Effects of changing criteria on improving interRAI assessment for elder abuse: analysis of a national dataset from Aotearoa New Zealand

Abstract

Objectives

Globally, one in six older adults in the community will be a victim of abuse (elder abuse; EA). Despite these horrific statistics, EA remains largely undetected and under-reported. Available screening methods and tools fail to accurately identify the phenomenon’s true prevalence. We aimed to test assessment capture rates by altering the criteria for suspicion of EA in the interRAI-HC (International Resident Assessment Instrument–Home Care) in a large national dataset.

Design

We employed secondary analyses of existing data to test a methodology to improve the detection of older adults at risk of EA using the interRAI-HC, which currently underestimates the extent of abuse.

Setting

The interRAI is a suite of clinical assessment instruments. In Aotearoa New Zealand, interRAI is mandatory in aged residential care and home and community services for older people living in the community. They are designed to show the assessor opportunities for improvement and any risks to the person’s health.

Outcome measure

Capture rates of individuals at risk of EA when the interRAI Abuse-Clinical Assessment Protocol (A-CAP) is changed to include the unable to determine abuse (UDA) group shown in a pilot study to increase capture rates of individuals at risk of EA.

Results

Analysis of 9 years of interRAI-HC data (July 2013–June 2022) was undertaken, encompassing 186 713 individual assessments consisting of 108 992 women (58.4%) and 77 469 men (41.5%). The mean age was 82.1 years (range: 65–109); the majority 161 378 were European New Zealanders (86.4%) and the most common minority ethnicity was Māori (6.1%). Those at high risk of abuse (A-CAP) tended to be male (2402; 51.0%), were 79.2 years old on average (range 65–105), with 49.6% (2335) living alone, 39.4% (1858) suffering from depression and a majority were assessed as not having independent decision making (2942; 62.5%). In comparison, the UDA group showed similar characteristics to the A-CAP group on some measures. They were slightly younger than the general sample, with a mean age 80.1 years (range 65–107), they had higher rates of depression (2123; 33.5%) compared with the general sample (25 936; 14.8%) and a majority were assessed as not having independent decision-making (3855; 60.9%). The UDA group is distinct from the general sample and the UDA group broadly has similar but less extreme characteristics to the A-CAP group. Through altering the criteria for suspicion of EA, capture rates of at-risk individuals could be more than doubled from 2.5% to 5.9%.

Conclusions

We propose that via adapting the interRAI-HC criteria to include the UDA category, the identification of older adults at risk of EA could be substantially improved, facilitating enhanced protection of this vulnerable population.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• In Aotearoa New Zealand, the International Resident Assessment Instrument (interRAI) home care assessment is a mandatory component of a routine needs assessment offering a unique screening opportunity for elder abuse (EA).

• Comprehensive collection of interRAI data for 186 713 assessments in the community was completed.

• Changing the criteria for at risk of EA was tested.

• Objective capture of EA case status is missing as this was a secondary analysis of an existing dataset.

Introduction

Elder abuse (EA) has devastating and costly effects on the victim and society, yet often goes unidentified or unreported. Healthcare professionals are in a unique position to identify and intervene in EA as they may be the only contact an older adult has outside of their home.^{1 2} Therefore, all members of the healthcare team need to be aware of risk factors and signs of EA as well as the systems in place to assist victims and families. There is agreement in the published literature that the issue of EA globally is not receiving attention adequate for the scale and severity of the problem. In a 2014 global survey to assess measures taken by countries to address interpersonal violence, EA was consistently addressed the least often.³ Further studies identified factors connected to the inherent complexity of the issue, pervasive ageism, lack of awareness and doubts about prevalence estimates, the intractability of the issue, inability to capitalise on policy windows and processes, disagreements on the nature of the problem and its solutions, and weakness of governance structures.⁴ There are opportunities for effecting change and increasing the priority of EA. Chief among these is the UN Decade of Healthy Ageing 2021–2030, which brings together governments, civil society, international agencies, professionals, academics, the media and the private sector for 10 years of concerted action to improve the lives of older people, their families and the communities in which they live, including by reducing EA.⁵

Globally, one in six people aged 60 years and older experience EA in the community annually, with potentially severe physical and mental health, financial and social consequences.^6–10 EA is a complex, multifaceted issue that is hard to measure accurately. Health professionals can enhance their abilities to address EA by being familiar with risk factors, precipitants and possible signs of EA and by integrating a screening tool into practice to help identify EA.¹¹ A comprehensive review of the efficacy and accuracy of tools administered to older people, intended to detect and measure EA identified 17 articles.¹² All were designed to be used by healthcare professionals. However, the authors were not able to recommend a single tool as clearly superior.¹² With growing calls for EA screening to be conducted by community-based service providers, improved screening is a valuable first step towards improving EA detection and response. However, evidence-based strategies for screening are needed.¹³ An article summarising the evidence gap and several brief screening tools for EA in the community focused on tools used in Australia. Community-based health professionals should be aware of available screening tools and consider how best to incorporate them into practice.² The authors concluded that more work is needed to identify a reliable screening tool that is acceptable to practitioners and their patients. No studies have examined their acceptability to older people, the psychometric properties of most of these tools require further testing, particularly as the sensitivity and specificity of most tools are limited.¹⁴

While the development of high-quality screening tools may improve the identification of EA, providers need to be willing to use these tools in their practice. The International Resident Assessment Instrument (interRAI) is a collaborative network of researchers and practitioners in over 35 countries committed to improving care for persons who are disabled or medically complex. interRAI instruments have been mandated by governments in several countries including Canada, New Zealand, Hong Kong, Singapore, Belgium, Ireland, Switzerland, Finland, as well as many US states. Thus, the interRAI suite of clinical assessment instruments lends itself to be utilised as a screening method for EA.¹⁵ Indeed, in a comprehensive systematic review the use of the interRAI Home Care (interRAI-HC) instrument, an internationally validated comprehensive geriatric assessment, as a base for the evaluation of home care projects was supported.¹⁶ Due to the evidence base of the interRAI-HC and its use by multiple health organisations worldwide, in New Zealand, an interRAI-HC is mandatory in the community as an assessment for required services for older people.

Research by our group has shown that the interRAI-HC is neither sufficiently sensitive nor specific to detect older adults at risk of EA, capturing only 3% from a population of increased frailty and thus at higher risk of abuse.¹⁷ We have then proceeded to improve the interRAI-HC capture rate in a pilot study analysing data from 18 884 assessments undertaken in the Otago region of New Zealand. We identified that by altering the criteria for suspicion of EA, capture rates of at-risk individuals could be nearly doubled from 2.6% to 4.8%.

The aim of this study was to test the effects of changing the interRAI criteria for suspicion of EA in a large national dataset in Aotearoa New Zealand.

Methods

Participants

Participants were New Zealanders aged 65 years and older who completed their first interRAI-HC assessment between July 2013 and June 2022 in Aotearoa New Zealand. If participants underwent multiple assessments, only data from the initial assessment were reviewed. Participant data were completely anonymised. Participant age was recorded at the time of assessment. Only interRAI-HC data from participants consenting to use in research (96%) was included, as per previous protocols for using interRAI-HC data.¹⁸ There were no changes in the interRAI protocol during the study period.¹⁹

Measurements

interRAI is a collaborative network of researchers and practitioners in over 35 countries committed to improving care for persons who are disabled or medically complex. interRAI instruments are designed to be compatible across health sectors. This improves continuity of care, promotes a person-centred approach and improves the capacity to measure clinical outcomes. Instruments are built on a ‘core’ set of items with identical definitions. Additional items are added to address issues unique to the population or setting. Most instruments use a format with a trained assessor. Comprehensive instruments identify key factors in the person’s life, including function, health, social support, service use, mood and behaviour. Status and outcome scales, and quality indicators are embedded in comprehensive instruments.

interRAI assessments are designed to show the assessor opportunities for improvement and any risks to the person’s health, which then form the basis of a care plan. A unique feature of the interRAI assessments is an algorithm that creates Clinical Assessment Protocols (CAPs). These support continuity of care planning by providing common protocols across settings.²⁰ An automated feature of interRAI systems, CAPs are based on systematic reviews of international literature and large data holdings. One of the interRAI-HC CAPs is the Abuse-CAP (A-CAP).

A health professional who is a trained assessor, a nurse, occupational therapist or social worker, has a structured conversation with the person and their family or carers, makes observations and refers to other clinical information. The assessor codes this information on software which creates a profile of the person’s needs and opportunities. Assessment information is stored electronically, in New Zealand’s national data warehouse. It is securely encrypted to protect patient privacy.

This study compared the characteristics of the general population undergoing interRAI assessment to those considered at high risk of abuse who triggered the A-CAP. In addition, characteristics of the general population and those at high risk were compared with those for whom the risk of abuse was considered ‘unable to be determined’ (UDA). We hypothesised that the UDA group will also be at risk for OAA. For the purposes of our analysis, the UDA group included any participant who did not trigger the A-CAP but did select ‘UDA’ as the answer to one or more of the A-CAP screening questions. Criteria for triggering the A-CAP and the relevant screening questions are detailed in table 1. The interRAI CAPs determine the frequency, duration, severity and likely consequences of the relevant to abuse observations. Higher values in table 1 represent increased frequency, duration and severity of recorded items.

Table 1.

Questions that contribute to the abuse CAP (A-CAP) group

Abusive relationship
Question	Trigger value	Actual value	Triggered
(Level 1) Moderate risk status persons who have one or more of the following direct indicators of abuse are present
(F1e) Fearful of family member	2,3,4…	Not answered
(F1f) Neglected or abused	2,3,4…	Not answered
(J2t) Poor hygiene	2,3,4…	Not answered
And less than 2 of the following ‘stressors’ are present
(E1i) Withdrawal	1,2,3…	Not answered
(E1j) Reduced social interactions	1,2,3…	Not answered
(F2) Lonely	1…	Not answered
(K2a) Weight loss	1…	Not answered
(K2c) Fluid intake	1…	Not answered
(sBMI) Body Mass Index Scale	1–18…	0
(J6a) Unstable conditions	1…	Not answered
(J7) Self-rated health	3…	Not answered
(M3) Drug adherence	1,2…	Not answered
(A13c) Better Living Elsewhere	1…	Not answered
(F1d) Openly expresses conflict w/family	2,3,4…	Not answered
(P2b) Informal helper stress	1…	Not answered
(sDRS) Depression Rating Scale indicates a depressive disorder	3–14…	0
(Level 2) Highest risk status triggered includes persons who have one or more of the following direct indicators of abuse are present
(F1e) Fearful of family member	2,3,4…	Not answered
(F1f) Neglected or abused	2,3,4…	Not answered
(J2t) Poor hygiene	2,3,4…	Not answered
And two or more of the following ‘stressors’ are present
(E1i) Withdrawal	1,2,3…	Not answered
(E1j) Reduced social interactions	1,2,3…	Not answered
(F2) Lonely	1…	Not answered
(K2a) Weight loss	1…	Not answered
(K2c) Fluid intake	1…	Not answered
(sBMI) Body Mass Index Scale	1–18…	0
(J6a) Unstable Health conditions	1…	Not answered
(J7) Self-rated health	3…	Not answered
(M3) Drug adherence	1,2…	Not answered
(A13c) Better Living Elsewhere	1…	Not answered
(F1d) Openly expresses conflict w/family	2,3,4…	Not answered
(P2b) Informal helper stress	1…	Not answered
(sDRS) Depression Rating Scale indicates a depressive disorder	3–14…	0

Trigger value=0-never; 1-more than 30 days ago; 2–8 to 30 days ago; 3–4 to 7 days ago; 4-in the last 3 days; 8-unable to determine. Higher numerical values represent increased frequency, duration and severity of recorded items.

Actual value=value assigned by the assessor.

Triggered=the summary computerised algorithm indicating ‘yes’ or ‘no’ for CAP triggering.

CAP, Clinical Assessment Protocol.

Six items from the interRAI-HC dataset were analysed for each of the comparison groups: age; gender; ethnicity, living situation; depression and independence in daily decision-making. These items were previously shown by our group to be associated with abuse potential.²¹ Our primary definition of depression was a Depression Rating Scale score of ≥3, as this is considered to reflect clinically meaningful depression.^{22 23} The interRAI-HC assesses the ability of the participant to make decisions regarding tasks of daily life, stratified into six categories. With respect to decision-making capacity, we collated participants with ‘independent’ or ‘modified independence’ into a single group termed ‘independent’ in decision-making. We considered participants in all other categories to have ‘impaired’ decision-making capabilities.

Statistical analyses

The demographic characteristics of the sample were summarised using number and per cent for categorical variables and mean and range for continuous variables. Age was also presented categorically in 10-year age groups. The mean age by gender was also presented to show the differences in age by gender. These summaries were also presented broken down by A-CAP, UDA and the remaining general sample.

A multinomial logistic regression was used to investigate which factors (age, gender, ethnicity, lives alone, depression, independence) were associated with the outcome of A-CAP, UDA and the general sample. This model essentially fits simultaneous logistic regression models to A-CAP compared with the general sample and to UDA compared with the general sample with odds ratios (and 95% CIs) showing the odds of being in the outcome group for a predictor variable compared with the reference group or per unit increase in a continuous variable. Wald p values were estimated and were also estimated for the comparison of the A-CAP to UDA group. All analyses were conducted in Stata, V.17.²⁴

Patient and public involvement

None.

Results

During the study period, 1 095 000 interRAI assessments were undertaken in New Zealand with an average of 30 000 interRAI-HC assessments undertaken annually. Of the 275 000 interRAI-HC first assessments undertaken during the study period, 88 287 participants were excluded from the final analysis including interviewees younger than 65 years and interviewees who either were in a coma or were recorded as having ‘null’ information relevant to this analysis.

Table 2 shows the demographic characteristics and key abuse indicators of the sample overall and by risk of abuse group. Indicators were chosen for analyses as they were the same ones used in our pilot study.²¹ There were 186 713 assessments included consisting of 108 992 women (58.4%) and 77 469 men (41.5%). The mean age was 82.1 years (range: 65–109) and the majority 161 378 were European New Zealanders (86.4%) and the most common minority ethnicity being Māori (6.1%). There were 87 201 (46.7%) interviewees living alone, a majority assessed as having independent decision-making capacity (117 655; 63.0%) and not suffering from depression (156 796; 84%). Looking at the A-CAP, UDA and general sample groups, those at high risk of abuse (A-CAP) tended to be male (2402; 51.0%), were 79.2 years old on average (range 65–105), with 49.6% (2335) living alone, 39.4% (1858) were suffering from depression and a majority were assessed as not having independent decision making (2942; 62.5%). In comparison, the UDA group showed similar characteristics to the A-CAP group on some measures. They were slightly younger than the general sample with a mean age 80.1 years old (range 65–107)) for UDA compared with 82.2 (range 65–109) for the general sample, they had higher rates of depression 33.5% (2123) compared with 14.8% (25 936) in the general sample and a majority were assessed as not having independent decision making (3855; 60.9%) compared with 35.4% (62 261) in the general sample. Missing data were minimal (<1% per variable).

Table 2.

Characteristics of the 186 713 interRAI assessments in total and by A-CAP, UDA and general sample

	Total sample		A-CAP		UDA		General sample
	Number	%	Number	%	Number	%	Number	%
	186 713	100.0	4710	2.5	6335	3.4	175 668	94.1
Gender
Male	77 469	41.5	2402	51.0	2696	42.6	72 371	41.2
Female	108 992	58.4	2304	48.9	3629	57.3	103 059	58.7
Other or missing	252	0.1	4	0.1	10	0.2	238	0.1
Age group
65 to <75 years	34 356	18.4	1447	30.7	1661	26.2	31 248	17.8
75 to <85 years	75 498	40.4	1967	41.8	2651	41.8	70 880	40.3
85 years and over	76 859	41.2	1296	27.5	2023	31.9	73 540	41.9
Mean age by gender	Mean	(Min, max)	Mean	(Min, max)	Mean	(Min, max)	Mean	(Min, max)
Male	81.4		78.7		79.3		81.6
Female	82.5		79.7		80.6		82.6
Other or missing	80.4		78.8		72.2		80.8
Total	82.1	(65, 109)	79.2	(65, 105)	80.1	(65, 107)	82.2	(65, 109)
Lives alone	Number	%	Number	%	Number	%	Number	%
Yes	87 201	46.7	2335	49.6	2545	40.2	82 321	46.9
No	99 512	53.3	2375	50.4	3790	59.8	93 347	53.1
Has depression
Yes	29 917	16.0	1858	39.4	2123	33.5	25 936	14.8
No	156 796	84.0	2852	60.6	4212	66.5	149 732	85.2
Is independent
Yes	117 655	63.0	1768	37.5	2480	39.1	113 407	64.6
No	69 058	37.0	2942	62.5	3855	60.9	62 261	35.4
Ethnicity
Asian	6491	3.5	134	2.8	266	4.2	6091	3.5
European	161 378	86.4	4028	85.5	5218	82.4	152 132	86.6
Māori	11 455	6.1	383	8.1	590	9.3	10 482	6.0
Middle Eastern/Latin American/African	679	0.4	14	0.3	33	0.5	632	0.4
Other ethnicity	687	0.4	23	0.5	30	0.5	634	0.4
Pacific peoples	5962	3.2	125	2.7	196	3.1	5641	3.2
Residual categories	61	0.0	3	0.1	2	0.0	56	0.0

A-CAP, Abuse-Clinical Assessment Protocol; interRAI, International Resident Assessment Instrument; UDA, unable to determine abuse.

The estimated unadjusted (univariable) and adjusted (multivariable) ORs from the multinomial logistic regression model are shown in online supplemental table 1. There were substantial unadjusted differences between the general sample, UDA and A-CAP groups on all measures reflecting the patterns seen in the summary statistics. For example, for every 10-year increase in age the odds of being in the UDA group was 0.70 (95% CI 0.68 to 0.72) times that of the general sample and was 0.60 (95% CI 0.58 to 0.63) for being in the A-CAP group. This reflects the pattern of younger mean age from the summary statistics. The odds of being in the UDA group and the ACAP group were much higher for those with depression (OR 2.91, 95% CI 2.76 to 3.07 and OR 3.76, 95% CI 3.54 to 3.99, respectively). Similarly, those in the UDA group (OR 0.35, 95% CI 0.34 to 0.37) and A-CAP group (OR 0.33, 95% CI 0.31 to 0.35) had lower odds of being assessed as living independently. Interestingly the UDA group had lower odds of living alone (OR 0.76, 95% CI 0.72 to 0.80) whereas the A-CAP had higher odds (OR 1.11, 95% CI 1.05 to 1.18) compared with the general sample, perhaps picking up a group at risk of abuse in a different setting. There were increased odds of being in the UDA group (OR 1.64, 95% CI 1.50 to 1.79) and A-CAP group (OR 1.38, 95% CI 1.24 to 1.54) for Māori compared with European. Those of Asian ethnicity were at increased odds of being in the UDA group (OR 1.27, 95% CI 1.12 to 1.44) but at decreased odds of being in the A-CAP group (OR 0.83, 95% CI 0.70 to 0.99).

After adjustment for the other variables in the model, the association with gender was removed for the UDA group but not for the A-CAP group. The association with those in the UDA group having lower odds of living alone was removed though the adjusted CI does not support any clinically important increased odds (95% CI 0.91 to 1.01). The pattern of the other associations remained showing that the UDA group is distinct from the general sample and while there were differences to the A-CAP group the ORs are in a similar direction indicating that the UDA group broadly has similar but less extreme characteristics to the A-CAP group.

Discussion

EA has been a worldwide major public health threat for decades, yet it remains a form of victimisation receiving limited attention, resources and research. EA has far-reaching and long-lasting impacts on older adults, their families and communities. By 2030, one in six people worldwide will be aged 60 or older, and approximately 16% will experience at least one form of maltreatment.²⁵ A recent review aiming to describe the status of EA prevention research on a global scale identified 72 articles published in the last 6 years. Several gaps were identified by the authors in the EA literature including the need for sensitive and valid tools to identify EA.⁴ These reports serve to support a call for health professionals to serve a very critical role in recognising, detecting and managing EA and it is recommended that guidelines for the detection and treatment of abuse and neglect be prepared and implemented.²⁶ Despite these critically important recommendations, the last decade saw only 18 controlled trials focusing on identifying and preventing EA. Only one of these clinical trials attempted to implement a screening tool for EA.²⁷

In this study, we were able to validate our previous pilot study and to establish that changing the interRAI criteria can increase older adults at risk of EA identification rates. We found similarities between the UDA group and the A-CAP group compared with the general sample. The A-CAP and UDA groups tended to include people who were younger, had higher rates of depression and were less likely to be assessed as living independently than the general sample. The A-CAP group was more likely to live alone but interestingly the UDA group was less likely to live alone. It is unclear whether this reflects the UDA identifying those people who were at higher risk of abuse but living in different settings to those in the A-CAP group or if there is some confounding with other factors, as this result weakened when confounders were included in the statistical model.

Some studies indicate higher rates of EA in ethnic and racial minority populations. A systematic review with 25 included studies focused on 5 racial categories in the USA and racial minorities in Canada. The review identified a lack of current research on EA in racial minority older women and race minority subgroups.²⁸ Yet EA is increasingly identified as a serious problem in Native American communities, Australian Aborigines,²⁹ Chinese minorities³⁰ and Canadian Inuits and immigrant communities.³¹ In New Zealand, an EA intervention with tribal communities was founded in 2007 to try and address EA in the Māori communities. It is based on a family conference intervention developed by the Māori people of New Zealand, who determined that Western European ways of working with welfare issues were undermining such family values as the definition and meaning of family, the importance of spirituality, the use of ritual and the value of non-interference. The Family Care Conference provides the opportunity for family members to come together to discuss and develop a plan for the well-being of their elders.³² In this study, we have found that older Māori people are at a higher risk of potentially being abused as captured by the UDA and A-CAP while Asian New Zealanders are more likely to be captured at a higher risk of potentially being abused by the UDA inclusion but not by the A-CAP. This finding is concerning as it reflects the relative insensitivity of the interRAI identification of abuse to cultural and ethnic differences. On the other hand, we are encouraged that the use of both A-CAP and the UDA while tapping into similar aspects of the abuse variables are still capturing different people potentially casting a net more sensitive to ethnic differences.

EA has been recognised as a serious problem for decades.³³ Yet rigorous studies are rare. Unfortunately, EA research has largely lacked the resources required to conduct the range of well-designed studies needed to obtain a clear picture of the problem and how to address it. One of the important recent attempts to overcome some of the major flaws of past studies is the use of a large, over 23 000 participants, representative dataset from the Canadian Longitudinal Study on Aging, to examine annual EA prevalence, as well as risk and protective factors. They found an annual rate of EA of 10% and reinforced findings from previous work that being physically, cognitively and or emotionally vulnerable increases the risk for EA.³⁴ In many ways, the findings of this study come to a similar inference.

There are limitations to this study that need to be acknowledged. This is a retrospective analysis of a cross-sectional dataset. There is no external validation of EA possible with the methodology herein applied. Thus, the objective capture of EA is not completed by an external source. Future studies will need to design a validation procedure to capture cases of EA and not those ‘…at high risk of abuse…’. This is in line with a recent attempt to map assessment tools for EA and determine their psychometric properties. However, this study showed psychometric limitation since the internal structure of the 17 scales examined was not evaluated by validity evidences.³⁵ The interRAI-HC recognises that it is ‘…difficult to assess the extent of financial abuse in the absence of detailed information on financial resources… and that not all economic trade-offs are due to abuse…’.³⁶ Finally, the interRAI-HC assessment was not designed to capture EA and indeed lacks sensitivity.

In an ideal world, we would have validated external measures of who was a victim of EA and who was not that we could compare these rules to. But there is no such absolute measure, if it did that is the analysis we would have done. There is no external standard that can be applied. The A-CAP appears to be highly insensitive, flagging a very small percentage of what we would expect based on the literature around EA. The UDA is an attempt to capture more potential cases in the absences of any ‘gold-standard’ measure. Without this external measure, we are left comparing the characteristics of the people identified to see if there is face validity to this approach and we would argue that there is strong face validity based on the characteristics of who the UDA captures. Given the percentage is still well under what we expect the real rate to be then changing thresholds will not help as we have used the most sensitive threshold possible by including any UDA, we are still well under what we expect the true EA rate to be.

We are also undertaking work to see if we can compare other health measures that may give insight into who is captured by this approach, however, this falls outside the scope of this paper and is an ongoing area of research.

In conclusion, most variables and indicators of EA analysed in this study show a similar pattern for UDA and the A-CAP. We suggest that increasing the capture rate of older adults at risk of EA by the interRAI through incorporation of the UDA cohort into the ‘…at a higher risk of potentially being abused…’ is an important step in improving suspicion of EA in a large number of vulnerable older adults. We call for further studies testing this strategy in countries using the interRAI.

Data availability statement

Data are available on reasonable request. Data may be obtained from a third party and are not publicly available.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

Ethical approval (HD22/043) was obtained from the University of Otago Ethics Committee and the Department of Psychological Medicine Ethics Committee.