Clinical and pharmacologic risk factors for neuroleptic malignant syndrome and their association with death


*Ümit Tural, MD, Kocaeli Üniversitesi Tıp Fakültesi, Psikiyatri Anabilim Dalı, Umuttepe, 41380 Kocaeli, Turkey. Email:


Aim:  The aim of the present study was to evaluate demographics, clinical features, psychiatric diagnoses and prognosis of neuroleptic malignant syndrome (NMS) reported in Turkey, and to assess their association with mortality.

Methods:  Data on all reported cases of NMS in the Turkish Psychiatric Index between 1985 and 2005 were collected. The type, dosage and administration period of neuroleptics, the clinical and laboratory findings; and prognosis were compared in terms of mortality.

Results:  Thirty-six patients with a mean age of 33.67 ± 16.98 years were identified. Fifteen (41.7%) were diagnosed as having schizophrenia or other psychotic disorders and the same number were diagnosed as having affective disorder. Remaining five (13.9%) were diagnosed with other psychiatric disorders and 1 (2.7%) had no psychiatric diagnosis. Twenty-two (61.1%) of the NMS cases were associated with high potency typical neuroleptics. Association between an atypical antipsychotic and NMS has been reported in one case. NMS appeared within 7 days after initiation of the antipsychotic medication in the majority of samples (n = 19, 52.8%). Several combinations of rescue treatments were used in the majority of cases (n = 19, 52.8%), although bromocriptine (n = 22, 61.1%) was the most frequently preferred rescue treatment for NMS. Benzodiazepines were significantly better than the other treatment options in preventing mortality. Five out of the 36 patients (13.9%) with NMS had died. Age was the only significant independent factor that was associated with mortality.

Conclusions:  Benzodiazepines may be included in the treatment of NMS. The mortality rate due to NMS in Turkey was lower than the previously reported rates from other developing countries.

NEUROLEPTIC MALIGNANT SYNDROME (NMS) is an uncommon, but life-threatening medical condition.1 Its prevalence is estimated to be 0.02–2.4% in developed countries.2–6 In developing countries such as China, India, Mexico and Turkey, the incidence of NMS among patients receiving neuroleptics was reported to be in a range of 1.23–6.0 per 1000.7–10 The mortality rates were found to be 10.5% in Russia,6 38% in India,8 and 55% in Spain.11 Somereports have found a relationship between demographic variables, psychiatric diagnosis, psychotrophics, previous central nervous system (CNS) lesions, other physical illnesses, dehydration, hot weather, exhaustion and i.m. administration of neuroleptics and NMS.1,12–17 The treatment options of NMS are electroconvulsive therapy (ECT), bromocriptine and dantrolene, but tend to be mainly supportive.1–5,12–17 Te mortality rate of NMS and preferred treatment options in Turkey, however, have not been reported as yet. The aim of the present study was therefore to evaluate demographics, mortality rate and clinical characteristics of NMS patients, and to explore the success rates of the preferred treatments in Turkey.


The Turkish Psychiatric Index and PubMed were searched for the key words ‘neuroleptic malignant syndrome’ and ‘Turkey’ between the years of 1985 and 2005. In total, 36 cases of NMS reported from Turkey were identified.18–43 All cases were reviewed according to demographics, psychiatric diagnoses, type, dosage and duration of neuroleptic treatment, clinical and laboratory findings, duration of treatment, and prognosis. Because some previous versions of DSM (such as DSM-III and III-R) were used to classify some of the patients, the diagnostic criteria in each case were reassessed and recoded according to the DSM-IV criteria for any comorbid axis I psychiatric disorder except NMS. We grouped the psychiatric disorders into affective disorders, psychotic disorders and other disorders to explore the relationship between diagnoses and mortality.

Data analysis

Variables were compared either on Mann–Whitney U-test for continuous variables or χ2 test for categorical variables. Fisher's exact test was used instead of χ2 test when a 2 × 2 table had a cell with an expected frequency of <5. Forward multiple logistic regression, which determines the relative contribution of each independent factor to variance explained, was used to examine factors for mortality due to NMS. Independent variables included in logistic regression analysis were those significantly associated with the outcome (died or recovered) bivariately. Statistical analyses were performed using SPSS 10 (SPSS, Chicago, IL, USA). Significance was defined as P < 0.05 with a two-tailed test.



General information, clinical signs and laboratory findings are listed, respectively, in Tables 1,2. The sample consisted of 18 women and 18 men. There was no relationship between sex and mortality due to NMS (Fischer's exact test, P = 1.000). The mean age was 33.67 ± 16.98 years. There was a significant difference in mean age between patients who had died due to NMS (49.00 ± 20.06 years) and who had survived (31.19 ± 15.40 years; Mann–Whitney U = 32.50, P = 0.037).

Table 1.  Turkish NMS cases
StudyPatient age (years), sexAxis I diagnosesPossible offending agent (duration)Rescue medication (days for recovery)Outcome
  1. ECT, electroconvulsive therapy; NMS, neuroleptic malignant syndrome.

Alpay et al.1840, FBipolar I disorder, manic episodeLithium + chlorpromazine (5 days)Biperiden + clonazepam (13 days)Recovery
Arkonaçet al.1934, FSchizophrenia, disorganized typeHaloperidol + biperiden (2 days)Biperiden + bromocriptine + dizaepam + ECT (48 days)Recovery
Arkonaçet al.1922, FBipolar I disorder, manic episodeHaloperidol + biperiden (3 days)Bromocriptine + diazepam + ECT (72 days)Recovery
Arpacıet al.2074, FDementia of Alzheimer typeChlorpromazine (7 days)Bromocriptine (20 days)Died
Aşkın et al.2119, MBipolar I disorder, manic episodeHaloperidol + lithium + biperiden (3 days)Bromocriptine + diazepam (19 days)Recovery
Aydın et al.2220, FMental retardationHaloperidol + chlorpromazine + biperiden (15 days)Bromocriptine + diazepam + ECT (25 days)Recovery
Bora et al.2356, MBipolar I disorder, manic episodeQuetiapine (8 days)ECT + lorazepam (21 days)Recovery
Erişik & Bayülkem2457, FSchizophrenia, paranoid typeHaloperidol + biperiden (6 days after dose increase of haloperidol)Bromocriptine (2 days)Died
Eryüce et al.2517, MSchizophrenia, paranoid typeECT + trifluoroperazine (28 days)Diazepam (22 days)Recovery
Gedizoğlu et al.2635, MDelirium due to alcohol withdrawalHaloperidol (15 days)Bromocriptine (40 days)Recovery
Gedizoğlu et al.2655, MDelusional disorder, persecutory typeHaloperidol (7 days)Bromocriptine (30 days)Recovery
Güney & Akbostancı2751, MDelusional disorder, persecutory typeHaloperidol + fluphenazine depot + chlorpromazine (10 days after dose increase of all)Bromocriptine + diazepam (31 days)Recovery
Herken et al.2820, MBipolar I disorder, manic episodeHaloperidol + lithium (13 days)Diazepam + bromocriptine (49 days)Recovery
Herken et al.2817, MBrief psychotic disorderHaloperidol (14 days)Bromocriptine + dizaepam + ECT (51 days)Recovery
Herken et al.2815, FBipolar I disorder, manic episodeHaloperidol (10 days)Bromocriptine + diazepam (54 days)Recovery
Herken et al.2822, FBipolar I disorder, manic episodeFluphenazine depot + lithium (15 days)Bromocriptine + diazepam (56 days)Recovery
Herken et al.2827, FBipolar I disorder, manic episodeHaloperidol + lithium (14 days)Bromocriptine + diazepam (38 days)Recovery
Işık & Karaağaç2975, FDementia, vascular typeMethyldopa + fluphenazine depot (45 days)Bromocriptine (15 days)Recovery
Orhan et al.3037, FBipolar I disorder, depressive episodeLithium + valproate + chlorpromazine + risperidone + sertraline (60 days)Bromocriptine + diazepam (8 days)Recovery
Ökten et al.3127, FDelirium due to postoperative state (cervical costa)Haloperidol (in operation)Only supportive (10 days)Recovery
Önder et al.3218, MSchizophrenia, catatonic typeECT + haloperidol (29 days)ECT (37 days)Recovery
Önder et al.3320, MSchizophreniform disorderChlorpromazine + haloperidol (2 days)Diazepam + ECT (15 days)Recovery
Önder et al.3335, MBrief psychotic disorderChlorpromazine + haloperidol + biperiden + ECT (5 days)Biperiden + ECT (5 days)Died
Önder et al.3316,MSchizophreniform disorderHaloperidol + biperiden (3 days)Diazepam + ECT (33 days)Recovery
Önder et al.3335, MDelusional disorder, mixed typeHaloperidol + fluphenazine depot + biperiden (3 days)Bromocriptine + propanolol (36 days)Recovery
Sır et al.3424, FDelirium due to normal-pressure hydrocephalusHaloperidol + biperiden (6 days)Bromocriptine (15 days)Recovery
Sürmeli et al.3543, MBrief psychotic disorderHaloperidol + chlorpromazine (15 days)Only supportive (10 days)Recovery
Şenol et al.3623, MDelirium due to head traumaHaloperidol + thioridazine (6 days)Only supportive (50 days)Died
Turhan et al.3764, KBipolar I disorder, manic episodeLithium (2 years) + hydroxyzine pamoate (2 months) + haloperidol injection 2 days agoBromocriptine (18 days)Recovery
Turhan et al.3719, KMajor depressive disorder, first episode with psychotic featuresHaloperidol + biperiden (7 days)Bromocriptine (9 days)Recovery
Ünal et al.3815, FSchizophrenia, paranoid typeHaloperidol + chlorpromazine (2 days)Biperiden + diazepam (20 days)Recovery
Ünal & Tunca3926, MSchizophrenia, paranoid typeHaloperidol + thioridazine + biperiden (25 days)Biperidene (2 days)Recovery
Yaluğet al.4056, MBipolar I disorder, depressive episodeHaloperidol + chlorpromazine + biperiden IM injection (3 days)Dantrolene (3 days)Died
Yıldız et al.4134, FBipolar I disorder, manic episode with psychotic featuresZuclopenthixol HCl (17 days) + zuclopenthixol acetate IM + thioridazine on day 17Bromocriptine + ECT (14 days)Recovery
Yumru et al.4237, FBipolar I disorder, manic episodeChlorpromazine single dose 25 mg i.m.ECT (7 days)Recovery
Zoroğlu et al.4327, MBipolar I disorder, manic episode with psychotic featuresHaloperidol + biperiden + metimazol (15 days)Bromocriptine + ECT (36 days)Recovery
Table 2.  Clinical signs and laboratory findings of NMS patients
Clinical signsReported n (%)Laboratory findingsNormal rangeReported n (%)
  1. ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; CT, computed tomography; CPK, creatine phosphokinase; CSF, cerebrospinal fluid; EEG, electroencephalography; ESR, erythrocyte sedimentation rate; LDH, lactate dehydrogenase; NMS, neuroleptic malignant syndrome.

Muscular rigidity36 (100.0)High CPK38–174 U/L30 (83.3)
High fever36 (100.0)Leukocytosis4.3–10.8 × 103/mm327 (75.0)
Altered consciousness36 (100.0)High AST7–27 units/L19 (52.8)
Diaphoresis31 (86.1)High ALT1–21 units/L18 (50.0)
Tachycardia29 (80.6)High LDH50–150 units/L14 (38.9)
Dysarthria and mutism22 (61.1)Proteinuria0 to 20 mg/dL10 (27.8)
Labile arterial blood pressure20 (55.6)Invasive cortical atrophy on CT6 (16.7)
Dysphagia19 (52.8)High ALP50–160 units/L6 (16.7)
Tremor15 (41.7)Slow wave pattern on EEG5 (13.9)
Urinary incontinence16 (44.4)Hypocalcaemia8.5–10.5 mg/dL5 (13.9)
Tachypnea13 (36.1)High fasting blood glucose70–110 mg/dL4 (11.1)
Hypertension10 (27.8)Hypokalaemia3.5–5.0 mEq/L3 (8.3)
Sialorea9 (25.0)High ESR1–20 mm/hr3 (8.3)
Pathological reflexes4 (11.1)Hematuria1–10 red blood cells per high-power microscopic field2 (5.6)
Inhibited deep tendon reflexes3 (8.3)High CSF protein0.18 to 0.58 g/L2 (5.6)
Low serum ferrum60–160 µg/dL2 (5.6)

Pre-NMS psychiatric disorders

The most common psychiatric disorders (Table 1) were schizophrenia and other psychotic disorders (n = 15, 41.7%), and bipolar disorder (n = 15, 41.7%). The rate of NMS during the treatment of other mental disorders was 16.7% (n = 6). NMS was reported in a non-psychiatric patient who had been exposed to a single dose of haloperidol i.m. injection after an operation for cervical costa. Because 3 × 2 (diagnosis × prognosis) comparison led to considerable high insufficient cell counts (66.7%), we performed 2 × 2 comparison by excluding the ‘other diagnoses’ group. Using that method, affective disorders and psychotic disorders were not significantly different in the death ratios (n = 1, 6.7% vs n = 3, 20.0%; Fischer exact test, P = 0.598).

Pre-NMS medications

Drugs that have been mostly implicated in NMS were high potency antipsychotics (n = 22, 61.1%). Low potency antipsychotics (n = 3, 8.3%) or a combination of high and low potency antipsychotics (n = 11, 30.6%) were also reported. Three (25%) of the patients on co-medication of antipsychotics and two on a single antipsychotic died (Fischer exact test, P = 0.307). One case of NMS (3%) was solely associated with quetiapine, which is an atypical antipsychotic.23 Seven patients (19.4%) were co-medicated with lithium. None of the patients co-medicated with lithium died, but there was no statistically significant association between lithium co-medication and mortality (Fischer exact test, P = 0.559). Although the mean time leading up to NMS onset after receiving the antipsychotic associated with the NMS was 11.73 ± 12.58 days, NMS appeared within 7 days of neuroleptic treatment initiation (n = 19, 52.8%) in the majority of the patients. There was no significant difference in mean neuroleptic use duration between the patients who died (5.40 ± 1.52 days) and those who survived (12.74 ± 13.28 days; Mann–Whitney U = 54.50, P = 0.306).

Clinical presentation of NMS

Hyperthermia, muscle rigidity, and altered consciousness were reported in all cases. Among other signs tachycardia, diaphoresis, dysarthria, dysphagia, labile arterial blood pressure, tremor, urinary incontinence, pathological reflexes and inhibited deep tendon reflexes were reported (Table 2). The most frequent laboratory findings were elevated creatine phosphokinase (CPK), leukocytosis, high aspartate aminotransferase (AST), alanine aminotransferase (ALT), lactate dehydrogenase (LDH) and proteinuria. Some previously unreported findings, such as cortical atrophy on computed tomography (CT), hematuria, slow wave pattern on electroencephalography (EEG), and high fasting blood glucose were also observed.

Course and type of rescue treatments

The mean treatment duration of NMS was 26.97 ± 19.04 days (range, 2–72 days). There was no significant difference in the mean treatment time between the patients who died (16.00 ± 20.36 days) and those who survived (27.55 ± 17.36 days; Mann–Whitney U = 43.00, P = 0.120). Median treatment times were 5 days and 22 days for mortality and survival arms, respectively. NMS had been treated with ECT, bromocriptine and/or benzodiazepine in general. Among the rescue medication options, only benzodiazepine was significantly associated with recovery (Table 3). A combination of rescue treatments such as benzodiazepines, bromocriptine, ECT, dantrolene was used for 19 patients (52.8%). Combined rescue treatments were not significantly successful in preventing death compared to single treatments (n = 18, 94.7% vs n = 13, 76.5%; Fischer's exact test P = 0.167). Five out of 36 patients (13.9%) have died.

Table 3.  Association between rescue treatment and outcome
Treatment modalityDeath, n (%)Recovery, n (%)P
  • Due to combined treatments the sum exceeds 100%.

  • Fischer exact test.

  • ECT, electroconvulsive therapy.

Benzodiazepine0 (0)17 (100)0.047
Bromocriptine2 (9.1)20 (90.9)0.357
ECT1 (8.3)10 (91.7)0.646
Biperidene1 (20.0)4 (80.0)0.549
Dantrolene1 (100)0 (0)0.139

Variables associated with death

We next performed a logistic regression analysis using the variables that were significantly associated with the treatment outcome (death or recovery) in the previous bivariate comparisons. There were two variables (age and benzodiazepine rescue treatment) associated with the treatment outcome bivariately. Thus, these two variables were entered into regression analysis. Multiple logistic regression analysis (Nagelkerke R2 = 0.226) showed that age was associated with the outcome with an overall 85.71% correct classification function (Exp β = 1.061, 95% confidence interval [CI] = 1.003–1.121, P = 0.039).


The three significant findings of this study are as follows: (i) the incidence of death due to NMS in Turkey is 13.9%; (ii) benzodiazepines may be preferred as a rescue agent in the treatment of NMS; and (iii) age is a significant risk factor for mortality. The mean age of patients reported from Turkey was 33.67 ± 16.98 years, although Pearlman reported that men younger than 40 years old appeared to be more susceptible.44 Haloperidol and/or chlorpromazine were the neuroleptics most frequently implicated in the present cases of NMS. This may in part reflect the fact that they are the two most prescribed neuroleptics in Turkey. Until 2000, classical antipsychotics were more preferred and prescribed than atypical antipsychotics in Turkey. Since 2000, however, the prescription of atypical antipsychotics has risen gradually and has overtaken that of classical antipsychotics. Rates of classical antipsychotic prescription are as follows: 1996, 89.8%; 2000, 51.9%; 2005, 15.5%.45 There are some case series that reported affective disorders as the dominant diagnosis,5,44 but this was not the case in the present study.

Clinical symptoms and NMS

Fever, tachycardia, and altered consciousness have been previously reported in all of the patients with NMS.15 Similarly, we have found that all of the present patients had high fever, muscle rigidity and altered consciousness, but diaphoresis and tachycardia were also observed in 86.1% and 80.6% of patients, respectively. It was previously reported that a high fever, tachycardia, muscle rigidity, altered consciousness, labile hypertension, tachypnea and diaphoresis were present in two-thirds of patients, but dysarthria, dysphagia, mutism, pathological reflexes and changes in deep tendon reflexes, hypersecretion, opisthotonus, and urinary incontinence were found to be rare.14 The symptoms and signs in the present study were consistent with this, except for a relatively high prevalence of dysarthria and dysphagia, which was found in 61.1% and 52.8% of the present patients, respectively.

Laboratory findings and NMS

A variety of laboratory abnormalities have been described in NMS. These are: high serum CPK, ALT, AST, LDH and alkaline phosphatase; leukocytosis; low serum iron, calcium, magnesium; proteinuria, myoglobulinuria; an increase in protein level of cerebrospinal fluid (CSF); and slow wave pattern on EEG. The results of laboratory assessments that could be obtained were similar to some in literature. Differing from the previous series, an increase in fasting blood glucose in four (11.1%) of the patients, hypokalaemia in three (8.3%) and cortical atrophy on CT in six patients (16.7%) were also determined. Likewise, hypocalcemia and hypokalaemia, high fasting blood glucose, hematuria, high CSF protein and slow wave pattern on EEG, which were observed in the present patients, have not been reported previously. Two hypotheses have been suggested for NMS pathogenesis: central dopaminergic blockage,46 and direct toxic effect on skeletal muscle.47 The latter hypothesis and the muscle rigidity could explain the hypocalcemia with a mechanism of depletion. There are also some studies in which is mentioned the probability of necrosis in hypothalamic nuclei that could account for the high CSF protein, slow wave pattern on EEG, and fasting blood glucose.48 In contrast, a previously published case report has suggested that NMS may occur as a complication of uncontrolled diabetes mellitus with dehydration or, conversely, NMS might precipitate diabetic coma in patients with previously well controlled blood glucose levels.49 Although low serum iron levels or iron deficiency anemia have been suggested as a risk factor for NMS in the literature,50 only two studies from Turkey measured and reported low iron levels in patients with NMS.23,41

Rescue medications and rate of mortality

Of the 36 patients with NMS, 13.9% (n = 5) died. Thus, the mortality rate from NMS in Turkey is lower than that of other developing countries.7,8 The studies from both developed and developing countries have reported that 10.5–55.0% of NMS patients have died.3,6–8,11 The present mortality rate is in the lower end of the global range. It was reported that death can occur within 3–30 days after the onset of symptoms or within the first 24–72 h.14,51 We found that a median treatment time to death was 5 days, which is similar to that of previous reports. Therefore, it can be said that death may mostly occur in the early phase of treatment in Turkey. Sodium dantrolene, bromocriptine, diazepam and ECT were found to be effective in previous case reports in the literature.52–55 It is also thought that the use of adjunctive anticholinergic drugs is essential in managing afebrile patients experiencing neuroleptic-induced parkinsonian symptoms, while it was not found to be directly correlated with benefit.56 In the cases reported from Turkey, different combinations of these treatments were used. Sodium dantrolene was used for one of the 36 patients, but that patient died. It should be noted, however, that this drug is not available in Turkey as yet (the sodium dantrolene used in the present study was provided by the patient's family from abroad). Treatment of NMS lasted for 26.97 days on average (range, 2–72 days). We achieved a statistically significant better recovery rate with diazepam compared to the other pharmacological treatment options on bivariate analysis. This finding may be helpful for clinicians who deal with NMS. Considering the lower mortality rate with benzodiazepines, it may be a rewarding choice in the treatment of NMS. Nevertheless, age was the only significant independent variable that was associated with mortality in NMS patients. An increase in age by 1 year increases the risk of death 1.061-fold. This may be explained by the impairment in the physical health status as a result of the aging process. In contrast, this finding conflicts partly with those studies reporting an increased prevalence for NMS in very young patients.1,16,44

The present study had some limitations. Because of the small number of deaths, there may be inflated type I error or an insufficient power to detect real differences, which might be related to the statistical method being used. The present finding of a possible increased risk of death by age needs to be interpreted cautiously, because the 95%CI of the risk coefficient is very close to 1 at the lower end. The authors have refrained from making strong conclusions and recommendations about the relationship between the use of benzodiazepines and mortality due to NMS. Despite a few previous studies suggesting beneficial effects of benzodiazepines,54,57,58 it may be too premature to reach such a conclusion based solely on the present results. Additionally, the recoding of diagnoses based on the old psychiatric classifications in an effort to adjust to the newest classification has the potential to cause a misdiagnosis or information loss. Some confounders should also be kept in mind, such as pre-NMS physical condition and seriousness of NMS. Because the study did not take into account a severity rating of NMS, the association of benzodiazepines with better outcome may be related to the possibility that those patients treated with benzodiazepines may have had milder NMS or had better pre-NMS physical condition.

In conclusion, clinical signs and laboratory findings, and mortality rates of NMS have overlapped with previous reports considerably. Age is a significant factor for mortality from NMS, and clinicians may consider using benzodiazepines in the treatment of NMS. The mortality rate due to NMS in Turkey is lower than in other developing countries, but death occurs mostly in the early phase of treatment.