Pet health insurance reduces the likelihood of pre-surgical euthanasia of dogs with gastric dilatation-volvulus in the emergency room of an Australian referral hospital

ABSTRACT

Aims

To determine the association between the presence of pet health insurance and the risk of euthanasia at the time of diagnosis for dogs with gastric dilatation-volvulus (GDV).

Methods

Insurance status at the time of GDV diagnosis was sought for a cohort of 147 non-referred, confirmed cases of GDV that presented to the emergency department of a university-based veterinary hospital in Australia between 2008 and 2017. Insurance status was obtained from the medical record (n=18) or after contacting the owners by phone using a standardised questionnaire (n=129). Animal, clinical and outcome data was retrospectively compiled in a research database. The primary outcome measure was whether or not the dog was euthanised before surgery. The Mantel-Haenszel procedure was used to quantify the association between the presence of pet health insurance and the risk of euthanasia at the time of diagnosis for dogs with GDV, adjusting for the confounding effect of age at the time of presentation using Bayesian methods.

Results

Of the 69 dogs for which insurance information could be obtained, 10 (14%) cases were insured at the time of the GDV event and 59 (86%) cases were not. The majority of non-insured dogs (37/59; 63 (95% CI=50–74)%) were euthanised before surgery, while none (0 (95% CI=0–28)%) of the insured dogs were euthanised at that time (p<0.001). Of the 32 insured and non-insured dogs that underwent surgery, four dogs (13 (95% CI=5–28)%) did not survive to hospital discharge. Three dogs (9%) were euthanised during or after surgery and one dog (3%) experienced cardiopulmonary arrest during treatment. The majority of dogs for which insurance status was known did not survive to hospital discharge (41/69; 59%), and 90 (95% CI=7–96)% of deaths were caused by euthanasia prior to surgery. Uninsured dogs were 5.0 (95% credible interval=1.8–26) times more likely to undergo presurgical euthanasia compared with insured dogs.

Conclusions

Euthanasia prior to treatment was most common cause of death in non-referred dogs with GDV; such euthanasia was entirely absent in the cohort of dogs that were insured.

Clinical relevance

Financial considerations significantly contribute to mortality of dogs with GDV presented to an emergency room. Financial instruments to reduce the out-of-pocket expense for pet owners confronted with unexpected veterinary expenses have potential to reduce pet mortality.

KEYWORDS:

• Gastric dilatation-volvulus
• pet insurance
• euthanasia
• veterinary cost
• emergency medicine

Introduction

Euthanasia of companion animals in the emergency room is commonly influenced by the financial limitations of pet owners (Aleccia 2009). Despite its apparent importance, the nature and the size of the problem of economic euthanasia in veterinary medicine is understudied. There is a paucity of systematic evaluation of financial interventions that have the potential to reduce the prevalence of economic euthanasia. Options available to clients to cover unexpected veterinary costs include payment plans, loans, pet health insurance, financial assistance from friends, family members and charitable organisations (Dunn 2006). We intuitively expect that increased rates of pet insurance would lead to higher rates of treatment for life-threatening conditions and thus reduced incidence of euthanasia for economic reasons.

Gastric dilatation-volvulus (GDV) is a common, acute, life-threatening condition in dogs which requires emergency surgery and peri-operative supportive care (Rudloff 2019). Without timely medical and surgical treatment, dogs with GDV will experience significant suffering and most likely die, thus humane euthanasia is considered the only ethically acceptable alternative intervention to prompt surgical treatment. However, surgery and perioperative care is expensive, and is a cost that some owners may be unable or unwilling to cover. We selected GDV to study economic euthanasia, as the condition is relatively inexpensive to diagnose and expensive to treat, but when treated surgically has a good prognosis with 80–90% of dogs surviving to hospital discharge (Mackenzie et al. 2010; O'Neill et al. 2017; Song et al. 2020).

In this study, we were interested in whether the decision by owners to euthanise dogs with GDV prior to surgery was economically motivated. Therefore the objective of this study was to determine the association between pet health insurance and the risk of euthanasia at the time of diagnosis for dogs with GDV presented to a university-based small animal emergency service. We hypothesised that in these dogs, the absence of pet insurance compared to the presence of pet insurance coverage would be associated with a higher risk of pre-surgery euthanasia.

Material and methods

Data source

An existing database containing all GDV cases that had presented to the Emergency and Critical Care department at U-Vet Animal Hospital since 1 January 2003 was searched for dogs that presented during the study period (1 January 2008–31 July 2018). This database was populated retrospectively and included all dogs with a diagnostic text of GDV included in their electronic medical record (EMR). Cases presented prior to 1 January 2008 were not considered as we anticipated extremely low insurance coverage, expiry of valid owner contact information and likely poor owner recall for cases occurring more than 10 years earlier. U-Vet Animal Hospital is an academic tertiary care veterinary hospital located in a less affluent, but rapidly developing part of suburban Melbourne, Australia. Hospital policy required a 50% deposit on estimated treatment costs and payment plans were not offered. However, financing through an external provider (VetPay Pty Ltd, Adelaide, SA, Australia) was available during the study period. The following information was recorded at the time of presentation: insurance status, referral status, breed, sex and age of dog, details of diagnostic tests performed (including radiographs and blood biochemistry analyses, if these were carried out), the occurrence of euthanasia prior to surgery and the final outcome of the case, if surgery was undertaken. Dog breeds recorded in the EMR were those reported by the pet owner at the time of presentation, and dogs reported as more than one breed were captured as mixed breed. A comorbidity was recorded as present if any pre-existing conditions were mentioned in the EMR, but the retrospective character of the study did not allow accurate collection of the nature and the severity of these pre-existing conditions. Blood lactate concentrations were determined from a citrated venous whole blood sample (ABL800 blood gas analyser; Radiometer Pacific Pty Ltd, Mt Waverley, VIC, Australia), and if several measurements were conducted prior to surgery, the one closest to presentation was recorded.

Study population

A series of exclusion criteria were applied to cases in the GDV database. Animals which were referred and those whose referral status was unknown were excluded, as owners of referred cases were likely informed of approximate costs of surgical treatment prior to referral and therefore prepared to pay for surgery. Referred and non-referred animals were those for which a clear referral statement was present in the medical records. In the absence of such a statement, the referral status was categorised as unknown. To be included in the study, cases required a clinical diagnosis of GDV by the attending clinician prior to surgery, based on history, physical examination and/or radiography. Radiographic confirmation of GDV was not mandatory for inclusion. Moreover, animals with characteristics considered to be clear confounders for a decision of presurgical euthanasia, were excluded such as GDV occurring in-hospital, ownership by government or organisations (e.g. police service dogs) or referral to another clinic for further treatment. In addition, we scrutinised the full text of the EMR for clear information on insurance status at the time of presentation for GDV.

Owner interview

Following the research protocol approval by the University of Melbourne Human Ethics Committee (ID 1851099), a single investigator (SA) attempted to contact all owners of dogs with unknown insurance status by phone. Owners were called in order starting from the most recent cases. A predefined calling regimen was followed. Owners were called repeatedly until they were contacted or until a total of five attempts were made. The purpose of the interview was to explain the aims of the research, to obtain oral informed consent to participate in the study, and to obtain insurance status. We used a standardised phone narrative (see Supplementary Information¹), with adaptive variations depending on the outcome of the case (i.e. death or survival) in order to convey empathy during the interview. The key question asked was whether the dog was covered by a pet health insurance policy at the time of GDV. The phone interviews were conducted over a 2-week period in June 2018.

A paper form was used to record client and animal details, the number of times and at what time of the day the client was called and responses to the key questions. If the client could not be reached, the reason for this was also noted. The data were then transferred to REDCap, a secure, web-based application designed to support data capture for research studies and hosted at the University of Melbourne (Harris et al. 2009).

Statistical analyses

Data were exported from REDCap to a commercial statistics package (JMP version 12; SAS Institute Inc., Cary, NC, USA) for further analysis. We used descriptive statistics to characterise variables recorded for insured and non-insured dogs and presented continuous data as mean and SD, or median, minimum and maximum and IQR, as appropriate after testing for normality using the Shapiro–Wilk test. Categorical data were presented as proportions and 95% CI. We used the Student’s $t$-test or Kruskal-Wallis test for continuous variables as appropriate, or the Pearson’s χ² test for categorical variables to identify differences in population characteristics (e.g. age, sex) between insured and non-insured dogs.

Unconditional associations between each of the patient characteristics hypothesised to influence the risk of euthanasia before surgery were quantified using the RR. Where the frequency of any cell of the 2×2 table was equal to zero we used a Bayesian approach to estimate the RR, in preference to the more commonly used Haldane-Anscombe correction (Haldane 1940; Anscombe 1956). In brief, the observed number of events in the insurance-positive and insurance-negative groups was assumed to follow a binomial distribution defined by parameters ${\pi }_{I+}$ and ${\pi }_{I-}$ (representing the probability of euthanasia for dogs in the insurance-positive and insurance-negative groups, respectively) and $n_{I+}$ and $n_{I-}$ (representing the number of dogs in the insurance-positive and insurance-negative groups, respectively). Priors for ${\pi }_{I+}$ and ${\pi }_{I-}$ were defined using beta distributions. Here, the point estimate and 95% CI of euthanasia prevalence for insurance-positive and insurance-negative dogs were used to select shape parameters for each beta distribution using the epi.betabuster function in the epiR package in R (Stevenson et al. 2020; R Core Team 2020, R Foundation for Statistical Computing, Vienna, Austria). The RR of interest was then calculated as ${\pi }_{I+}÷{\pi }_{I-}$. A joint probability distribution was formed by combining the prior distributions with the likelihood, given the observed data 2×2 contingency table data. Markov chain Monte Carlo methods implemented within the OpenBUGS software package (Lunn et al. 2009) were used to obtain samples from the joint posterior distribution to estimate the RR. Uncertainty around the computed risk ratio was expressed as a 95% credible interval.

Patient age and concentrations of lactate in blood were dichotomised for the analysis. Animals were classified as geriatric if their age met or exceeded reported median longevity for each breed (Michell 1999; O'Neill et al. 2013). Concentration of lactate was categorised as unmeasured, if a measurement was not present in the EMR, low if <6.0 mmol/L, or high if ≥6.0 mmol/L, as this cut-off has been shown to be associated with outcome (De Papp et al. 1999; Green et al. 2011; Rosenstein et al. 2018). The categorical explanatory variables tested included insurance status (present, absent), sex (male, female), age (non-geriatric, geriatric), breed category (purebred, mixed-breed), comorbidity status (present, absent) and blood lactate category (low, high, not measured).

Since age was causally associated with euthanasia and with insurance status, and age and insurance status were on two separate pathways as causes of euthanasia, we reasoned that age confounded the association between insurance status and euthanasia risk. Given the relatively small numbers of observations available for analysis we adjusted the RR estimate of the association between insurance status and euthanasia risk for the effect of age using the Mantel-Haenszel procedure (Mantel and Haenszel 1959). A Bayesian approach was used for the Mantel-Haenszel analyses, as described earlier (Efthimiou 2018).

Results

The GDV database included 279 dogs that presented within the study period, of which 107 were excluded due to their referral status (Figure 1). Review of the EMR for the remaining 172 cases, led to exclusion of 25 animals due to duplicate entry, erroneous GDV classification, GDV occurrence while hospitalised for another reason, referral to another clinic for treatment, death on arrival, or ownership conditions that constrained decision making at the time of presentation (i.e. rescue organisation involvement, animals in government care). Eighteen dogs for which insurance status could be affirmed from the EMR were directly included in the study. We attempted to contact the owners of the remaining 129 dogs by phone to obtain insurance status. This was successful for 51 dogs. Therefore together with the 18 cases identified by EMR review, a total of 69 dogs with GDV had confirmed insurance status and were included in the data analysis. For the remaining 78 dogs, either the pet owners could not be contacted (56/129; 43%), or consent to participate in the study was not granted (22/129; 17%).

Figure 1. Flow diagram describing inclusion and exclusion of cases in a retrospective analysis of the medical records (EMR) of a veterinary hospital for outcomes for dogs presenting to the emergency department with gastric-dilatation volvulus (GDV). Ownership indicates special ownership circumstances that constrain the euthanasia decision (i.e. rescue organisation, government).

Population characteristics, rates of pre-surgical euthanasia and survival to discharge of the 69 dogs with known insurance status were similar to the 78 dogs where insurance status was unknown (Table 1). Of the 69 dogs for which insurance information could be obtained, the five most prevalent breeds were Great Dane (10; 15%), German Shepherd (8; 12%), Boxer (5; 7%), Labrador Retriever (3; 4%) and Saint Bernard (3; 4%).

Table 1. Median (min, max) age and blood lactate concentration, and number (95% CI^a) of male, desexed, and purebred dogs that had co-morbidities and were euthanised or survived in a retrospective study of dogs presenting to a veterinary hospital emergency department Australia with gastric dilatation-volvulus, categorised by whether their insurance status was known or unknown.

	Insurance status
Variable	Known (n=69)	Unknown (n=78)	P-value
Age (years)	9.0 (2.0, 15)	9.0 (1.0, 15)	0.61^b
Lactate (mmol/L)^c	5.7 (1.3, 12)	5.4 (1.2, 17)	0.55^d
Male	37 (54, 41–66)	40 (51, 40–63)	0.78^e
Desexed	42 (61, 48–72)	48 (62, 50–72)	0.93^e
Purebred	65 (94, 86–98)	71 (91, 82–96)	0.46^e
Comorbidities^f	24 (35, 24–47)	20 (26, 16–37)	0.23^e
Euthanised	37 (54, 41–66)	48 (62, 50–72)	0.33^e
Survived	28 (41, 29–43)	24 (31, 21–42)	0.21^e

^a Exact CI are reported (Collett 1991).

^b Two-sided Kruskal-Wallis rank sum test.

^c Lactate values were missing for 19 (28%) dogs with known and 24 (31%) dogs with unknown insurance status.

^d Two-sided Student’s $t$-test assuming unequal variances.

^e Pearson’s χ² test.

^f Number of animals with at least one comorbidity recorded in their medical record.

Ten (14%) dogs were insured at the time of the GDV event and 59 (86%) cases were not. The majority of the non-insured dogs (37; 63 (95% CI=50–74)%) were euthanised before surgery, while none (0 (95% CI=0–28)%) of the insured dogs were euthanised at that time (p<0.001). Of the 32 dogs that underwent surgery, four (13 (95% CI=5–28)%) dogs, all uninsured, did not survive to hospital discharge. Three (9%) dogs were euthanised during or after the surgical treatment and one (3%) dog experienced cardiopulmonary arrest post-operatively. Of all 69 animals with known insurance status, the majority (41; 59%) did not survive to hospital discharge, with 90 (95% CI=78–96)% of deaths caused by euthanasia prior to surgery (Figure 2). The median costs of treatment for animals that were euthanised prior to surgery or underwent surgery was AUD$589 (quartile 1 = $384; quartile 3 = $793) or AUD$5,670 (quartile 1 = $4,626; quartile 3 = $6,743), respectively.

Figure 2. Proportions of dogs alive at various stages of hospitalisation for dogs that presented to the emergency room (ER) of a veterinary hospital in Australia with gastric dilatation-volvulus.

The median age of dogs in the insured group was lower than in the uninsured group, while sex, de-sexing status, the frequency of comorbid conditions and lactate concentrations did not differ between groups (Table 2). In the univariable analyses, age category and insurance status, but not sex, presence of at least one comorbidity or lactate category were associated with euthanasia before surgery (Table 3). After adjusting for the effect of age using the Mantel-Haenszel procedure, uninsured dogs were 5.0 (95% CI=1.8–26) times more likely to undergo presurgical euthanasia compared to insured dogs.

Table 2. Median (min, max) age and blood lactate concentration, and number (95% CI^a) of male, desexed, and purebred dogs and number of dogs with co-morbidities, in a retrospective study of dogs presenting to a veterinary hospital emergency department Australia with gastric dilatation-volvulus, categorised by whether they were insured or uninsured.

Variable	Insured (n=10)	Uninsured (n=59)	P-value
Age (years)	5.5 (2.0, 11.0)	9.0 (2.0, 15.0)	0.009^b
Lactate (mmol/L)^c	4.8 (1.3, 10.6)	6.1 (1.6, 12.3)	0.444^d
Male	8 (80, 44–97)	29 (49, 36–63)	0.070^e
Desexed	4 (40, 12–74)	38 (64, 51–76)	0.144^e
Purebred	10 (100, 69–100)	55 (93, 84–98)	0.396^e
Comorbidities^f	3 (30, 7–65)	21 (36, 24–39)	0.731^e

^a Exact CI are reported (Collett 1991).

^b Two-sided Kruskal-Wallis rank sum test.

^c Lactate values were missing for one insured and 18 uninsured animals.

^d Two-sided Student’s $t$-test assuming unequal variances.

^e Pearson’s χ² test.

^f Number of animals with at least one comorbidity recorded in their medical record.

Table 3. Results of univariable analyses of factors associated with the risk of euthanasia of dogs with gastric dilatation-volvulus after presentation to a veterinary hospital emergency department in Australia.

	Euthanised
Risk factor	Yes	No	Total	IR^a	RR (95% CrI or CI)	P-value^b
Insurance status
Yes	0	10	10	0	Reference
No	37	22	59	63	9.2 (2.1–197)^c	–
Total	37	32	69
Sex
Male	16	21	37	43	Reference
Female	21	11	32	66	1.5 (0.97–2.4)	0.06
Total	37	32	69
Age category^d
Non-geriatric	11	24	35	31	Reference
Geriatric	26	8	34	76	2.4 (1.4–4.1)	<0.01
Total	37	32	69
Breed category
Mixed breed	1	3	4	25	Reference
Purebred	36	29	65	55	2.2 (0.40–12)	0.24
Total	37	32	69
Comorbidities
Not recorded	23	22	45	51	Reference
Yes	14	10	24	58	1.1 (0.73–1.8)	0.57
Total	37	32	69
Lactate category
Low (<6 mmol/L)	8	18	26	31	Reference
High (≥6 mmol/L)	11	13	24	46	1.5 (0.72–3.1)	0.27
Not measured	18	1	19	95	3.1 (1.7–5.5)	<0.01
Total	37	32	69

Note: CrI: credible interval.

^a Incidence risk (IR) of euthanasia (number of euthanised dogs per 100 dogs at risk).

^b Significance of likelihood-ratio test.

^c Bayesian estimate of the RR, accounting for zero cell contingency table frequencies as described in the text.

^d Animals were classified as geriatric if their age met or exceeded reported median longevity for each breed.

Discussion

We found that the presence of pet insurance was associated with a significant reduction in risk of presurgical euthanasia of dogs with GDV. Only 40% of the dogs survived to hospital discharge and death was due to euthanasia prior to surgery in 90% of non-survivors. Once surgery was undertaken, the mortality rate was 12%, which is commensurate with reported data in recent studies (Song et al. 2020; Sharp et al. 2020). The veterinary literature indicates that outcomes for dogs undergoing GDV surgery has remained largely unchanged over the past two decades, with reported rates of survival from 80–90%. (Beck et al. 2006; Mackenzie et al. 2010; Song et al. 2020). Therefore further advances in perioperative care may have some, though limited, impact on improving survival rates for dogs with GDV. Instead, our findings raise the possibility that improved outcomes for dogs with GDV may also result from reducing the frequency of pre-surgical euthanasia performed on economic grounds, as shown in Figure 2.

Our observation that a high proportion of dogs with GDV were euthanised prior to surgery is in accordance with findings reported by others. Recent studies presented pre-treatment euthanasia rates of 38% at two hospitals in Australia (Song et al. 2020), 23% at a single hospital in the United States of America (Sharp et al. 2020) and 37% at emergency veterinary practices in the United Kingdom (O'Neill et al. 2017). Factors contributing to owners’ decisions for euthanasia vary and may include severe illness conferring a perceived poor prognosis or lengthy recovery, pre-existing comorbidities impacting on quality of life, older age of the animal and cost (Sharp et al. 2020). We believe that the latter factor leading to economic euthanasia may be most amenable to remediation and thus offers a realistic potential to reduce preventable deaths of dogs with GDV. Pet insurance is an instrument to alleviate the economic burden of unexpected out-of-pocket veterinary expenses. We used pet insurance as an example of a financial intervention as it is clearly identifiable as present or absent and other financing opportunities were either uniformly available to all owners (i.e. VetPay) or not at all (e.g. payment plans) or were impossible to identify (e.g. line of credit available to pet owners at time of GDV). Although not studied herein, these other financial instruments may nevertheless be effective in alleviating economic euthanasia. In our study, insurance was an effective intervention to counteract the risk of euthanasia of dogs with GDV in the emergency room. This suggests first, that euthanasia prior to surgery is, to a large part, financially motivated, and second, that pre-surgical euthanasia rates (and thereby GDV mortality rate) can be reduced with this economic intervention. Given the impact of insurance on patient outcomes in this study, we suggest that insurance status should be consistently recorded in medical records as this enables interpretation of mortality changes over time both for hospital quality control purposes, and to allow insurance status to be analysed as an explanatory variable in observational studies where survival is a critical outcome.

Put into its larger context, our study findings are reflective of the tension that the veterinary profession is operating in, where the disparity between cost of care and affordability is a major source of distress for veterinarians and pet owners alike (Batchelor and McKeegan 2012). Economic conflicts surrounding veterinary care are the most common source of ethical dilemmas, moral distress and professional burnout in veterinarians (Batchelor and McKeegan 2012; Moses et al. 2018; Kondrup et al. 2016). Alongside high workloads, relatively low salaries, accumulation of student debt, and an overall high compassion burden, this contributes to the well-documented poor psychological health of veterinary professionals around the world (Moir and Van den Brink 2020). Likewise, companion animal owners may experience their pet’s euthanasia on economic grounds in the presence of viable medical options as more stressful than where no such medical options are available (Bussolari et al. 2018). The burden of high out-of-pocket expenses, described as financial toxicity in human critical care and cancer care, may lead to financial distress and material hardship of pet owners (Zafar and Abernethy 2013; Hauschildt et al. 2020). The acute occurrence of a large, unexpected expense combined with a life-threatening illness that does not allow much time for considering all options, may exacerbate these economic conflicts in emergency veterinary medicine. Our study, although limited to a single veterinary teaching hospital, provides evidence for the size of this problem, but also suggests that financial instruments maybe effective in alleviating economic conflicts.

The percentage of insured cases in our data set (14%) is within the range of reported insurance rates for dogs in Australia that vary from 7.1% to 30%, depending on source (Anonymous 2018; Anonymous 2019). The primary reasons for non-insurance relate to cost, perceived value or unawareness of the offerings (Anonymous 2019). Pet owners that have a higher income, are in employment and live in inner-city areas are more likely to obtain insurance for their animals (Anonymous 2019). The decision to treat insured dogs with GDV may therefore be further amplified by the higher socioeconomic status of their owners.

Several study limitations deserve consideration. As insurance status is not routinely recorded in the medical records of the hospital we studied, we attempted to obtain this information retrospectively in a manner as free of bias as possible. Nevertheless, it is possible that pet owners whose animals died from GDV were less likely to contribute to the study and that some euthanised animals may have been recognised as insured if the entire insured cohort was able to be captured. However, this is also true for the non-insured population. The cases in this retrospective study represent a convenience sample including approximately 50% of all non-referred dogs presented to our hospital with GDV over the study period, which may not be representative of the whole population. However, the dogs with GDV that were included had similar age, sex, comorbidities, pre-surgical euthanasia rates and overall mortality as animals excluded from analysis. Moreover, insured and non-insured animals were not well matched for age and some confounders, such as significant comorbidities, may have been missed due to record quality. Due to the presence of cells with zero frequencies in the contingency table analyses of the association between insurance status and euthanasia, we used a Bayesian approach to quantify the association between insurance presence and the risk of euthanasia, in preference to the more commonly used Haldane-Anscombe correction which is known to produced biased estimates of RR (Weber et al. 2020).

The findings of this study originate from a single veterinary hospital in Australia and may not represent the situation in the broader veterinary industry either within Australia or internationally. Larger, prospective studies including several veterinary hospitals spanning across a wide socioeconomic range are required to further characterise the impact of insurance on euthanasia in veterinary emergency medicine, and to validate pre-treatment euthanasia rates of dogs with GDV as a marker of economic euthanasia.

In conclusion, euthanasia prior to treatment was the most common cause of death in non-referred dogs with GDV and was significantly reduced in animals with pet insurance compared to uninsured animals: no insured animal underwent pre-surgical euthanasia. This finding suggests that economic euthanasia is common in dogs with GDV and that interventions to reduce such euthanasia could limit mortality in dogs with GDV and possibly other life-threatening but treatable diseases. As this study only includes dogs from a single veterinary emergency centre, future multicentre studies including hospitals in diverse geographic and socioeconomic settings should be conducted to validate the impact of insurance on presurgical euthanasia of dogs with GDV.

Acknowledgements

The authors wish to recognise the contributions of Dr Elise Boller in the refinement and training of the phone interview script.

Notes

1 https://doi.org/10.1080/00480169.2021.1920512