Pathological narcissism: An analysis of interpersonal dysfunction within intimate relationships.

Pathological narcissism is marked by deficits in psychosocial functioning. Difficulties in relationships include instances of aggression, devaluation and control; however, few studies have examined these relationships from the perspective of partners and family members. We studied participants who were in relationships with relatives high in narcissistic traits (N = 436; current romantic partners [57.3%]; former romantic partners [21.1%]; family members [15.4%]). Participant responses were analysed thematically, and their underlying mental health problems were also measured. Thematic analysis of participant responses indicated themes of abuse from the relative with narcissism (physical, verbal, emotional and sexual) as well as the relative imposing challenging financial and sexual behaviours. There were complex interpersonal themes of mutual idealization but also devaluation. In response, participants reported high levels of anxiety, depression, self-aggression, sickness and somatic concerns. Further, participants expressed overt outward hostility towards their relative with narcissism, but also dependency strivings and frustrated dependency themes. Partners and their relative with narcissism appeared locked into interpersonal and intrapersonal dynamic conflicts. Clinical implications include specific attendance to alliance issues, dependency themes, and a focus on limit setting to establish personal safety.

BACKGROUND

Interpersonal dysfunction is a well‐documented aspect of pathological narcissism (Byrne & O'Brien, 2014; Grenyer, 2013; Kealy & Ogrodniczuk, 2011) with some authors suggesting that pathological narcissism and interpersonal dysfunction go ‘hand in hand’ (Ogrodniczuk & Kealy, 2013, p. 114). Such dysfunctional patterns have involved controlling, vindictive and intrusive behaviours (Cheek et al., 2018; Ogrodniczuk et al., 2009), displaying dispositional and reactive anger and hostility (Czarna et al., 2019; Hyatt et al., 2018). Specifically within romantic domains, people with narcissistic traits have been described as using ‘game playing tactics’ (Campbell et al., 2002), showing self‐centred, materialistic, deceptive or controlling behaviours (Brunell & Campbell, 2011), which may also include stalking behaviour and interpersonal violence (Green & Charles, 2019; Menard et al., 2021; Menard & Pincus, 2012). Correspondingly, romantic partners and family members in relationship with individuals with pathologically narcissistic traits report significant levels of burden, grief and psychological distress (Bailey & Grenyer, 2014; Day et al., 2019). A recent study by Day et al. (2020) investigated the reported characteristics of individuals with pathological narcissism from the perspective of those in an intimate relationship. Results reflected the proposed related features of pathological narcissism, ‘grandiosity’ and ‘vulnerability’ (Pincus & Lukowitsky, 2010), with the majority (69%) of the sample describing both of these aspects in their relative. Within these relationships, challenging interpersonal themes were also described such as ‘devaluation’, ‘narcissistic rage’ and ‘vengefulness’. Examined through the lens of interpersonal theory, Edershile and Wright (2019) report narcissistic grandiosity as associated with interpersonal dominance and coldness, whereas narcissistic vulnerability was associated with both displaying interpersonal coldness to others and perceiving others as cold. Similarly, Wright et al. (2017) report that perceptions of dominance predicted quarrelsome behaviours for individuals with pathological narcissism, mediated by negative affect. In this way, antagonistic and quarrelsome interpersonal behaviours may serve a regulatory or defensive function for individuals with pathological narcissism, consistent with findings that highlight the links between emotional dysregulation, compromised empathic capability and impaired social functioning (Lee et al., 2020; Ronningstam, 2016; Ronningstam, 2020).

Clinically, individuals are unlikely to present to treatment directly seeking help regarding their narcissistic pathology. Rather, as highlighted by Ronningstam and Weinberg (2013), narcissistic patients may seek treatment along more interpersonal themes, such as difficulty maintaining work due to frequent interpersonal conflict with co‐workers, or due to receiving a relationship ultimatum due to issues of infidelity or lack of intimacy. Indeed, the prominence of interpersonal dysfunction was clearly reflected in early editions of the Diagnostic and Statistical Manual of Mental Disorders (DSM) for narcissistic personality disorder (NPD) (American Psychiatric Association, 1980). Such classification systems overtly required the presence of significant interpersonal dysfunction (Criterion E), as relating to entitlement and non‐reciprocation, interpersonal exploitativeness, idealization and devaluation and lack of empathy (Levy et al., 2011; Levy et al., 2013; Reynolds & Lejuez, 2011). The current categorical criteria for NPD do not explicitly require the presence of interpersonal dysfunction in the same way, with interpersonal dysfunction being explicitly outlined in one criterion (e.g., Criterion 6: Is interpersonally exploitative) and implicit in a number of others (American Psychiatric Association, 2013a). However, the DSM's newly introduced alternate model of personality disorders (AMPDs) offers a more coherent conceptualisation of narcissism (Fossati et al., 2017; Pincus et al., 2016; Skodol et al., 2014) and has again prioritized interpersonal functioning as a core component of personality disorder criteria as relating to difficulties in empathy and intimacy, along with the pathological personality trait of antagonism (American Psychiatric Association, 2013b).

Individuals with pathological narcissism are known to evoke strong, negative, reactions from those with whom they are in a relationship (Tanzilli et al., 2017). Studying this interpersonal situation from the perspective of close informants may provide additional insight into narcissistic functioning. As such, this study aims to investigate the behavioural and relational characteristics of individuals with pathological narcissism as informed by those in a close personal relationship with them. The use of informant ratings have found to be a valid methodology to assess aspects of personality pathology, including pathological narcissism (Lukowitsky & Pincus, 2013; Oltmanns et al., 2018), given the documented limitations of self‐report research for this population (Klonsky & Oltmanns, 2002). For this research, partners and family members will be referred to as ‘participants’. Individuals with pathological narcissism will be referred to as the ‘relative’.

METHOD

Recruitment

Participants provided written informed consent to participate following institutional review board approval. The participants were recruited through invitations posted on various mental health websites that provide information and support that is narcissism specific (e.g., ‘Narcissistic Family Support Group’). In an effort to ensure that included participants were appropriate to the research, three criteria were applied. First, participants had to identify as having a close personal relationship with someone who was very narcissistic. Second, participants had to complete relevant questions to meet inclusion criteria for the study. Relevant questions included basic demographic information (age, gender, and relationship type) and answers to qualitative questions under investigation. Third, the relative had to have a cumulative score of 36 or above (average score of 3 or above) on a narcissism screening measure (SB‐PNI‐CV, described in measures section), as informed by participants (consistent with previous methodology, see Day et al., 2019). Participants who took part in this study were drawn from the same participant pool as those presented in the results of related research (Day et al., 2019; Day et al., 2020).

Participants

The inclusion criteria for this study were as follows: (a) having a relative with narcissistic traits, (b) relatives scores met threshold of a narcissism screening measure, (c) participants provided at least a 70‐word narrative about their relative and their relationship together, (d) participant completed most of the survey (at least questions 1–5). Applying these inclusion criteria, a sample of 436 was studied. In reaching this sample, we began with a potential sample pool of 2219 who had initially clicked on the consent to participate link; however, many did not proceed beyond this point (n = 955). We then applied the above criteria to the remaining 1264 participants. First, participants were removed who indicated that they did not have a ‘close’ (i.e., intimate) personal relationship with someone who was narcissistic (n = 129). Second, participants who clicked on the link to begin the survey but dropped out within the first 1–5 questions were deemed ‘non‐serious’ and were removed (n = 51). Third, participants identified as rating relatives' narcissism below summed cut off score of 36 (average score of 3) on a narcissism screening measure (SB‐PNI‐CV, described in Section 2.3) were removed (n = 249). Fourth, participants whose text sample was too brief, that is, less than 70 words, as specified by Gottschalk et al. (1969), were excluded from analysis (n = 399).

While included participants required their relative to have elevated scores on a narcissism screening measure as described, subsequent analysis found a high proportion of pathologically narcissistic characteristics in participant descriptions. Themes of ‘grandiosity’ were found in 70% of participant responses, ‘vulnerability’ themes in 81% of participant responses, and descriptions of both grandiose and vulnerable descriptions in 69% of responses (see Pincus & Lukowitsky, 2010, for more information). Table 1 outlines the demographic information of participants and the relative included in the study.

TABLE 1

Demographics for participants (partners and family) and their relatives (people high in pathological narcissism) (N = 436)

	Participants (n = 436)	Relative (n = 436)
Mean age in years (SD)	43.9 (10.1)	48.7 (11.9)
Gender
Male	4.2%	77.7%
Female	79.9%	22.3%
Not specified	15.9%	—
Employment
Full time	45.2%	53.4%
Part time	15.1%	9.2%
Unemployed	9.9%	12.7%
Other	13.9%	24.3%
Support pension	3%	4.2%
Self‐employed	2.5%	8.7%
Retired	4%	7%
Student	1.7%	0.2%
Other	2.7%	4.2%
Not stated	15.9%	0.5%

Note: ‘Other’ relationship type category consisted of ‘close friend’, a non‐blood relative, or was left unspecified. Familial relationships listed reflect the relationship of the relative with narcissistic traits.

Measures

Pathological Narcissism Inventory (carer version)

Schoenleber et al. (2015) developed a short version of the Pathological Narcissism Inventory (SB‐PNI; ‘super brief’) as a 12‐item measure consisting of the six best performing items of the Grandiosity and Vulnerability scales of the Pathological Narcissism Inventory (Pincus et al., 2009). This measure was then adapted into a carer version (SB‐PNI‐CV) in the current research by changing all self‐referential terms (i.e., ‘I’) to refer to the relative with pathological narcissism (i.e., ‘my relative’). This was done for all items covering domains of both ‘grandiosity’ (item example: ‘My relative often fantasizes about performing heroic deeds’) and ‘vulnerability’ (item example: ‘My relative finds it hard to feel good about themselves unless they know other people admire them’). The scale wording allows for both rating the presence (i.e., ‘very much like my relative’, a score of 5) and absence (i.e., ‘not at all like my relative’, a score of 0) of narcissism. The SB‐PNI‐CV demonstrated strong internal consistency (α = 0.80). Subscales of the measure also demonstrated internal consistency for both grandiose (α = 0.73) and vulnerable (α = 0.75) items. Participants were required to meet a summed threshold cut‐off of 36 (average score of 3 for each item) or above for inclusion in the study. This cut‐off captures only participants who endorse the presence of pathologically narcissistic features in their relative and screens out those who do not. Schoenleber et al. (2015) report descriptive statistics for the SB‐PNI within a large university based sample (N = 2,862) of M = 2.38, SD = 0.97. Importantly, by using a cut off of 3, our sample was almost two standard deviations above this published population mean (our informant M = 4, SD = 0.54).

Qualitative analysis of interpersonal themes

Participants who met inclusion criteria were asked to describe their relative using the Wynne‐Gift speech sample procedure as outlined by Gift et al. (1986). This included participants responding to the question:

What is your relative like, how do you get on together?

Participants were given a textbox to respond to this question in as much detail as they would like. As described above, participants whose text responses were too brief (<70 words), were removed from analysis as specified by Gottschalk et al. (1969). Mean response length was 237 words, with a standard deviation of 193 words. Text responses ranged from 70 to 1279 words.

A phenomenological orientation was adopted in understanding the data, which places primacy on understanding the ‘lived experience’ of participant responses (Smith et al., 2009). The data analysis process followed the steps outlined by Braun et al. (2019) in conducting thematic analysis. This involved familiarization with the data, generating initial codes, searching for themes, reviewing themes, defining and naming themes, and writing up the analysis (Braun & Clarke, 2006). This methodology of data analysis via phenomenologically analysing and grouping themes is a well‐documented and regularly utilized qualitative approach (e.g., Ng et al., 2019; White & Grenyer, 1999). To do this, significant statements were extracted and coded into themes reflecting their content (e.g., ‘physical abuse’ and ‘infidelity’) using Nvivo 11. Statements were free to be coded into multiple themes as appropriate, as statements provided by participants may contain multiple meanings. Themes were then grouped together in an overarching dimension (e.g., ‘abuse’ and ‘sexual behaviours’). For instance, the themes ‘infidelity’, ‘pornography’ and ‘sexually inappropriate’ were all grouped together under the theme of ‘sexual behaviours’ as these themes were seen to be related to a common phenomenon.

Once the data had been analysed by the first author, a second researcher completed coding for inter‐rater reliability analysis on 10% of data. Cohen's kappa coefficient was used to index inter‐rater reliability by calculating the similarity of themes identified by the two researchers. This method takes into consideration the agreement between the researchers (observed agreement) and compares it with how much agreement would be expected by chance alone (chance agreement). Inter‐rater reliability for the whole dataset was calculated as κ = 0.80, which reflects a very high level of agreement between researchers that is not due to chance alone (Viera & Garrett, 2005).

Quantitative analysis of psychological states

We used thematic analysis of narratives of interactions with the relative, and then scored psychiatric content analysis scales to assess the resultant psychological symptoms of participants. We used the Psychiatric Content Analysis and Diagnosis (PCAD‐3) to assess underlying psychological states in participants. PCAD‐3 is a computer software program based on the Gottschalk–Gleser content analysis method for measuring the magnitude of various psychological states and traits from the content analysis of verbal behaviour (Gottschalk et al., 1969; Gottschalk & Gleser, 1969). The most recent version of content analysis software was utilized (PCAD‐3, Gottschalk & Bechtel, 2016). Scoring of these scales is done via software analysis of text‐based data against word‐based dictionaries, with analysis conducted at the clause level (as opposed to individual word level). Clauses are identified by the dictionary as reflecting the presence or absence of psychiatric content reflected in the scales described, with varying degrees of severity. For instance, self‐accusation (a subscale within the depression dimension) is scored by the presence of ridicule, shame, embarrassment, condemnation or moral disapproval in the text and is differentially weighted if it is experienced as coming from the self (+3), others (+2) or as expressed denial (+1). Validity and reliability of the content analysis scales have been demonstrated though corroboration with theoretically related variables and sound inter‐rater and test–retest coefficients (Gottschalk, 1995; Viney, 1983). Computerized scoring of content scales has demonstrated validity and reliability (Gottschalk & Bechtel, 1995).

RESULTS

Qualitative analysis

A total of 795 theme expressions were coded from participant responses (n = 436), with a total of 1284 references. This means participant responses were coded with an average of two individual theme expressions (e.g., ‘emotional abuse’ and ‘infidelity’) and that there were on average three expressions of each theme in the text. Four different overarching dimensions were identified from participant responses, these included: abusive behaviours, financial problems, sexual behaviours and idealization and devaluation.

Overarching dimension: Abusive behaviours from the relative

Abusive behaviours were spontaneously described by 43.9% of participants (n = 177). This dimension was made up of four themes: ‘emotional abuse’ (present in 20.6% of responses, n = 83), ‘physical abuse’ (present in 17.1% of responses, n = 69), ‘sexual abuse’ (present in 5.7% of responses, n = 23) and ‘verbal abuse’ (present in 16.6% of responses, n = 67). Table 2 displays the themes and sample text examples that demonstrate this dimension.

TABLE 2

Themes of abuse and representative text examples as reported by partners and family members in a close relationship with an individual with pathological narcissism

Theme	Text example
Emotional abuse	‘He was emotionally abusive, [he] made me believe that it was all my fault and I was the crazy one and I was told that if I ever left, he would take my children, make sure he destroyed me in court and that I would end up with nothing because I was a useless waste of skin who could do nothing right and had no skills’ (#1689) ‘Able to withhold emotions and affection for months… periods of great conversation and affection… slides bit by bit until back to [being] cold, unloving, spiteful, mean’ (#2183) ‘In his house you are his property and he can do anything to you. If you start crumbling he makes it clear that this is your fault and he does that to make you better because he loves you very badly’ (#346)
Physical abuse	‘He's got a very violent temper and has assaulted me several times during our relationship including choking me, breaking my finger, thick lip, bloody nose, bruises all over me, he's also tried to bite my face and stab me with keys. He locks me in the house to prevent me from leaving him takes my mobile so I cannot call anyone’ (#1350) ‘Growing up, it was typical for him to strike me… He stopped hitting me when I was 15 because [child protection services] got involved, but it's still not unheard of for him to threaten violence if he does not get his way. He will violently shake his fist next to his victims head or make a motion like he's going to strike someone’ (#1078) ‘She is violent and abusive. The attacks happen out of the blue, no provocation, no indication of it coming… I have been strangled twice, with deadly force [but] I am strong enough to force her off me’ (#441)
Sexual abuse	‘The last straw came last summer when he returned home black out drunk and raped me’ (#1296) ‘Forces sex. No intimacy… I finally decided to leave after he raped me twice’ (#1488) ‘He has admitted to me that he masturbated while lying next to [daughter]—he was fantasizing about her (she was 17 at the time)’ (#1105) ‘He thinks it's ok to touch his children sexually for his own satisfaction’ (#1181)
Verbal abuse	‘He has rages which are brutally cruel, with verbal tirades that include shouting, swearing, name calling, and using my most private vulnerabilities as a weapon to hurt me and mock me’ (#634) ‘He is extremely verbally abusive. He called me every name you could think of (loser, asshole, dumbass, idiot, etc.) on a daily basis’ (#806) ‘My dad yelled at me, calling me names and belittling me… I was told I was lazy, ugly and that if I kept it up like that, I would never find a husband, but who would want to marry me anyway’ (#996)

Overarching dimension: Imposition of financial burden

Participants described various behaviours involving their relatives use and misuse of finances; this occurred in 32% of participant responses (n = 129). This dimension was made up of five themes: ‘debt’, ‘stealing’, ‘controlling’, ‘dependent’ and ‘irresponsible’. Table 3 displays the themes and sample text examples that demonstrate this dimension.

TABLE 3

Themes of financial burden and representative text examples as reported by partners and family members in a close relationship with an individual with pathological narcissism

Theme	Text example
Debt	‘We always had money problems and debts but to the outside world we appeared very well… Money was always borrowed or credit cards. He had a bad gambling problem where we lost everything’ [#246] ‘He has been in bankruptcy because he does not pay bills, he does not pay people that do work for him’ [#860] ‘He is currently bankrupt, owes huge tax debts and child support arrears’ [#1119]
Stealing	‘He used my computer… to transfer $66,500 from my account’ [#122] ‘[Stole] $25,000… from the joint account’ [#1476] ‘He cheated on taxes and we owed $40,000’ [#1727]
Controlling	‘He controlled everything. … I had to justify every penny spent but he was able to spend what he wanted when he wanted’ [#1689] ‘He was extremely controlling. Controlled finances, made all the financial decisions’ [#1316] ‘I never knew where all the money went. He had nothing to show for it and would not discuss it with me… He lied to me about how much money we had and did not pay our bills. Eviction notices piled up’ [#1891]
Dependent	‘He does not have a job and expects me to pay for everything’ [#1211] ‘He is financially dependent on whichever woman he is with at the time’ [#1009]
Irresponsible	‘No self‐control with money. Refuses to live on a budget’ [#1944] ‘Believes he deserves the best of everything and will spend money on fancy cars and trips instead of paying bills or buying groceries’ [#788]

Overarching dimension: Imposition of unwanted sexual behaviours

Participants described various problematic sexual behaviours of their relative, occurring in 34.2% of participant responses (n = 138). This theme was made up of six themes: ‘infidelity’, ‘addiction’, ‘selfish’, ‘demanding’, ‘inappropriate’ and ‘withholding’. Table 4 displays the themes and sample text examples that demonstrate this dimension.

TABLE 4

Themes of sexual behaviours and representative text examples as reported by partners and family members in a close relationship with an individual with pathological narcissism

Theme	Text example
Infidelity	‘Had an affair with my best friend when I was pregnant with his son and told me the entire time I was imagining things because I was emotional from being pregnant’ [#1619] ‘He is a serial cheater with at least a dozen local sex and dating website accounts, and when I stumbled onto proof of any of them he threatened me with physical violence’ [#1688]
Addiction	‘He is addicted to pornography’ [#600] ‘He kept trying to talk me into threesomes which disgusted me. He was obsessed with porn’ [#241] ‘She was obsessed with sex… it was obviously not a normal obsession; she was forever talking about sex and it was almost impossible to have a conversation about anything else without her butting in and starting some kind of sexual talk’ [#466]
Selfish	‘He is like a robot in bed. It is only about him.’ [#1183] ‘Sex was very strange and odd. Often I would have to remind him that I was there too, not just him’ [#116] ‘He is addicted to masturbating because he loves himself so much, no one else can give him as much pleasure as he can give himself’ [#956]
Demanding	‘He expects sex 3 times a week and will sulk if he does not get it’ [#283] ‘If he did not get sex for more than 2 days he would give the silent treatment for days and then verbally abuse me’ [#1727]
Inappropriate	‘There almost always had to be an element of some sort of perversion for him to get [sexually] excited’ [#116] ‘He is an inappropriately sexual human being and is constantly making gross jokes and unnecessarily telling others about his sex life’ [#1565]
Withholding	‘He started withholding sex and intimacy because it mattered to me’ [#1681] ‘Uses sex as a tool to gain power’ [#1186] ‘Used intimacy as a punishment; would not have relations with me after I got sick’ [#1287]

Overarching dimension: Mutual idealization and devaluation from relative

Participants described the pattern of interactions with their relative as alternating between extremes of idealization and devaluation, occurring in 31% of participant responses (n = 125). Typically, at the beginning of the relationship, there was a period of mutual idealization, in which their relative presented themselves as very appealing while at the same time heavily idealizing participants. For instance,

Our early relationship felt like a fairy tale; I'd never been adored and idealized before and was totally sucked in ( #1046).

[he] was very charming in the beginning. He pursued me hard and fast and I didn't quite know what was happening… He complimented me, put me on a pedestal, and told me he loved me really early on in the game. I was flattered ( #1419).

However, participants also described how this idealization was inevitably followed by devaluation. For example,

At first, it was great. He made it seem like he was my saviour. He was kind, loving and attentive. He pressured me into getting married very quickly. After we got married he changed [and] became prone to extreme anger if I didn't compliment him enough. He is explosive, seems totally unemotional, and unstable ( #1910).

When we first met he drew me in fast… I was so taken in with this guy. He made himself to be everything I had ever wanted. After several months the lectures started… he would spend hours criticizing me, blaming me for everything. I had no local family or friends and the loneliness was horrible… Over the next years the lectures became more frequent and more harsh with increased name calling and blame. Anytime he was in a bad mood or had a bad day, where something didn't go his way, he would spend the rest of the night lecturing me. He would use sex as a means to get the lectures to stop, saying that he would stop talking if I sexually gratified him ( #1750).

Psychological symptoms in participants

As described in the Section 2.3, PCAD‐3 is a computer software program that measures the magnitude of various psychological states and traits as identified in participants verbal content. Table 5 displays the selected scores of elevated psychiatric content from analysis of our participant's text samples. Participant output scores are compared with normative scores drawn from Gottschalk et al. (1969). Results demonstrate that participants had elevated scores in areas relating to impaired personal wellbeing (‘total anxiety’, ‘total depression’, ‘hostility inward’, ‘somatic concerns’ and ‘sickness’) as well as interpersonal difficulties (‘hostility directed outward’, ‘dependency strivings’ and ‘frustrated dependency’).

TABLE 5

Psychological states of participants in a relationship with a relative with pathological narcissism

	Comparison norm (SD)	Partner (n = 230)	Ex‐partner (n = 85)	Family (n = 65)	Total (n = 436)
Total anxiety	1.48 (0.70)	2.34 a	2.40 a	2.20 a	2.26 a
Total depression	5.39 (1.53)	8.53 b	8.54 b	8.34 a	8.40 a
Hostility directed outward	0.77 (0.33)	1.33 a	1.33 a	1.37 a	1.35 a
Hostility inward	0.60 (0.35)	0.99 a	0.96 a	0.99 a	0.98 a
Somatic concerns	0.46 (0.17)	0.79 a	0.81 b	0.79 a	0.79 a
Sickness	0.46 (0.34)	2.46 c	2.31 c	2.26 c	2.30 c
Dependency strivings	0.54 (0.42)	1.28 a	1.10 a	1.32 a	1.28 a
Frustrated dependency	0.11 (0.18)	0.54 b	0.62 b	0.50 b	0.53 b

Note: Unless indicated, scores fall within the ‘normal range’.

Indicates score is ‘slightly high’.

Indicates score is ‘moderately high’.

Indicates score is ‘very high’. Norms of these scales are outlined by PCAD Manual (2016) and Gottschalk et al. (1969).

DISCUSSION

This study aimed to investigate the behavioural and relational characteristics of individuals with pathologically narcissistic traits from the perspective of those in a close personal relationship with them. Analysis of participant responses indicated themes of abuse (physical, verbal, emotional and sexual), instances of idealization and devaluation, and challenging financial and sexual behaviours from narcissistic relatives. Psychological states of participants included elevated feelings of hostility and dependency, as well as anxious, somatic and depressive symptomatology.

Narcissistic abuse and its impact on partners and family members

Recognizing ‘narcissistic’ abuse has been highlighted as a priority area for effective mental health care practice (Howard, 2019). Investigating the links between narcissism and abuse perpetration, Lowenstein et al. (2016) report on the roles of emotion dysregulation and narcissistic grandiosity which can ‘present a direct pathway to serious violence’ (p. 8). The authors describe that personality comorbidities involving narcissism significantly increases the risk of serious physical violence, consistent with the severe forms of violence described in our participant sample. Day et al. (2020) report on features of affective instability, hypersensitivity and rage for individuals with pathological narcissism. Related features, such as anger, hostility and aggression, have been argued to inform significant interpersonal dysfunction for individuals with pathological narcissism (Czarna et al., 2019; Krizan & Johar, 2015; Maciantowicz et al., 2019; Reardon et al., 2020). These findings help explain the presence of such severe forms of violence described by participants in our sample.

Our findings also present descriptions of covert forms of abuse, such as emotional and psychological abuse. This is noteworthy as majority of abuse research focuses on overt manifestations occurring within these relationships (Green & Charles, 2019; Ponti et al., 2020). Further, while most research has also focused on romantic relationships, Määttä and Uusiautti (2018) describe narcissistic abuse as occurring within familial relationships and the importance of recognizing and supporting these patient groups—a perspective supported by our sample and results. Our results also identified the presence of burdensome financial and sexual behaviours. Research has suggested the link between narcissism and the problematic use (and loss) of others' money (Jones, 2013). Further findings have highlighted the link between narcissism, sexual coercion, infidelity and sexual aggression within romantic relationships (Altinok & Kilic, 2020; Lamarche & Seery, 2019; Moradi et al., 2019). However, while the majority of research has focused on male narcissistic samples, research has also demonstrated the presence of sexual aggression, coercion and intimate partner violence in females with pathological narcissism (Blinkhorn et al., 2015; Green et al., 2020).

These themes of abuse and burdensome behaviours inform the impaired psychological states of participants in our sample. Consistent with findings of Day et al. (2019), participants in this sample were identified as having impaired mental health in both anxious and depressive symptomatology, however the current sample also reported elevated degrees self‐blame, self‐recrimination and hostility. Further, the elevated PCAD scores of dependency alongside identified themes describing patterns of idealization and devaluation may highlight the difficulty of participants to leave such relationships, despite its destructiveness (Brunell & Campbell, 2011). For instance, within the idealization and devaluation theme, one participant (#210) described the interpersonal pattern as ‘addicting’ stating that they ‘need him in my life, [and to] play by his roles. He is outgoing and fun, and I want to be part of that, I don't want to see the bad things, the things that are bad for me’ (#210). Another (#1229) described how the cycles of ‘constant negative/positive reinforcements lead to traumatic bonding which lead me to continue to take him back despite the mistreatment.’. As such, these results indicate the patterns of interpersonal dysfunction in this sample whereby participants feel both controlled or attacked by their relative and simultaneously dependent on them.

Implications for personality assessment, diagnosis and treatment

First, these results highlight the high prevalence of interpersonal dysfunction for individuals with pathological narcissism and support approaches that incorporate this factor as a key component of both assessment and diagnosis, for instance, the DSM's AMPDs, which conceptualize personality relating to key areas in both self and interpersonal functioning (American Psychiatric Association, 2013b). Consistent with the AMPD, these results clearly indicate relational deficits in both empathy and intimacy for individuals with pathological narcissism towards their partners and family. These results also support the proposed superordinate pathological personality trait domain of antagonism within the alternate model as involving the presence of challenging interpersonal behaviours. However, beyond grandiosity and attention seeking, these results suggest potential for meaningful expansion of additional traits within the antagonism domain to indicate the severity of pathology in interpersonal functioning (e.g., manipulativeness, callousness and hostility), such as that described in the ‘malignant narcissism’ subtype (Kernberg, 2008; Lenzenweger et al., 2018; Russ et al., 2008). Further, trait domains of detachment (withdrawal, intimacy avoidance and depressivity) or negative affectivity (emotional lability and hostility) may also be of relevance (Pincus et al., 2016), given links between negative affect and quarrelsome behaviours (Wright et al., 2017), and interpersonal coldness (Edershile & Wright, 2019), for individuals with pathological narcissism. Finally, these results also implicate interpersonal patterns of idealization and devaluation for individuals with narcissistic pathology. While early DSM criteria also included this for NPD (e.g., American Psychiatric Association, 1980), it was subsequently removed in order to reduce overlap with other personality disorders (Levy et al., 2011; Levy et al., 2013); however, these results suggest that it may remain a potentially salient feature of narcissistic functioning as has been suggested in alternate diagnostic and theoretical frameworks (Lingiardi & McWilliams, 2017).

These results inform approaches to treatment that consider significant interpersonal dysfunction as relevant, both internally and externally, to the treatment. First, this study highlights the importance for clinicians who are working with individuals with a partner with suspected narcissistic traits to conduct a direct assessment of abuse perpetration and current safety for these individuals. Second, these findings may also provide avenues for therapeutic interventions, such as the systematic exploration of the identified ‘fragile’ or ‘dependent’ self that partners of individuals with pathologically narcissistic traits may identify with, as this may perpetuate such individuals to remain within destructive relationships.

Regarding the treatment of individuals with pathological narcissism, interventions to promote interpersonal safety may involve the creation of a ‘treatment contract’. The treatment contract establishes clear expectations and consequences that inform treatment progression, such as those described in transference focused psychotherapy (Caligor et al., 2018), which has specific modifications for the treatment of pathological narcissism (Diamond et al., 2021; Diamond & Hersh, 2020; Stern et al., 2017). For instance, a treatment contract may include the fact that treatment progression is contingent on the client not acting out violent urges against intimate partners, or even the therapist, and rather treatment would involve exploring these impulses in therapy in a safe way, with specific consequences (e.g., contacting authorities, therapy termination) if the contract is significantly or repeatedly violated. Further, therapists need to be adequately prepared to tolerate strong countertransference reactions as related to patterns of idealization and devaluation that may occur in the therapeutic alliance (Crisp & Gabbard, 2020; Tanzilli et al., 2017; Tanzilli & Gualco, 2020).

Limitations

First, as we relied on informant ratings for both endorsement of relative's narcissism and their described behaviours, the possibility of biased reporting is increased. While the common nomenclature of ‘narcissistic’ behaviours may be highly variable across individuals, research has demonstrated the reliability of informant‐based methods of assessing narcissism (Lukowitsky & Pincus, 2013; Oltmanns et al., 2018). Second, as participants were reporting on a specific relationship at a specific time, it is unknown if the relational characteristics of participants are specific to the relationship with their relative or if they are also observable in current or previous social or romantic relationships (for instance, regarding hostility, dependency strivings, idealization and devaluation). A potential avenue for future research may be to investigate the quality (e.g., attachment) and features (e.g., patterns or schemas) of an individual's interpersonal interactions with their relative with narcissistic features compared with their wider relationships. Third, there was significant gender disparity in this sample, with the majority of participants being female and majority of relatives with pathological narcissism being male. This disparity was not unexpected, as narcissistic personality has a high gender imbalance in diagnosis and research (American Psychiatric Association, 2013b; Grijalva et al., 2015) and most participants in our sample were in a romantic, heterosexual relationship. As such, this imbalance does not preclude its relevance to the study of narcissism as typically examined; however, it does highlight the need for broader research efforts to examine diverse narcissistic presentations, such as those in females. Fourth, while use of a narcissism screening measure was utilized, there were no exclusion criteria implemented to screen out participants with comorbid or alternate diagnoses. While these results indicate the presence of pathological narcissism and co‐occurring interpersonal dysfunction, the specific function of pathological narcissism is unable to be specified against other potential personality features (e.g., antisocial personality disorder) in this sample and is a suggested avenue for future research. It was clear that for a small number of participants in this study antisocial themes were also present, meaning the narcissism was severe (or ‘malignant’ (American Psychiatric Association, 2013b; Kernberg, 2014) and included a personality constellation such as ‘paranoia, psychopathic features, sadism, and, especially, aggression’ (Lenzenweger et al., 2018, p. 319). In this way, our results are consistent with taxonomies of psychopathology such as the HiTOP (Kotov et al., 2017) in which antisocial and narcissistic personalities share a higher order dimension of antagonistic‐externalizing spectra. As such, while it is the case that only a minority of participants in this sample may also have antisocial or psychopathic personality functioning, our findings highlight the fact that these results may hold relevance beyond only that of the narcissism construct. Finally, although this study was strengthened by its large sample size, a limitation is the relatively brief length of text supplied by participants. As such, it is open to interpretation the degree of generalizability of the descriptions of relationships provided. For instance, it is unclear whether a participant who focused on describing a pattern of idealization and devaluation would have also described instances of overt physical abuse if they had provided more text. However, as participants were not asked specifically to describe dysfunctional aspects of their relationship, it is noteworthy that such descriptions were provided with regularity.

CONCLUSIONS

This study examined interpersonal behaviours of relatives with pathological narcissism from the perspective of partners and family members. Themes of abuse from the relative were described, involving physical, verbal, emotional and sexual abuse, as well as descriptions of imposed financial and sexual burden from the relative. Complex interpersonal themes were also present, such as participants and relatives engaging in mutual idealization, with subsequent devaluation from the relative. Participants psychological state was measured, revealing heightened levels of anxiety, depression, as well as heightened dependent longings. Interpersonal dysfunction is a prominent feature of pathological narcissism, and these findings provide clear examples within the context of intimate relationships. These findings also inform clinical interventions, such as the need to assess for interpersonal violence in the treatment of individuals with pathological narcissism, as well as attending to potential conflicts around dependency for partners and family members with a narcissistic relative. Treating clinicians may also need to carefully examine the therapeutic alliance with individuals with pathological narcissism, attending to themes of idealization and devaluation, as well as potentially needing to set limits and establish a sense of personal safety in the treatment.

ETHICS STATEMENT

Ethics approval and consent to participate: University of Wollongong Institutional Review Board approval was received from the University of Wollongong Human Research Ethics Committee (16/079). All participants provided informed consent to participation. All procedures were carried out in accordance with the approved procedures as stipulated by the ethics committee.