Effects of a Schizophrenia Pay-for-Performance Program on Health Care Resource Utilization

ABSTRACT

Objectives: This study aims to assess the impacts of a pay-for-performance (P4P) program on health care resource utilization for patients with schizophrenia in Taiwan.

Study Design: This study employed a difference-in-differences (DID) approach to compare outcome measures between the exposed (P4P program enrollees) and unexposed cohorts across 2 time points (pre and post intervention). Patients were selected from Taiwan’s National Health Insurance Research Database between 2009 and 2017, then divided into the exposed and the unexposed cohorts using the double propensity score adjustment technique with a ratio of 1 patient in the exposed cohort to 4 patients in the unexposed cohort, based on the matching criteria of age, sex, and the index date.

Methods: Generalized estimating equation with a Poisson distribution and a negative binomial distribution, along with the DID technique, was utilized to estimate the effects of the P4P program on health care resource utilization among enrollees.

Results: Patients in the exposed cohort were less likely to have outpatient visits than their counterparts, indicating a reduced demand for outpatient health care (incidence rate ratio [IRR], 0.8357; P < .0001). However, the P4P program was not effective in reducing emergency department visits (IRR, 1.0234; P = .6687), hospital admissions (IRR, 0.9146; P = .3728), or lengths of hospital stay (IRR, 1.1253; P = .2128).

Conclusions: The study demonstrated that the P4P program had mixed effects on health care resource utilization for patients with schizophrenia, particularly regarding the program’s effectiveness in reducing reliance on inpatient and emergency services, which was not significant.

Am J Manag Care. 2026;32(9):In Press

doi:10.37765/ajmc.2026.90011

Takeaway Points

This study aims to assess the impact of a pay-for-performance (P4P) program for schizophrenia on health care resource utilization among enrollees in Taiwan.

• Results indicated that, compared with patients with schizophrenia not enrolled in the P4P program, patients participating in the P4P program were less likely to have outpatient visits.
• However, the findings revealed that the P4P program was not effective in reducing emergency department visits, hospital admissions, or lengths of stay.
• These insights may guide future efforts to improve P4P schemes, particularly focusing on implementing more needs-driven, rather than supply-guided, health care policies and intervention strategies.

Mental illness, along with mortality and morbidity, negatively impacts direct health care costs and indirect costs, such as productivity loss.¹ In 2015, the United Nations identified mental health as a global development priority for the next 15 years, emphasizing its link to physical health, employment, and economic growth.² The Global Burden of Diseases, Injuries, and Risk Factors Study (GBD) revealed a significant increase in the prevalence of mental disorders worldwide, increasing from 654.8 million cases in 1990 to 970.1 million cases in 2019.³ The GBD study also found that mental disorders accounted for 125.3 million disability-adjusted life-years (DALYs) in 2019, with depressive disorders comprising 37.3%, anxiety disorders 22.9%, and schizophrenia 12.2%. Although depressive and anxiety disorders were the leading contributors in the GBD study, schizophrenia had the highest disability weight in an acute state.

Approximately 24 million individuals worldwide are affected by schizophrenia, with a prevalence rate of 1 in 222 people (0.45%) among adults.⁴ In Taiwan, the estimated schizophrenia prevalence was 0.45% in 2023.⁵ Schizophrenia placessubstantial health care, personal, societal, and economic burdens on patients, providers, payers, and society.^6,7 The number of psychiatric inpatients and emergency department visits is higher among individuals with schizophrenia than those with major depressive disorder or bipolar disorder.⁸ Additionally, patients with schizophrenia experience increased functional disability during the follow-up assessments.⁹

In 2001, the Ministry of Health and Welfare of Taiwan launched the pay-for-performance (P4P) voluntary programs for 5 main diseases: pulmonary tuberculosis, diabetes, asthma, breast cancer, and cervical cancer. Since then, the program list saw some additions, including the P4P program for schizophrenia in January 2010.¹⁰

According to the program’s inclusion criteria, Taiwan’s National Health Insurance Administration (NHIA) classifies patients with schizophrenia on the voluntary list into 3 subgroups based on their health care resource utilization in the previous year: (1) the regular subgroup: patients who were prescribed medications by psychiatrists at least 8 times and visited a specific facility for more than 60% of their total physician visits for mental disorder treatments, (2) the irregular subgroup: patients who did not receive regular treatments at a specific facility, and (3) the never-treated subgroup: patients who had no medical records of psychiatric treatments for 6 consecutive months in the previous year.¹⁰

The P4P program for schizophrenia comprises 4 key interventions: (1) creating and maintaining a care plan, (2) promoting and strengthening collaboration among interprofessional care teams, (3) assigning a case manager, and (4) offering financial incentives to encourage and reward health care facilities for improved performance.¹⁰ The P4P program prioritizes pay-for-performance (ie, rewarding positive outcomes and quality) over the fee-for-service model (ie, payment based on the quantity of services, which can lead to unnecessary health care services).

Although the effects of various P4P programs have been evaluated concerning health care resource utilization, quality of care, and health care inequalities, the results remain inconclusive.^11-15 Several systematic reviews and meta-syntheses have reported either no effect or only very modest effects for most P4P programs.^16,17 Moreover, compared with P4P programs for other diseases (such as diabetes),^18-25 the effects of the P4P program for schizophrenia on health care resource utilization have rarely been examined. It has been suggested that investigations should place greater emphasis on P4P design features and contextual factors.²¹ Given the complexity of schizophrenia management, this study aims to explore the effects of Taiwan’s schizophrenia P4P program using a rigorous quasi-experimental methodology.

METHODS

Data Sources and Study Cohorts

Data for this study were retrieved using unique national identification numbers from the Taiwan National Health Insurance Research Database (NHIRD), which covers 99% of Taiwan’s population, totaling more than 23 million people. The study population comprised patients holding a schizophrenia catastrophic illness card (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] code 295 or ICD-10-CM codes F20, F25) recorded in the NHIRD between 2009 and 2017. As noted, the P4P program for schizophrenia was initiated in January 2010, and the NHIA categorized patients into 3 subgroups based on their health care resource utilization in the previous year. Therefore, to determine the eligibility of potential patients in the sample, 2009 was used as a washout period. Newly enrolled patients from 2010 to 2017 were then included in the analysis. With a 1-year follow-up for every participant, we observed the study cohorts until the end of 2018.

The exposed cohort consisted of patients in the regular subgroup who received treatment at health care institutions (either inpatient or outpatient) participating in the P4P program for schizophrenia. Patients who were in the regular group in the previous year but received treatment at nonparticipating health care facilities at baseline were classified into the non-P4P (unexposed) cohort. Furthermore, in accordance with P4P program guidelines, patients were excluded if they held more than 1 type of catastrophic illness card or were admitted for chronic treatment to psychiatric wards in the previous year.¹⁰

Outcome Measures

The primary outcome variable for this investigation was health care resource utilization among the study population, aimed at assessing the effectiveness of the P4P program for schizophrenia. This was analyzed using 4 specific outcome measures for sample patients within 1 year after the index date: outpatient visits, emergency department visits, hospital admissions, and lengths of hospital stay.

Confounding Variables

In addition to the main explanatory variable (the P4P program), we considered several potential confounders in this analysis based on relevant literature reviews and the availability of variables in the data set used for this study. These included patient characteristics (eg, comorbidities, geographic region, and monthly insurance premium in lieu of monthly income), attending physician characteristics (eg, age, sex, and specialist seniority), and hospital characteristics (eg, ownership, accreditation level, and teaching status).

Study Design

To select patients for the exposed and unexposed cohorts, we performed propensity score matching to adjust for selection based on observable characteristics, ensuring comparability. Specifically, to minimize potential effects from incomplete matching and reduce bias from residual confounding, we conducted double propensity score adjustment.²⁶ We first employed an optimal matching algorithm based on patients’ age, sex, and the index date (the first date of enrollment in the P4P program), followed by regression covariate adjustment using propensity scores. For each patient in the exposed cohort, 4 patients were selected from the unexposed cohort.

Additionally, to assess the balance of covariate distribution between the 2 cohorts after propensity score matching, we calculated absolute standardized mean differences, setting the threshold for mean difference at 0.1.^27,28 Each qualified sample patient in both cohorts was tracked 1 year before and 1 year after the index date to evaluate the effects of the P4P program for schizophrenia, with the last data collection on December 31, 2018.

Furthermore, to effectively assess the impact of the P4P program, we employed a difference-in-differences (DID) approach. The DID method is a quasi-experimental approach that enables us to adjust for time-invariant unobserved confounders related to the outcome of interest (ie, health care resource utilization) between the exposed and unexposed cohorts, when randomization is
not feasible.^29,30

Statistical Analyses

We performed generalized estimating equations (GEEs) to adjust standard errors for clustering at the physician level. A GEE with a Poisson distribution was used for the first 3 outcome measures: outpatient visits, emergency department visits, and hospital admissions. For the final outcome measure—length of hospital stay—we applied GEEs with a negative binomial distribution because the outcome was skewed. As mentioned previously, we employed the DID technique in this study; therefore, we fitted the GEE models by including an exposure × time interaction term, allowing the effects of the P4P program to be captured through its coefficients. The following base regression model was used:

Y_it= β₀+ β₁ P4P_i+ β₂ T_t+ β₃ P4P_i T_t+ β₄ X_it+ ε_it

where:

Y_it is the outcome measure of interest for respondent i in year t,

the dummy variable P4P_i is 1 if respondent i was enrolled in the P4P program and 0 otherwise,

the dummy variable T_t is 1 if it is an observation after the P4P enrollment and 0 otherwise,

the vector X_it incorporates a set of confounding variables (eg, comorbidities, geographic region, and monthly insurance premium of respondents),

ε_it is the error term,

and the coefficient β₃ reflects the DID parameter.

Finally, we included the E-values to assess the potential impact of unmeasured confounders.³¹ All analyses were performed using the SAS software version 9.4, with statistical significance set at P < .05 (2-tailed).

RESULTS

Cohort Characteristics

This study included 48,145 patients with schizophrenia who met the inclusion criteria during the study period. Among them, 2275 patients were in the P4P group (the exposed cohort), and 45,870 patients were in the non-P4P group (the unexposed group) before matching. There were more female patients than male patients in both cohorts (55.30% in the exposed cohort and 52.93% in the unexposed cohort), with mean (SD) ages of 45.78 (12.46) years and 45.09 (14.02) years in the exposed and unexposed cohorts, respectively. After matching on patients’ age, sex, and the index date, there were 2118 patients in the exposed cohort and 6299 patients in the unexposed cohort (Table 1).

Effects of the P4P Program

Table 2 summarizes the results of the GEE modeling estimates, representing the effects of the P4P program on outpatient visits among patients with schizophrenia. To effectively isolate the impacts of the P4P program, we adopted a DID technique and included a time*exposure interaction term in the models, where time is a dummy variable representing the intervention period (0 = preintervention period, 1 = postintervention period), with the intervention referring to the P4P program. We observed a lower likelihood of outpatient visits among P4P enrollees (incidence rate ratio [IRR], 0.8357; P < .0001). The P4P program was associated with a 16.43% relative reduction in the rate of outpatient visits, calculated as [(1 – 0.8357) × 100] 1 year after the intervention.

On the contrary, patients in the P4P group did not have fewer emergency department visits than their counterparts over time. As shown in Table 3, this result was not statistically significant (IRR, 1.0234; P = .6687). Similarly, we did not find that the P4P program had a statistically significant impact on the likelihoods of hospital admission (IRR, 0.9146; P = .3728; Table 4) or on the lengths of hospital stay (IRR, 1.1253; P = .2128; Table 5). Notably, longer lengths of stay in hospitals were observed among patients with more comorbidities (Table 5).

Lastly, the parallel trends assumption is a key component of the DID approach. However, this assumption cannot be statistically tested directly, as it relates to the unobserved counterfactual post intervention.^32,33 It is suggested that controlling for unobserved heterogeneity can be achieved by ensuring statistical comparability between individuals with and without intervention during the baseline period.^34,35Therefore, to validate the parallel trends assumption, we performed propensity score matching for the baseline year before conducting the DID analysis, which improved the comparability of the 2 cohorts and helped satisfy this key assumption.^34,36 Moreover, we examined the pretreatment trends between both cohorts by rerunning the DID model, which included an interaction between the time and the exposure variables in the preintervention period.³⁷ This combined approach can reduce potential time-invariant differences between the exposed and unexposed cohorts, allowing for selection based on observed and unobserved factors that remain constant over time. The results suggested that the parallel trends assumption held since all coefficients were statistically insignificant (outpatient visits: IRR, 1.0169; P = .8987; emergency department visits: IRR, 1.0258; P = .6573; hospital admissions: IRR, 0.9476; P = .8150; and lengths of hospital stay: IRR, 1.0304; P = .4592).

DISCUSSION

Main Research Findings

Numerous studies have examined the effects of P4P schemes in health care, but the findings are heterogeneous, underscoring the need for more empirical evidence to support the claim that P4P programs are an effective strategy for achieving value-based care. Because schizophrenia is one of the most severe chronic psychiatric disorders and is associated with high health care resource utilization, significant productivity loss, social burden, and premature death,^4,38,39 effective policies and strategies—such as P4P programs—are needed to improve care outcomes and resource efficiency.

Notably, the real-world evidence supports our research hypothesis that the P4P program for schizophrenia would positively impact health care resource utilization, specifically regarding outpatient visits. The results of this analysis revealed that participation in the P4P program was associated with fewer outpatient visits among enrollees compared with those in the unexposed cohort. This inverse association is consistent with the conclusion of prior studies.^40,41 In Taiwan, schizophrenia is the third most common reason for outpatient visits, following neurotic disorders and affective disorders.⁴² Therefore, the positive effect observed from the P4P program is particularly encouraging.

Regarding other outcome measures, including emergency department visits, hospital admissions, and lengths of stay, this study yielded disappointing results, indicating that the P4P program had no impact on reducing usage among enrollees. Previous research assessing the P4P program for schizophrenia in Taiwan has produced inconsistent conclusions. Some research demonstrated positive effects of the program in terms of reductions in hospitalizations, lengths of stay, and emergency department visits,⁴¹ whereas another investigation showed no association,⁴⁰ consistent with the findings of this study.

In this analysis, we found that participation in the P4P program was associated with fewer outpatient visits (IRR, 0.8357; P < .0001). The reduction may indicate substantial cost savings, enhanced care coordination, and effective mitigation of unplanned or preventable outpatient service use. Consequently, these findings could have significant implications for resource allocation and patient access. Importantly, this decrease in outpatient visits was not accompanied by an increase in emergency department visits (IRR, 1.0234; P = .6687), hospitalizations (IRR, 0.9146; P = .3728), or lengths of stay (IRR, 1.1253; P = .2128). This suggests that the intervention may have effectively reduced unnecessary outpatient visits rather than merely shifted care to other settings.

The findings of this study regarding emergency department visits and hospitalizations warrant discussion. Prior research has shown that approximately half of individuals presenting to emergency departments have at least 1 mental health diagnosis,⁴³ and patients with mental illness often have higher rates of emergency department visits than those without.⁴⁴ Furthermore, emergency department visits for schizophrenia-related diagnoses are associated with more hospital admissions than visits for other reasons.⁴⁵ Reducing preventable emergency department visits and avoidable hospitalizations is a critical goal for the health care system. Therefore, it is particularly noteworthy that the P4P program did not achieve this objective. Previous studies have also indicated that P4P programs alone may not be sufficient to enhance the quality of medical care.^46,47

The relevance of Taiwan’s P4P program for schizophrenia to other countries warrants discussion. Adopting a holistic approach to disease management for chronic illnesses, such as schizophrenia, has emerged as a promising strategy.⁴⁸ In light of this study’s findings that having more morbidities contributes to longer hospital stays, there is an evident need for more comprehensive disease management systems. Although community intervention services, such as vocational peer support, have shown positive effects in improving psychiatric symptoms and reducing rehospitalizations, these services remain limited in Taiwan.⁴⁹ Furthermore, holistic health promotion is not a core component of the P4P program for schizophrenia. Therefore, integrating holistic approaches that not only address symptoms but also incorporate the social and behavioral health needs of patients should be considered when implementing or modifying existing P4P programs for individuals with chronic illnesses to achieve better patient outcomes.

Strengths and Limitations of the Study

The primary strength of this study lies in its use of a nationwide registry database over a relatively long study period, which enhances the robustness of our results. This approach minimizes potential validity threats related to selection bias, recall bias, and information bias that are commonly found in cross-sectional or regional studies. Another advantage is the application of the DID approach, which allows us to account for temporal trends in the outcome measures and time-fixed differences between the exposed and unexposed cohorts. Finally, we employed double propensity score adjustments to mitigate potential effects of incomplete matching and residual or unmeasured confounding.

Despite these strengths, several limitations warrant discussion. First, not all patients with schizophrenia in Taiwan are enrolled in this P4P program, as participation is voluntary. This may introduce selection effects at both the patient and provider levels. Furthermore, caution should be exercised when generalizing the findings to other health care systems due to variations in metrics and implementation of health care interventions. Another limitation is that the study relied on secondary data analyses. Consequently, our findings may be constrained by a lack of information on certain important risk factors in the NHIRD that could adversely affect health care outcomes but could not be adjusted for in this analysis, such as lifestyle factors (eg, alcohol consumption), family support, and traumatic life events. Moreover, we could not utilize validated psychiatric rating instruments, such as the Clinical Global Impression scale or the Clinical Global Impression-Schizophrenia scale, to measure symptom severity in schizophrenia. Additionally, this study employed a 1-year tracking period, which may restrict our ability to observe the tangible impacts of the P4P program, as some effects may take longer to manifest. Future studies should consider a longer follow-up period to adequately capture these developments. Lastly, despite utilizing a rigorous DID propensity score matching approach, we cannot account for many key unobserved provider traits, leaving unmeasured differences between participating and nonparticipating providers a potential source of residual confounding.

CONCLUSIONS

In summary, this study contributes to the literature by providing new real-world evidence regarding P4P programs. However, we acknowledge that the impacts of the P4P program for schizophrenia on health care resource utilization among enrollees are somewhat mixed. Therefore, the benefits of adopting a P4P scheme or similar value-based initiatives in mental health settings warrant further investigation. Additional work is needed to identify opportunities for enhancing P4P schemes to improve outcomes for patients with psychiatric disorders and other chronic diseases, particularly by focusing on implementing needs-driven, rather than supply-guided, health care policies and intervention strategies.⁵⁰

Author Affiliations: Department of Health Administration and Informatics, Governors State University (NL), University Park, IL; School of Health Care Administration, Taipei Medical University (VNL, JCC, KCH), Taipei, Taiwan; Department of Psychiatry, Taipei Medical University (MCMT), Taipei, Taiwan; Institute of Health and Welfare Policy, National Yang Ming Chiao Tung University (LNC), Taipei, Taiwan.

Source of Funding: The National Science and Technology Council in Taiwan (MOST 111-2410-H-038-004-MY2)

Author Disclosures: The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (NL, MCMT, LNC, KCH); acquisition of data (KCH); analysis and interpretation of data (NL, VNL, JCC, MCMT, LNC, KCH); drafting of the manuscript (NL, KCH); critical revision of the manuscript for significant intellectual content (NL, VNL, JCC, MCMT, LNC, KCH); statistical analysis (VNL, JCC, KCH); obtaining funding (KCH); and supervision (KCH).

Address Correspondence to: Kuo-Cherh Huang, DrPH, MBA, School of Health Care Administration, College of Management, Taipei Medical University, 11 Floor, Biomedical Technology Building, No. 301, Yuantong Road, Zhonghe District, New Taipei City 235, Taiwan. Email: kchuang@tmu.edu.tw.

Ethical Approval Statement: This study had been approved by the Institutional Review Board of Taipei Medical University, Taiwan (TMU-JIRB No. N202205046).

Note: Ning Lu, PhD, MPH, and Valeska N. Laksono, MBA, contribute equally to this work and are listed as co-first authors.

REFERENCES

1. Investing in Mental Health: Evidence for Action. World Health Organization. 2013. Accessed December 7, 2025. https://iris.who.int/server/api/core/bitstreams/79cb0722-c661-40f8-8b46-3c230687581e/content

2. Votruba N, Thornicroft G; FundaMentalSDG Steering Group. Sustainable development goals and mental health: learnings from the contribution of the FundaMentalSDG global initiative. Glob Ment Health (Camb). 2016;3:e26. doi:10.1017/gmh.2016.20

3. GBD 2019 Mental Disorders Collaborators. Global, regional, and national burden of 12 mental disorders in 204 countries and territories, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet Psychiatry.2022;9(2):137-150. doi:10.1016/S2215-0366(21)00395-3

4. Schizophrenia. World Health Organization. 2022. Accessed January 4, 2026. https://www.who.int/news-room/fact-sheets/detail/schizophrenia

5. Share of the population with schizophrenia 2023. Our World in Data. Accessed August 14, 2026.
https://ourworldindata.org/grapher/schizophrenia-prevalence

6. Kotzeva A, Mittal D, Desai S, Judge D, Samanta K. Socioeconomic burden of schizophrenia: a targeted literature review of types of costs and associated drivers across 10 countries. J Med Econ. 2023;26(1):70-83.
doi:10.1080/13696998.2022.2157596

7. Chong HY, Teoh SL, Wu DB, Kotirum S, Chiou CF, Chaiyakunapruk N. Global economic burden of schizophrenia: a systematic review. Neuropsychiatr Dis Treat. 2016;12:357-373. doi:10.2147/NDT.S96649

8. Pan YJ, Kuo KH, Yeh LL. Healthcare cost, service use and mortality in major psychiatric disorders in Taiwan. J Affect Disord. 2019;246:112-120. doi:10.1016/j.jad.2018.12.046

9. Srisurapanont M, Kunchanaphongphan T, Chokemaitree N, et al. Course and predictors of disability in Thai patients with schizophrenia: a 2-year, multi-center, prospective, observational study. Asian J Psychiatr. 2022;70:103044. doi:10.1016/j.ajp.2022.103044

10. National Health Insurance Administration, Ministry of Health and Welfare, Taiwan. The Pay-for-Performance Program for Schizophrenia. 6th ed. National Health Insurance Administration, Ministry of Health and Welfare, Taiwan; 2023. Accessed August 26, 2026. https://www.nhi.gov.tw/ch/cp-6007-71119-2830-1.html

11. Khan N, Rudoler D, McDiarmid M, Peckham S. A pay for performance scheme in primary care: meta-synthesis of qualitative studies on the provider experiences of the quality and outcomes framework in the UK. BMC Fam Pract. 2020;21(1):142. doi:10.1186/s12875-020-01208-8

12. Lalloué B, Jiang S, Girault A, Ferrua M, Loirat P, Minvielle E. Evaluation of the effects of the French pay-for-performance program-IFAQ pilot study. Int J Qual Health Care. 2017;29(6):833-837. doi:10.1093/intqhc/mzx111

13. Eijkenaar F. Key issues in the design of pay for performance programs. Eur J Health Econ. 2013;14(1):117-131. doi:10.1007/s10198-011-0347-6

14. Jha AK, Joynt KE, Orav EJ, Epstein AM. The long-term effect of premier pay for performance on patient outcomes. N Engl J Med. 2012;366(17):1606-1615. doi:10.1056/NEJMsa1112351

15. Ryan AM, Blustein J, Doran T, Michelow MD, Casalino LP. The effect of phase 2 of the Premier Hospital Quality Incentive Demonstration on incentive payments to hospitals caring for disadvantaged patients. Health Serv Res. 2012;47(4):1418-1436. doi:10.1111/j.1475-6773.2012.01393.x

16. Ho L, Mercer SW, Henderson D, Donaghy E, Guthrie B. Effect of UK Quality and Outcomes Framework pay-for-performance programme on quality of primary care: systematic review with quantitative synthesis. BMJ. 2025;389:e083424. doi:10.1136/bmj-2024-083424

17. Mathes T, Pieper D, Morche J, Polus S, Jaschinski T, Eikermann M. Pay for performance for hospitals. Cochrane Database Syst Rev. 2019;7(7):CD011156. doi:10.1002/14651858.CD011156.pub2

18. Lu JR, Chen YI, Eggleston K, Chen CH, Chen B. Assessing Taiwan’s pay-for-performance program for diabetes care: a cost-benefit net value approach. Eur J Health Econ. 2023;24(5):717-733. doi:10.1007/s10198-022-01504-3

19. Lian WC, Livneh H, Huang HJ, Lu MC, Guo HR, Tsai TY. Adding pay-for-performance program to routine care was related to a lower risk of depression among type 2 diabetes patients in Taiwan. Front Public Health. 2021;9:650452. doi:10.3389/fpubh.2021.650452

20. Cheng SW, Wang CY, Ko Y. Costs and length of stay of hospitalizations due to diabetes-related complications. J Diabetes Res. 2019;2019:2363292. doi:10.1155/2019/2363292

21. Tsai YS, Kung PT, Ku MC, Wang YH, Tsai WC. Effects of pay for performance on risk incidence of infection and of revision after total knee arthroplasty in type 2 diabetic patients: a nationwide matched cohort study. PLoS One. 2018;13(11):e0206797. doi:10.1371/journal.pone.0206797

22. Chen CC, Cheng SH. Does pay-for-performance benefit patients with multiple chronic conditions? evidence from a universal coverage health care system. Health Policy Plan. 2016;31(1):83-90. doi:10.1093/heapol/czv024

23. Chi MJ, Chou KR, Pei D, et al. Effects and factors related to adherence to a diabetes pay-for-performance program: analyses of a national health insurance claims database. J Am Med Dir Assoc. 2016;17(7):613-619. doi:10.1016/j.jamda.2016.02.033

24. Lin TY, Chen CY, Huang YT, Ting MK, Huang JC, Hsu KH. The effectiveness of a pay for performance program on diabetes care in Taiwan: a nationwide population-based longitudinal study. Health Policy. 2016;120(11):1313-1321. doi:10.1016/j.healthpol.2016.09.014

25. Yu HC, Tsai WC, Kung PT. Does the pay-for-performance programme reduce the emergency department visits for hypoglycaemia in type 2 diabetic patients? Health Policy Plan. 2014;29(6):732-741. doi:10.1093/heapol/czt056

26. Austin PC. Double propensity-score adjustment: a solution to design bias or bias due to incomplete matching. Stat Methods Med Res.2017;26(1):201-222. doi:10.1177/0962280214543508

27. Stuart EA, Lee BK, Leacy FP. Prognostic score-based balance measures can be a useful diagnostic for propensity score methods in comparative effectiveness research. J Clin Epidemiol. 2013;66(suppl 8):S84-S90.e1. doi:10.1016/j.jclinepi.2013.01.013

28. Austin PC. Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples. Stat Med. 2009;28(25):3083-3107. doi:10.1002/sim.3697

29. Fu AZ, Pesa JA, Lakey S, Benson C. Healthcare resource utilization and costs before and after long-acting injectable antipsychotic initiation in commercially insured young adults with schizophrenia. BMC Psychiatry. 2022;22(1):250. doi:10.1186/s12888-022-03895-2

30. Fredriksson A, Oliveira G. Impact evaluation using difference-in-differences. RAUSP Manage J.2019;54(4):519-532. doi:10.1108/RAUSP-05-2019-0112

31. VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introducing the E-value. Ann Intern Med. 2017;167(4):268-274. doi:10.7326/M16-2607

32. Angrist JD, Pischke JS. Parallel worlds: fixed effects, differences-in-differences, and panel data. In: Angrist JD, Pischke JS, eds. Mostly Harmless Econometrics: An Empiricist’s Companion. Princeton University Press; 2009:221-248.

33. Blundell R, Costa Dias M. Evaluation methods for non-experimental data. Fiscal Studies. 2000;21(4):427-468. doi:10.1111/j.1475-5890.2000.tb00031.x

34. d’Albis H, El Mekkaoui N, Legendre B. Health accidents and wealth decline in old age. Soc Sci Med. 2023;332:116117. doi:10.1016/j.socscimed.2023.116117

35. Khandker SR, Koolwal GB, Samad HA. Handbook on Impact Evaluation: Quantitative Methods and Practices. World Bank; 2010.

36. Basu S, Meghani A, Siddiqi A. Evaluating the health impact of large-scale public policy changes: classical and novel approaches. Annu Rev Public Health. 2017;38:351-370. doi:10.1146/annurev-publhealth-031816-044208

37. Riveros-Gavilanes JM. Testing parallel trends in differences-in-differences and event study designs: a research approach based on pre-treatment period significance. J Res Innov Technol. 2023;2(4):226-237. doi:10.57017/jorit.v2.2(4).07

38. Charlson FJ, Ferrari AJ, Santomauro DF, et al. Global epidemiology and burden of schizophrenia: findings from the Global Burden of Disease Study 2016. Schizophr Bull. 2018;44(6):1195-1203. doi:10.1093/schbul/sby058

39. Laursen TM, Nordentoft M, Mortensen PB. Excess early mortality in schizophrenia. Annu Rev Clin Psychol. 2014;10:425-448. doi:10.1146/annurev-clinpsy-032813-153657

40. Chen TT, Yang JJ, Hsueh YA, Wang V. The effects of a schizophrenia pay-for-performance program on patient outcomes in Taiwan. Health Serv Res. 2019;54(5):1119-1125. doi:10.1111/1475-6773.13174

41. Chen LH, Chang FC, Chien IC, Day GI. Effects of a pay-for-performance program for schizophrenia at a psychiatric hospital in northern Taiwan. Taiwan J Public Health. 2016;35(2):173-186. doi:10.6288/TJPH201635104058

42. Dai YX, Chen MH, Chen TJ, Lin MH. Patterns of psychiatric outpatient practice in Taiwan: a nationwide survey. Int J Environ Res Public Health. 2016;13(10):955. doi:10.3390/ijerph13100955

43. Hunt KA, Weber EJ, Showstack JA, Colby DC, Callaham ML. Characteristics of frequent users of emergency departments. Ann Emerg Med. 2006;48(1):1-8. doi:10.1016/j.annemergmed.2005.12.030

44. Niedzwiecki MJ, Sharma PJ, Kanzaria HK, McConville S, Hsia RY. Factors associated with emergency department use by patients with and without mental health diagnoses. JAMA Netw Open. 2018;1(6):e183528. doi:10.1001/jamanetworkopen.2018.3528

45. Albert M, McCaig LF. Emergency department visits related to schizophrenia among adults aged 18-64: United States, 2009-2011. NCHS Data Brief. 2015;215:1-8.

46. Shih T, Nicholas LH, Thumma JR, Birkmeyer JD, Dimick JB. Does pay-for-performance improve surgical outcomes? an evaluation of phase 2 of the Premier Hospital Quality Incentive Demonstration. Ann Surg. 2014;259(4):677-681. doi:10.1097/SLA.0000000000000425

47. Ryan AM, Blustein J, Casalino LP. Medicare’s flagship test of pay-for-performance did not spur more rapid quality improvement among low-performing hospitals. Health Aff (Millwood). 2012;31(4):797-805. doi:10.1377/hlthaff.2011.0626

48. Rado JT. Management of comorbid medical conditions in schizophrenia. In: Janicak PG, Marder SR, Tandon R, Goldman M, eds. Schizophrenia: Recent Advances in Diagnosis and Treatment. Springer; 2014:175-204.

49. Cheng KY, Yen CF. The social support, mental health, psychiatric symptoms, and functioning of persons with schizophrenia participating in peer co-delivered vocational rehabilitation: a pilot study in Taiwan. BMC Psychiatry. 2021;21(1):268. doi:10.1186/s12888-021-03277-0

50. Maertens de Noordhout C, Levy M, Claerman R, et al. Identification of health-related needs: the needs examination, evaluation and dissemination (NEED) assessment framework. Health Policy. 2025;155:105263. doi:10.1016/j.healthpol.2025.105263