Abstract
Background:
The objective of this study was to investigate the concordance in attitudes of psychiatrists towards the doses of antipsychotics given to stable outpatients with schizophrenia and to examine the psychiatrists’ estimates of equally potent doses of haloperidol and olanzapine.
Methods:
We asked all 22 doctors serving at the psychiatry department of Jönköping County Hospital if they considered the combined dose of antipsychotics for 20 individual patients to be ‘low’, ‘medium’ or ‘high’. We also asked each doctor to state the dose of haloperidol that they considered to be clinically equivalent to 20 mg/day of olanzapine.
Results:
The inter-rater reliability (Krippendorff’s alpha (α)) was 0.50, and the mean estimated dose haloperidol considered clinically equivalent to 20 mg/day of olanzapine was 4.45 mg/day.
Conclusions:
The inter-rater reliability (Krippendorff’s α) was low, suggesting lack of agreement. The dose of antipsychotics given to a patient might thus be more influenced by which doctor they meet than the severity of the disease. The respondents in this study considered a mean dose of 4.45 mg/day of haloperidol to be clinically equivalent to 20 mg/day of olanzapine. This is a considerably lower dose than was determined by an international consensus study of antipsychotic dosing, and more in line with the available PET studies measuring central dopamine receptor blockage of optimal clinical doses.
Introduction
Antipsychotics are effective in the acute and long-term management of schizophrenia, psychosis and bipolar affective disorder [Foussias and Remington, 2010] and have been widely prescribed not only for these conditions, but also for the management of other psychiatric conditions, e.g. behavioral disturbance, anxiety and sleeping difficulties. In contrast to national recommendations [Lindefors, 2009; NICE, 2014], the prevalence of combined antipsychotics (two or more prescribed antipsychotic drugs) is high in clinical practice [Suokas et al. 2013; McCue et al. 2003; Faries et al. 2005] and combined antipsychotics can contribute to high total doses [Centorrino et al. 2004; Harrington et al. 2002]. Despite the widespread use of combined antipsychotics, there is no firm evidence that high doses are more effective than standard doses [Paton et al. 2008; Centorrino et al. 2004; Davis and Chen, 2004]. It is humbling that, more than 50 years after the introduction of antipsychotics, we are still struggling to use them appropriately [Centorrino et al. 2004].
The majority of adverse effects are dose-related, and the use of high doses is associated with an increased risk of adverse effects [Barnes and Paton, 2011; Weinmann et al. 2009; Correll et al. 2007]. These range from an increased likelihood of extrapyramidal symptoms (e.g. dystonia, bradykinesia, tremor, akathisia, dyskinesia), sedation, sexual dysfunction, type 2 diabetes mellitus, weight gain, and hypotension, to more severe conditions such as cardiac arrhythmia and neuroleptic malignant syndrome [Gardner et al. 2005]. Therefore, the use of antipsychotics entails a difficult trade-off between its benefits and its adverse effects.
Studies in individuals suffering from schizophrenia have shown that a central dopamine D2 receptor blockade of 65–80% will provide an antipsychotic effect with a minimum of adverse effects [Nordstrom et al. 1993; Kapur et al. 1997], but the positron emission tomography (PET) method used in these studies is not suitable for clinical use, due to high costs and low accessibility. Determination of drug serum concentration is useful to determine compliance, but the therapeutic interval is wide [Stroup, 2007], and side effects may occur within standard concentrations. In most cases, prescribers rely upon clinical assessment in determining the dose required. Experience-based opinions on dosing are susceptible to many influences, including national and local dosing practices, manufacturers’ recommendations, and personal views on what constitutes a high or a low dose. Prescribers may also adjust doses depending on patient-related factors such as age, sex, nationality, hepatic function impairment, renal function and the severity of psychiatric symptoms [Gardner et al. 2010]. Lifestyle-related factors such as diet, exercise habits, smoking, and consumption of alcohol, may also contribute to changes in the dose [Stroup, 2007]. Due to the inherently subjective nature of dosing, it is possible that clinicians recommend doses outside the narrow range of central D2 blockage of approximately 15% that has been shown to be optimal for treatment. For example, an international consensus study [Gardner et al. 2010] was conducted to assess clinical opinions on the optimal daily dose of haloperidol for the treatment of a reference case. The reference case was an adult man with DSM-IV schizophrenia, showing moderate symptoms, with at least 2 years of antipsychotic treatment, and untreated during the previous month. The responders in the study agreed that the target dose ought to be 5–10 mg of haloperidol per day. However, numerous PET studies have shown that this daily dose will block more than 85% of central dopamine D2 receptors and thus make side effects highly likely [Kapur et al. 1997, 2000]. Furthermore, due to the subjective nature of dosing, one would expect high variability among the opinions of doctors over what constitutes a ‘low’ or ‘high’ dose of any given drug.
The aim of this exploratory study was to examine the concordance, that is, the inter-rater reliability, in the attitudes of 22 psychiatrists towards the doses of antipsychotics given to 20 stable outpatients with schizophrenia. More specifically, we asked the doctors if each patient’s total individual dose of antipsychotics should, in their opinion, be regarded as a ‘high’, ‘medium’ or ‘low’ daily dose. To highlight any discrepancy between the doctors’ estimates of an equally potent dose and a previously undertaken international consensus study [Gardner et al. 2010], we also asked each doctor to estimate the dose of haloperidol that he or she considered to be clinically equivalent to a dose of 20 mg/day of olanzapine.
Method
We asked all 22 doctors (16 senior psychiatrists and 6 psychiatry residents) serving at the psychiatry department of the Ryhov County Hospital (with a catchment area of a little more than 147,000 inhabitants during the day and 340,000 inhabitants at night) whether they considered the combined dose of antipsychotics for twenty individual patients to be ‘low’, ‘medium’ or ‘high’. A total of 20 consecutive stable patients were selected by a secretary during a 2-week period in the fall of 2013 from the psychosis outpatient service where the authors were working on a regular basis as clinicians. The doctors participating were informed that the patients regularly attended the psychosis outpatient clinic, were well and in a stable condition. They were also informed of the patients’ age, sex and current antipsychotics dosage. No other information was given. Finally, each doctor was asked to state the dose of haloperidol that they considered to be clinically equivalent to 20 mg/day of olanzapine.
This study was undertaken as part of the general improvement assignment in the psychosis outpatient clinic and therefore ethical approval was not sought. The doctors’ answers were anonymized and stored on a computer with password protection.
We used Krippendorff’s alpha (α) [Krippendorff, 2013] to measure the inter-rater reliability of the doctors’ subjective classification of dose. A bootstrap procedure was used with 10 000 resamples [Krippendorff, 2013] to obtain a 95% confidence interval for the α value. All calculations were made using the statistical software R (version 3.2.2) with the irr package (version 8.84). We also calculated the mean and median dose of haloperidol judged clinically equipotent to 20 mg of olanzapine.
Results
The inter-rater reliabilities (Krippendorff’s α) were 0.48, 0.52 and 0.50, respectively, for the 16 senior psychiatrists, the 6 psychiatry residents and for all 22 doctors. The 95% confidence intervals (CIs) for α are presented in Table 1.
The inter-rater reliabilities and mean and median doses of haloperidol considered clinically equivalent to 20 mg/day of olanzapine among the twenty-two doctors.
The 16 senior psychiatrists considered 4.2 mg/day (mean value) of haloperidol to be clinically equivalent to 20 mg/day of olanzapine, whereas the 6 psychiatry residents considered 5.3 mg/day of haloperidol to be the appropriate figure. The median answer was 4 mg/day in both groups. Taken together, the 22 doctors considered 4.45 mg/day (mean value) of haloperidol was clinically equivalent to 20 mg/day of olanzapine. See Table 1.
Discussion
The aim of this study was to gain a greater understanding of the agreement of psychiatrists concerning dose levels of antipsychotics. To the best of the authors’ knowledge, there is no prior study on inter-rater reliability on psychiatrists’ attitude to patients’ total antipsychotic dose level. Krippendorff’s α is applicable to all levels of measurement, including ordinal data, that is, data that can by ranked as the doctors’ responses in this study [Krippendorff, 2013]. Fleiss kappa (κ) [Sim and Wright, 2005] is a more widely used measure of agreement but is restricted to nominal data, that is, categorical data (e.g. sex or nationality). Both measures calculate the degree of agreement; a value of 0 implies total random agreement and a value of 1 implies perfect agreement. The inter-rater reliability among doctors was Krippendorff’s α 0.50 (95% CI 0.34–0.59). Krippendorff suggests the following for interpreting his coefficient: ‘Rely only on variables with reliabilities above α = .800. Consider variables with reliabilities between α = .667 and α = .800 only for drawing tentative conclusions’ [Krippendorff, 2013]. Thus, according to Krippendorff, an inter-rater reliability of 0.5 would suggest lack of agreement. However, the sample size in this study was small and therefore the results must be interpreted with caution and confirmed in other studies.
The risks of high doses of antipsychotics are well known [Centorrino et al. 2004; Barnes and Paton, 2011; Weinmann et al. 2009; Correll et al. 2007] and the majority of adverse effects are dose-related. In practice, doses are often increased when patients respond inadequately to treatment, or when a more immediate response is required. Today, there is no clinically objective method of determining the minimum effective dose of antipsychotics, and therefore we have to rely entirely on clinical assessment. Our results show lack of agreement between doctors as to whether a dose-level of antipsychotics is regarded as ‘high’ or ‘low’. This could be a problem for our patients. It might be that the dose of antipsychotics given to a patient is influenced more by which doctor they meet than by the severity of the disease. The results of this study show that this risk is substantial.
Chronically ill individuals tend not to take medicine as prescribed, and compliance is often low [Tiihonen et al. 2011; De Geest and Sabate, 2003; Culig et al. 2011]. However, Remington and colleagues have shown that patients, who take their dose of antipsychotics every other day instead of daily, still respond well to the treatment [Remington et al. 2011]. It may be the case that patients intuitively know their minimal effective dose more accurately than their psychiatrists do, and are therefore consciously noncompliant with the doctor’s prescription in order to minimize unwanted side effects. In this case, a reasonable approach would be to work with the patient in gradually reducing the dose to the minimal effective dose: a dose by which the patient does not suffer from side effects, but does not relapse. On the other hand, if the lack of compliance is unconscious and a sign of impaired judgment, the lower dose of antipsychotics would increase the risk of a relapse into psychosis. As symptoms often develop slowly, this approach might therefore be risky [Nyberg et al. 1997]. It would require a well-functioning outpatient service, continuous supervision and the means to increase the dose, if needed. We suspect that this approach is uncommon and that patients are prescribed high doses of antipsychotics to minimize the risk of relapse.
The respondents in this study considered 4.45 mg/day of haloperidol to be clinically equivalent to 20 mg/day of olanzapine. This is a considerably lower dose than determined by an international consensus study of antipsychotic dosing [Gardner et al. 2010], in which the respondents finally agreed that 10 mg/day of haloperidol is clinically equivalent to 20 mg/day of olanzapine. PET studies have shown that a dose of 10 mg/day of haloperidol corresponds to more than 90% of central dopamine D2 blockade [Kapur et al. 2000; Lim et al. 2013]. In all probability, this high blockade produces adverse effects and may lead to reluctance on the part of the patient to continue medication. It is not unusual for 10 mg/day of haloperidol to be used as the reference substance in head-to-head comparisons [Beasley et al. 1996; Hamilton et al. 1998; Montgomery et al. 2004]. The high dopamine D2 blockade produced by the reference substance could induce side effects which would be a disadvantage and would favor the study compound. Nyberg and Farde and others have proposed that studies should use doses that produce comparable central dopamine D2 blockade. Such equipotent doses would lead to more equitable comparison between the compounds [Lim et al. 2013; Nyberg and Farde, 2000; Hugenholtz et al. 2006].
Micrographia is an early fine-motor extrapyramidal symptom seen during antipsychotic treatment [Haase, 1980]. Using single-photon emission-computed tomography (SPECT), Kuenstler and colleagues have shown that micrographia induced by antipsychotics correlates with the central dopamine D2 receptor occupancy, and quantification of micrographia during antipsychotic treatment could function as a proxy for receptor occupancy [Kuenstler et al. 1999]. Measuring the reduction of handwriting area (i.e. micrographia) during treatment could thus be one unbiased method for antipsychotic dosing. This approach needs to be investigated more in detail in the clinical setting, and could be a tool for clinical dosing in the future.
It has been recognized that there is considerable variability in patients’ responses to different antipsychotics and doses [Stroup, 2007]. The overall effectiveness of antipsychotic medication is limited by drug efficacy and tolerability. Antipsychotic drug choice and dosage should be based on the individual’s clinical situation, history and preferences. Listening to our patients when they want to lower the dose, and to find an effective way to do so, must be considered a part of the ars medicina within psychiatry. Analysis of micrographia during treatment could be one way to accomplish this.
Significant outcomes
Due to a lack of a clinically objective method of finding the optimal antipsychotic dose in a specific patient, there is a risk that a patient will be prescribed different doses of antipsychotics, depending on which doctor the patient meets.
The low inter-rater reliability for the dose estimates of antipsychotics found in this study highlights the importance of collegial meetings in which treatment and doses can be discussed and the prescribing culture can be improved.
There was a discrepancy between the views of a distinguished international board of research psychiatrists and the views of clinically active psychiatrists on what constitutes equipotent doses of haloperidol and olanzapine. As haloperidol is often used as a comparative drug in head-to-head comparisons with new drugs, an excessive dose of haloperidol might favor the drug investigated in such trials.
Limitations
Although the psychiatrists are working in the sixth largest county in Sweden in terms of population, the group of respondents was small in number.
The doctors’ answers were conveyed via a questionnaire, as opposed to in an interview in which answers could have been followed up and clarified.
There is no universally accepted consensus on inter-rater agreement values.
Footnotes
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Conflict of interest statement
The authors declare that there is no conflict of interest.