Neuroleptic-induced movement disorders in a naturalistic schizophrenia population: diagnostic value of actometric movement patterns

Globally, many schizophrenia patients still suffer from neuroleptic-induced movement disorders (NIMDs) [1–3] in spite of the increasing use of second generation antipsychotics, which in general have a lower propensity for NIMDs [4]. Although in earlier years movement adverse effects were even considered to be an indicator of antipsychotic action [5], today they are seen as burdening and stigmatizing phenomena that should be avoided by careful choice of treatment [6]. NIMDs, which are investigated in this study and classified in DSM-IV [7], include neuroleptic-induced akathisia (NIA), neuroleptic-induced parkinsonism (NIP) and tardive dyskinesia (TD). Pseudoakathisia (PsA) is a movement disorder with objective signs of akathisia, but without the subjective feeling of restlessness [8]. PsA is not included in the DSM-IV classification [7].

Besides motor inconvenience, NIMD can cause subjective suffering through cognitive or emotional disturbance [9]. Patients have described rigidity and dullness of thinking in association with parkinsonism [10]. Akathisia causes significant distress and is reported to be associated with suicide attempts [11] and aggression [12]. It results in reduced compliance with antipsychotic drug treatment and thus a cause of psychotic relapses [13].

The interest in the instrumental measurement of NIMDs in addition to clinical examination and observer-based rating scales has long traditions. Instrumental measurement has been reported 1) to identify more NIMDs patients than observer-based ratings [14], 2) to be more sensitive to sub-clinical motor changes [15], 3) to exhibit greater linearity with regard to severity [15–17], and 4) to need less training to achieve sufficient inter-rater reliability and test-retest reliability [16].

Previous studies have mostly been conducted with selected patients and non-NIMD controls, but in chronic patients using long-term antipsychotics, different NIMD often occur together [1–3].

We recruited 99 chronic schizophrenic institutionalized adult patients from a state nursing home in Estonia [3]. Inclusion criteria were a DSM-IV diagnosis of schizophrenia or schizoaffective disorder, stable antipsychotic medication (for at least one month), and an age of 18–65 years. Diagnosis was made using a semi-structured interview according to DSM-IV criteria for schizophrenia by a psychiatrist (SJ) and medical records. Patients with severe somatic illness or neurological illness were excluded. Written informed consent was obtained from the subjects and the study was approved by the Ethics Review Committee on Human Research of the University of Tartu.

An experienced psychiatrist (SJ) assessed all subjects to identify NIMDs (NIA, NIP, and NITD) in accordance with DSM-IV criteria [7]. PsA diagnosis was established according to Barnes and Braude [8]. NIMDs and PsA patients are in this report defined as the movement disorders group.

The temporal connection between NIMD and PsA with a neuroleptic medication was established retrospectively by interview and medical records.

Seventy nine (79.8%) patients were receiving conventional antipsychotics (mainly haloperidol, cyclopentixol, perphenazine, levomepromazine, chlorpromazine, and also thioridazine, sulpiride, chlorprotixen, fluphenazine), while 20 (20.2%) were receiving clozapine (one of whom used clozapine combined with sulpiride). No new atypical antipsychotics were used. The mean daily chlorpromazine equivalent dose [38] was 328 mg (SD 221, range 1417).

The actometric recording was performed while sitting in a standardized clinical interview for 30 minutes between 9 and 11 AM, a method described previously as measuring "controlled rest activity" [21]. Controlled rest activity is a parameter of motor activity in a situation where sitting still is adequate and expected, but not instructed or required. The actometers (PAM3, IM-systems, Baltimore, USA) were attached to the ankles of the subjects to measure lower limb motor activity. Actometers are small computerized movement detectors of match-box-size which do not influence normal moving of the patient. The mode of data collection was digital integration, and the sampling rate was 40 Hz and the chosen epoch was 0.1s. PAM3 records acceleration signals exceeding 0.1G. The actometry and controlled rest activity method have been described previously by Tuisku et al. [21] and Janno et al. [24].

To test the feasibility and properties of actometric recording in a normal clinical setting we trained raters without previous experience of evaluating actometric recordings. A team of five neuropsychiatrists developed rater instructions and a data collecting form. Data evaluation training comprised two hours, followed by supervised evaluation of ten actometric recordings. Two raters (BA and AV) trained according to this procedure, and achieved an appropriate level of inter-rater reliability (0.44 to 1.0, mean 0.82) during their training phase. Raters were blinded and had no access to patient data other than the actometric recording.

The two raters evaluated all study subjects' actometric activity recordings for the existence of activity periods, the duration of activity periods (activity for at least 10 seconds), the existence of rhythmical activity. Raters calculated from persistent rhythmical activities three most dominant frequencies for every patient (if patients had rhythmical activity with different frequencies). Raters found the highest acceleration peaks in activity periods in the scope of 10 seconds. After calculating inter-rater reliability, a meeting between raters was held to establish consensus values for the estimated frequencies of activity periods for 27 patients. The extracted data was then assessed by KT and SJ to find any patterns for individual NIMDs and PsA. Answers to the subjective question were analyzed for a correlation with movement disorders diagnoses.

The inter-rater reliability was measured by kappa coefficients for categorical values and intra-class correlation (ICC) coefficients for continuous values. A two-way ANOVA mixed model was used to calculate ICC, so as to estimate the reliability of a single rating [39]. One-way ANOVA was performed to analyze the ability to discriminate different qualities of movement patterns. Differences between the movement disorder (NIA, NIP, TD and PsA) and the non-movement disorder groups in activity periods were analyzed by the Mann-Whitney two-tailed U-test for continuous variables (frequency, amount of periods). Chi-square analyze was used for dichotomous variables (presence of activity periods, rhythmical activity). Where necessary, Fisher's exact test was used for calculating significance. The performances of movement patterns in case identification were evaluated by receiver operating characteristics (ROC) analyses. The software used in analyses was SPSS 12.0 (SPSS Inc. Chicago, Illinois, USA).

The presence of rhythmical activity, amount of activity periods, maximal frequency, and highest values of acceleration peaks differentiated between the subgroups of NIMDs and PsA. The other parameters of movement patterns showed no ability to discriminate between movement disorders groups by one-way ANOVA analysis (data not shown). Presence of activity periods and rhythmical activity in actometric recordings had the best abilities to discriminate the patient group of pooled NIA and PsA (Table 2).

The differences between particular movement disorder and non-movement disorder groups in the highest acceleration peak medians are presented in Table 4. Highest acceleration peak medians (i.e. greatest digital integration of acceleration) in activity periods were almost significantly (p = 0.053) different (Mann-Whitney U = 444) between NIP (203, interquartile range (IQ) 89, 250) and non-NIP (116, IQ 67, 190) groups. No significant differences were found between other groups.

Median values of third frequency and minimal frequency were not statistically significantly different between groups.

To evaluate screening properties of actometry against DSM-IV and PsA diagnostic criteria we used ROC analysis. The area under the ROC curve (AUC) of the lower limb activity count was 0.80 for PsA and 0.84 for pooled akathisia (NIA and PsA).

Answers to the subjective question differed between the movement disorders and the non-movement disorder group (Pearson chi-square 8.209, p = 0.004). The ROC curve for the screening performance of the subjective question in NIA is presented in Figure 1. ROC analysis of the subjective question for differentiating abilities showed an AUC value of 0.67 for NIMD and PsA, 0.87 for NIA, 0.52 for NIP, 0.46 for TD, and 0.25 for PsA.

Figure 1. Receiver Operating Characteristic (ROC) curves for subjective question and lower limb activity count against DSM-IV defined neuroleptic-induced akathisia (NIA).

Our main finding was that actometry could objectively differentiate NIA and PsA patients from the non-movement disorder group and from all the other NIMD patients. Also, the subjective question on movement problems was able to detect NIA patients from all other subjects. Answers to this question were highly selective for NIA cases, but not for other NIMDs and PsA.

The strength of the study is the naturalistic non-selected patient sample. Therefore, we can generalize results to other chronic schizophrenia in-patients. It is possible that co-morbid movement disorders, existing in this real-life sample, compromise the previously reported characteristics of individual NIMDs or PsA. Therefore, a loss of differentiating ability of actometry is seen in a clinical setting like ours.

A limitation of the study is that for practical reasons, the clinical diagnosis of NIMD or PsA was made and the subjective question was asked by the same clinician, which could have contaminated the investigator's judgement.

Another limitation of our study was that it could not detect the sub-clinical movement disorders, because we used the DSM-IV [7] and PsA [8, 20] diagnostic criteria as the "gold standard": These criteria do not identify sub-clinical cases. Thus, the non-movement disorder subgroup may include sub-clinical movement disorder subjects, who were not detectable by DSM-IV criteria for NIMD. The latter is supported by the following findings: 1) even 75% of non-movement disorder subjects showed rhythmical activity which can be related to movement disorders, seldom a normal activity during rest; and 2) the non-movement disorder patients had less activity periods in recordings, but when they had, then frequencies were quite similar to NIMD or PsA (mostly TD) subjects.

The actometric recording and analysing of the standardized left-leg motor activity is moderately time-consuming, but it may have limited use. The method can be used in NIMD and PsA patients for differentiating akathisia (pooled NIA and PsA) from other patients, but not for discriminating between NIA and PsA. Our study employed ankle recording only, and actometric recording from the ankle and wrist simultaneously may give more information for detecting NIMD in the schizophrenia patient population. Actometry allows a quantifiable recording of current state of movement disorders and is thus useful in research settings. Further research should focus on the ability of actometry to show change in the patient condition. Initial studies in this direction have already been done [21, 35].

Actometry is useful for measuring change in the overall movement count or patterns after a change of risk factors (dosage, antipsychotic type, time course etc) in experimental conditions, and further research can show the ability of actometric data to drive treatment decisions in a clinical setting.

The present study showed that pooled NIA and PsA patients had different patterns in lower limb descriptive actometry than other patients in a non-selected sample. Careful questioning of patients is a useful and clinically significant method of diagnosing NIA in a clinical setting.