Article Text

Download PDFPDF

Survey on conduct of anaesthetic monitoring in small animal practice in the UK
  1. Elisabeth Marie Richardson and
  2. Matthew McMillan
  1. Department of Veterinary Medicine, University of Cambridge, Cambridge, UK
  1. Correspondence to Dr Elisabeth Marie Richardson, Davies Veterinary Specialists Limited, Higham Gobion SG5 3HR, UK; richardsonliz40{at}


Background An online survey was used to investigate current anaesthetic monitoring practices and the confidence level of personnel monitoring anaesthetics in small animal practices within the UK.

Methods Veterinary surgeons (VSs), registered veterinary nurses (RVNs) and student veterinary nurses (SVNs) working in the UK were invited to participate in an anonymous, internet-based survey. To gather data, the questions used free text, multiple choice or scales measuring respondent attitude or opinion. No questions were mandatory and data were analysed with descriptive statistics or inductive thematic analysis.

Results 524 valid surveys were completed and included in the data analysis (VS n=136, RVN n=307, SVN n=81). The results indicated mainly RVNs perform pre-anaesthetic monitoring equipment checks, set-up the monitoring equipment and monitor anaesthesia and are more confident than VSs monitoring anaesthetics. VSs, RVNs and SVNs were all recognised to interpret and address changes in parameters monitored. Critical tasks pertaining to anaesthetic monitoring are being performed by personnel other than a VS, RVN or SVN. Respondents recognised the importance of monitoring in relation to patient outcome; however, a considerable proportion of respondents indicated that improving standards of monitoring was not a priority in their practice. Most respondents felt that standards of monitoring could be improved and that financial constraints were the major factor limiting improvement. Most respondents felt they would benefit from further training in anaesthetic monitoring.

Conclusion Variability exists in how anaesthetic monitoring is conducted. Workplace pressures afflicting veterinary staff can influence the conduct of anaesthetic monitoring and initiating change within a veterinary practice can be difficult.

  • Anaesthesia
  • Monitoring
  • Conduct
  • General practice
View Full Text

Statistics from


It is well recognised that monitoring plays a key role in the conduct of safe anaesthesia. Monitoring can range from hands-on techniques (such as assessing jaw tone and palpation of pulses) to the use of advanced electronic equipment. Monitoring reduces the risk of incidents and accidents by detecting consequences of errors and providing early warning signs of patient deterioration.1 Analysis of reported peri-anaesthetic events in human anaesthesia suggested a combination of pulse oximetry, capnography and blood pressure would have allowed earlier identification of 93% of incidents.2 3 In veterinary medicine, the Confidential Enquiry into Perioperative Small Animal Fatalities (CEPSAF) found that pulse monitoring and pulse oximetry were associated with reduced odds of anaesthetic-related death and suggested closer monitoring might aid a further reduction in mortality rates.4

It is, however, important to note that monitoring in isolation cannot improve patient safety. The information monitoring is only useful if correctly analysed and interpreted so that action may be taken when a problem is recognised. Therefore, the ability and confidence of the anaesthetist to interpret and act on changes in physiological variables are perhaps more important that the monitoring itself.

Various organisations such as the Royal College of Veterinary Surgeons (RCVS) provide guidelines for anaesthetic monitoring which aim to improve the level of anaesthetic care for veterinary patients (see online supplementary appendix 1). These guidelines do not supersede national laws, but are considered essential for veterinary surgeons (VS) in maintaining their professional responsibilities and regard for animal health and welfare. The RCVS Codes of Professional Conduct (CPC) state that VS must maintain minimum practice standards equivalent to the Core Standards of the RCVS Practice Standards Scheme (PSS).5 The PSS makes no specific requirements for specific modalities of monitoring required to meet the minimum practice standards. It does, however, within the higher levels of the PSS, provide more specific requirements, such as ‘at least one monitoring device must be available e.g. oesophageal stethoscope, pulse oximeter, capnograph or electrocardiogram (ECG)’.5 Core standards are relevant to all veterinary practices and reflect primarily legal requirements which must be followed when running a veterinary practice, together with guidance as set out in the CPC. The Association of Veterinary Anaesthetists (AVA) provides monitoring specific guidelines for safer anaesthesia.6 These include guidelines on anaesthetic case planning, staff who monitor anaesthetics, monitoring equipment to be used, patient support during anaesthesia, patient care during recovery, clinical training for staff involved with anaesthesia and anaesthetic record keeping (see online supplementary appendix 2).

Little is known about the how the conduct of anaesthetic monitoring in small animal practice has changed since CEPSAF. This study was designed to assess aspects of anaesthetic monitoring in small animal practices within the UK. Primary aims were to survey VS, registered veterinary nurses (RVN) and student veterinary nurses (SVN) to ascertain current anaesthetic monitoring practices and the confidence level of personnel monitoring anaesthetics. The overall objective was to establish baseline data on the conduct of monitoring to promote professional development through targeted education.

Materials and methods

A cross-sectional, opportunistic, voluntary survey of veterinary personnel engaged in small animal practice in the UK was conducted.

The commercial online program Qualtrics was used to construct and host the survey. A preliminary survey of 10 VS and 10 RVN was performed to optimise survey format. The final survey consisted of 33 questions (see online supplementary appendix 3 for full survey). Consent was obtained for the data submitted to be analysed and reported, including the use of direct quotes. No questions were mandatory. All data were submitted anonymously, managed in a confidential fashion and stored on the University of Cambridge’s Qualtrics Server.

The questions were designed to obtain demographic data of respondents and their experiences of the conduct of anaesthetic monitoring in practice.

To gather data, the questions used free text, multiple choice or scales measuring respondent attitude or opinion. Qualitative data were gathered in two questions using free-text boxes, where respondents were asked to outline any difficulties or challenges they had encountered monitoring anaesthesia, and express any additional comments they felt important when considering anaesthetic monitoring in their practice.

Small animal VS, RVN and SVN practising in the UK were invited to participate in the survey which was distributed via social media and direct email communication. A letter outlining the purpose of the study was published in the Vet Record7 and made accessible via Uniform Resource Locator link and Quick-response code. The survey was open during October and November 2017.

Questions were checked for clarity and validity through interrogation of the results with the questions: did respondents understand the question, and, did the responses consistently answer the question as intended?

Statistical and qualitative analysis

Demographic and quantitative data were analysed with descriptive statistics.

Closed-ended questions which allowed respondents to select single answers were presented as total response counts per respondent category. Closed-ended questions that allowed respondents to select multiple answers were presented as total choice counts per respondent category.

Qualitative data were analysed using inductive thematic analysis, in which patterns are identified through a rigorous process of data familiarisation, data coding, and theme development and revision. These patterns were identified as data having meaning across the dataset which directly related to the research questions being addressed. The methodology for carrying out the thematic analysis was that prescribed by Braun and Clarke8 ,8 (see online supplementary appendix 4 for details).

Where responses of individual groups could be clearly differentiated, the data were reported as being from RVN, VS or SVN. Where the responses came from the global dataset, then they were reported as respondents.


In all, 524 valid surveys were completed and included in the data analysis. As all questions were optional, respondent number per question was variable. Question 15 (see online supplementary appendix 3) was removed from subsequent analysis as respondent comments indicated a lack of detail specifying what maintenance, calibration and renewal of equipment entailed.

Demographic data

The results of the surveyed demographical characteristics of the respondents are summarised in tables 1 and 2.

Table 1

Demographic distribution and practice data of respondents presented as response counts per answer category

Table 2

Demographic data from closed-ended questions that allowed respondents to select single answers (presented as total response counts)

Quantitative data

Results of the questions gathering quantitative data are summarised in tables 3–5.

Table 3

Data from closed-ended questions that allowed respondents to select single answers (presented as total response counts per respondent category)

Table 4

Data from closed-ended questions that allowed respondents to select multiple answers (presented as total choice counts per respondent category)

Table 5

Percentage proportion of respondents using various hands-on and electronic monitoring techniques, in their healthy compared to their sick patients

A histogram displaying the comfort levels of VS, RVN and SVN when interpreting haemoglobin oxygen saturation (SpO2), anaesthetic depth, Doppler non-invasive blood pressure (NIBP), oscillometric NIBP, capnograms, end-tidal carbon dioxide tensions (PETCO2), pulse oximeter waveform and ECG is presented in figure 1.

Figure 1

Histograms representing respondent comfort levels when interpreting haemoglobin oxygen saturation (SpO2), anaesthetic depth (Depth), Doppler NIBP, oscillometric NIBP, capnograms (capnograph waveforms), end-tidal carbon dioxide tensions (PETCO2), pulse oximeter waveform (pulse waveforms) and ECG. Y-axes represent choice counts per answer category. NIBP, non-invasive blood pressure; RVN, registered veterinary nurse; SVN, student veterinary nurse; VS, veterinary surgeon.

Global comfort levels (VS, RVN and SVN responses combined) in using and interpreting SpO2, depth of anaesthesia, Doppler NIBP, oscillometric NIBP, PETCO2, capnograms, ECG and pulse oximeter waveform are presented in figure 2.

Figure 2

Comfort level of combined VS, RVN and SVN responses in using and interpreting haemoglobin oxygen saturation (SpO2), depth of anaesthesia (Depth), oscillometric blood pressure (Oscillometric NIBP), Doppler blood pressure (Doppler NIBP), end-tidal carbon dioxide tensions (PETCO2), capnograms (capnograph waveforms), ECG and pulse oximeter waveform (pulse ox waveforms). Data are presented as percentage of choice counts per answer category. NIBP, non-invasive blood pressure; RVN, registered veterinary nurse; SVN, student veterinary nurse; VS, veterinary surgeon.

A histogram displaying confidence levels of the respondents who actively monitor anaesthetics in performing various tasks is presented in figure 3.

Figure 3

Histogram displaying the confidence levels of respondents who actively monitor anaesthetics in performing various tasks (0=no confidence to 100=fully confident). RVN, registered veterinary nurse; SVN, student veterinary nurse; VS, veterinary surgeon.

Factors investigated in both the CEPSAF and present study are compared and displayed in table 6.

Table 6

Comparison of data obtained from CEPSAF to the present study

Qualitative data

There were 343 (VSs n=102, RVNs n=186, SVNs n=55) responses entered when respondents were asked to outline any difficulties and challenges they had encountered when monitoring anaesthesia. Data analysed from this question are displayed in figure 4.

Figure 4

Coding frame organising data obtained from respondents outlining difficulties and challenges encountered when monitoring anaesthesia in small animal practice. The global theme (blue), subthemes (grey) and codes (orange) are displayed. Number in brackets indicate frequency of code occurrence and example quotes (yellow) are also displayed. RVN, registered veterinary nurse; VS, veterinary surgeon.

There were 70 (VSs n=27, RVNs n=38, SVNs n=5) responses entered when respondents were asked to enter any additional comments they felt important. Data analysis from this question is displayed in table 7.

Table 7

Data gathered from additional comments left by respondents after thematic analysis


The results of this survey indicate how anaesthetic monitoring is currently conducted in UK small animal practice, as well as the confidence of the staff responsible for recording and interpreting the data provided. Our online survey asked VS, RVN and SVN to share details of their experiences, using a combination of quantitative and qualitative questions. The results revealed variability in how anaesthetic monitoring is conducted, as well as the associated pressures encountered in a practice environment, stressors affecting veterinary staff as a consequence and that initiating change can be difficult.

The most recent study of anaesthetic deaths reported incidence of anaesthetic and sedation‐related deaths to be 0.17% in dogs and 0.24% in cats.9 This is higher than the 0.01%–0.02% reported in human anaesthesia in developed countries.10 11 Brodbelt hypothesised that ‘such a difference is likely to reflect differences in standards of anaesthesia in human and veterinary anaesthesia more than species differences, and that anaesthetist expertise and resources available in the medical setting are significantly greater than that routinely available in veterinary practice’.9 Interestingly, a human study concluded overall rates of perioperative and anaesthetic-related mortality to be two to three times higher in developing countries.12 This suggests there is substantial scope for further improvements in veterinary practice. The comparison of data displayed in table 6 suggests that use of pulse oximeters, ECG, NIBP and capnography, and the routine monitoring of body temperature of patients has increased from the time of the CEPSAF to the present study. These increases are likely to be due to awareness raised to the importance of patient monitoring by the CEPSAF. In contrast, routine use of the oesophageal stethoscopes appears to have decreased since the time of the CEPSAF. This might suggest an increase in reliance on electronic monitoring equipment. However, the present study found around half of respondents routinely monitor pulse rate manually, and the high percentages of respondents using ‘hands-on’ techniques (eg, assessing mucous membrane colour, jaw tone, palpebral reflex) might be an indication they felt most comfortable with assessing depth of anaesthesia in this way. The basic senses of a trained anaesthetist can provide invaluable information and are available at all times without the need for specialised equipment. Our data suggest respondents are using their senses to monitor patients and focus on the animal itself, rather than relying on electric monitoring modalities alone.

In human medicine, the introduction of routine monitoring in anaesthesia coincided with numerous improvements in clinical facilities, training and other factors likely to affect patient outcomes.13 The progressive reduction in anaesthesia-related morbidity and mortality is therefore linked to instrumental monitoring by association rather than proof from prospective randomised trials. This circumstantial evidence was viewed as a clear indication that the use of such monitoring improves patient safety. Consequently, the Association of Anaesthetist of Great Britain and Ireland(AAGBI) produced clear recommendations about the standards of monitoring which human anaesthetists must use.13 Recently, The AVA has followed suit, albeit with less stringent and detailed guidelines.6 There was strong agreement by respondents that monitoring is important to patient outcome. Greater than 80% of all respondents felt patients were being adequately monitored, conversely around half of all respondents agreed standard of monitoring needed to be improved.

These results indicate that in small animal practice, it is mainly RVN who perform pre-anaesthetic monitoring equipment checks, set-up the monitoring equipment and monitor anaesthesia. Both VS and RVN agreed nurses are more confident than VS in monitoring anaesthetics. The RCVS CPC states ‘monitoring a patient during anaesthesia and the recovery period is the responsibility of the veterinary surgeon, but may be carried out on his or her behalf by a suitably trained person’ and ‘the most suitable person to assist a veterinary surgeon to monitor and maintain anaesthesia is a veterinary nurse or, under supervision, a student veterinary nurse’ (see online supplementary appendix 1).14 Our survey found that critical tasks pertaining to anaesthetic monitoring are performed by personnel other than a VS, RVN or SVN. Unfortunately, the survey did not allow respondents to elaborate further.

An interesting finding in this survey was that, RVN VS, SVN and other personnel were all acknowledged to interpret and address changes in parameters monitored. The RCVS CPC advises that diagnosis and treatment of a problem requires a VS to perform or authorise. It would be interesting to conduct further studies to identify who makes the minute-to-minute decisions and who makes the more complex, higher-level or executive decisions.

Almost all respondents agreed monitoring is important for patient outcome; however, a considerable proportion of respondents suggested that improving standards of monitoring was not a priority in their practice. Most respondents agreed standards of monitoring could be improved; about one half stated that the standard of monitoring needed to be improved; and the majority of respondents felt they would benefit from further training. Financial constraints were the major factor limiting improvements of anaesthetic monitoring, although a lack of priority, knowledge and awareness were also mentioned. These findings imply that a degree of resistance exists towards the concept of improving standard of monitoring. This was supported by comments made by respondents in free-text boxes (see figure 4). Improving standards is technically achievable in most settings. It is important to highlight that early detection of untoward trends or events during anaesthesia will not only result in prevention or mitigation of patient injury but also that this, in turn, may help counter potential for anesthesia-related malpractice actions.

Respondents suggested that experience enhances knowledge and that undergraduate training may be inadequate. It was also suggested that a lack of understanding and knowledge of monitoring leads to a lack of confidence, which, in turn, leads to stress. Some respondents suggested that practices may possess certain pieces of monitoring equipment, but do not use it during anaesthesia due to a lack of adequate training. These comments applied to both VS, RVN and SVN. The AAGBI states the presence of an appropriately trained and experienced anaesthetist is the main determinant of patient safety during anaesthesia.13

Respondents stated they were more comfortable interpreting SpO2 than pulse oximeter waveforms. Interpretation of the pulse oximeter waveform is important because it can provide a pulse rate, information on changes in pulse volume and indicate the reliability of the SpO2 value provided. The waveform produced by a pulse oximeter must be assessed by the anaesthetist to confirm the monitor is determining the correct rate—waveform and rate should be assessed in conjunction where possible. It is possible some respondents possessed pulse oximeters which display only SpO2 and pulse rate but not a pulse waveform, and so are not as experienced with waveform interpretation.

Monitoring modalities least often employed by respondents were capnography and ECG, which correlated with VS feeling least comfortable using and interpreting capnograms, and with RVNs least comfortable with ECG. Respondents indicated they were more comfortable interpreting PETCO2 than the capnogram itself. This is important as together these components provide extremely valuable information about metabolism, pulmonary perfusion, alveolar ventilation, respiratory patterns and elimination of carbon dioxide from the anaesthetic breathing system. In hindsight, it would have been useful to know whether respondents fully understand the origins of the normal capnogram, as this provides the foundation for interpretation of the abnormal.

RVN indicated they were more comfortable using and interpreting Doppler compared to oscillometric NIBP measurements, whereas VS were more comfortable with oscillometry. Both techniques have been shown to produce an approximation of intra-arterial blood pressure (IBP), and are easier to perform than IBP. However, these NIBP techniques do not allow interpretation of the arterial pressure waveform, are inherently less accurate than the IBP technique and may fail to give readings in certain circumstances such as hypotension or vasoconstriction.15 There is evidence to show that the Doppler technique may provide more reliable measurements than the oscillometric.16 17 Doppler requires some operator experience and extent of training in use of both NIBP techniques may be a factor to consider when assessing comfort levels of their use and interpretation.18 A possible explanation for the differences between VS and RVN may be differences in experience or training in using these modalities.

Respondents indicated that use of all the monitoring techniques listed to be increased in anaesthesia of sick compared to healthy patients, and IBP to be monitored in 8.7% of their sick patients. As critically ill patients are at greater risk of haemodynamic instability, accurate blood pressure monitoring is of even greater importance. Arterial blood pressure is monitored as a guide to organ perfusion during anaesthesia.19 In our study, NIBP (both oscillometric and Doppler) was monitored in 61.2% of patients considered healthy and in 84.7% of those considered sick. IBP measurement is more accurate, but requires more expensive equipment, technical skills and experience for use. In this study, the difference seen between use of IBP and NIBP in sick patients may result from respondents not having access to equipment, technical skills or the experience necessary for IBP monitoring.

This survey found temperature was monitored in 70.1% of healthy and 84.7% of sick patients, demonstrating that although some monitoring equipment is not only readily available and simple to interpret, it is not always used. Intraoperative hypothermia can result in prolonged recovery and increase risk of postoperative complications.20 All anaesthetic drugs depress the thermoregulatory centre and cause hypothermia, involving an initial rapid decrease in core body temperature within the first 20 min of anaesthesia due to redistribution of heat from the core to periphery.21 It is therefore important to monitor a patient’s body temperature from pre-anaesthetic medication administration and throughout anaesthesia, whatever the duration. Corrective measures may then be instigated when necessary to prevent further heat loss. It is of concern that the findings of this study suggest temperature during anaesthesia is not being monitored routinely in all cases. Respondents highlighted a lack of access to patients during anaesthesia (see figure 4). This, and the possibility that respondents were considering short duration anaesthetics for routine procedures, might explain the apparent lack of attention to patient body temperature indicated.

Respondents also highlighted that patients may not be adequately monitored during pre-anaesthetic sedation and recovery (see figure 4). Most anaesthetic deaths occur in the postoperative period22 and the main contributing factors are lack of monitoring and observation. Loss of body heat during the post-anaesthetic period is one factor contributing to the risk of fatality. It is therefore critical that body temperature is monitored until the patient is normothermic and fully recovered.

Responses from within all respondent groups stated a ‘lack of staff’ and increased ‘time pressures’, indicating that RVN and SVN are having to juggle numerous jobs, which ‘distracts’ them from monitoring and can lead to ‘poor anaesthetic record keeping’ and ‘undertrained personnel’ having to step in to monitor anaesthesia. It also leads to a lack of attention to patients who are sedated and recovering from anaesthesia. In the study of Clarke et al,23 the major factor contributing to anaesthetic deaths was lack of direct observation when problems first arose; they concluded that monitoring is essential from the time drugs are administered until recovery to full consciousness. Respondents to our survey reported that they had other tasks to perform, either all of some of the time while monitoring anaesthesia. SVNs, however, seem to be left undisturbed and allowed to focus on the job in hand. An interesting finding was that 36.8% of the VS respondents did not have other jobs to perform.

Poor staffing meant inexperienced SVN are not always supported appropriately according to some respondents. A ‘flippant’ attitude towards monitoring, or ‘overconfident’ anaesthetists was reported by others. These individuals were said to have outdated knowledge or exhibit an ‘old-school’ approach which may have been compounded by a lack of continuing professional education, a failure to update their knowledge and a reluctance to change. A ‘lack of interest or priority’, a ‘reluctance to change or invest’ and ‘cost or budget limitations’ were emphasised. All respondents involved in monitoring anaesthetics indicated they are left with either a lack of or faulty pieces of monitoring equipment. RVN, VS and SVN felt the differing attitudes towards anaesthetic monitoring led to poor teamwork, staff relationships and lack of trust between members of the team. In contrast, a number of respondents reported to have ‘no concerns’ with their experiences in anaesthetic monitoring and acknowledged its importance; a high proportion expressed enthusiasm to enhance their knowledge of the subject. Respondents agreed that lack of interest in monitoring was the least significant factor limiting improving standards.

Flaws in survey design as highlighted by respondents are possible limitations. Better wording and formatting make questions less ambiguous and so may have enabled the generation of more consistent and accurate responses. We cannot be certain whether the data reported is representative of the UK population. The most recent figures published by the RCVS24 showed there were 24 852 VS, 16 351 RVN and 5812 SVN active in the UK in 2016. Therefore, respondents to the present survey represent 1.1% of the UK total (0.5% of VS, 1.9% of RVN and 1.4% of SVN). We acknowledge these population totals do not represent only those in UK small animal practice; however, it was not possible to find numbers of VS, RVN and SVN in UK small animal practice alone. Also, as a greater proportion of RVN and SVN completed the survey, it is possible the data provided by them might be more representative of their population groups than that for VS. A modification of response bias might exist due to respondent awareness of being observed and responses may have been modified to sound more positive or appropriate.

Sending out paper copies to practices might have increased the response rate, but it was decided that a more environmentally friendly, cost-effective and efficient way of circulating the study was to make it available online. Responses were accepted for 8 weeks, which may also have contributed to the low return rate. It is also possible the time taken to undertake the survey discouraged some individuals from completing it. The total number of surveys started (n=1,266) compared to those completed (n=524) suggests this was indeed the case. Making all questions mandatory might have provided more data for analysis, but the authors felt this would discourage respondents.

Distribution of the survey via direct communication with corporate veterinary groups and charitable organisations means it was possible their staff may have been over-represented in the results. The demographic data revealed 40% of respondents reported to be from corporately owned veterinary practices and 60% of respondents from independently owned veterinary practices. This evidence suggests that the survey was circulated beyond cooperate groups. However, the extent of representation by charitable organisations is not known.

The data were verified by a second person to strengthen validity; respondent counts and statistical calculations were checked for accuracy and the organisation of qualitative data into themes and codes were checked for appropriacy. However, a perception bias may still exist due to the limitations of self-interpretation and reporting of the results.

Individuals with a special interest in anaesthesia may have been more likely to complete the survey; therefore, a volunteer bias may exist towards respondents more likely to conduct high standards of anaesthetic monitoring. This means results might be interpreted as a ‘best-case’ scenario. Conversely, a volunteer bias may exist towards respondents with more negative experiences of anaesthetic monitoring, and so it could be argued the results might be interpreted as a ‘worst-case’ scenario.

Despite these limitations and flaws, we consider the data to be a valuable as an initial exploration into the practice of anaesthetic monitoring in small animal practice. We hope that the data reported will raise awareness and interest and lay the groundwork that will lead to future studies.

In conclusion, this study suggests that it is predominantly nurses who monitor, interpret and address changes they encounter. It also suggests that although comfortable using electronic monitoring equipment, respondents feel they would benefit from further training and that anaesthetic monitoring is often performed alongside other tasks.

This study provides data on the conduct of anaesthetic monitoring in small animal practice to raise awareness among the veterinary profession.


The authors would like to thank the Veterinary Record, those involved in circulating the survey and all respondents for participating.


View Abstract


  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Ethics approval Ethics approval was obtained via the University of Cambridge’s Department of Veterinary Medicine Ethics and Welfare Committee (reference CR279).

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement All data relevant to the study are included in the article.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.