicians, and technician assistants. An in-person meeting was appropriate for the size of clinic but for larger clinics, or those in which veterinarians work in shifts, more than 1 meeting, or use of e-mail (with a deadline for comments) or posters (for example) could help ensure information was widely communicated.

3. A general discussion achieved a consensus to: a. measure rectal temperature hourly when hypothermia was identified at extubation (included in recovery area monitoring by technicians: Who and Where); b. implement a practice-wide target temperature of 37.5°C (What); c. use forced air warmers and continue temperature measurements until the target temperature was achieved; and d. move unused forced air warmers from the surgical floor to the recovery area at the end of the day to ensure equipment availability. Data collection was planned for 1 mo (When) with a view to interpreting the data at this stage (Step 3: “Study”) and deciding the next step(s) (Step 4: “Act”).

Step 2: “Do” (carrying out the plan)

This is the data collection phase. Once the audit is in progress, monitoring the data collection process itself helps ensure that data are being collected properly (staff included in data collection should be appropriately instructed and supported) (28). Time to collect data should be budgeted into the workload of participants to maximize the likelihood of complete and accurate reporting. There should be the opportunity to record problems encountered as these will help identify causes of audit failure, or limited improvement gains, and can be incorporated in future audit cycles. This could be achieved informally, such as asking for feedback during regular clinic meetings/rounds, or by making available a box in which suggestions can be placed, or providing an e-mail contact (specific person or dedicated e-mail address). Ongoing visualization of data can provide rapid feedback on progress towards achieving a change and serve to motivate participants if goals are set.

In large-scale, multicenter/national audits it can be invaluable to have people dedicated to collecting data, monitoring progress, and providing feedback (29). This approach facilitates early identification of factors contributing to likely failure to implement change(s), which can then be addressed. Software can also be used to facilitate data collection (30).

In the hypothermia example, a poster summarizing the changes being implemented was prominently displayed. Additionally, the line in the postoperative record sheet for recording temperature was highlighted. To minimize data loss, a designated person was responsible for inputting data into an electronic spreadsheet weekly.

Step 3: “Study” (data analysis and interpretation)

Following data collection, graphs are often used to visualize and interpret the data. The data should be discussed among colleagues to obtain a wide range of perspectives (9). This discussion is essential to complete the current audit cycle (moving to the “Act” phase) and prepare for future audits.

A common method of displaying collected clinical audit data, which facilitates tracking of progress, is the run chart (Figure 2) (31). Run charts are an example of control charts and more sophisticated forms, such as cumulative summation plots, are used in statistical process control, a formal technique to evaluate quality control/improvement (3,32,33). Run charts facilitate rapid assessment of the data as they are collected, allowing users to see emerging patterns, including deviations from an audit goal. They are simple and fast to compile, and easy to interpret. The x-axis of a run chart represents a measure of time, such as days or visit/procedure number. The y-axis represents the subject of study e.g., client time in waiting room (Figure 2). Targets to achieve or avoid can be plotted on the y-axis (e.g., Figure 2 shows a desired maximum wait time of 10 min after scheduled appointment time), enabling a rapid assessment of progress and motivation.

Figure 2.

Run chart of fictitious data illustrating percentage of clients meeting a goal of 10 min or less (dashed horizontal line) for time spent waiting for appointment to start. The historical median wait time is 15 min (dotted horizontal line). A trend (5 or more consecutive points moving in the same direction) and an astronomical point (deviant data point) are identified. These could result from the knock-on effect of an appointment running late and a cancellation, respectively. The impact of a new DVM requiring longer appointments contrasts with an experienced practitioner (cases 32 onwards).

Beyond subjective observations, several standardized rules can be applied to aid run chart interpretation, including identification of a “trend” and “astronomical point” (13,31). These rules help identify non-random changes in the data and limit the risk of over-interpreting small changes. A trend is identified when 5 (or more) consecutive data points all move in the same direction (either upwards or downwards). Identifying an astronomical point is based on subjective evaluation of the run chart and is indicated by a datum that is striking by its deviation from the adjacent data. More complex interpretations of run charts can also be applied using statistical methods (3,13,31,34).

For the hypothermia clinical audit, a box plot was used to track the percentage of animals achieving normothermia within 4 h over the 4 wk following the introduction of the temperature management changes (Figure 3a). Data were compiled weekly as this fitted in with the work schedule of the person responsible for managing data. A run chart, reflecting rectal temperatures at the target 4-hour time point for individual dogs could also be applied to provide detail at an individual case level (Figure 3b).

Figure 3.

a — Circles indicate percentage of animals normothermic at 4 h after recovery for 6 days of the week (Monday–Saturday). Box plot (box limits are range, central horizontal line is median: calculated from daily percentage values). The initial positive effects of changes introduced during the audit begin to wane in weeks 3 and 4 (lower median and daily percentages and increased data variability). Tracking performance identifies this change and can be used to trigger a more detailed investigation to understand the basis of the change (e.g., individual case tracking, Figure 3b) or a further audit cycle. b — Run chart showing rectal temperature for sequential cases anesthetized before (cases 1–9) and after (cases 10–28) warming management changes introduced (“Temp mx change”). Changes result in more cases achieving the target temperature and less variability between cases.

The duration of data collection is somewhat arbitrary, often based on a prediction of caseload so that enough cases are collected to be representative. In general, though collecting more data is informative, compliance in maintaining introduced changes and/or continuing data collection is likely to wane over time (especially if data are not visualized on an ongoing basis). In contrast, too short a period of data collection may not provide an accurate reflection of reality and improvements may reflect the Hawthorne effect (a change in behavior as a result of being observed), with a return to previous performance levels once data collection stops (35,36).

Step 4: “Act” (decision for further action based on results)

After data collection and interpretion of results, the current audit cycle is completed by defining and implementing further action to continue improvement (or address identified problems). This could be as simple as intermittent periods of data collection to confirm improvements are maintained. Further audit cycles may be planned, with the goal of continuous improvement (or to test alternative approaches in the case of an unsuccessful outcome). The results of the hypothermia audit were presented to clinic staff, providing the opportunity to reflect on the process and discuss strategies for further improvement (the initial target was achieved but a small number of animals remained hypothermic beyond the 4-hour goal).

Factors associated with failure of clinical audit

Successful audit depends on a careful selection of the audit subject (see preparatory questions) and a supportive environment (9,27). Environment can be considered as structure and culture (8). Structure refers to ensuring adequate time and resources are available to conduct an audit (from preparation through to data collection and presentation). This could include time to identify relevant guidelines when establishing outcomes/processes to measure and audit targets as well as the time and resources to collect and interpret data. Culture refers to a collective desire to promote and sustain improvement, including supporting openness, so that errors/failures can be identified without blame (37,38). All phases of the audit cycle are more likely to be successful if the process is inclusive, with staff involvement. The extent of involvement varies, from consultation and discussion during the planning phase to participation in data collection and review of results (39). There is an important distinction between a successful audit cycle, that is, completion of the audit cycle, and success in generating an improvement. Failure to achieve improvement does not necessarily constitute a failed audit process: it may be that the action taken had limited impact on the outcome of interest or was not closely related to the outcome. Failure to achieve improvement still provides useful information to be incorporated in subsequent audit cycles.

Ethical review for clinical audits

The overarching goal of clinical audit is quality improvement, resulting from observing and evaluating processes and outcomes against explicit criteria. Thus, the end result should be an improvement in the quality of care. Understanding the audit cycle process provides a framework for ethical review. It is simplistic to assume that clinical audit does not require ethical review (28,40–43). Though clinical audit often differs from prospective clinical research in that animals are not randomized to receive different treatments (with the possibility that a subject receives a less effective/placebo treatment), there are several areas that warrant ethical review: 1. There may be randomization to assess the effect of different interventions in improving standard of care; 2. The change introduced with the goal of improving care may be associated with risk (perceived or actual); 3. Safeguarding of patient data; 4. A radical change in care/practice that could have unforeseen effects; and 5. Re-allocation of resources from other patients as a result of a change in care.

Publishing quality improvement studies

The number of published clinical audits in veterinary medicine is limited, although it is unclear if this reflects the number of audits performed, the selection process of journals or the number of audits submitted for publication (5). The publication of clinical audits should be encouraged. This is particularly important in veterinary medicine, where evidence of criteria associated with improvement is limited: published audits will disseminate information on the success/failure of changes implemented. Furthermore, as evidence-based guidelines are developed, clinical audit can help evaluate if guidelines are achievable in practice and written in a way that the relationship between process(es) and outcome is clear. To facilitate clinical audit publishing and standardize reporting, reporting guidelines are available: SQUIRE (Standards for QUality Improvement Reporting Excellence) (44, 45). These guidelines are also helpful when planning clinical audits, providing a framework of items to consider.

The quality of clinical audit is reflected by completion of the audit (PDSA) cycle, as well as the selection of measurable criteria, reliable and accurate data collection and analysis, and reporting and interpretation of the results (8).

In conclusion, clinical audit has the potential to promote and maintain high standards of patient care. Available literature shows that clinical audit is being discussed and applied in veterinary medicine (1,4,5,7,30,46–48), with evidence that complete audit cycles are being performed (1,3,7,47–49). A current barrier to the widespread adoption of clinical audit may be a relative lack of evidence-based guidelines in veterinary medicine. Nevertheless, clinical audit can and should be used to assess performance against widely accepted (best-practice) standards of care. The importance of clinical audit cannot be overstated. With the increase in volume of published literature, a gradual move towards evidence-based veterinary medicine, increased focus on the relationship between human and veterinary medicine (“One Health”), and increasing public awareness, steady improvement of practice standards is a challenging yet critical task. Clinical audit is a useful, accessible, under-utilized tool to achieve this goal.

Figure 4.

a — Bar graph of fictitious data showing relationship between weekly cystitis cases and number of cultures performed. b — Run chart documenting duration of antimicrobial prescriptions for fictional cases of cystitis. Upper and lower broken horizontal lines indicate target prescription duration window. Case 17 could be classified as an astronomical point (see text for details), warranting investigation.