یک مطالعه آینده نگر از چربی و سروتونین به عنوان نشانگر خطر بروز خشونت و خودآسیبی در بیماران روانی حاد

Abstract Cross-sectional studies have reported an association between lipids and serotonin levels and aggression, but a literature search revealed a paucity of prospective studies. Subjects of the present naturalistic study were 254 of all (489) involuntary and voluntary acutely admitted patients to a psychiatric hospital during 1 year. Serum lipids and platelet serotonin at admission were prospectively compared with recorded intra-institutional and 1-year post-discharge violence and self-harm. Total cholesterol had a significant negative relationship to inpatient suicidal behaviour and inpatient violent behaviour and to 3-month post-discharge violent behaviour. Triglycerides were a significant marker of inpatient self-mutilation and of self-mutilation in combination with suicidal behaviour at 3 and 12 months of follow-up. High-density lipoprotein (HDL) had a significant negative relationship to violence at 12-months, and to repeated violence in seven patients with two or more admissions. The post-discharge relationships between total cholesterol and violence and between triglycerides and self-harm remained significant even when controlling for other possible explanatory variables in a multivariate model. Results did not change after controlling for current medication at admission. There was no association between platelet serotonin and violence or self-harm. Future research may examine if lipid measurements add incremental validity to established clinical risk assessment procedures of violent and self-harm behaviour.

1. Introduction A significant association between low cholesterol concentrations and violence has been found across many types of studies (Golomb, 1998 and Hillbrand & Spitz, 1999). In a health-screening project, low cholesterol was associated with increased criminal behaviour (Golomb et al., 2000). Studies in psychiatric settings have given similar results (Mufti et al., 1998, Hillbrand et al., 2000, Paavola et al., 2002 and Diaz-Sastre et al., 2007). Serum cholesterol has been shown to be lower in suicidal attempters than in non-attempters (Kunugi et al., 1997, Florkowski et al., 2001, Lee & Kim, 2003, Vevera et al., 2003 and Peres-Rodriguez et al., 2008) and in violent attempters compared with non-violent (Atmaca et al., 2003 and Atmaca et al., 2008). An association between low cholesterol and impulsive violence or impulsive suicidal attempts has been shown by several authors (Paavola et al., 2002, Vevera et al., 2003 and Agargun et al., 2004). Increased cholesterol levels have been found in premeditated aggression in contrast to decreased cholesterol in impulsive aggression and anxiety (Agargun, 2002, Conklin, 2006 and Conklin & Stanford, 2008). Lower cholesterol and platelet serotonin concentrations were found in suicide attempters characterised by violence and impulsiveness (Alvarez et al., 1999 and Crowell et al., 2008). In another study, self-harm patients had significantly lower mean total cholesterol than controls, but platelet serotonergic measures did not differ (Garland et al., 2007). Aggression is often categorised as either ‘premeditated’ (predatory, instrumental) or ‘impulsive’ (reactive, affective) (Meloy, 2006 and Siever, 2008). Insufficient serotonergic facilitation of ‘top-down’ control including the 5-hydroxytryptamine2A receptors (Coccaro et al., 1997 and Krakowski et al., 2006) and 5-HT2C receptors (Winstanley et al., 2004), catecholaminergic stimulation, the gamma aminobutyric acid (GABA)-glutaminergic system (Lieving et al., 2008) and pathology in neuropeptide systems (Coccaro et al., 1998, Kirsch et al., 2005, Coccaro et al., 2007a, Coccaro et al., 2007b, Hermans et al., 2008 and Ditzen et al., 2009) may contribute to this. Heredity contributes substantially to impulsive aggression (Beitchman et al., 2006, Marks et al., 2007 and Mann et al., 2009). In most hypotheses, low cholesterol is related to low serotonin and, in turn, linked to violence, suicidal behaviour and impulsivity (Siever, 2008). A search of the literature revealed that prospective studies are scarce (Deisenhammer et al., 2004 and Fiedorowicz & Coryell, 2007) and that there seems, in particular, to be a paucity of studies conducted in acute psychiatric wards (psychiatric emergency wards). The scope of our prospective study was to examine serotonin and lipids as risk markers of violent and self-harm behaviour in patients admitted to an acute psychiatric ward over a 1-year period.

3. Results Comparisons of demographic and clinical data are shown in Table 1. Lipid concentrations are shown in Table 2. There were no significant differences in mean lipid levels between gender or for statin users (n = 9). Table 2. Lipid concentrations in mmol/l (95% CI) and in mg/dl (95% CI) for all patients, for readmitted patients, and for patients that repeat violence or self-harm after readmissions. All patients Readmitted patients All Post-discharge suicidal patients Post-discharge self-mutilating Violent patients1 n 254 45 12 7 7 T. cholesterol 5.04 (4.89-5.20) 5.11 (4.77-5.44) 4.98 (4.35-5.61) 4.89 (3.46-6.31) 4.90 (3.70-6.10) mmol/l 196 (190-202) 199 (185-211) 194 (169-218) 190 (134-245) 190 (144-237) mg/dl LDL 3.30 (3.15-3.45) 3.40 (3.15-3.66) 3.02 (2.57-3.47) 3.00 (2.05-3.95) 3.53 (2.51-4.54) mmol/l 127 (122-133) 133 (122-141) 117 (99.2-134) 116 (79.1-152) 136 (81.1-175) mg/dl HDL 1.33 (1.26-1.31) 1.25 (1.13-1.37) 1.34 (1.04-1.64) 1.18 (0.74-1.6) 0.83 (0.69-0.96) mmol/l 51.1 (48.4-50.3) 48.0 (43.4-52.6) 51.5 (40.0-63.0) 43.3 (28.4-61.4) 31.9 (26.5-36.9) mg/dl TG 1.37 (1.28-1.47) 1.50 (1.22-1.77) 2.06 (1.10-3.49) 2.00 (0.86-3.14) 1.84 (1.03-2.65) mmol/l 125 (116-133) 136 (111-161) 187 (100-317) 182 (78.1-285) 167 (93.6-241) mg/dl 1 = 3 patients during hospital stay and 4 patients after discharge. Table options The mean platelet serotonin content in patients using serotonin reuptake inhibitors (SRI) was, in nmol per 109 platelets (95% confidence interval (CI)): 0.86 (0.65–1.1). This was significantly different (P < 0.001) from the other patients, 3.3 (2.9–3.7), and patients without current medication or substance misuse at admission, 2.9 (2.5–3.3). Within the SRI-using group of patients, mean serotonin content showed no significant difference between selective serotonin reuptake inhibitor (SSRI), serotonin–norepinephrine reuptake inhibitor (SNRI) and tricyclic antidepressant (TCA) users. When non-SRI users were categorised into drug groups (other antidepressants, secondary antipsychotics, lithium, neuroleptics and stabilisers), no significant between-groups differences were found. In the further statistics of serotonin, the sample was divided into SRI and non-SRI users. 3.1. Prevalence of violence and self-harm The rates of inpatient and post-discharge violence and self-harm are shown in Table 3. Table 3. Recorded violent and self-harming patients during hospital stay and in the 0-3 and 0-12 months period after discharge. Inpatient,n = 254 3 months post-discharge, n = 178 1 year post-discharge, n = 199 Threats + acts (%) Threats (%) Less severe acts (%) Severe acts (%) Threats (%) Less severe acts (%) Severe acts (%) Violence 17 (6.8) 16 (9.0) 5 (2.8) 0 23 (12) 8 (4.1) 5 (2.8) Suicidal 2 91 (3.4) 16 (9.0) 2 (1.1) 3 (1.7) 23 (12) 5 (2.6) 8 (4.1) Self-mutilation3 61 (2.4) 2 (1.1) 5 (2.8) 0 4 (2.0) 4 (2.0) 1 (0.5) Suicidal + Self 4 0 3 (1.7) 3 (1.7) 8 (4.5) 3 (1.5) 7 (3.5) 15 (7.6) 1 acts only, 2 suicidal behaviour only, 3 self-mutilation only, 4 both suicidal and self-mutilating behaviour. Table options Consenting patients had significantly lower rates of violence after discharge than non-consenting patients (3 months: 14% vs. 27%, P = 0.005; 1 year: 19% vs. 33%, P = 0.002). No significant differences were found for self-harm. Some patients were recorded with both suicidal and self-mutilating behaviour at 3- (n = 11) and 12-month (n = 23) follow-up. Subsequently, patients were categorised into three self-harm groups: suicidal only, self-mutilation only and both suicidal and self-mutilation. Four of the seven patients with inpatient self-mutilation were ‘suicidal and self mutilation’ after 3 months. The ‘suicidal and self-mutilation’ group had a mean of 3.0 re-admissions preceded by self-harm during the 0–3 months’ period, while the mean number of re-admissions in the suicidal group was 2.2 and, in the self-mutilation group, 1.6. Corresponding figures for the three groups for the 0–12 months’ period were 2.4, 1.4 and 1.6, respectively. 3.2. Predictive validity No associations between platelet serotonin and violence, suicidal behaviour or self-mutilation were found. Predictive validity estimates of lipids are presented in Table 4. After 12 months, HDL predicted violence (AUC = 0.62, 95% CI = 0.53–0.72, P = 0.021) and TG predicted ‘suicidal + self-mutilation’ (AUC = 0.66, 95% CI = 0.53–0.79, P = 0.016). Table 4. Predictive values of lipids for any violence, any self-mutilation, and any suicidal behaviour. Inpatient 3 months outpatient AUC (95%CI) p AUC (95%CI) p Violence Total cholesterol 0.681 (0.55-0.82) 0.013 0.72 (0.61-0.81) 0.001 HDL 0.621 (0.48-0.75) 0.112 0.64 (0.52-0.76) 0.027 LDL 0.631 (0.51-0.76) 0.067 0.69 (0.59-0.80) 0.003 Suicidal behaviour 3 Total cholesterol 0.761 (0.63-0.89) 0.009 0.582 (0.45-0.72) 0.199 Self-mutilating behaviour 3 Triglycerides 0.832 (0.72-0.94) 0.006 0.511 (0.26 – 0.77) 0.899 Suicidal + self-mutilation 4 (no inpatients) Triglycerides 0.752 (0.58-0.93) 0.007 HDL1 0.691 (0.54-0.84) 0.034 1negative prediction, 2 positive prediction, 3 only suicidal or self-mutilating behaviour, 4 patients recorded with both suicidal and self-mutilating behaviour, 3, 4 four of 7 self-mutilating inpatients were “suicidal + self-mutilation” 3 mo after discharge. Table options For the subsample of involuntary admitted patients (n = 46), total cholesterol AUC of inpatient violence was 0.67 (0.51–0.83, P = 0.064). These patients constituted 18% of the total sample, and 14 (82%) of the 17 violent patients during hospital stay were involuntary admitted. For patients with repeated inpatient or outpatient violent behaviour after two or more admissions, the AUCs for the low HDL and violent behaviour were 0.85 (0.75–0.96, P < 0.001) after 3 months (n = 12) and 0.74 (0.61–0.87, P = 0.001) after 1 year (n = 17). When tested for a possible gender interaction effect, TG level was a significantly more accurate predictor for women than men concerning inpatient self-mutilation: odds ratio (OR) = 1.4 (95% CI = 1.1–1.7, P = 0.002); 3 and 12 months ‘suicidal and self-mutilation’: 2.8 (1.5–5.2, P = 0.002) and 1.9 (1.3–2.7, P = 0.002), and 3 and 12 months ‘self-mutilation only’: 2.2 (1.2–4.1, P = 0.013) and 1.9 (1.1–3.4, P = 0.018). No significant gender interactions were found for total cholesterol. Across genders and 3 months post-discharge, total cholesterol was significant for violent behaviour and TG was significant for ‘suicidal and self-mutilation’. 3.3. Univariate and Multivariate Analyses A ‘positive’ correlation was found between total cholesterol and suicidal behaviour after 3 months, OR = 1.5 (1.0–2.1), P = 0.042. When the significant lipids and the corresponding univariate significant demographic and clinical variables were entered into multivariate analyses, no variables remained significant for the inpatient period. However, TG was borderline non-significant (P = 0.068) in multivariate analysis of inpatient self-mutilation. Multivariate analyses of the significant factors of the outcome measures after discharge are shown in Table 5. Entering cholesterol after the significant factors of 3 months’ violence yielded an χ² value of 11.2 (df = 9, P = 0.001), and model fit estimates increased from 12–22% to 18–32%. Entering TG after the significant factors of 3 months’ ‘suicidal and self-mutilation’ yielded a χ² of 7.0 (df = 1, P = 0.008), and the corresponding 12 months’ χ² was 9.1 (df = 1, P = 0.003). The model fit increased from 16–41% to 20–51%, and from 17–30% to 22–39%, respectively. Table 5. Multivariate significant predictors of self-harm and violence 3 and 12 months after discharge. 3 months 12 months OR (95% CI) p OR (95% CI) p Violent behavior:1 Total cholesterol 0.38 (0.20-0.72) 0.003 0.82 (0.55-1.3) 0.332 Gender (male) 2.7 (1.1-12) 0.087 2.8 (1.1-6.9) 0.028 Inpatient violence 5.1 (0.91-28) 0.064 4.5 (1.1-19) 0.041 Affective disorder 0.33 (0.09-1.3) 0.106 0.33 (0.11-0.97) 0.045 Suicidal behavior + self-mutilation:2 Triglycerides 4.8 (1.4-17) 0.014 3.4 (1.5-7.7) 0.003 Personality disorders 8.1 (0.55-120) 0.128 20 (2.8-147.0) 0.004 Inpatient suicidal behavior 44 (1.9-1003.0) 0.028 4.3 (0.27-70) 0.301 Alcohol abuse 0.000 8.9 (2.0-40) 0.004 1, 2 Univariate significant, but not significant in multivariate analyses: 1 Involuntary hospitalization, mandatory aftercare, F1x substance abuse, F2x psychotic disorders (one year), F3x affective disorders (inverse), F4x anxiety disorders (inverse, 3 months). 2 Inpatient self-mutilation, inpatient violence, age, female, 3 Coc&SnellR2-NagelkerkeR2. Table options Controlling for current medication at admission did not affect the significant relationships between total cholesterol and subsequent violent or suicidal behaviours, or between TG and self-harm behaviours. 3.4. Other predictive characteristics and medication effect The PPVs, NPVs and NNDs with cut-off at mean value are demonstrated in Table 6. With cut-off at mean, the corresponding false positives and false negatives were 126 patients (91%) and four patients (3.5%) for total cholesterol and inpatient violence, and 69 (77%) and three (4%) for total cholesterol and 3-months’ post-discharge violence. Results for TGs and inpatient self-mutilation were 94 (95%) and one (17%), and the results for 3-months’ post-discharge ‘suicidal and self-mutilation’ were 47 (87%) and three (30%). Post hoc analyses were also performed with ‘optimal’ cut-off values (see Section 2.5 Statistics). These values were lower than mean for total cholesterol and HDL, and higher than mean for TG. With the exception of an optimal HDL cut-off at 1.00 mmol l–1 (38.5 mg dl–1) for violence repeaters at 3 months, results showed no significant improvement over the results when mean values were used as cut-offs. Table 6. Positive predictive values (PPV), negative predictive values (NPV), and numbers needed to detain (NND) pertaining to significant lipid predictors. Lipids Recorded episodes PPV NPV NND Total cholesterol1 Inpatient violence 0.09 0.97 11 Inpatient violence by involuntary admittances 3 0.36 0.85 2.8 Inpatient suicidality 0.06 0.99 17 Violence 3 months post-discharge 0.23 0.96 4.3 TG 2 Inpatient self-mutilation 0.05 0.99 20 Suicidal + self-mutilation 3 mo post-discharge 0.13 0.97 7.7 Suicidal + self-mutilation 12 mo post-discharge 0.23 0.88 4.3 HDL1 Violence one year post-discharge 0.17 0.90 5.9 Violence repeaters at 3 months 4 0.23 1.00 4.3 1, 2 cut off values 1 lower and 2 higher than mean, 3p = 0.064, 4 patients with repeated inpatient or outpatient violent behaviour after two or more admissions.