Psychiatric adverse drug reactions (ADRs) have been reported with statin use, but the literature regarding statin-associated mood/behavioral changes remains limited. We sought to elicit information germane to natural history and characteristics of central nervous system/behavioral changes in apparent connection with statin and/or cholesterol-lowering drug use, and delineate mechanisms that may bear on an association. Participants (and/or proxies) self-referred with behavioral and/or mood changes in apparent association with statins completed a survey eliciting cholesterol-lowering drug history, character and impact of behavioral/mood effect, time-course of onset and recovery in relation to drug use/modification, co-occurrence of recognized statin-associated ADRs, and factors relevant to ADR causality determination. Naranjo presumptive ADR causality criteria were assessed. Participants (n = 12) reported mood/behavior change that commenced following statin initiation and persisted or progressed with continued use. Reported problems included violent ideation, irritability, depression, and suicide. Problems resolved with drug discontinuation and recurred with rechallenge where attempted. Eight met presumptive criteria for “probable” or “definite” causality; others had additional factors not considered in Naranjo criteria that bear on casual likelihood. (1) Simvastatin 80 mg was followed in 5 days by irritability/depression culminating in suicide in a man in his 40s (Naranjo criteria: possible causality). (2) Simvastatin 10 mg was followed within 2 weeks by depression in a woman in her 50s (probable causality). (3) Atorvastatin 20 mg was followed in ~1 month by depression and irritability/aggression in a male in his 50s (probable causality). (4) Atorvastatin 10 mg was followed in several months by aggression/irritability and depression culminating in suicide in a man in his 40s (possible causality). (5) Fenofibrate + rosuvastatin (unknown dose), later combined with atorvastatin were followed in 1 month by aggression/irritability in a male in his 30s (probable causality). (6) Lovastatin (unknown dose and time-course to reaction) was followed by depression, dyscontrol of bipolar disorder, and suicide attempts in a male in his 40s (possible causality). (7) Atorvastatin 20 mg was followed within 2 weeks by cognitive compromise, and nightmares, depression, and anxiety culminating in suicide in a man in his teens (definite causality). (8) Simvastatin 10 mg was followed (time-course not recalled) by depression, aggression/irritability culminating in suicide in a man in his 60s (possible causality). (9) Simvastatin 20 mg then atorvastatin 10 mg were followed (time-course not provided) by irritability/aggression in a man in his 60s (definite causality). (10) Atorvastatin 10 then 20 then 40 mg were followed shortly after the dose increase by violent ideation and anxiety in a man in his 30s (probable causality). (11) Atorvastatin 20 mg and then simvastatin 20 mg were followed in 2 weeks by aggression/irritability in a man in his 50s (definite causality). (12) Lovastatin, rosuvastatin, atorvastatin, and simvastatin at varying doses were followed as quickly as 1 day by aggression, irritability, and violent ideation in a man in his 40s (definite causality). Most had risk factors for statin ADRs, and co-occurrence of other, recognized statin ADRs. ADRs had implications for marriages, careers, and safety of self and others. These observations support the potential for adverse mood and behavioral change in some individuals with statin use, extend the limited literature on such effects, and provide impetus for further investigation into these presumptive ADRs. Potential mechanisms are reviewed, including hypothesized mechanisms related to oxidative stress and bioenergetics.
Psychiatric adverse effects, altering mood, personality, and behavior, sometimes arise in patients receiving statins.
Statin psychiatric effects can include irritability/aggression, anxiety or depressed mood, violent ideation, sleep problems including nightmares, and possibly suicide attempt and completion.
Most adverse drug reaction (ADR) reporting focuses on non-behavioral health risks to the medication-taking individual; however, attention is increasingly given to drug-induced behavioral and mood changes that may affect self or others. Drugs and medications with behavioral concerns include alcohol (best recognized) , but also varenicline [2–4], loratadine , mefloquine , tramadol [7, 8], isotretinoin , tricyclics , benzodiazepines [11, 12], and selective serotonin reuptake inhibitors (SSRIs) [13–16], among others . Emerging evidence suggests such problems may occasionally arise with cholesterol-lowering drugs [18–20]. These drugs are widely prescribed and most prominently include statins (3-hydroxy-3-methylglutaryl coenzyme-A reductase inhibitors), which held the place of best-selling class of prescription drugs in the world and include the best-selling prescription drug in history [21–24].
Neuropsychiatric ADRs of statins, including suicide and aggression, have been reported in pharmacovigilance databases [20, 25] and in adverse event reports and series [18, 19, 26]. Moreover, adverse behaviors have been reported in settings of low cholesterol [27–29]; and of lower omega-3 fatty acid levels (the omega-3 to omega-6 ratio is reportedly reduced with statins) [30–32]. Both naturally low cholesterol and randomized assignment to cholesterol reduction in the pre-statin era have been reported to be linked to increased violent deaths [28, 33, 34], though statin randomization has not . Recent randomized controlled trial (RCT) evidence indicating that statins can have bidirectional effects on aggression may be germane here: different mediating factors, including increased sleep problems and perhaps reduced testosterone, appear to drive effects in differing directions, and bidirectional effects on oxidative stress may also be speculated to do so. Nonetheless, the literature relating mood and behavioral changes to cholesterol-lowering drugs, and depicting the character and potential implications of adverse psychiatric effects with statins, remains relatively sparse [18, 20, 25, 36, 37]. Here we present 12 cases of mood and behavioral change apparently associated with lipid-lowering agents.
A total of 12 subjects and/or their family members (if the subject was deceased) who reportedly experienced mood and behavioral changes while receiving one or more statin, as identified by the subject and/or by family members, contacted our study group and provided survey information. Written informed consent was obtained from each participant (or proxy for deceased subjects) for inclusion of their case in this case series. These 12 represent a convenience sample, chosen because the neuropsychiatric problem was the primary complaint, because the nature or severity appeared to warrant representation in the literature, because the participant or proxy was amenable to inclusion (with proper de-identification), and because the aggregate number, 12, was small enough to allow inclusion of some individual detail, yet sufficient to illustrate a suite of potential issues.
Elicited information included demographic characteristics, drug(s) used (statin and concomitant medications), dose(s), time-course of mood and behavior change relative to drug use (onset, duration, recovery), character of symptoms, and open-ended narrative of the impact of behavior, mood, and/or personality changes. We inquired whether a modification to the treatment regimen occurred—such as changes in dose, drug discontinuation, and drug rechallenge—and the impact on symptoms. Recognized risk factors for statin ADRs, and development of other recognized statin ADRs, were also elicited. Information for each case is presented in tabular form.
Cases were assessed for adherence to presumptive causality criteria using the published Naranjo drug ADR causality classification. This employs a point system of positive and negative causality points to estimate the probability of an ADR, with a score ≥9 deemed to indicate “definite” ADR causality, 5–8 “probable”, 1–4 “possible”, and ≤0 “doubtful” . For all participants, adverse event causality was at least “possible”, since psychiatric and behavioral reactions to statins or cholesterol reduction bear biological plausibility and prior reports. “Probable” causality was limited to those who experienced occurrence following drug initiation, had the drug discontinued, and improved following drug discontinuation. “Definite” causality assignments were limited to those who, in addition to recovery with discontinuation, were rechallenged with the drug, and experienced symptom recurrence.
Information on drug, drug dose, reported mood and behavioral effect, factors supporting a causal connection, other statin symptoms, family history of psychiatric problems associated with statin use, and presence of risk factors for statin adverse effects are listed in Table 1. Cases commonly involved more than one psychiatric element (Table 2). An expanded description of each case is provided (Table 3). Marked change in mood and/or behavior was commonly noted, in some instances leading to tragic consequences. In some cases, proximal mood or behavioral changes arising with cholesterol-lowering agents led to the addition of psychiatric medications, and a role for these psychotropic medications in behavioral sequelae cannot be excluded. Behavioral/mood findings often accompanied muscle, cognitive, or other better recognized statin adverse effects [39–41]. Some participants had family members who also had experienced psychiatric adverse effects attributed to statins. Several participants exhibited compelling on-off-on reproducibility of findings.
Based on Naranjo presumptive criteria for ADR causality, eight cases qualified as bearing a “probable” or “definite” causality designation. The four cases designated “possible” are included because of factors not considered in Naranjo criteria that bear on likelihood of a causal connection. (As was said for a drug bearing a similar spectrum of behavioral adverse behaviors, “the clear temporal relationship, lack of prior history of this behavior, and unusual nature of these events strengthens the accumulating scientific evidence” ). Each patient exhibited persistent absence of the symptom prior to administration of the statin, followed by persistent presence while receiving the statin (days to years). One possible exception was a man with bipolar disorder; however, he had manifested years of stability and good control since initiation of lithium until statin initiation, which resulted in loss of psychiatric stability persisting for the years he was receiving statins until his death. Prospects for a causal connection were buttressed by an adverse behavioral change while receiving statins in a first-degree relative who also experienced dechallenge–rechallenge support (family history and genetics are risk factors for statin problems) [46–52]. A total of 75 % of cases were accompanied by other symptoms with a confirmed relation to statin use, including muscle symptoms [41, 53–60], cognitive problems [40, 61–63], and dermatologic reactions . Additionally, 50 % of cases had factors previously shown to be linked to an elevated risk of statin ADRs, such as thyroid conditions .
Behavioral and psychiatric changes in the cases presented range from violent nightmares to aggression, mood/personality change, violent or homicidal ideation (in some instances culminating in suicide), each in apparent association with statin use. The temporal association between the drug initiation and mood and behavior change, and again between drug discontinuation and resolution of symptoms where this occurred, suggests a causal connection in a number of these cases. Notable mood and behavioral changes for all patients or introduction of serious psychiatric events began after drug initiation. The latency profile is consistent with that for other statin ADRs that share common mechanisms [40, 41], and bear RCT support [39, 59, 61, 62]. Symptoms persisted or progressed with continued use in all cases. Those able to discontinue the drug experienced resolution of symptoms. For those in whom rechallenge was possible, symptoms recurred. The presence of ADRs and risk factors with a confirmed relation to statin ADRs is consistent with common pathophysiological factors, hypothesized to underlie many statin adverse effects, extending to behavioral effects . (12 cases are included; for those interested in knowledge of other cases of this kind, a 13th case, involving a physician with behavioral changes while receiving statins and leading to professional review, is planned for inclusion in a separate manuscript on statin adverse effects in physicians).
Explanation of Findings; Comparison/Contrast with Other Findings in Literature
Though a relation of statins to instances of behavior alteration may seem counterintuitive, it fits with a body of existing literature. Observationally, lower cholesterol has been linked to greater violence (including aggression, suicide, homicide, violence) in many studies, including prospective longitudinal studies [28, 29, 64, 65]. Behavioral consequences of lipid-lowering medications have been reported for non-statin treatments, including fibrates  (implicated in case 5), with support extending to meta-analyses of RCTs showing significantly increased violent death [28, 33, 34]. RCT meta-analysis in the statin era did not support an increase in violent death on average (indeed the point estimate was lower, though not significantly) . RCT patient selection may be one factor: excess violent deaths on pre-statin lipid drugs were preferentially evident in those with risk factors for violence—e.g., alcohol, psychiatric history, and non-compliance . Of note, there was no evidence of more patients with these characteristics in the cholesterol-lowering group, but the excess of events emphasized these patients. Psychiatric histories, alcohol/substance use, and low conscientiousness are risk factors for lower compliance, so may lead to exclusion with compliance run-ins, but are also risk factors for adverse behaviors, so vulnerable individuals might be preferentially excluded . Relative exclusion may affect detection of risk in two ways. First, the same fractional (relative) risk change will lead to a greater number affected, and more power to show a change, in those at higher risk (a doubling of nothing is still nothing) . Second, the fractional risk change itself may be greater in a behaviorally vulnerable subset (effect modification): illustrating this point, the odds ratios risk of psychiatric events with mefloquine use was reported to be 3.8 among those without a psychiatric history, versus apparently double that (8.0) in those with a psychiatric history .
Other factors may also explain this. Randomized trial evidence examining statin effects on aggression provides potential insights: statins (vs. placebo) promoted both average significant increases and reduction in aggression, in different groups . Typical effects in men (particularly in men who were both younger and less aggressive at baseline) were toward reduced aggression , with older age and female sex shifting the distribution. Simvastatin has been shown to both significantly lower testosterone  and worsen sleep problems on average . (It may also promote sleep apnea in some ). For men receiving simvastatin, the magnitude of each of these effects significantly predicted the change in aggression, in opposite directions . Sleep problems and sleep apnea are elsewhere linked to irritability [72–74], as well as aggression and violence . Lower testosterone in some settings is linked to lower aggression and violence [76, 77].
Other mechanisms may be theorized. Low central serotonin has been linked to low cholesterol and to aggression [28, 78, 79]. However, whole blood serotonin (which correlates inversely to central serotonin) did not predict aggression in mediation analysis within a randomized trial . We also hypothesize that effects on aggression, and perhaps mood problems on statins, may be linked to oxidative stress  and inter-related mitochondrial dysfunction (which has been linked to temper disorders) . The explanation accords with our finding that behavioral adverse effects commonly co-occur with better known statin adverse effects, with the documented relation of better known statin adverse effects to oxidative stress and mitochondrial dysfunction , and with the documented connection of mitochondrial dysfunction to a range of psychiatric problems [82–87].
Other hypothetical mechanisms of potential relevance, such as the role of cholesterol in synapse formation or membrane function, and myelin production have been reviewed elsewhere [40, 88], but in contrast to mechanisms cited above, triangulating evidence for a role is currently lacking.
The explanation accords with our finding that behavioral adverse effects commonly co-occur with better known statin adverse effects, for which a relation to oxidative stress and mitochondrial dysfunction has been elucidated . Mitochondrial dysfunction has a documented connection to a range of psychiatric problems [82–87], and may also contribute to behavioral change with statin-induced oxidative damage.
More lipophilic statins have better brain penetration [89, 90], though all statins have some ability to cross the blood–brain barrier . Whether this is important for central effects of the kinds described is unknown, since peripheral effects can have central consequences—and since prooxidant effects of statins, which are linked to the occurrence of statin adverse effects , also raise blood–brain barrier penetrability [93, 94]. Each of the patients in our report experienced problems while receiving at least one of simvastatin, atorvastatin, or lovastatin, which are lipophilic statins. However, most involved atorvastatin and simvastatin, which have also historically been the most frequently prescribed statins. This, and the lack of a defined base population in which statin prescribing rates are known, obviates our ability to comment on whether lipophilicity is relevant to these effects.
This study has the limitations inherent to all case series: data are observational, which constrains causal inference. However, a profound change in state while receiving a drug, particularly with dechallenge–rechallenge support, strengthens the case for causality. There is no defined base population or control group, precluding calculation of rates and risk ratios, and occurrence of adverse neuropsychiatric effects in these cases, even if presumed causal, has no inherent implications for usual effects of statins on the outcomes reported—nor do usual effects have necessary implications to individual ones. Rather, this case series underscores that statins may in specific individuals promote adverse behavioral, mood, and personality changes, irrespective of whether behavioral (or mood) effects are on average favorable, adverse, or neutral. Even reporting rates relative to other ADRs may be a challenge to gauge meaningfully for neuropsychiatric problems, because these may be particularly sensitive, and may go unreported. In an analysis of emails abstracting statin ADRs mentioned spontaneously, mood or personality problems were spontaneously related as part of the ADR complex in a minority, but in a survey directly asking about each of a list of symptoms the participant attributed to statins, among those self-selected for having at least one such symptom, a majority cited a neuropsychiatric effect. (However, in the latter, there is no gauge of severity; and physical problems can themselves contribute to some level of psychic distress).
Some of our participants had underlying conditions (such as a psychiatric disorder) that can themselves lead to an adverse outcome; however, as above, the odds ratios for risk of psychiatric events due to a drug can be magnified in the presence of a psychiatric history . Consistent with this, most of the excess cases of violent death arising in the active treatment, cholesterol-lowering arms in a pre-statin-era analysis of randomized trials had other risk factors for violent death ; since those risk factors themselves were not shown to be increased in the active treatment group, this underscores that individuals with such issues may represent a vulnerable subgroup (as with mefloquine). As a further analogy, an analysis of rhabdomyolysis cases in the San Diego area showed that typical instances involved the confluence of more than one risk factor—consistent again with an increase in the odds of an event “due to” each agent, in the presence of the other . In our cases, factors such as the temporal relation to statin use, and concurrence of other statin-compatible ADRs increases the prospects for causality with statins.
Data rely on self-report. However, patient self-reports of ADRs can be a valuable and reliable tool [96–100], and—if heeded—may hasten recognition of important ADRs ; such benefits have led to standardized implementation of patient reporting for EU-based pharmacovigilance databases . Self-selection of participants is inherent to studies with volunteer patients; however, for studying those with ADRs, whether or not they reflect typical effects, observed effects are important.
Though statins are widely tolerated, they may be among the growing list of prescription agents that, in some participants, may increase the risk of serious psychiatric events and/or behavioral changes. In the cases cited here, these adverse experiences posed risks to the safety of self and others—sometimes, tragically, adversely affecting marriages and careers, or culminating in death. The possibility of such ADRs, even if rare, should be recognized by physicians who prescribe cholesterol-lowering drugs, such that if personality or behavior changes arise, the drug can be included in considerations of etiology and treatment. This series extends the modest literature on behavior and psychiatric changes apparently associated with cholesterol-lowering treatment. These findings further the evidence that cholesterol-lowering drugs should be added to the list of agents that bear consideration when new irritability, or aggressive or violent behavioral changes arise.