Abstract
Background:
Antipsychotic therapy can bring a great deal of benefit to individuals with schizophrenia and other psychiatric diagnoses, though this class of medications is also associated with intolerabilities due to possible adverse events. Notable adverse events commonly experienced include movement disorders and metabolic concerns that can occur even at low doses and minimal exposure.
Objectives:
Describe the prescribing patterns of antipsychotics in the inpatient behavioral health setting, and report on characteristics that lead to populations having higher or lower dosing exposures to antipsychotic therapy that follows FDA guidelines.
Design:
Multicentered, retrospective, cross-sectional chart review.
Methods:
The study was conducted at two academic hospitals in the inpatient behavioral health setting. Information was collected on 785 cases of patients who were hospitalized during the study period with a psychiatric diagnosis that necessitated a prescription for an antipsychotic medication at the time of discharge. Differences in antipsychotic drug (dose, type, number, and duration of treatment) were characterized, and data were collected on clinician adherence to FDA guidance, as listed in the package insert of each antipsychotic drug. Primary outcomes were the amount of Total Chlorpromazine Equivalent Dosing doses and the clinician’s adherence to FDA guidance.
Results:
This study found that the male gender (p < 0.0001), no social support (p < 0.0025), more previous antipsychotic trials (p < 0.0001), and longer hospital stays (p < 0.0001) were statistically significant for patients receiving higher antipsychotic doses at discharge. Additionally, patients with comorbid diabetes (p < 0.0001), more antipsychotics at discharge (p = 0.001), and Medicare insurance (p = 0.005) were less likely to receive antipsychotic medication following FDA recommendations and current guidelines.
Conclusion:
The results of this study can inform practice regarding antipsychotic prescribing patterns in different populations. Conducting prospective studies across a broader range of institutions, it is recommended to better understand the relationship between patient demographics and prescribing patterns during periods of acute illness.
Keywords
Background
Antipsychotics are a commonly prescribed class of medications for psychiatric disorders. Antipsychotics are often administered along with non-pharmacologic therapies to assist in improving patient prognosis in a multitude of psychiatric disorders. 1 While medications may have similar rates of efficacy, 2 and considering that individual response and tolerability often cannot be predicted, it is usually required to follow a trial-and-error treatment selection strategy. 3 Hospitalized patients in the acute phase of illness often warrant an aggressive strategy to aid in symptom resolution, though this can come at the cost of adverse events.
Though antipsychotic therapy can impose a great deal of benefit in the acute setting, including lower rehospitalization rates and lower risk of mortality, many common adverse events can occur with this class of medications in a dose-dependent manner.4,5 The notable dose-dependent adverse events, such as movement disorders and metabolic concerns that are commonly experienced with these medications, can remain even when controlling for dose or number of antipsychotic medications.6–8 Though each antipsychotic has a unique adverse events profile that affects each patient differently, studies have shown that up to 70% of patients experience at least one debilitating adverse event from antipsychotic therapy. 9 Of these adverse events, the annual incidence has ranged from 37% to 44% for pseudo-parkinsonism, 26% to 35% for akathisia, and 8% to 10% for tardive dyskinesia. 10 Other reported adverse events commonly experienced by those receiving antipsychotic therapy include weight gain, excessive sleep, insomnia, sexual dysfunction, dry mouth, constipation, urinary problems, and dizziness. 11
This report aims to describe the prescribing patterns of antipsychotics in hospitalized patients to report on characteristics that lead to populations having higher dosing exposures to antipsychotic therapy, as well as which populations are less likely to be exposed to antipsychotic therapy that follows FDA guidance. A previous report by the current authors 12 had found that involuntary admission status, past medication trials, a diagnosis of schizophrenia or bipolar disorder, and those who lacked social support had higher total daily doses of antipsychotics upon discharge. It also found that minority populations, increasing age, and patients with a diagnosis of insomnia were more commonly prescribed inappropriate medications per FDA guidance. 12 Considering that the previous study was limited by being performed at a single (public) institution, this new study included a second (private) institution to expand the number of patients studied, confirm results of the previous study, and incorporate other institutional factors that may affect the prescribing patterns in each institution.
Methods
The reporting of this study conforms to the STROBE guidelines for cross-sectional studies, 13 and the checklist is attached as Supplemental Material.
Study design
This multi-centered, retrospective, cross-sectional chart review was conducted at two academic hospitals, Ochsner Medical Center (OMC) and University Medical Center of New Orleans (UMCNO) in the inpatient behavioral health setting. UMCNO is an institution that is part of LCMC Health System, the nonprofit health system of Southern Louisiana hospitals, provides services to a high percentage of participants using Medicaid and Medicare. 12 OMC is a nonprofit healthcare organization of the Gulf South that provides services to a blend of private insurance, managed care plans, and government insurance such as Medicare and Medicaid.
Outcomes
The primary outcomes (response variables) are the amount of Total Chlorpromazine Equivalent Dosing (TCPZE) doses and appropriateness of medications prescribed to patients. The criteria used for clinician adherence followed the FDA guidance for the prescription of antipsychotics (diagnosis, dosages, polypharmacy, contraindications, and allergies); while “treatment-refractory disease” was indicated for patients with two or more documented antipsychotic trial failures, “inappropriate polypharmacy” was indicated for patients who have no documentation of a history of clozapine use but were prescribed multiple antipsychotics at discharge. 12
Inclusion/exclusion criteria
Similar inclusion and exclusion criteria were used as in the previous study by the authors, and data collected remained constant as well. 12 Accordingly, only UMCNO or OMC adult patients with a psychiatric diagnosis and receiving antipsychotic medication (oral formulation or a long-acting injectable formulation) at discharge were included in the study. Patients who became medically destabilized or transferred to another unit of the hospital during the study timeline were excluded.
Sample size and data collection
Information was collected, from medical charts review, on a total of 785 cases of patients who met the inclusion criteria during the study period (October 1, 2015–December 31, 2017), and incorporated the patient population from the previous study at UMCNO. Demographic data were collected on each patient. All patients who joined the study from both institutions were admitted to the inpatient setting of behavioral health, indicating that their illness was severe enough to qualify as harm to self, harm to others, or gravely disabled. Differences in antipsychotic drug, dose, type, number, and duration of treatment were characterized. Dosing was standardized to “Total Chlorpromazine Equivalent doses (TCPZE),” as this is a common means of classifying antipsychotic dosing equivalency.14,15 Data were also collected on clinician adherence to FDA guidance as listed in the package insert of each antipsychotic drug.
Statistical analysis
The data were analyzed using the R statistical programming language. Analysis of variance was used to examine the factors impacting TCPZE. TCPZE was calculated using the classical mean dose method. 15 The explanatory variables considered included race, ethnicity, diagnosis, insomnia, length of stay in hospital, age, weight, gender, insurance, voluntary or involuntary admission, housing, social support from family/friends and assertive care teams (ACT), history of type II diabetes mellitus (DMII), history of hyperlipidemia (HLD), number of antipsychotics prescribed at discharge, total past antipsychotic trials, alcohol use, and drug use. Categories with only a few individuals were combined into larger categories (e.g., an “Other” category) to maximize statistically significant results while ensuring that the division into categories was still useful and logical. An “unknown” category was used when there were a few individuals missing data in specific variables.
A square root transformation was applied to TCZPE after a preliminary analysis indicated the need to have equal variance and normally distributed residuals. As a precaution, one patient was deleted because it had a large influential point in the data (Cook’s distance), and five patients were not considered since there was no data regarding their TCPZE.
Factors influencing whether patients received appropriate medicines or not were analyzed using logistic regression, and the same explanatory variables as above.
All explanatory variables were tested for statistical significance after controlling for all other statistically significant variables. T-tests and proportion tests were also used to analyze data. The Holm and Bonferroni multiple comparison corrections were used. Using a square root transformation on the length of stay was considered, but it did not significantly change the results.
Results
Charts reviewed during this study included a total of 785 patients (398 from UMCNO and 387 from OMC). Table 1 shows a summary of their demographic characteristics. More relevant results regarding differences in TCPZE doses and appropriate medications are presented below.
Demographics of patients in the sample (n = 785).
Asian, Native American, other groups, or no response.
Totals include total past trials, not patient cases.
BZD, benzodiazepines; EPS, extrapyramidal symptoms; OMC, Ochsner Medical Center; THC, tetrahydrocannabinol; UMCNO, University Medical Center of New Orleans.
Total chlorpromazine equivalent dose
Main patient demographics and treatment characteristics having a significant impact on TCPZE were gender, social support, total of antipsychotics prescribed at discharge, length of stay, and number of past antipsychotic trials (Table 2), and diagnosis type (Table 3).
Factors significantly affecting TCPZE.
Number of past antipsychotics—there was a significant reduction for each antipsychotic removed.
CI, confidence interval; TCPZE, Total Chlorpromazine Equivalent Dosing.
Letter plot of the impact of diagnosis on TCPZE.
Table shows the impact of diagnosis on TCPZE. In the table, diagnoses that share a letter do not have a statistically significant difference between them. For example, SCZ spectrum disorders and bipolar mania share the letter A, so the difference between them is not statistically significant.
SCZ, Schizophrenia; TCPZE, Total Chlorpromazine Equivalent Dosing.
Gender had a statistically significant impact on TCPZE (p = 0.0002), with women having significantly lower TCPZE than men (p < 0.0001). A 95% confidence interval is that the sqrt (chlorpromazine equivalent dose) is predicted to be between (0.96, 2.84) lower for women than for men.
Having support from family and/or friends was associated with significantly lower TCPZE (p = 0.0025). A 95% confidence interval is that it predicts a lower sqrt (total chlorpromazine equivalent dose) for this group by between (0.5, 2.5).
Having more antipsychotics prescribed at discharge had a strong impact on the TCPZE (p < 0.0001). A 95% confidence interval is that each additional antipsychotic prescribed at discharge (up to three) predicts an increase in sqrt (chlorpromazine equivalent dose) by between (4.21, 6.57).
Longer stays (p < 0.0001) and a higher number of past antipsychotic trials (p < 0.0001) were both associated with higher TCPZE. There was also a statistically significant interaction between the effect of these variables on TCPZE (p = 0.0004). The interaction was an interference interaction, meaning that the combined impact of longer stays and more past antipsychotic trials was less than the sum of the parts. However, the predicted combined impact on TCPZE remained positive for the entire range of the data.
Diagnosis also had a statistically significant impact on TCPZE (Table 3). Patients with a schizophrenia (SCZ) spectrum disorder had a significantly higher TCPZE than patients with bipolar depression (p = 0.0173), major depressive disorder (MDD) (p < 0.0001), or “other” disorders (p < 0.0001), such as traumatic brain injury, intellectual dysfunction, among others. Patients with bipolar mania had significantly higher TCPZE than patients with MDD or “other” diagnoses (p < 0.0001 for both). Finally, patients with bipolar depression (p = 0.0007) and “other” diagnoses (p = 0.019) both had significantly higher TCPZE than patients with MDD.
No other variable had a statistically significant impact on TCPZE, once the statistically significant variables were taken into consideration. Patients with voluntary admission had significantly lower TCPZE (p < 0.0001). However, the effect was only marginally significant (p = 0.108) once the other statistically significant variables in the model were taken into consideration.
Appropriate medication
The appropriateness of medications prescribed to patients in this study was significantly influenced by the type of diagnosis, the number of antipsychotics prescribed at discharge, the type of insurance, and the hospital where the patients received care (Table 4).
Some 95% CI for odds ratios on whether patients receive appropriate medication or not.
CI, confidence interval; DMII, Diabetes Mellitus type 2.
Patients with type II diabetes mellitus (DMII) were less likely to be prescribed appropriate medications (p < 0.0001). A 95% confidence interval for the difference in rates is that the percentage of patients receiving appropriate medications is between 9.3% and 33.8% lower for patients with diabetes. A 95% confidence interval for the odds ratio of the impact of diabetes on receiving appropriate medication is (0.20, 0.55).
Patients with more antipsychotics prescribed at discharge also had lower odds of receiving appropriate medication (p = 0.001). A 95% confidence interval of the odds ratio for each additional antipsychotic prescribed at discharge (up to three) is between 0.06 and 0.16.
Insurance also had a statistically significant impact on whether patients received appropriate medication (p = 0.005). Specifically, Medicare patients had lower odds of receiving appropriate medication than patients without insurance (p = 0.0236) and patients with Medicaid (p = 0.0067).
Finally, the hospital had a statistically significant impact on patients receiving appropriate medication (p = 0.0005). There were also statistically significant interactions between the effect of hospital, the effect of diabetes (p = 0.0075), and the number of antipsychotics prescribed at discharge (p < 0.0001).
The hospitals had similar rates of appropriate medication for patients with one antipsychotic prescribed at discharge. However, UMCNO had significantly lower rates of use of appropriate medications for patients with more than one antipsychotic prescribed at discharge (p < 0.0001). In contrast, UMCNO had higher rates of use of appropriate medications for patients with diabetes; however, this result was only marginally significant (p = 0.0536) due to the relatively small number of patients with diabetes (n = 79).
No other variable had a statistically significant impact on whether patients received appropriate medications after controlling for all statistically significant variables. Older patients were significantly less likely to receive appropriate medications (p = 0.006). However, the impact was not statistically significant once the statistically significant variables in our model were considered (p = 0.386). There was also a marginally significant interaction between the hospital and insomnia variables (p = 0.0672), with insomnia leading to lower rates of appropriate medications at UMCNO, but not at OMC.
Discussion
This study was an expansion of a previous study 12 to a second hospital to increase the sample population and describe the antipsychotic prescribing patterns outside of one system. The setting was the same, as well as the acuity of illness among the populations; the only difference was that care was delivered at a different hospital system, with some demographic differences between the patient populations at the hospitals. Most of the changes in results from our previous study were due to improvements in the study design made for this study, including the consideration of several new variables, which proved to be relevant. These included the number of antipsychotics prescribed at discharge and whether the patients had DMII. The social support variable was also subdivided to indicate whether the patients had support from family or friends. These new variables had statistically significant impacts on the TCPZE and/or whether patients received a medication regimen that was in concordance with FDA labeling. In addition, the inclusion of these new variables resulted in a more accurate assessment of the impact of the other variables considered. For example, legal hospitalization status (voluntary or involuntary) had a statistically significant impact on TCPZE, and age had a statistically significant impact on whether patients received appropriate medications or not, but both ceased to be statistically significant once the other variables were considered. Of note, one interesting finding was the difference in the two hospitals in the number of patients who were positive for THC (tetrahydrocannabinol) on their urine drug screen (33 at OMC vs 118 at UMCNO). At UMCNO, there was a significantly higher proportion of patients reporting overall illicit substance use (p < 0.0001). UMCNO also had a higher proportion of involuntary admissions (p < 0.0001) and schizophrenia (p < 0.0001). There was no significant relationship between involuntary admission and illicit substance use, or between SCZ and illicit substance use. While there was no link between legal hospitalization status and illicit substance use, the possibility remains that there could be a relationship between the severity of symptoms and illicit substance use, as this was not captured in the data. Previous literature has explored the relationship between marijuana use and schizophrenia, finding that marijuana use can worsen symptoms of schizophrenia and make the symptoms more difficult to treat with common doses of antipsychotics.16–19 Future studies could target antipsychotic prescribing patterns in individuals who use THC versus those who do not. Factors such as race, socioeconomic status, and environment/culture could also influence this relationship due to the effect on the use patterns of marijuana.
Another way we improved on our previous study was by nearly doubling our sample size, which allowed us to find more significant results. For example, gender had only a marginally significant impact on TCPZE in the first study; however, it was found to have a statistically significant impact on TCPZE with the larger sample size in this study.
The addition of a second institution led to other interesting findings, as well. There was no evidence that the hospital where the patients received care had an impact on TCPZE, either directly or by interacting with another variable. However, the hospital did have a statistically significant impact and statistically significant interactions for whether patients received a medication regimen that followed FDA guidance. This suggests that adding a second hospital increased the applicability of our results for whether patients receive appropriate medication. It also suggests that a future study might get even more applicable results for whether patients receive appropriate medication by considering several randomly selected hospitals.
Limitations
Some limitations should be stated, as they may have an impact on the validity of the results. First, Data was not always clearly documented in patient charts, and the data was collected in slightly different ways between the institutions, which may have affected some of the results. Second, while chlorpromazine equivalents are the recognized means of standardizing antipsychotic dosing, there continues to be debate in the literature about how to optimally calculate dose equivalence between antipsychotics.14,15,20–22 Because blood plasma levels do not correlate with the effect in patients, there is no perfect model for considering dose equivalency in antipsychotics. The classical dose method was used in this study as it is not reliant on limited fixed-dose data and utilizes the dosing ranges selected in the original trials found in the package inserts for the individual medications. 15 Third, while treatment is recommended to follow FDA-indication, this does not necessarily correlate with real-world practice, as it can be difficult to ascertain a proper diagnosis during the acute episode of illness, and antipsychotics are often used off-label in the acute treatment of other disease states. There are also times that either higher-than-normal doses, polypharmacy, or medications of other classes are required to address refractory symptoms.23–25 This could have led to mislabeling the practice as “inappropriate use of medication.” Of note, there were varying levels of support by a pharmacist in the care for the individuals in this study. This could have an impact on outcomes related to the inappropriateness of medications. 26 It is also important to note that maximum dose recommendations in FDA labeling for each agent can be higher for certain diagnoses. As an example, the maximum dose for aripiprazole is 30 mg daily versus 15 mg in major depressive disorder. 27 Lastly, as this study was descriptive and focused more on hypothesis-generating than testing, no prior power analysis was conducted to determine the appropriate sample size to detect meaningful effects. However, the study had a fairly large sample size of 785 participants and generated a large number of statistically significant results, demonstrating that it had sufficient power. A post hoc analysis of achieved power of a two-tailed t-test comparing means for two independent groups, given effect size d = 0.5 and alpha level of 0.05, resulted in a power of 0.999.
Conclusion
This study identified patterns of antipsychotic prescribing in patients across two institutions. Results showed that male patients and those with no social support, more previous antipsychotic trials, longer hospital stays, and some diagnoses (see Table 3) received statistically significantly higher antipsychotic doses at discharge. In addition, patients with comorbid diabetes, more antipsychotics at discharge, and Medicare insurance were less likely to receive antipsychotic medication following FDA recommendations and current guidelines.
Revealing characteristics that lead to populations having higher dosing exposures to antipsychotic therapy may assist with informing practice on antipsychotic prescribing patterns in various populations, as well as lead to the ideal pharmacotherapy, along with guidelines or guidance. While it is possible the results will apply only to the specific hospitals involved or the greater New Orleans area, they could be generalized to patients at all similar hospitals in the United States. As further investigation is often needed to confirm their wider applicability, prospective studies across a larger spectrum of institutions are recommended to elucidate the relationship between patient demographics and antipsychotic prescribing patterns during periods of acute illness. Future studies can use this data to explore how to make improvements in antipsychotic prescribing patterns to improve overall patient outcomes on antipsychotic therapy.
Supplemental Material
sj-docx-1-tpp-10.1177_20451253251394296 – Supplemental material for A description of antipsychotic prescribing patterns on the inpatient behavioral health setting: a multicenter cross-sectional analysis
Supplemental material, sj-docx-1-tpp-10.1177_20451253251394296 for A description of antipsychotic prescribing patterns on the inpatient behavioral health setting: a multicenter cross-sectional analysis by Thomas Maestri, David Anderson and Margarita Echeverri in Therapeutic Advances in Psychopharmacology
Footnotes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
