Pharmacoeconomics of long-acting atypical antipsychotics for acutely relapsed chronic schizophrenia in Finland

Abstract

Background:

Atypical long-acting injectable (LAI) antipsychotics are increasingly available for treating chronic schizophrenia in patients chronically non-adherent to prescribed regimens. Few economic studies have compared these products.

Purpose:

To determine the cost-effectiveness of aripiprazole (ARI-LAI), paliperidone (PP-LAI), olanzapine (OLZ-LAI), and risperidone (RIS-LAI) in patients with chronic schizophrenia in Finland.

Methods:

A 1-year decision tree model was adapted with guidance from an expert panel. Patients started hospitalized in relapse; those who responded continued treatment, others were switched to secondary drugs, then clozapine in the event of 2nd line failure. Rates of adherence, stable disease, relapse, and hospitalization were taken from pivotal trials, and utilities from published research. Included were direct costs paid by the Finnish Ministry of Health, in 2015 euros. Outcomes included quality-adjusted life-years (QALYs), hospitalization rates, and rates of relapse not requiring hospitalization. Model robustness was assessed using a series of 1-way and multivariate sensitivity analyses.

Results:

Expected costs were lowest for PP-LAI at 41,148€, followed by 41,543€ for ARI-LAI, 42,067€ for RIS-LAI and 45,406€ for OLZ-LAI. Respective QALYs were 0.683, 0.671, 0.666, and 0.672. Re-hospitalization rates and non-admitted relapses were 23.6% and 3.9% for PP-LAI, 28.5% and 4.1% for ARI-LAI, 28.8% and 5.0% for RIS-LAI, 28.3% and 5.2% for OLZ-LAI. PP-LAI treatment was associated with the most days with stable disease (132.0), followed by OLZ-LAI (125.5), ARI-LAI (122.6), and RIS-LAI (114.4). Sensitive inputs between PP-LAI and ARI-LAI included rates of adherence, dropouts, and relapses plus drug prices; dropout and relapse rates for RIS-LAI; OLZ-LAI results were insensitive. In probability sensitivity analyses, PP-LAI dominated ARI-LAI in 75.8% of the 10,000 iterations, RIS-LAI in 83.1% and OLZ-LAI in 95.7%.

Conclusions:

PP-LAI dominated the other atypicals. It appears to be the preferred option for treating chronic relapsing schizophrenia.

Keywords::

• Long-acting injectable
• Antipsychotic
• Atypical
• Chronic schizophrenia
• Paliperidone
• Aripiprazole
• Risperidone
• Olanzapine
• Cost-effective

Introduction

Chronic schizophrenia remains a challenge for clinicians and healthcare providers1. Treating schizophrenia is challenging and a costly disease to treat, impacting patients, families, and friends, as well as the healthcare system1–4. The largest proportion of the costs of care are due to relapses of symptoms, which result in hospital visits5–7. Non-adherence to medications has been identified as a major contributor to these relapses8^,9. Long-acting injectable (LAI) antipsychotics are becoming increasingly more available in clinical practice, as they help to decrease both intentional and unintentional non-adherence. At the same time, they have had a beneficial effect on clinical and economic outcomes10. Bera et al.11 also found that, along with other benefits, costs were lower with LAIs (44% of which were RIS-LAI and 56% typical depots) than with oral antipsychotics.

Presently, there are four atypical LAI antipsychotics available in Finland for use in patients requiring such products, including aripiprazole (ARI-LAI)12, olanzapine (OLZ-LAI)13, paliperidone (PP-LAI)14, and risperidone (RIS-LAI)15. The purpose of this research was to determine the cost-effectiveness of these atypical LAIs for treating chronic schizophrenia in Finland.

Methods

A 1-year decision tree previously developed for Finland was modified to reflect the current LAI landscape16. Figure 1 depicts the clinical pathway. One major difference between the original and modified model was that the original model started with patients in remission, while the modified model starts with patients in relapse. Another modification was the addition of ARI-LAI, which has recently been approved12. Treatment patterns and adaptations were determined in conjunction with a panel of clinical experts. The process was undertaken according to the Finnish Ministry of Social Affairs and Health guidelines for preparing a health economic evaluation17.

Figure 1. Decision tree used in this analysis.

Base case scenario

Patients would be treated for the acute phase of the relapse in hospital as per recommendations and standard procedure in Finland. They would either respond to LAI antipsychotics after 13 weeks and receive maintenance therapy, or would continue for another 13 weeks with acute therapy. In cases of drug failure or intolerance, patients would be switched to alternative treatments. Those failing ARI-LAI, PP-LAI, or RIS-LAI would be switched to OLZ-LAI, while those on OLZ-LAI would be switched to RIS-LAI. Patients who failed second line treatment were assigned to clozapine, as per NICE guidelines18. Once stabilized, patients would proceed with treatment, with adherence rates taken into consideration. Further relapses would result in re-starting the drug or switching to another antipsychotic.

Clinical rates were obtained from published randomized controlled trials or from large databases with data from comparable patients. To avoid bias, we used all data currently available on a special group of patients; i.e., persons with chronic, relapsing schizophrenia, starting in relapse. These were patients who had difficulty with adherence to prescribed regimens of oral drugs and, therefore, required LAIs. Table 1 summarizes the input rates and the sources of those values19–63. Clinical states of interest included stable disease and two forms of relapse, which were those requiring hospitalization and those that could be treated in the emergency room (ER) as outpatients. Rates of response to acute treatment were determined from RCTs, all of which defined response as a ≥30% decrease in PANSS score. As placebo rates varied among studies, we first combined placebo rates across trials using a weighted average, then adjusted rates using the approach of Bucher et al.24.

Table 1. Model inputs and their sources.

Patient state	Drug	Item	Value	Source
Hospitalized	All	Utility	0.490	Average of Lenert et al.19 and Revicki et al.20
	ARI-LAI	Regimen	2 weeks of oral aripiprazole 12.8 mg + 400 mg ARI-LAI, then every 4 weeks	Abilify Maintena Product Summary21, Kane et al.22
		Response @ 13 weeks	0.429	Kane et al.23, placebo adjusted using Bucher et al.’s24 method
		Response @ 26 weeks	0.167	Calculation based on 26-week rate from Kane et al.23 and 13-week response rate
	OLZ-LAI	Regimen	300 mg every 2 weeks ×3, then 473 mg every 4 weeks	Zypadhera Product Summary25, Lauriello et al.26
		Response @ 13 weeks	0.427	Lauriello et al.26, placebo adjusted using Bucher et al.’s24 method
	PP-LAI	Regimen	150 mg week 1, 100 mg week 2, then 82.8 mg monthly maintenance	Xeplion Summary of Product Characteristics27, Hough et al.28
		Response @ 13 weeks	0.360	Average of Gopal et al.29, Kramer et al.30, Nasrallah et al.31, Pandina et al.32, Takahashi et al.33, placebo adjusted using Bucher et al.’s24 method
		Response @ 26 weeks	0.455	Calculation based 13-week response rate and assume 26-week rate equals OLZ-LAI
	RIS-LAI	Regimen	50 mg every 2 weeks	Maximum dose as per Risperdal Consta EMA Product Summary34; Pandina et al.35; Meltzer et al.36
		Response @ 13 weeks	0.337	Pro-rated from PP-LAI using the average from head-to-head trials by Fu et al.37, Li et al.38, Pandina et al.35
		Response @ 26 weeks	0.412	Calculation based on pro-rated PP-LAI rate and 13-week response rate
Relapsed, treated in ER	All	Utility	0.659	Briggs et al.39, Cummins et al.40, Lenert et al.19
	ARI-LAI	Regimen	2 weeks of oral aripiprazole 12.8 mg + 400 mg ARI-LAI, then every 4 weeks	Abilify Maintena Product Summary21, Kane et al.22
	OLZ-LAI	Regimen	473 mg every 4 weeks	Lauriello et al.26
	PP-LAI	Regimen	84.9 mg monthly	Average of Gopal et al.29, Hough et al.41, Nasrallah et al.31, Pandina et al.32^,35
	RIS-LAI	Regimen	50 mg every 2 weeks	Maintenance dose pro-rated as per PP-LAI
Stable disease	All	Utility	0.890	Average of Briggs et al.34, Cummins et al.35, Lenert et al.19, Oh et al.42, Revicki et al.20
	ARI-LAI	Adherence rate	0.730	Average of other three drugs
		Rate of stable disease	0.887	Kane et al.43; placebo adjusted as per Bucher et al.24
		Hospitalization rate	0.079	Relapse rate from Kane et al.43 adjusted using Ascher-Svanum et al.44 ratio of Hospitalizations:ER visits
		ER visit rate	0.034	Relapse rate from Kane48 adjusted using Ascher Svanum44 ratio of ER visits:Hospitalizations
		Dropout rate (LoE + AEs)	0.204	Average from all placebo-controlled trials of acute treatment (Kane et al.22, Potkin et al.45, McEvoy et al.46, Kane et al.47); placebo adjusted as per Bucher et al.24
		Regimen	396.3 mg every 4 weeks	Kane et al.43 (96.3% received 400 mg, 3.7% received 300 mg); Fleischhacker et al.48
	OLZ-LAI	Adherence rate	0.731	Detke et al.49
		Rate of stable disease	0.826	Kane et al.50
		Hospitalization rate	0.122	Kane et al.50; adjusted for adherence51
		ER visit rate	0.052	Calculation: 1 – (other rates)
		Dropout rate (LoE + AEs)	0.212	Lauriello et al.26 rate, placebo adjusted
		Regimen	432 mg every 4 weeks Adjunct: oral 10.8 mg/day for 31 days (21% of patients)	Kane et al.50; Ascher-Svanum et al.52
	PP-LAI	Adherence rate	0.749	Asseburg et al.53 rate for RIS-LAI adjusted via Mehnert and Diels54
		Rate of stable disease	0.836	Calculation: 1 - Relapse rate from Hough et al.28, adjusted for placebo using Bucher et al.’s24 method
		Hospitalization rate	0.115	Relapse rate from Hough et al.28 adjusted using Ascher Svanum et al.44 ratio of Hospitalizations:ER visits
		ER visit rate	0.049	Relapse rate from Hough et al.28 adjusted using Ascher Svanum et al.44 ratio of ER visits: Hospitalizations
		Dropout rate (LoE + AEs)	0.202	Rates from placebo controlled acute treatment: Gopal et al.29, Pandina et al.35, placebo adjusted24
		Regimen	69.3 mg monthly	Gopal et al.55, Fleischhacker et al.56
	RIS-LAI	Adherence rate	0.709	Asseburg et al.53
		Rate of stable disease	0.763	Calculation: 1 – (Hospitalization rate + ER visit rate) adjusted for adherence via Weiden et al.51
		Hospitalization rate	0.166	Olivares et al.57 adjusted for adherence via Weiden et al.51
		ER visit rate	0.071	Calculation: 1 – (Hospitalization + Stable rates)
		Dropout rate (LoE + AEs)	0.117	OLZ-LAI rate x ratio from Ascher-Svanum et al.58
		Regimen	40.3 mg every 2 weeks + adjunct: 3.2 mg/day for 43 days (22% of patients)	Average of Fleischhacker et al.56, Kissling et al.59, Lee et al.60, Lindenmayer et al.61, Möller et al.62, and Olivares et al.63
Non-adherent	All drugs	Stable disease	0.234	Average of Hough et al.28 and Kane et al.43 placebo non-relapse rates at 1 year
		Relapses (hospitalized)	0.556	Adjust relapse rate via Ascher-Svanum et al.44 Hospitalization:ER ratio for non-adherent patients
		Relapses treated in ER	0.209	Adjust relapse rate via Ascher-Svanum et al.44 ER: Hospitalization ratio for non-adherent patients

AE, adverse event; ARI, aripiprazole; ER, emergency room; LAI, long-acting injectable; LoE, lack of efficacy; OLZ, olanzapine; PP, paliperidone; RIS, risperidone.

Only direct costs paid by the Finnish Ministry of Health were considered, including drugs, health professionals, and institutional care. Table 2 lists the unit costs and sources64–67. We did not include the costs associated with treating adverse events because they tend to be similar across drugs, as found in the CATIE trial68. Also, the impact of discontinuing and switching drugs due to adverse events was incorporated into the model, thereby capturing most of the associated costs. All costs were calculated in 2014 euros; those from other years were adjusted to 2014 using the Consumer Price Index for Finland69. Because the model was 1 year in length, discounting was not applied to either costs or outcomes.

Table 2. Cost inputs used in the model (2015 euros).

Cost centre	Resource	Cost/unit	Data source
Drugs*	PP-LAI	4.25	SLD price list64
	RIS-LAI	4.63	SLD price list64
	OLZ-LAI	0.887	SLD price list64
	ARI-LAI	0.824	SLD price list64
	Risperidone	0.2674	SLD price list64
	Olanzapine	0.024	SLD price list64
	Aripiprazole	0.2291	SLD price list64
	Clozapine	0.0064	SLD price list64
	Lorazepam	0.0646	SLD price list64
	Clozapine level	12.64	Kliinisen Kemian Laboratoriotutkimusten65
	Blood count	7.00	Kliinisen Kemian Laboratoriotutkimusten65
Health professionals	Psychiatrist-outpatient visit	171.25	Kapiainen et al.66
	Primary care physician	116.28	Kapiainen et al.66
	Psychiatric nurse	50.74	Kapiainen et al.66
Hospitals/facilities	Emergency room visit	256.87	Kapiainen et al.66
	Acute psychiatric ward	731.50	Kapiainen et al.66
	Psychiatric rehab ward	431.29	Kapiainen et al.66
	Long-term ward	201.90	Kapiainen et al.66
	General medical ward	225.16	Kapiainen et al.66
	Day hospital	206	HUS Palveluhinnasto 201467
	Day care centre visit	162	HUS Palveluhinnasto 201467

*All drugs are costed using the average price per milligram across all strengths.

ARI, aripiprazole; LAI, long acting injectable; OLZ, olanzapine; PP, paliperidone palmitate; RIS, risperidone.

Utilities were those used in our previous model16, which were derived from the literature. We assigned a utility of 0.890 to stable disease, 0.659 to an exacerbation that could be managed at the hospital emergency room or on an outpatient basis, and 0.490 to hospitalization. Table 1 indicates the sources of these values.

We conducted both cost-utility and cost-effectiveness analyses. In the primary analysis, outcomes assessed included the expected cost per patient treated as well as quality-adjusted life-years (QALYs) associated with each treatment. The economic outcomes was the incremental cost per QALY gained. For the cost-effectiveness analyses, outcomes were hospitalization rate, rate of relapses that were treated in the emergency room (ER), and days with stable disease.

Sensitivity of the results was examined using 1-way (threshold and break-even) analyses on all important inputs. Also, we conducted pairwise multivariate probability sensitivity analyses, varying the input values of all parameters using standard distributions (i.e., gamma for costs and beta for clinical rates) and over plausible ranges for 10,000 simulations with each drug.

Results

Table 3 presents the results from the base case analysis. PP-2LAI had the lowest overall expected cost per patent treated and the most QALYs. At the same time, it was associated with the highest number of days with stable disease and the lowest number of days in relapse, ER visits, and hospitalizations. Consequently, it dominated all other LAIs in the base case.

Table 3. Clinical and economic outcomes.

Long acting antipsychotic	Cost/patient*	QALY	Economic outcome	Days with stable disease	ER visits	Hospital re-admissions
Paliperidone	€41,148	0.683	Dominant	132.0	0.039	0.236
Aripiprazole	€41,543	0.671	Dominated	122.6	0.041	0.285
Risperidone	€42,067	0.666	Dominated	114.4	0.050	0.288
Olanzapine	€45,406	0.672	Dominated	125.5	0.052	0.283

ER, emergency room; QALY, quality adjusted life-year.

*All costs are in 2015 euros.

The cost driver was hospitalization, which accounted for 82–85% of the total cost. Drugs accounted for 10–11% and medical care 4–7%. In a sensitivity analysis, hospitalization was reduced to acute care only (14 days), which resulted in drastically reduced costs by 32–36%. Nonetheless, the rank order of costs remained the same, with PP-LAI dominating all other drugs.

One-way sensitivity analyses indicated that results between PP-LAI and ARI-LAI were sensitive to changes in rates of adherence, dropouts, relapses, and possibly drug prices. Between PP-LAI and RIS-LAI, results were sensitive to changes in relapse rates; OLZ-LAI results were insensitive to all changes. Break-even points for dominance are presented in the Appendix. For PP-LAI to become not cost-effective, differences in inputs would have to be much greater. For example, PP-LAI would remain cost-effective over ARI-LAI, with monthly maintenance doses of up to 121–132 mg to match the NICE minimum (£20,000 ≈ 24,800€) and maximum thresholds (£30,000 ≈ 37,200€). In probability sensitivity analyses, PP-LAI dominated ARI-LAI in 75.8% of the 10,000 iterations, RIS-LAI in 83.1%, and OLZ-LAI in 95.7%. As well, it was cost-effective in 77.2%, 86.1%, and 96.3% of the 10,000 pairwise iterations when compared with ARI-LAI, RIS-LAI, and OLZ-LAI, respectively, when using the NICE threshold of £20,000/QALY gained (∼24,800€ using the average exchange rate for 2014)70.

Discussion

This model is somewhat different from the original Finnish model, in that this model was initiated with patients in relapse. Several published economic analyses have started with patients in relapse, using a variety of analytical models including decision trees71^,72, discrete event simulations73–76, and clinical trial models77. As a result, outcomes would take into consideration the cost of hospitalization for all patients and lower QALY score for that initial time period. We hypothesized that the overall cost per patient would be considerably higher when compared to models that started with patients having stable disease, which was observed.

The results reported here generally validate what has been previously reported. A search of Medline identified 12 economic analyses involving comparisons between different atypical LAIs other than clozapine16^,72^,78–87. In nine of those analyses, PP-LAI was compared with RIS-LAI (and other LAIs)16^,78^,79^,81–86, eight compared PP-LAI with OLZ-LAI (and others)16^,78^,80–82^,84–86, one examined OLZ-LAI against RIS-LAI72, and one studied ARI-LAI vs PP-LAI87. In eight out of nine analyses, PP-LAI had a lower cost and more QALYs than RIS-LAI. The same situation occurred in seven out of eight comparisons with OLZ-LAI, with PP-LAI dominating. In both cases, the discrepant study did not use data obtained from studies of PP-LAI, but assumed that PP-LAI and RIS-LAI were equal in every way82. Thus, the only difference was in the drug acquisition cost. In Norway, Carroll et al.72 found that OLZ-LAI dominated RIS-LAI, having a lower cost and higher QALYs. Finally, Citrome et al.87 recently published a pharmacoeconomic analysis comparing ARI-LAI with PP-LAI in the US. In their analysis, ARI-LAI was cost-effective, despite having a higher cost, due to a lower relapse rate for ARI-LAI. The present study found the same overall relapse rate for PP-LAI (27.5%), but a somewhat higher rate for ARI-LAI (32.6%). This latter rate is very close to the 33.8% relapse rate reported by Pigott et al.88 after 26 weeks on oral aripiprazole. Since no real world data have been located for ARI-LAI, we must wait to see if these results can be validated.

Many economic analyses have used clinical rates derived from independent studies, which can introduce bias due to differences in patients, protocols, or environments. To produce comparable clinical rates, we adjusted them through placebo whenever possible using the method of Bucher et al.24.

On the other hand, we did not include the costs of adverse events because they comprise only a very small proportion of the costs and overall tend to be quite similar across drugs. As examples from other European countries, the Lindström study reported that those costs constituted 1.2–1.7% of the total costs for four different oral atypical antipsychotics89. Another analysis from Sweden from Mehnert et al.85 found comparable rates of 0.2–1.4% for three atypical LAIs, including PP-LAI, RIS-LAI, and OLZ-LAI. In a study from Slovenia90, adverse events associated with atypicals (including one depot, RIS-LAI) comprised 0.3–0.7% of the overall cost and, in Greece91, they constituted 0.3–0.4% of the total costs of six oral atypicals. Thus, adverse events exert a negligible impact on the final results and do not differ substantially between drugs.

We also found that QALYs were very similar between the drugs examined after 1 year. The difference between the highest (PP-LAI) and the lowest (RIS-LAI) was small, at 0.017 QALY (2.5%). However, even less variation was reported in a recent publication by Park and Kuntz92, who modeled the CATIE trial over 10 years. They compared 12 different strategies using four oral atypicals (olanzapine, quetiapine, risperidone, and ziprasidone), varying the second line treatment in the different scenarios. Over the 10-year time horizon, those authors recorded a difference of only 1% in QALYs between the highest (7.351 QALYs) and the lowest (7.275 QALYs)92.

Limitations

As with all models, this one represents the average results for the average patient with chronic schizophrenia who is in relapse. They would not apply to other forms of the disease (e.g., first episode schizophrenia), other states (e.g., in remission), different forms of psychosis, or to persons with comorbidities. Also, results may not apply to other healthcare systems or jurisdictions due to differences in treatment patterns, availability of resources, or methods of financing. Results might vary in places where traditional depots and other approaches are still commonly used.

Also, the model had a 1-year time horizon; results could vary with a longer duration. Because of the duration, a large proportion of time was spent in hospital under highly supervised conditions; hence, there was very little opportunity for differences to emerge between drugs with respect to QALYs.

There was often a lack of specific input data for the analysis, especially for ARI-LAI, as it is a fairly new drug with very few published studies providing information. Also, little real world information has been published, as there has been limited clinical experience. In the absence of directly applicable data, assumptions had to be made. In all such cases, we took a conservative approach to avoid introducing bias. For example, we assumed that the adherence rate for ARI-LAI and OLZ-LAI would be comparable to that of PP-LAI as all can be dosed on a monthly basis. However, in three 52-week trials starting with patients in relapse, 57% of patients taking oral aripiprazole discontinued before the trials ended, suggesting that adherence rates in these patients may not be so high93–95. More accurate estimates with the LAI formulation could alter results.

Conclusions

In this analysis, PP-LAI had the lowest overall cost per patient and was associated with the most QALYs and lowest rates of adverse outcomes (i.e., days in relapse, hospitalizations, and ER visits). In the base case analysis, it dominated the other alternatives; in sensitivity analyses, results were fairly robust across a wide range of possible values. It appears to be the preferred option for treating chronic relapsing schizophrenia in patients who require LAIs.

Transparency

Declaration of funding

TRE and BGB received funding for this study from Janssen-Cilag. HP, PG and KvI are employees of Janssen-Cilag.

Appendix

Results of the one-way (break-even) sensitivity analyses.

Parameter	Drug rate changed	Base case	Comparison drug	Value where PP-LAI cost not lowest	% difference	Value where PP-LAI QALYs not highest	% difference
Adherence	PP-LAI	74.9%	ARI-LAI	66.6%	−8.3%	44.7%	−30.2%
			RIS-LAI	55.5%	−11%	31.8%	−43.1%
			OLZ-LAI	resistant	n/a	46.3%	−28.6%
	ARI-LAI	73.0%	PP-LAI	78.0%	+5.0%	99.5%	+26.5%
	RIS-LAI	70.9%	PP-LAI	88.5%	+17.6%	resistant	n/a
	OLZ-LAI	73.1%	PP-LAI	resistant	n/a	98.3%	+25.2%
Dropout rate (LoE + AEs)	PP-LAI	20.2%	ARI-LAI	26.2%	+6.0%	42.6%	+22.4%
			RIS-LAI	34.2%	+14.0%	52.2%	+32.0%
			OLZ-LAI	84.8%	+64.6	41.4%	+21.2%
	ARI-LAI	20.4%	PP-LAI	14.3%	−6.1%	resistant	n/a
	RIS-LAI	30.4%	PP-LAI	14.3%	−16.1%	resistant	n/a
	OLZ-LAI	21.2%	PP-LAI	resistant	n/a	resistant	n/a
Stable disease/relapse	PP-LAI	83.6%	ARI-LAI	80.9%	−2.7%	63.0%	−23.6%
			RIS-LAI	77.4%	−6.2%	54.2%	−29.4%
			OLZ-LAI	55.1%	−28.5%	67.1%	−19.5%
	ARI-LAI	88.7%	PP-LAI	93.5%	+4.8%	resistant	n/a
	RIS-LAI	76.3%	PP-LAI	92.1%	+15.8%	resistant	n/a
	OLZ-LAI	82.6%	PP-LAI	resistant	n/a	resistant	n/a
Drug price (per mg)	PP-LAI	4.251€	ARI-LAI	4.73€	+11.3%	n/a*	n/a
			RIS-LAI	5.35€	+25.9%	n/a*	n/a
			OLZ-LAI	9.33€	+113%	n/a*	n/a
	ARI-LAI	0.824€	PP-LAI	0.705€	−14.4%	n/a*	n/a
	RIS-LAI	4.632€	PP-LAI	2.60€	−43.9%	n/a*	n/a
	OLZ-LAI	0.887€	PP-LAI	resistant	n/a	n/a*	n/a
Institutional costs	All costs		ARI-LAI		−56%†	n/a*	n/a
			RIS-LAI	resistant	n/a
			OLZ-LAI	resistant	n/a
Adherence (2nd line)	all	40.0%	ARI-LAI	all resistant	n/a	all resistant	n/a
			RIS-LAI
			OLZ-LAI
Clozapine success	clozapine	40.2%	ARI-LAI	53.2%	+13.0%	all resistant	n/a
			RIS-LAI	resistant	n/a
			OLZ-LAI	resistant	n/a
Drug maintenance dose	PP-LAI	69.3 mg	ARI-LAI	99 mg	+42.4%	n/a	n/a
			RIS-LAI	138 mg	+99.1%	n/a	n/a
			OLZ-LAI	386 mg	+457%	n/a	n/a
	ARI-LAI	396.3 mg	PP-LAI	228 mg	−42.5%	n/a	n/a
	RIS-LAI	40.3 mg	PP-LAI	2.6 mg	−93.5%	n/a	n/a
	OLZ-LAI	432 mg	PP-LAI	resistant	n/a	n/a	n/a

*Costs have no impact on outcomes.

†All costs resistant to increases; all hospital and related costs would have to be reduced by that percentage for comparison drug costs to be lower than PP-LAI.

AEs, adverse events; ARI, aripiprazole; LAI, long acting injectable; LoE, lack of efficacy; n/a, not applicable; OLZ, olanzapine pamoate; PP, paliperidone palmitate; QALY, quality adjusted life year; RIS, risperidone.