Steroid psychosis: a review

3.1. Symptomatic profile 

Two clinical profiles (Table 1) came out from the 130 cases tabulated for this review: an affective profile and a toxic-organic profile. For the sake of clarity, all cases available within the three time slots were lumped together here, including 79 collated by Lewis in the literature, 14 additional provided by Lewis [13], and 37 new cases retrieved for this review [Villareal [15], Wada [16], Wada [17], Sharfstein [18], Stoudemire [19] and Finkenbine [20]. Table 1 shows the relative frequency of both types.

Table 1.

	Symptoms	N.	%	Lr	Lc	V	W1	W2	Sh	St	F
	mania	:47	(35%)	22	7	9	5	1	2	—	1
	depression	:37	(28%)	32	1	1	3	—	—	—	—
	mania & depression	:15	(12%)	6	—	—	—	8	1	—	—
	delirium	:17	(13%)	8	4	4	—	—	—	1	—
	psychosis	:14	(11%)	11	2	1	—	—	—	—	—

Lr = Lewis’ review; Lc Lewis’ cases; V = Villareal; Wada1 = Wada2001; Wada2 = Wada 2000; Sh = Shafstein; St = Stoudemire; F = Finkenbine

The affective profile includes mania, depression, and mania and depression; these comprise 75% of cases. Psychotic (hallucinations) symptoms are often associated, up to half the cases. Mania is more frequent than depression. Early lead signs for affective symptoms are mainly insomnia, distractibilty and excitation; and later, hyperactivity, euphoria, pressured speech and irritability might be present, while apathy and mutism will appear for depression. The toxic-organic profile of delirium and psychosis makes for 25% of cases. Early lead signs for the toxic-organic profile include confusion, perplexity, agitation; hallucinations, delusions and thought impairment are present in full-blown psychosis.

3.2. Clinical course 

Median onset of psychiatric symptoms after the start of steroid treatment is an unsettled aspect: 3–4 days as reported by Hall [12] and 11 days as reported by Lewis [13]. Inspection of Lewis’ fourteen cases shows a bimodal pattern: an early onset (about 4 days) similar to the data from Hall in seven cases, and a late onset, about 3 weeks into the treatment, in seven cases. Drug dosage and clinical profile are not discriminant for time of onset. Duration of symptoms is ill-defined in the literature and the use of the mean as a measure is misleading. Data from Lewis’ cases show a difference according to the clinical profile. Delirium lasts a few days, psychosis around a week, mania about 2–3 weeks and depression longer. But the numbers are small and only rough estimates and the severity of symptoms within each group is not discussed. Moreover, althought the toxic-organic profile makes for 6 out of 14 cases, Lewis reports that 50% of patients were found to recover within 2 weeks and 90% within six weeks. Outcome is complete recovery in over 90% of cases. Hall’s data show that half cases improved within 4 days, half in about two weeks. A tail-off period of a few additional days were needed in some patients while others had their symptoms abated quicker. There is no indication in Hall’s data that the toxic-organic group improved faster than the affective group.

3.3. Patient factors 

According to Lewis, age appears not significant (39.6 yr in cases and 42.7 yr in controls) and is related to the underlying medical pathology to be treated. Distribution of cases by sex shows that twice as many female as men are found to have psychiatric reactions to corticosteroid treatment. In the 130 cases collated for this review, data about 115 (Villareal’s data are incomplete) show 77 females and 38 males, a 2/1 ratio. It may only imply that more women are affected with diseases that require steroid treatment. Lewis tried to check this point; first by excluding systemic lupus erytematosus and rheumatoid arthritis patients where the F/M ratio is around 9/1 and 3/1, respectively, resulting in a 60/40 F/M ratio; second by reviewing clinical trials where the incidence by sex was reported resulting in a 19.7% to 3.3% female to male ratio in such studies. The incidence of psychiatric reactions to corticosteroid treatment according to the underlying medical conditions to be treated is not significantly higher than the expected one in ulcerative colitis, rheumatoid arthritis, lymphoma. However, patients with SLE do have a higher incidence of corticosteroid-induced psychiatric symptoms. History of psychosis during previous steroid treatment does not predict later relapses with further treatment [12], [13]. Where relapses occurred the clinical profile was different from first reactions [12]. Data about history of psychiatric illness unrelated to corticosteroid treatment show that 7 out of 55 cases (12%) where information is available (41 retrieved by Lewis and his own 14 cases) have such a history. As no data are available for controls neither Hall nor Lewis found it significant, confirming earlier findings by Lewis and Fleminger [4] in 1954.

3.4. Treatment factors 

Lewis’ review confirmed the increased incidence of psychiatric reactions to steroid treatment with increasing average daily doses of prednisone. The observation first made in 1972 by the Boston Collaborative Drug Surveillance Project [21] was confirmed in the collated cases retrieved by Lewis, as well as studies by Sergent [22], Rosenberg [23], and Hall [12] as 77% of patients with psychiatric reactions had been receiving doses greater than 40 mg/day. However, the dose-response effect could not be demonstrated to apply to earlier onset of symptoms or longer duration of such reactions. Lewis examined data about 17 cases that had had a previous course of corticosteroid treatment. There was no evidence that the presence or absence of a psychiatric disturbance during the initial course predicted the response to a subsequent course. Contrary to early reports, Lewis [13] further reported that a small percentage of patients have persistent symptoms. He first reported that one of 14 patients had a mild permanent impairment of cognitive function.

3.5. Management 

Tapering of corticosteroid dosage and administration of psychiatric drugs for symptom relief were the two main options (13, p. 327). Lewis found that tapering only brought recovery in 92% of 36 patients, neuroleptics only brought recovery in 84%, and 100% of 8 patients managed both ways recovered. Psychiatric drugs used for symptom relief included neuroleptics and tricyclic antidepressants. Tricyclic antidepressants used in patients with depressive reactions worsened the condition that was reversed with phenothiazines [12]. Dosage of neuroleptics used to bring improvement had to be ascertained case by case. Some patients improved rapidly with with low dosage while others needed higher dosage for a longer period. Most patients improved in the first three days with a tail-off period to bring complete recovery (12, p.232). In 1979, Falk [24] used lithium as prophylaxis for corticosteroid-induced psychosis with beneficial effects.

4.1. Clinical aspects 

The area of affective responses to corticosteroids has gained scant additional evidence. Comparision has been made with excessive endogenous corticosteroids in Cushing’s syndrome patients; over half of these cases were found to be depressed and about 30% had manic symptoms, some antedating the depressive symptoms [27], [28]. Simple euphoria, as distinct from hypomania, was regarded as a reduction of dysphoria secondary to clinical improvement subsequent to corticosteroid treatment. However, Swinburn [29] showed that mood improvement preceded improvement in physiological measures in a series of patients with chronic obstructive airway disease, suggesting a primary effect on mood. Brown [30] also recently showed that mood changes were not correlated with improvement in airway obstruction in asthma patients. The association of psychotic symptoms to affective ones remain an unsettled question. Some investigators like Naber [31] found none in a series of ophtalmologic patients while early reports by Clark [5] described the association as frequent. Wada [16] studied patients with recurrent corticosteroid mood disorders and reported subacute onset, predominance of mania and more frequent accompanying psychotic symptoms than those found in single episode cases. Brown [30] found that mild mood changes occurred early (3–7 days) into the treatment and returned to baseline 10 days off prednisone, in the context of a short course (14 days) of treatment in asthma patients.

Cognitive deficits have been reported associated with either short-term (Naber [31]) or long-term (Keenan [32]) steroid treatment. The specific deficit is memory impairment of the declarative (verbal) type, and is related to involment of the hippocampal areas (Wolkowitz [33], p.1301). That impairment may be reversed with the reduction or withdrawal of corticosteroids (Newcomer [34]). However, Varney [35] reported on six cases where “steroid dementia” lasted many months and recovered very slowly from their memory deficits. Four of these patients had not developed an acute “steroid psychosis”; the two other cases showed intellectual impairment after the steroid psychosis had resolved.

Delirium has been also reported but confusing evidence still shadows proper ascertaiment of such reaction as associated with corticosteroid treatment. The evidence is based only on occasional case reports and the following confounding variables are present: 1) the general clinical state: reported in cancer patients (Stiefel [36]) where the toxic state or concurrent treatment is not given; 2) reported as a withdrawal syndrome (Campbell [37], Baloch [38]); 3) reported as corticosteroid-induced (Stoudemire [19]) in an addicted patient; and 4) could be considered as part of an acute psychotic reaction in a severely ill patient (Clark [5], Case 6). Patten [26] suggested also that some of the cases described by Hall [12] as worsening with tricyclic antidepressants treatment of the steroid psychosis might be occurrence of a worsening delirium aggravated by the anticholinergic effects of such drugs. The four cases of delirium in Lewis’ series are not explicitely described. The symptomatic configuration of delirium needs therefore further detailed global clinical assessment.

Withdrawal symptoms have been reported with cessation of corticosteroid therapy, basically with long-term treatment. As early as 1951, Freyberg [39] had described a corticosteroid withdrawal syndrome characterized by fatigue, anorexia, discouragement and depression in 20% of patients with rhumatoid arhtritis. Wolkowitz [40] reported on two cases with depersonalization, fatigue and decreased energy. Dixon [41] had given earlier (1980) a clear outlook of the condition. He gave evidence of three types of withdrawal: type 1 is linked with biochemical evidence of HPA (hypothalamic-pituitary-adrenocortical) function; type 2 is associated with a recrudescence of the symptoms of the disease for which corticosteroids were prescribed; type 3 is defined as the true drug dependence, whether physical or psychological. The symptomatic psychiatric pattern of steroid withdrawal is diverse : delirium (Campbell [37]), mania (Venkatarangam [42]), and depression (Wolkowitz [40]).

4.2. Patient factors 

There are very few leads in the literature to identify individuals at risk for steroid psychosis. According to Patten [26], there are no studies providing clear evidence that a previous history of psychiatric disorder increases the risk of psychiatric adverse effects of steroid treatment. Data provided by early studies stood still. However, Brown [30] in a recent study of asthmatic patients suggested ″that psychiatric history might predict mood responses to corticosteroid treatment” (p.59). Subjects (N=16) with a history of depressive disease had a significantly smaller increase in mania (as measured by scale and not clinically), a few days into the treatment, as opposed to subjects (N=10) without such history. Also, 50% of subjects (N=6) with a history of posttraumatic stress disorder had a worsening of depressive aspects (measured by scale). Wada [16] studied nine patients with recurrent corticosteroid-induced mood disorders. None had had previous psychiatric episodes, all had their first episode as a result of steroid treatment, and none had a family history of psychiatric disorder. But a bipolar mood disorder seemed to have been activated by corticosteroid treatment as later non steroid-induced episodes occurred. All but two had at some later point episodes unrelated to steroid therapy. Seventeen of the manic bouts occurred after an increase in corticosteroid dosage while only 3 of 11 episodes were corticosteroid-induced. It appears as if steroid treatment triggered in these patients a genuine bipolar mood disorder. This would be consistent with Pies’ observation [43] of a persistent bipolar ilness after corticosteroid administration. Three of these patients had Systemic Lupus.

4.3. Management of psychiatric reactions to corticosteroids 

The classic stand is to try to taper steroid treatment and administer phenothiazines (Lewis [13], p.327). Hall [49] had reported earlier that patients given tricyclic antidepressants did not respond well and worsened their clinical psychiatric condition. The observation still stands. Trials have been made to prevent psychiatric reactions to steroid psychosis by lithium prophylaxis. Earlier also, Falk [24] had reported that none of 27 patients given such treatment during corticosteroid therapy developed a psychotic reaction; in a control group of 44 multiple sclerosis and retrobulbar neuritis patients not given lithium 6 cases of affective psychoses occurred. Goggans [50] and Siegal [51] reported similar results. However, in clinical practice quite a few diseases treated with corticosteroids provoke renal dysfunction, also corticosteroid-induced changes in sodium balance might increase the risk of lithium intoxication. Wada [16] advocated the use of mood stabilizers in such occurrence, carbamazepine or valproate, as does Abbas [52]. But carbamazepine has the capacity to decrease serum concentrations of prednisolone [52]. A study [53] reported on the beneficial effect of fluoxetine, a serotonin reuptake inhibitor (SSRI), on corticosteroid-induced depression. While olanzapine [54] and risperidone [55] have been used in the symptomatic treatment of psychic reactions to corticosteroid treatment, its use appears limited to subacute cases. Up to now haloperidol remains a widely used neuroleptic to control most psychotic reactions to steroid therapy because its versatile mode of administration (po;sc;im;IV) makes it easier to adjust to both acute and subacute clinical situations.