Participation and Outcomes among Disabled and Non-Disabled People in the Diabetes Pay-for-Performance Program

Objectives: This study’s objectives were to compare the participation rates of people with and without disabilities who had type 2 diabetes in a diabetes pay-for-performance (DM P4P) program, as well as their care outcomes after participation. Methods: This was a retrospective cohort study. The data came from the disability registry file, cause of death file, and national health insurance research database of Taiwan. The subjects included patients newly diagnosed with type 2 diabetes between 2001 and 2013 who were followed up with until 2014 and categorized as disabled and non-disabled patients. The propensity score matching method was used to match the disabled with the non-disabled patients at a 1:1 ratio. Conditional logistic regression analysis was used to determine the odds ratio between the disabled and non-disabled patients who joined the P4P program. The Cox hazard model was used to compare the risk of dialysis and death between the disabled and non-disabled patients participating in the P4P program. Results: There were 110,645 disabled and 110,645 non-disabled individuals after matching. After controlling for confounding factors, it was found that the disabled individuals were significantly less likely (odds ratio = 0.89) to be enrolled in the P4P program than the non-disabled individuals. The risk of dialysis was 1.08 times higher for people with disabilities than those without, regardless of their participation in the P4P program. After enrollment in the P4P program, the risk of death for people with disabilities decreased from 1.32 to 1.16 times that of persons without disabilities. Among the people with disabilities, the risk of death for those enrolled in the P4P program was 0.41 times higher than that of those not enrolled. The risk of death was reduced to a greater extent for people with disabilities than for those without disabilities upon enrollment in the DM P4P program. Conclusion: People with disabilities are less likely to be enrolled in the P4P program in Taiwan and have unequal access to care. However, the P4P program was more effective at reducing mortality among people with disabilities than among those without.


Introduction
Approximately 537 million people worldwide have diabetes, with a prevalence rate of 9.8%, and 6.7 million deaths have been caused by diabetes [1].Proper glycemic control can reduce the risk of neuropathy by 60%, retinopathy by 76%, and nephropathy by 35-56% [2].The United States Renal Data System 2020 reported a 47.1% rate of hemodialysis among patients with kidney disease caused by diabetes [3], indicating that renal failure is one of the most serious complications of diabetes, which may eventually necessitate dialysis or kidney transplantation.The related literature indicates that the incidence of end-stage renal disease (ESRD) in diabetic patients is 3-12 times higher than that in non-diabetic patients [4,5].
People with disabilities are prone to having multiple chronic diseases and poor overall health [6]; therefore, they tend to utilize healthcare more and consequently incur greater costs [6][7][8].However, the quality of care they receive and the prevention of complications (such as diabetes) are poor, meaning that they suffer from more severe disease and higher mortality compared to the general population.
Taiwan's national health insurance covers 99.9% of the population [9].The diabetes pay-for-performance plan (DM P4P) was implemented in November 2001.The DM P4P program provides complete services, including diagnosis, examination, health education, and follow-up, to reduce or delay the complications of diabetes [10].Diabetic patients can only be enrolled if they have been diagnosed with diabetes mellitus (the first three codes of ICD9CM: 250) more than twice within 90 days by the same physician at the same institution or if they have been hospitalized for diabetes mellitus [10].Physicians need to provide diabetes self-management education to patients enrolled in the P4P program and regularly follow up with them, at least three times a year, with biochemical tests measuring HbA1c, serum creatinine, and LDL cholesterol [10].In addition to the regular physician consultation fees and management care fees, physicians who rank in the top 25% in terms of quality of care can receive additional case incentives [11].The percentage of patients enrolled in the DM P4P program in 2013 was 35.1%, which increased to 51.3% in 2018 [12].However, the situations of people with disabilities in the P4P program have not been explored.
Most studies have confirmed that the implementation of the P4P program improves the quality of care for diabetic patients while controlling medical costs [13,14].An American study [15] examined diabetes disease management programs (DDMPs) and found that diabetic patients who joined the management programs had fewer hospitalization days and fewer expenses than those who did not.Their glycated hemoglobin tests, blood lipid tests, retinal tests, microprotein tests, and smoking habits were better than those of nonenrollees.Another British study also showed that the quality of medical care for diabetes has significantly improved since the pay-for-performance contract was introduced [16].
Research in Taiwan has also shown that the diabetes pay-for-performance program can improve the quality of care for diabetic patients and control medical costs at the same time [17].In addition, enrollment in diabetes pay-for-performance programs can improve the continuity of care and reduce hospitalization and emergency room utilization for diabetes-related diseases [18,19].However, there is no research on the care outcomes of people with disabilities participating in P4P programs.
Disabled people can apply for a disability certificate for obtaining social welfare and medical subsidies from the Ministry of the Interior (MOI) after being diagnosed by a qualified specialist from a public hospital [20].According to the Ministry of Health and Welfare (MOHW) statistics from 2023, there were 1,196,654 disabled people in Taiwan in 2022, accounting for approximately 5.14% of the total population [21].People with disabilities have a higher risk of developing diabetes than the general population [22][23][24].The agestandardized prevalence of diabetes among the disabled is 15.8%, which is much higher than the age-standardized prevalence of 7.2% among all adults in the United States [25].Furthermore, people with intellectual disabilities have a higher risk of developing diabetes than those with non-intellectual disabilities, with prevalence ranging from 7.1% to 14% [22,23].
Many adults with disabilities have more sedentary lifestyles [26], low levels of physical activity [27], and high-fat diets [28], which may lead to obesity and, consequently, the development of diabetes.Another study confirmed that the use of antipsychotic drugs (clozapine) is positively related to obesity (BMI > 30) [29], resulting in increased risks of dyslipidemia and diabetes [30,31].The poor overall health of people with disabilities puts them at risk for more severe comorbidities and higher mortality rates compared to the general population [7,[32][33][34].
Few studies have examined the enrollment rates and care outcomes for people with disabilities.Previous studies have shown that patients with more severe comorbidities [35][36][37], more severe diabetes complications, and more complex chronic illnesses [10,35,37] are less likely to enroll in the DM P4P program.This suggests that the Taiwan DM P4P program tends to screen patients and that diabetic patients enrolled in the program have significantly less severe conditions than those not enrolled.Thus, this study aimed to investigate the differences between disabled and non-disabled diabetic patients enrolled in the DM P4P program and their associated factors.This study also aimed to analyze the differences in care outcomes, including the risks of dialysis and mortality, when disabled diabetic patients enrolled in the DM P4P program.

Data Sources and Participants
This is a retrospective cohort study.The data sources included the people registered as having disabilities by Taiwan's MOI from 2000 to 2008, the National Health Insurance Research Database maintained by Taiwan's Ministry of Health and Welfare, the Registry for Catastrophic Illness Patients, and cause of death data.The study population included patients who were newly diagnosed with type 2 diabetes between 1 January 2001 and 31 December 2013.The period from 2000 served as the wash-out period for newly diagnosed diabetics.With reference to previous related studies [19], in this study, diabetic patients were defined as those with three primary or secondary outpatient diagnoses of diabetes (ICD9: 250, A181) or one hospitalization within 365 consecutive days [36,[38][39][40], using the date of the first diabetes visit as the date of confirmation (index date) while excluding gestational diabetes (ICD9CM: 648.0, 648.8) and type 1 diabetes (ICD9CM: 250.x1, 250.x3, 250.1x).
In this study, patients with diabetes were divided into those with disabilities and those without disabilities.The study subjects were defined as disabled according to data from the registry of people with disabilities from Taiwan's MOI.Furthermore, the occurrence of dialysis or death was tracked up to 31 December 2014.The following were excluded: (1) those who had diabetes prior to acquiring a disability; (2) those who were in a vegetative state; (3) those with disabilities as well as congenital metabolic abnormalities related to metabolic diseases; and (4) those who had a kidney transplant or dialysis before developing diabetes.
Many studies on the DM P4P program have shown that the program tends to screen patients; in particular, those with a higher Charlson Comorbidity Index (CCI) and Diabetes Complications Severity Index (DCSI) scores were less likely to be enrolled in the P4P program [10,35,37].The disabled people in the studies have poorer health than those without disabilities [6] and tend to have more severe CCI scores [7,34].Therefore, to prevent the effects of the inherent differences between the disabled and non-disabled participants from influencing the study's results, because our research was a national database analysis, the number of disabled people selected was 110,645.For avoidance, the total number of people after matching was too large, which would have easily led to statistically significant differences during analysis.Therefore, we used a 1:1 propensity score matching (PSM) method using the OneToManyMTCH macro proposed by Parsons, Ovation Research Group, Seattle, WA, USA [41].Logistic regression analysis was used to calculate the propensity scores of disabled and non-disabled patients who enrolled in DM P4P program; then, we matched them in the hierarchical order of eight digits to one digit in the greedy algorithm (Figure 1).
Before 2012, persons with disabilities were evaluated by qualified medical specialists from public hospitals and their information was sent to the Social Bureau in each city for registration before being sent to the Ministry of the Interior.There are sixteen categories of disability and four levels of severity (mild, moderate, severe, and very severe).The data used in this study were from the Disability Registration Files, and include the type, severity, cause of disability, and date of disability.
exemptions for people with serious illnesses who meet the NHI's definition of catastrophic illnesses as diagnosed by physicians, including cancer, stroke, dialysis, longterm respirator dependence, and another 26 categories.Dialysis patients are diagnosed by a nephrologist as requiring long-term dialysis due to kidney failure and are provided with indications for dialysis and laboratory data.The date of application (date of illness), types of major illness, and ICD-9-CM diagnosis are recorded in the Registry for Catastrophic Illness Patients database.
Figure 1.Screening process for the study participants.

Description of Variables
Subject to the regulations of Taiwan's DM P4P program and the previous related literature [19], those with a primary diagnosis of diabetes, outpatient case classification of "E1", specific treatment item code of "E4", or inpatient case category of "C" were defined as having enrolled in the DM P4P program, with the date of first report recorded as the index date.To reduce the financial burden of long-term healthcare for people with serious illnesses, the Taiwan National Health Insurance Administration (NHI) provides financial exemptions for people with serious illnesses who meet the NHI's definition of catastrophic illnesses as diagnosed by physicians, including cancer, stroke, dialysis, long-term respirator dependence, and another 26 categories.Dialysis patients are diagnosed by a nephrologist as requiring long-term dialysis due to kidney failure and are provided with indications for dialysis and laboratory data.The date of application (date of illness), types of major illness, and ICD-9-CM diagnosis are recorded in the Registry for Catastrophic Illness Patients database.

Description of Variables
Subject to the regulations of Taiwan's DM P4P program and the previous related literature [19], those with a primary diagnosis of diabetes, outpatient case classification of "E1", specific treatment item code of "E4", or inpatient case category of "C" were defined as having enrolled in the DM P4P program, with the date of first report recorded as the index date.
Dialysis patients were defined as patients recorded as being on dialysis in the Catastrophic Illness database.If a patient died, the date of death was determined from the cause of death file.The rest of the variables and their definitions are listed below.
Disabled: Those who are registered as disabled, including their gender and age at the index date.CCI: The patient's disease diagnosis codes within two years prior to the index date were used to calculate the CCI score based on the Charlson Comorbidity Index as modified by Dey et al. [42].DCSI: In 2008, Young et al. classified diabetic complications into seven categories, with diagnoses of complications within two years before the index date [43].Hypertension (ICD9CM: 402, 405) and hyperlipidemia (ICD9CM: 272.0-272.1):Two outpatient visits or one diagnosis due to hospitalization.Monthly salary: Salary income at the end of the index year.Level of urbanization: All townships were categorized according to seven levels of urbanization, the first of which was the most urbanized and the seventh of which was the least urbanized [44].Characteristics of health providers: The primary healthcare provider was defined as the institution and physician who received the largest number of medical visits from diabetic patients (primary or secondary diagnosis of 250 or A181).Service volume of physicians: The annual service volume of the primary physician was ranked among the annual service volumes of all physicians who provided outpatient care for diabetic patients.The annual service volume of each primary care physician for diabetes was categorized as high service volume (>Q3), medium service volume (Q3 to Q1), or low service volume (<Q1).Level of primary care facilities: Divided into medical centers, regional hospitals, district hospitals, and primary care clinics.Ownership of institution: Categorized as public hospitals and nonpublic hospitals.

Statistical Analysis
Statistical analysis was conducted using SAS 9.4 (SAS Institute, Cary, NC, USA) software with a confidence level of α = 0.05 to determine the differences between the dependent variables (enrollment in the P4P program) according to the independent variables, including status (disability and non-disability), basic personal characteristics (gender and age), health status (CCI, DCSI, hypertension, and hyperlipidemia), economic factors (monthly salary), environmental factors (level of urbanization), and the characteristics of the primary health providers (service volume of the primary physician, level of primary care facility, and ownership).
A conditional logistic regression analysis was then conducted to examine the odds ratios between the disabled and the non-disabled and whether they enrolled in the P4P program or not, controlling for basic personal characteristics, health conditions, economic factors, environmental factors, and the characteristics of primary health providers.
Finally, we evaluated the effects of enrollment in the P4P program on the care outcomes of disabled and non-disabled people by examining the risk of dialysis and the risk of death.As the care outcomes of disabled and non-disabled patients after joining the P4P program may vary, the Mantel-Haenszel chi-square was used to examine these interactions.The Cox proportional hazard model (Cox PH model) was then used to control for relevant factors to examine the differences in the risk of dialysis and the risk of death between disabled and non-disabled patients who did or did not enroll in the P4P program.

Results of Matching between Disabled and Non-Disabled Diabetic Patients
There was a total of 2,474,399 diabetic patients diagnosed between 2000 and 2013.After exclusion, 1,730,891 people from 2001 to 2013 remained, of whom 1,620,246 were non-disabled and 110,645 were disabled.The percentage of disabled males was 56.2%, with an average age of 63.3 years, whereas the proportion of non-disabled males was 53.4%, with an average age of 57.9 years.Gender, age, CCI, and DCSI significantly differed between disabled and non-disabled patients (p < 0.05).After 1:1 matching through PSM, there were 110,645 non-disabled patients (average age of 62.1 years) and 110,645 disabled patients (average age of 62.4 years) included, resulting in a total of 221,290 study participants (Table 1).There were no statistically significant differences in gender, age, CCI, or DCSI (p > 0.05).(25,691) of the non-disabled patients were enrolled in the P4P program.The percentage of people with disabilities who joined the P4P program was lower than that of those without disabilities, regardless of gender and age, and only 6.94% of people with disabilities aged 75 years and older enrolled in the P4P program.The average age of the disabled people who were enrolled in P4P was significantly lower.Only 10.8% of disabled people with CCI scores greater than three were enrolled in the P4P program, while 14.5% of non-disabled people were enrolled.Likewise, only 9.12% of disabled people with DCSI scores greater than three were enrolled in the P4P program, whereas this statistic for non-disabled people was 13.2%.As a result, in the context of the same personal characteristics, health status, and socioeconomic status, fewer disabled people were enrolled in the P4P program than non-disabled people (Table 2).
After further controlling for relevant variables using conditional logistic regression, the probability of enrollment in the P4P program among the disabled was found to be 0.89 times higher than that among the non-disabled (95% CI: 0.87-0.91)(Table 2).This shows that people with disabilities were less likely to enroll in the P4P program.In addition, the rate of enrollment in the P4P program was lower among patients with a higher age, catastrophic illness, higher CCI scores, and higher DCSI scores (p < 0.05) (Table 2); however, those with hypertension and hyperlipidemia had a higher rate of enrollment in the P4P program (p < 0.05).Table 3 shows that the proportion of dialysis occurring in both disabled and nondisabled patients was approximately 1.82%.In total, 690 (1.44%) patients who were enrolled in the DM P4P program had undergone dialysis, which is significantly lower than the proportion of those who did not enroll in the P4P program and had undergone dialysis (3345, 1.93%) (p < 0.05).Additionally, the Mantel-Haenszel chi-square test was used to identify any interaction effects between the presence or absence of disabilities and the P4P program participation status on the risk of dialysis; the results show that there was no interaction effect (p = 0.839).Finally, the Cox PH model was used to analyze the risk of dialysis, considering the time factor, to explore the risk of dialysis from the time of diagnosis of diabetes or from the time of enrollment in the P4P program until the end of 2014.As shown in Table 3 and Figure 2, after controlling for relevant factors, the risk of dialysis was found to be significantly higher for people with disabilities than for those without disabilities (HR = 1.08, 95% CI: 1.01-1.15).Furthermore, those enrolled in the P4P program were less likely to have undergone dialysis (HR = 0.66, 95% CI: 0.60-0.72).

Effects of the DM P4P Program on the Mortality in Disabled and Non-Disabled Diabetic Patients
As shown in Table 3, 37.97% of the disabled patients died, which is significantly higher than the proportion of the non-disabled diabetics who died (28.02%) (p < 0.05).Of the 47,945 people enrolled in the P4P program, 5900 (12.3%) eventually died, significantly fewer than those (38.72%)who died and were not enrolled in the P4P program (p < 0.05).
As the effects of the P4P program on the care of non-disabled and disabled people may differ, the Mantel-Haenszel chi-square test revealed an interaction between the presence or absence of disability and enrollment or lack of enrollment in the P4P program in terms of the risk of death (p < 0.05) (Table 3).The Cox PH model was conducted by considering the interaction and the duration of diabetes, and a stratified analysis was carried out based on the presence or absence of disability as well as whether or not the patient was enrolled in the P4P program.The results are shown in Table 3 and Figure 3.In summary, based on the interaction term (Table 3), the risk of death was reduced more effectively among disabled patients compared to non-disabled patients when enrolled in the P4P program (HR = 0.87, 95% CI: 0.82-0.92).Furthermore, we conducted a stratified analysis and found that P4P care was more effective at reducing the risk of death for those with disabilities (HR = 0.41, 95% CI: 0.39-0.42)than for those without disabilities (HR = 0.47, 95% CI: 0.46-0.49)(Figure 3).Further analyses of those enrolled in the P4P program show that the risk of death of those with disabilities was significantly higher than that for those without disabilities (HR = 1.16, 95% CI: 1.10-1.22),even when enrolled in the P4P program (Figure 3).
Table 3 shows that the proportion of dialysis occurring in both disabled and nondisabled patients was approximately 1.82%.In total, 690 (1.44%) patients who were enrolled in the DM P4P program had undergone dialysis, which is significantly lower than the proportion of those who did not enroll in the P4P program and had undergone dialysis (3345, 1.93%) (p < 0.05).Additionally, the Mantel-Haenszel chi-square test was used to identify any interaction effects between the presence or absence of disabilities and the P4P program participation status on the risk of dialysis; the results show that there was no interaction effect (p = 0.839).
Finally, the Cox PH model was used to analyze the risk of dialysis, considering the time factor, to explore the risk of dialysis from the time of diagnosis of diabetes or from the time of enrollment in the P4P program until the end of 2014.As shown in Table 3 and Figure 2, after controlling for relevant factors, the risk of dialysis was found to be significantly higher for people with disabilities than for those without disabilities (HR = 1.08, 95% CI: 1.01-1.15).Furthermore, those enrolled in the P4P program were less likely to have undergone dialysis (HR = 0.66, 95% CI: 0.60-0.72).In previous research, there was no comparison between the disabled and non-  In previous research, there was no comparison between the disabled and non-disabled in terms of their enrollment in the P4P program, nor was there exploration of their care outcomes after enrollment.In this study, the rate of enrollment in the P4P program of disabled people was significantly lower than that of non-disabled people (Table 2).According to the regulations of the P4P program in Taiwan, healthcare providers must withdraw from the program if they do not meet NHI criteria during the care process (e.g., number of follow-ups [10,11]), and this affects the ranking of the quality of care (top 25%), resulting in the loss of additional incentives.Thus, providers are inadvertently pressured to select patients who are more likely to cooperate with the P4P program.Previous studies have also shown that P4P programs tend to exclude older patients, patients with more comorbidities [36,37], or patients with more severe conditions, leading to health inequities [10,35].Therefore, healthcare providers are less likely to include people with disabilities in the P4P program due to their poorer health statuses and the difficulty in providing them with care compared to non-disabled people [6][7][8], which may result in difficulties in achieving the relevant quality control indicators required to participate in the P4P program.
Previous studies have shown that diabetic patients' enrollment in the DM P4P program can improve the continuity of care and reduce hospitalization utilization for diabetes-related diseases [18]; research from South Korea has shown that greater continuity of care (COC) was associated with fewer preventable hospitalizations in people with type 2 diabetes [45].Another article adopted the definition of hospitalization from the Prevention Quality Indicator (PQI) proposed by the Agency for Healthcare Research and Quality (AHRQ) and showed that patients with type 2 diabetes who participated in the P4P program had a lower chance of preventable hospitalization [46].Therefore, if disabled people with diabetes experience greater opportunities to enroll in the DM P4P program, similar to non-disabled people, this may help reduce preventable hospitalizations due to diabetes among people with disabilities.

The Similar Effects of the P4P Program on Reducing the Risk of Dialysis among DM Patients with Disabilities and Those without Disabilities
In this study, the risk of dialysis for people with disabilities was 1.08 times higher than that for people without disabilities (Table 3, Figure 2).People with disabilities have poorer health statuses [6], weaker skills in communication with healthcare providers [47,48], and more difficultly in accessing healthcare [49], resulting in poorer glycemic control and thus a higher risk of dialysis.In addition, people with disabilities may have unhealthy lifestyles, which increases the risk of weight gain and further contributes to obesity and metabolic syndromes [28, [50][51][52]; these explain why diabetic patients with disabilities have a higher risk of requiring dialysis compared to non-disabled diabetic patients.
Poor glycemic control is a risk factor for diabetic nephropathy [28, [50][51][52], and previous studies have shown that diabetic patients enrolled in the P4P program have more stable glycemic control [47,48].The risk of proteinuria was reduced in those enrolled in the P4P program [47,48], as was the risk of nephropathy [28, [50][51][52].Our findings show that the risk of dialysis was lower among diabetic patients enrolled in the P4P program, regardless of whether they were disabled; their risk was significantly lower than that of those where were not enrolled (HR = 0.66, p < 0.05).In addition, the effectiveness of the P4P program in reducing the risk of dialysis among people with disabilities compared to that among those without disabilities when they were enrolled was not significantly different (Table 3).

P4P Program Enrollment Reduced the Risk of Death More for Disabled Individuals than for Non-Disabled
The ultimate goal of all healthcare is to prevent death.Table 3 shows that 12.3% of the diabetic patients who were enrolled in the P4P program died, while 38.7% of those who were not enrolled in the P4P program died.Those who were enrolled in the P4P program also had a much lower risk of dying (HR = 0.45) after controlling for relevant factors (Table 3, Figure 3).This coheres with previous studies, in which the risk of death among diabetics enrolled in the P4P program was 0.43-0.89times lower than that of those not enrolled [19,53].This study suggests that the reduction in the risk of death may be due to the P4P program requiring regular visits and periodic related tests.It has been revealed that diabetic patients enrolled in the P4P program have more outpatient visits where they undergo related tests for diabetes (such as HbA1c, Cr, and fundus examinations) [54], better continuity of care [19], and fewer emergency visits for infections [55], thus reducing the risk of death.
The present study has shown that the risk of death among patients with disabilities enrolled in the P4P program was significantly lower than that among patients with disabilities not enrolled in the P4P program.After controlling for other relevant factors using the Cox PH Model (Table 3, Figure 3), it was found that, among those who joined the P4P program, the risk of death for people with disabilities was 1.16 times higher than that of those without disabilities, and among those who did not participate in the P4P program, the risk of death for people with disabilities was 1.32 times higher than that of those without disabilities.The mortality rate of the physically and mentally disabled is higher than that of the non-disabled, which is consistent with the results of previous studies [7,[32][33][34].Among the disabled diabetics, the risk of death was 0.41 times lower in those enrolled in the P4P program than in those who were not, suggesting that enrollment in the P4P program is quite beneficial in mitigating the risk of death.Furthermore, P4P program enrollment reduced the risk of death to a greater extent among disabled individuals compared to non-disabled individuals, indicating that disabled individuals gain more benefits from enrolling in the P4P program.Gillani's study found that diabetic patients with disabilities could not sufficiently self-monitor their blood glucose due to a lack of diabetes-related knowledge, but this could be effectively addressed with the assistance of medical professionals who focused more on improving patients' knowledge and behaviors [56].This echoes the findings of this study, wherein enrollment in the P4P program, which integrates healthcare, diet education, weight control, and the regular return monitoring of diabetic patients through the collaboration of healthcare providers and dietitians, could effectively prevent the deterioration of disabled patients due to diabetes.

Conclusions
The rates of enrollment in the P4P program were lower for people with disabilities compared to those without disabilities (OR = 0.89).Those with disabilities were at risk of rejection from the P4P program, resulting in unequal access to care.The dialysis risk was 1.08 times higher for people with disabilities than for those without disabilities, regardless of their participation in the P4P program.However, after enrollment in the DM P4P program, the mortality risk for people with disabilities was improved more than that of those without disabilities.After enrollment in the P4P program, the risk of death for people with disabilities decreased from 1.32 to 1.16 times that of persons without disabilities.For patients who enrolled in the P4P program, the risk of death was reduced significantly more for those with disabilities than for those without disabilities.Therefore, this study recommends that, in the future, the government should provide incentives to strengthen doctors' willingness to enroll people with disabilities in the P4P program and encourage physicians to pay more attention to this disadvantaged group, thus improving the outcomes of diabetes care for people with disabilities.In future research, we suggest that studies assess the impact of P4P program enrollment on the risk of amputation and the risk of preventable hospitalization in people with disabilities and those without disabilities.

Limitations
Regarding the study's limitations, personal health behaviors and physiological parameters such as HbA1c were not available and could not be analyzed as variables in the study.Previous studies have indicated that blood sugar control significantly affects the risk of future dialysis and death in diabetic patients.However, blood sugar data in diabetic patients was not available in this study and could not be used as a control variable for disabled and non-disabled diabetic patients.This is another limitation of this study.

Figure 1 .
Figure 1.Screening process for the study participants.

3. 3 .
Effects of Enrollment in the DM P4P Program on the Risk of Dialysis in Disabled and Non-Disabled Diabetic Patients

Figure 2 .
Figure 2. Comparing the effects of P4P program participation status and disability status on the dialysis risk in type 2 diabetic patients via a Cox proportional hazard model (after controlling for sex, age, monthly salary, urbanization of residence area, CCI, DCSI, hypertension, hyperlipidemia, physician volume, healthcare organization level, and healthcare organization type).

Figure 3 .
Figure 3. Stratified analysis: comparing the effects of P4P program participation status and disability status on the risk of death in diabetic patients via a Cox proportional hazard model (after controlling for sex, age, monthly salary, urbanization of residence area, CCI, DCSI, hypertension, hyperlipidemia, physician volume, healthcare organization level, and healthcare organization type).

1 .
Disabled People Had a Lower Rate of Enrollment in the P4P Program than Non-Disabled People

Figure 3 .
Figure 3. Stratified analysis: comparing the effects of P4P program participation status and disability status on the risk of death in diabetic patients via a Cox proportional hazard model (after controlling for sex, age, monthly salary, urbanization of residence area, CCI, DCSI, hypertension, hyperlipidemia, physician volume, healthcare organization level, and healthcare organization type).

4 . Discussion 4 . 1 .
Disabled People Had a Lower Rate of Enrollment in the P4P Program than Non-Disabled People

Table 1 .
Bivariate analysis of type 2 diabetes patients with and without a disability before and after matching.
Among the 221,290 diabetic patients, 47,945 (21.7%) were enrolled in the P4P program and 173,345 (78.3%) were not enrolled (Table2).A total of 22,254 (20.1%) of the disabled patients were enrolled in the P4P program, while 23.2% * SMD: standardized mean difference.3.2.Comparison of Enrollment in the P4P Program among Diabetic Patients with and without Disabilities

Table 2 .
Bivariate analysis and logistic regression analysis of variables in type 2 diabetes patients with or without a disability and P4P program participation status.

Table 3 .
The effects of disability status and P4P participation status on the risks of dialysis and death.
# All models have been controlled for sex, age, monthly salary, urbanization of residence area, CCI, DCSI, hypertension, hyperlipidemia, physician volume, healthcare organization level, and healthcare organization type.aHR: adjusted hazard ratio.

Table 3 .
The effects of disability status and P4P participation status on the risks of dialysis and death.Comparing the effects of P4P program participation status and disability status on the dialysis risk in type 2 diabetic patients via a Cox proportional hazard model (after controlling for sex, age, monthly salary, urbanization of residence area, CCI, DCSI, hypertension, hyperlipidemia, physician volume, healthcare organization level, and healthcare organization type).