Court-ordered assessments and routine access to confidential health information: findings from a regional forensic mental health service

Abstract

Health Assessors undertaking court ordered assessments are granted access to material related to the alleged They do not have access to defendant’s confidential health records. Defendant’s consent must be obtained for their health information to be accessed. One hundred files of court ordered reports were audited, to see whether Health Assessors in a New Zealand regional forensic mental health service had obtained such consent. The results were categorised as follows:

Category 1: Consent was provided and health record accessed, c (n = 76).

Category 2: Consent was not provided and health information not accessed (n = 5) – included defendants deemed unfit to stand trial and unable to give informed consent (n = 2).

Category 3: Consent was not provided but health information accessed (n = 19).

Ethical and medico-legal implications of Health Assessors accessing defendant’s confidential health records are discussed. Recommendations are made for improving services so that public trust in the integrity of confidential information can be maintained.

Health Assessors (HA), consisting largely of psychiatrists and psychologists, routinely undertake court ordered assessments on defendants accused of various crimes. In New Zealand, section 38[1] of the Criminal Procedure Mentally Impaired Persons [CPMIP] Act 2003, authorises courts to order reports to determine whether the person is unfit to stand trial:

1. Whether the person is insane within the meaning of section 23 of the Crimes Act 1961;

2. The type and length of sentence that might be imposed on the person;

3. And, the nature of a requirement that the court may impose on the person as part of, or as condition of, a sentence or order.

Under the CPMIP Act, a HA is expected, where practicable, to consult with a caregiver or welfare guardian of the subject, to obtain information regarding the condition and background of the subject. HA in conducting such assessments act as experts in helping the court to understand the defendant’s mental state or their clinical needs either during the trial or at the material time when the alleged offending occurred. In order for the assessment to be complete and meaningful, HA need access to the defendant’s health records containing information on their health and psychiatric history. Such information is crucial while making recommendations for meeting the defendant’s healthcare needs or to reduce the risk of reoffending. Contemporaneous information about the defendant’s mental health at the time of offending is often found in health records and can play an integral role in enabling the HA to provide the opinion required by the court. The wording of Section 38 of the CPMIP Act 2003 does not specify whether HA have access to the defendant’s health records without their consent. Royal Australian New Zealand College of Psychiatrists (RANZCP) ethical guidelines for independent medical examination require consent to be obtained from the person undergoing assessment for their health information to be accessed.

A defendant being ordered to undergo assessment under s 38 may feel deprived of choice and forget they still have the choice to decline in the assessment. They may feel coerced, and are sometimes without legal representation. They are thus at risk of misperceiving the situation and unduly cooperating with the HA in the belief that anything they reveal could be in their favour (Allnut & Chaplow, 2000). They may not appreciate that reports, containing confidential health information, could be made public (Allnut & Chaplow, 2000). Court ordered assessments create settings in which the potential for power imbalance between defendants and the HA can be further accentuated (Arboleda-Flórez & Stuart, 2012; Evans, 2005; Pargiter & Bloch, 1997). Defendants may fail to protect the privacy of their health information that was collected in the context of a trusting therapeutic relationship.

Being court appointed, some HA may ignore respect for autonomy of the defendant, and proceed to access their confidential health information (Niveau & Welle, 2018) despite the moral or ethical expectations against such practice (Rappeport, 1982; Waithe et al., 1982). Some well-intentioned HA may access defendant’s health information out of a sense of professional obligation to provide a comprehensive report. Others may harbour an implicit belief that the legislation entitles them access to such information. Susceptibility of HA to experience bias and being influenced by outside influences has been well studied (Gudjonsson & Hayward, 1998; Murrie et al., 2013). Onus is placed on the HA to advise defendants that the report will not be confidential and will be accessed by different parties, and to record in the report that this information was conveyed to and understood by the defendant (Ackerman, 2006).

Despite such widespread awareness and explicit understanding around consent processes an alarming number of court ordered reports do not incorporate discussions regarding these points with the defendant (Demier, 2012). Essentially, court ordered health assessments create a context in which defendant’s rights to protect their private health information has the potential to be breached with major consequences. Neglecting to advise or indeed remind the defendant that the HA does not have an automatic access to their sensitive health information can undermine their confidence in medicine and in justice in general (Niveau & Welle, 2018).

New Zealand legislation offers some protection to defendants’ autonomy and privacy in a court-ordered assessment by making provisions that the s 38 of CPMIP Act order does not routinely authorise the HA to access a defendant’s health records prepared in a therapeutic context (Dawson, 2012). The collection, use, disclosure and access of health information of a person in New Zealand is protected by the Privacy Act 1993. The purpose of this Act is to promote and protect individual privacy by establishing principles relating to the collection, use and disclosure of information relating to individuals and access by individuals to information held about them. Extra protection to health information is given through the Health Information Privacy Code 1994, which ensures that health information obtained in one context cannot be used in another. Breaching these two legislative codes by using health information for a court ordered assessment can damage a person’s trust in the health system, which is built on the principles of autonomy, beneficence, non-maleficence and justice (Beauchamp & Childress, 2001; Evans, 2005). Such a breach can also result in unnecessary health information reaching authorities who may not need such information (Wettstein, 2002). An important distinction must, however, be made in this context because potential breaches of privacy law may be different to breaching professional guidelines. The information that defendants share with HA is not confidential whereas what they share with their health professionals is. There are good legal and ethical reasons for the two processes (confidential health information and court ordered assessments) to be kept separate.

In routine practice, it is often left to the HA to decide what health information is to be accessed and how to go about accessing that information while respecting the autonomy of the defendant (Kumar et al., 2014). Given the accentuated power imbalance between HA and defendants and the potential for variance between different HA in adhering to professional and legal guidelines, it is important for all forensic mental health services to monitor existing processes around obtaining informed consent to access health information in court ordered reports. A gap can always exist between what should be done and what occurs in reality. With this in mind, one hundred files of court ordered reports were audited retrospectively in a New Zealand regional forensic mental health service. We conducted a retrospective descriptive study in order to avoid any modification in the behaviour or practice of HA. It was our expectation that informed consent for accessing health information would be obtained in all court ordered assessment in keeping with the ethical guidelines and legal requirements outlined above. We did not have data from other forensic mental health services to compare our findings at the time of writing this paper.

Method

Study setting

This study was conducted at Puawai: Midland Regional Forensic Psychiatric Services, established in 1998 in the city of Hamilton, New Zealand. Puawai operates under the public health system and delivers Faculty of Medical and Health Sciences (FMHS) to a catchment population of approximately 772,443 spread across five district health boards (DHBs) in the Midland region of New Zealand. Overall, the Midland region has a higher percentage of people identifying as Maori and a lower percentage of people identifying as Pacific or Asian than the national average. FMHS delivered by Puawai include court-based services, prison-based services, inpatient psychiatric assessment, treatment and rehabilitation services, and community-based services. A significant proportion of Puawai’s catchment area is rural and is serviced by small urban centres situated at considerable distances from Hamilton. Puawai serves three prisons in the region with a total of 2343 prison beds and two ‘Detention Hubs’ administered by the Police and Department of Corrections, respectively. Patients enter the service primarily via the acute ward (Puna Maatai) from a variety of avenues, including prison, courts, community forensic psychiatric service and General Adult Mental Health Services (GAMHs). Courts situated in the region order assessments under s 38 of the CPMIP Act, which are conducted in the community, prison and forensic mental health inpatient setting by psychiatrists and psychologists. The service has a written consent form that each defendant is expected to read and sign in order for the HA to access their health information. In most cases such consent is obtained in the court by the court liaison nurse. Otherwise HA are expected to obtain a signature on the consent form at the time of conducting the assessment. On receipt of the court order a secretary assigns the reports to individual HA depending on their availability in the roster with some oversight from a senior psychiatrist.

Study procedure

One hundred consecutive files with a completed court ordered report prior to 1 January 2020 were retrospectively reviewed. One of the authors was involved in writing reports and did not participate in data extraction. The defendant’s National Hospital Identification (NHI) number, age, gender and ethnicity were extracted from the database of reports completed by the service during the study period. Court files were manually reviewed to examine consent provided by the defendant for accessing their health records. Completed reports were then reviewed to extract the type of the report (outpatient, prison or ward), current charges and diagnosis, and to establish whether the HA had used information from health records both in the body of the report and while enlisting sources of information. Files that did not contain a signed consent were further reviewed to examine whether a reason for not obtaining consent had been given by the HA. Files containing signed consent were reviewed to examine the type of consent, which in turn were categorised as obtained consent, verbal consent, consent without witness and consent without appropriate identification information. Finally, we also reviewed the reports that did not contain any information from health records. We elected not to look at individual practice trends among HA with regard to completing reports without seeking consent as the audit was completed in the spirit of quality assurance where ‘shame and blame’ is often counterproductive (Duckett et al., 2008).

No ethics approval was sought or deemed necessary due to the retrospective nature of this file-based audit.

Statistical analysis

Statistical analysis was carried out using STATA (Stata Statistical Software: Release 16; StataCorp, College Station, TX, USA). Means, levels of significance and standard deviations were obtained for continuous demographic variables.

Results

From the 100 files reviewed, 24 subjects had not given consent to have their health records to be accessed by the HA. We attempted to establish from the files whether the defendants had declined to give consent, were unable to give consent or simply consent had not been obtained by reviewing the reports that referred to discussion between HA and the defendant about confidentiality and consent. Five of these 24 reports had no consent for accessing health records, and HA did not access the defendant’s health records, explicitly noting that such information could not be recorded due to absent consent. Further scrutiny of these five reports showed the following: three reports did not cite information contained in health records because no consent had been obtained; one report stated that health information had been accessed despite the lack of consent but had not been used in the report as the subject lacked capacity to give informed consent; and in one report the subject had not given consent to access health records but as a previous court ordered report contained the necessary and relevant information the HA cited the information and completed the report. Health records had been accessed by HA for the remaining 19 subjects, without having obtained signed consent.

Among the 76 subjects who gave consent for their health information to be accessed, 11 gave verbal consent (including seven through a no-contact booth; two via audio–video-link; and two refused to sign the consent form in face to face contact but gave verbal consent. and this was recorded by HA as such). Among the remaining 65 signed consent forms, one had missing identification details (date of birth and national hospital identification number (NHI), and six were solely signed by the subject and had not been witnessed. Despite these missing details we included them in the ‘consented’ category because the Health Assessors had mentioned difficulties in obtaining consent. This breakdown gave us three categories:

• Category 1: Consent provided and health record accessed (n = 76).

• Category 2: Consent not provided and health information not accessed (n = 5). This category included defendants who were deemed unfit to stand trial and unable to give informed consent (n = 2).

• Category 3: Consent not provided but health information accessed (n = 19).

The studied variables, including sociodemographic details, type of report, diagnosis and nature of offence, are outlined in Table 1. No significant differences in these variables were found across the three groups.

Table 1.

Variables studied.

Variables	Total (N = 100)		Consented		Not consented
			Health information contained in report (N = 76)		Health information not contained in report (N = 5)		But health information contained in report (N = 19)
	M (SD)	n	M (SD)	n	M (SD)	n	M (SD)	n	p
Age	33.4 (12.4)		34.5 (13.2)		29.6 (7.7)		30.3 (9.9)		1.0
Gender
Male		85		64		4		17	.805
Female		15		12		1		2
Ethnicity
Maori		60		45		2		13	.711
European		35		27		3		5
Other		5		4		0		1
Type of report
OP		50		43		2		5	.105
Prison		44		30		2		12
Ward		6		3		1		2
Diagnosis
Psychosis		22		17		1		4	.021
Psychosis/drug		14		8		1		5
BPAD		9		6		1		2
Depression		6		6		0		0
Substance		3		0		0		3
PD		11		8		2		1
Cog impair		4		4		0		0
Other/no diagnosis		31		27		0		4
Offence
Violence		48		36		2		10	.582
Sexual		18		14		1		3
Alcohol/drug		4		2		1		1
Dishonesty		14		13		0		1
Others		16		11		1		4

Note: OP = outpatient; BPAD = bipolar affective disorder; PD = Personality Disorder; cog impair = cognitive impairment.

The significant difference (p = .021) identified across diagnostic categories is difficult to interpret given the small numbers involved. It does appear though, compared to other diagnoses, those with substance use were more likely to have their health files accessed without consent.

Data were further analysed to compare the responses between Categories 2 and 3 using Fisher’s exact test. No statistically significant differences emerged between the two categories on the following variables: gender (p = .52), ethnicity (p = .44), assessment location (p = .475), diagnosis (p = .42) and offence type (p = .9). A two-sample t test did not find any statistically significant differences in the ages of subjects in Category 2 (n = 5, M = 29.6, 95% confidence interval, CI [20.03, 39.16]) and Category 3 (n = 19, M = 30.26 years, 95% CI [25.45, 34.1]) (p = .55). Category 3 included five defendants who were found unfit to stand trial and unable to give informed consent for accessing their health records but their records had been accessed. Given the small numbers no further statistical analyses were carried out.

Discussion

This descriptive study was conducted in a naturalistic and non-interventional context. This study for the first time describes the variables associated with consent to access health information in court ordered reports in a regional forensic mental health service in New Zealand. Due to the retrospective nature, variables impacting Health Assessor’s efforts to seek consent were not impacted by the study. This is particularly important while studying matters of ethical significance as awareness of research can modify a participant’s behaviour (McCambridge & Kypri, 2011).

For reasons cited previously it is crucial for forensic mental health services and Court appointed HA to have robust systems in place for obtaining informed consent from defendants for accessing their health records. Such robustness is vital for maintaining trust in the system while preserving the subject’s autonomy by making them aware to what extent their health information will be made public (Niveau & Welle, 2018). It is reassuring to see in this audit that HA had accessed health information after seeking valid consent in 76% of the cases. We were also reassured to find in 5% of the cases report writers did not access health records and documented in the reports that the defendants had not given their consent or were deemed unable to consent. These findings are consistent with expectations around HA seeking a defendant’s consent for accessing their health records. Having said this, safeguards may be necessary where verbal consent is obtained. For instance, in 18 of the 78 cases in this study a verbal consent was obtained by the HA but explanations provided as to why written consent could not be obtained were inadequate.

The findings in Category 3 where HA had accessed a defendant’s health records without their consent is of concern. This category also included a small number (n = 5) where defendants were deemed unfit to stand trial and unable to give informed consent for accessing their health records and yet HA had incorporated some findings from their records. In quantitative terms this was only a small proportion of the overall population. No statistically significant differences emerged when compared to the groups of defendants who gave their consent and others who did not consent and report writers did not access records. From a clinical and ethical practice standpoint, however, this is a significant finding as ideally not even one person’s health records should have been accessed without their consent. Health professionals are already faced with increasing reports of the public losing faith in the system as health information is lost in the public and private sector (Nicholas & Nicholas, 2010). While statutory release of health information when summoned by the law does not constitute violation (Blightman et al., 2014), HA are not granted automatic access to a defendant’s health records. While to date there has been no criminal conviction resulting from breach of confidence, civil claims in negligence have occurred and damages awarded in cases where health information without explicit consent has been revealed (Blightman et al., 2014).

The characteristics of subjects in terms of age, gender, ethnicity, diagnosis, the type of report (outpatient, prison or forensic psychiatric ward) and nature of offences did not show any statistically significant differences between the three groups (consented and health records accessed, did not consent and health records not accessed, and did not consent but health records accessed). The possibility that in some instances defendants can mislead the HA if they had no access to health information does need to be considered. Compared to other diagnoses, those with substance use were more likely to have their health files accessed without consent. It is understandable that many defendants with substance use issues may deliberately withhold information given significant concerns about the knowledge of a person’s substance use may result in negative consequences on multiple domains including employment, housing, child custody, discrimination by medical professionals and insurers, arrest, prosecution and incarceration (Substance Abuse and Mental Health Services Administration, SAMHSA, 2017).

Another group of defendants worth mentioning here are defendants who had their reports prepared while they were imprisoned. Twelve of the 19 reports for defendants who had their health information accessed by HA and used in the report came from prison as opposed to only two in the outpatient setting and five in the secure ward. Imprisoned defendants are more likely to feel coerced, have fewer choices and challenge when their health information is accessed and used in court ordered reports without their consent (van der Kaap-Deeder et al., 2017). HA may need to be extra vigilant while preparing reports for this group of defendants. This, however, did not occur in this study. The difference did not reach statistical significance but it is difficult to draw any far-reaching conclusions from this finding given the small sample size of the study. We collected demographic and diagnostic variables in order to elucidate whether they were likely to be associated with the likelihood of health information being accessed without consent. Variables such as ethnicity and certain diagnostic rubrics are reported to be associated with increased risk of marginalisation (Farrelly et al., 2014). In our study we did not find any association between demographic and diagnosis with health information being accessed without consent.

We had expected to see a higher proportion of cases in Category 3 (consent not provided but health information accessed) to occur in ward-based reports as such subjects would be assumed to be more unwell, or less able or willing to give consent, and assessors found it easier to access health information. It is difficult to say whether this occurred or not from this study or whether assessors accessed information and simply did not mention it in the report. It was reassuring to find that even if this occurred assessors did not use such information in the reports that were reviewed. Face to face interviews in a prospective study with a large sample size may provide more meaningful data in this regard.

This study does have limitations, which go beyond the small sample size and its retrospective design. Simply because written consent for accessing health information was on the file it does not mean such consent was ‘informed’ unless the HA made a specific reference to how the subject’s capacity was assessed. This was difficult to assess from this audit and is a limitation of its retrospective nature. Routine inclusion of details of explanation provided and the grounds for refusal on the consent form or the subject being deemed to lack capacity to give consent would be helpful in this regard. We did not record how many of the defendants found unfit to stand trial had given consent for their health records to be accessed although some HA had recorded that consent was either not obtained or not valid among those who were unfit to stand trial. Exploration of the relationship between being unfit to stand trial and being unable to give consent for health information to be accessed warrants a well-designed study. It is likely that the vast majority of defendants who are unfit to stand trial may also lack the capacity to give informed consent for their health records to be accessed.

In the absence of any existing guidelines we make some recommendations that may assist forensic mental health services to develop robust systems for obtaining consent for accessing health information. Provisions should be made to clearly identify whether consent was obtained for general health or psychiatric records. Consent forms should have a specific mention of the defendant’s capacity to give consent. Should HA use health information in reports they should state clearly whether defendants gave consent and had the capacity to do so. In cases where health information was accessed without consent or if the defendant was unable to give consent, a clear rationale should be given. HA may find being required to provide such rationale in the body of the report will prompt them to actively seek consent from defendants while preparing reports.

Conclusion

Obtaining consent for a defendant’s health information to be used in court ordered health assessment is vital and a legal requirement in New Zealand. This occurred in the vast majority of cases reviewed in this audit but in a small but significant proportion of cases it did not. Additional safeguards may include capturing a defendant’s capacity to give informed consent, how the consent was obtained and the reasons for subject’s refusal or inability to sign the consent form. Forensic mental health systems should have provisions for recording the type of consent (verbal or written) and should have additional safeguards, especially when verbal consent was obtained.

Ethical standards

Declaration of conflicts of interest

Claudia van Kan has declared no conflicts of interest

Shailesh Kumar has declared no conflicts of interest

The authors were both employed by the District Health Board where the audit was conducted as part of Quality Assurance Activity.

Ethical approval

Formal approval from the ethics committee was not needed as this was an audit.

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed consent

Informed consent was not needed from all individual participants included in the study as this was an audit