Factors Affecting Health-Care Costs and Hospitalizations among Diabetic Patients in Thai Public Hospitals

Authors

  • Usa Chaikledkaew PhD,

    Corresponding author
    1. Department of Pharmacy, Faculty of Pharmacy, Mahidol University, Bangkok, Thailand;
    2. Health Intervention and Technology Assessment Program, Ministry of Public Health, Nonthaburi, Thailand;
    Search for more papers by this author
  • Petcharat Pongchareonsuk PhD,

    1. Department of Pharmacy, Faculty of Pharmacy, Mahidol University, Bangkok, Thailand;
    Search for more papers by this author
  • Nathorn Chaiyakunapruk PharmD, PhD,

    1. Department of Pharmacy Practice, School of Pharmacy, Phitsanulok, Naresuan University, Thailand;
    2. Setting Priority Using Cost-effectiveness Analysis, Ministry of Public Health Nonthaburi, Thailand;
    3. School of Population Health, The University of Queensland, Brisbane, Australia;
    Search for more papers by this author
  • Boonsong Ongphiphadhanakul MD, PhD

    1. Faculty of Medicine, Ramathibodi Hospital, Mahidol University, Bangkok, Thailand
    Search for more papers by this author

Usa Chaikledkaew, Department of Pharmacy, Faculty of Pharmacy, Mahidol University, Bangkok 10400, Thailand. E-mail: pyuck@mahidol.ac.th

ABSTRACT

Objective:  The study investigated the factors affecting health-care costs and hospitalizations among diabetic patients in Thai public hospitals.

Methods:  A retrospective study was conducted by using administrative claims data obtained from diabetic patients during October 1, 2002 and September 30, 2003. Dependent variables were total health-care costs and the occurrence of hospitalizations. Independent variables included demographic factors, health-care utilizations, complications, comorbidities, and payment methods. Multivariate statistical analyses were applied.

Results:  The results of this study suggested that demographic factors of patients (i.e., age and male sex), payment methods (i.e., capitation, fee-for-service, and out-of-pocket) were significantly associated with higher health-care costs and probability of hospitalization. Patients receiving treatment from teaching hospitals significantly consumed higher health-care costs. In addition, the more health-care utilizations (i.e., occurrence of hospitalization, number of outpatient visit, and insulin utilization), the higher health-care costs the patients significantly had. Diabetic patients taking insulin had significantly higher health-care costs and risk of hospitalization. Furthermore, comorbidities (e.g., hypertension and cancer) and diabetes-related complications (e.g., nephropathy, neuropathy, retinopathy, coronary artery disease, cardiovascular disease, and peripheral vascular disease) were significantly associated with an increase in health-care costs and hospitalization.

Conclusion:  Factors affecting health-care costs and hospitalizations may help health-care providers intervene to improve patient management and possibly reduce health-care costs in the future.

Introduction

Diabetes is a common, serious, and chronic disease causing major long-term complications and comorbidities. For all age groups worldwide, the prevalence of diabetes was estimated at 2.8% in 2000 and 4.4% in 2030 [1]. Especially in the economically developing countries, it is predicted to have the greatest increase [2]. Among Thai people, the prevalence of diabetes was estimated at 2.4% in 1995 and 3.5% in 2025 [2]. The rise in prevalence of diabetes leads to an increase in prevalence of diabetic complications (e.g., retinopathy [23%], nephropathy [24%], amputation [1.6%], coronary disease [8.2%], and stroke [4.4%]) and diabetic comorbidities (e.g., hypertension [63.6%] and dyslipidemia [73.3%]) [3]. Diabetic-related complications and comorbidities largely affect patient outcomes and health-care costs. In the United States, the total cost of diabetes was $132 billion (i.e., direct [69.7%] and indirect cost [30.3%]) [4]. In Thailand, there have been few studies estimating the cost of diabetes. Based on the study determining the costs of patients with diabetes in seven Thai government hospitals located in four regions of Thailand and Bangkok, the annual average direct medical cost per diabetic patient was 6017 baht, which was significantly higher than those without diabetes [5]. In addition, the annual average total health-care cost per diabetic patient was 13,751 baht (i.e., direct medical and nonmedical cost [82.26%] and indirect cost [17.74%]) [6]. The average direct medical cost per outpatient visit was about 1206 baht per diabetic patient [7]. Recently, seven studies performed in the United States or Taiwan have investigated the impact of factors such as demographic characteristics, number of diabetic complications, number of health-care utilization, length of stay (LOS), and payment methods on health-care costs or hospitalizations [8–14]. In Thailand, only two studies have determined the factors associated with direct medical costs, but no study has ever investigated the association between factors and the occurrence of hospitalization [6,15]. Therefore, the objective of this study was to investigate the factors associated with total health-care costs and the occurrence of hospitalization. Knowledge of these factors may help health-care providers intervene to improve patient management and possibly reduce health-care costs in the future.

Methods

Data Source

A retrospective study was conducted by using administrative databases obtained from four Thai government hospitals during October 1, 2002 and September 30, 2003. These data were allowed to be used in this study by hospital administrators. Data included demographic characteristics, medical history of illness, health-care utilizations, and medical costs. Medical cost data were all charges of patients' underpayment methods such as capitation (i.e., social security scheme [SSS] and universal coverage [UC]), fee-for-service (FFS) (i.e., civil servant medical benefits scheme [CSMBS]), and out-of-pocket. The social security office pays a fixed amount of money per year to hospitals for covering health-care benefits of employees who enrolled under SSS. Under CSMBS, the government provides full health-care coverage for government officers and their dependents (e.g., parents, spouse, and up to three children). Regarding the UC, the national health security office pays a fixed amount of money per year to hospitals for covering health-care benefits of patients who enrolled under UC, and patients also pay 30 baht per visit ($US 1 = 35 baht) [16]. Out-of-pocket means that patients pay all health-care costs by themselves. Although patients under capitation payment method did not actually pay for their total charges, their medical charge data were still recorded on hospitals' databases.

Patient Selection

Diabetic patients must have at least one claim with the primary, secondary, or tertiary diagnostic code of diabetes mellitus based on the International Statistics Classification Diagnostics and Health Problem tenth revision (ICD-10 codes = E10-E14).

Statistical Analysis

Data were transformed to a patient-level or cross-sectional data. Univariate and multivariate statistical analyses (i.e., ordinary least square (OLS) regression and logistic regression analyses) were applied. SPSS program version 11.0 was used for statistical analyses. Multiple linear regression analysis and log trans formation were used when a dependent variable was total health-care costs. Total health-care costs were the summation costs of both diabetic and nondiabetic-related resource use (e.g., drugs, medical supplies, laboratory tests, surgeries, hospitalizations, and other health-care services) incurred by patients with diabetes. Logistic regression analysis was applied when the occurrence of hospitalization was a dependent variable. Nevertheless, the occurrence of hospitalization was used as one of independent variables when a dependent variable was total health-care costs. Other independent variables included demographic factors (e.g., age, female), payment methods such as capitation (i.e., SSS and UC), FFS (i.e., CSMBS), and out-of-pocket, hospital characteristics (e.g., teaching hospital), health care and drug utilizations (e.g., outpatient visits and insulin utilization), comorbidities (e.g., hypertension, hyperlipidemia, and cancer), microvascular complications (e.g., retinopathy, nephropathy, and neuropathy), and macrovascular complications (e.g., coronary artery disease [CAD], cardiovascular disease [CVD], and peripheral vascular disease [PVD]). Comorbidities and complications were identified by ICD-10 codes. All variables used in the analysis and the reference categories of dummy variables are presented in Table 1.

Table 1.  Variables used in the analysis
VariablesType (reference category)
  1. CAD, coronary artery disease; CVD, cardiovascular disease; CSMBS, civil servant medical benefits scheme; FFS, fee-for-service; PVD, peripheral vascular disease; SSS, social security scheme; UC, universal coverage.

Dependent variables
 Health-care costsContinuous (baht)
 HospitalizationDummy (yes = 1, no = 0)
Independent variables
 Demographics:
  AgeContinuous (years)
  FemaleDummy (female = 1, male = 0)
 Payment method:
  Capitation (i.e., SSS or UC)Dummy (yes = 1, no = 0)
  FFS (i.e., CSMBS)Dummy (yes = 1, no = 0)
  Out-of-pocketDummy (yes = 1, no = 0)
 Hospital characteristics:
  Teaching hospitalDummy (teaching hospital = 1, nonteaching hospital = 0)
 Health-care utilization:
  Number of outpatient visitsContinuous
  Insulin utilizationDummy (insulin users = 1, noninsulin users = 0)
 Comorbidity:
  HypertensionDummy (yes = 1, no = 0)
  HyperlipidemiaDummy (yes = 1, no = 0)
  CancerDummy (yes = 1, no = 0)
 Microvascular complications:
  RetinopathyDummy (yes = 1, no = 0)
  NephropathyDummy (yes = 1, no = 0)
  NeuropathyDummy (yes = 1, no = 0)
 Macrovascular complications:
  CADDummy (yes = 1, no = 0)
  CVDDummy (yes = 1, no = 0)
  PVDDummy (yes = 1, no = 0)

Results

Table 2 shows the results of the descriptive statistics of the sample. There were 24,051 patients with diabetes with average age of 59 years old (standard deviation [SD] 13.14). Sixty-six percent of patients with diabetes were female and 99% had type II diabetes. In this study, diabetic patients were under capitation (34%) (i.e., SSS [6%] and UC [28%]), FFS (i.e., CSMBS [19%]), and out-of-pocket (47%). Moreover, 61% of patients received their treatment at teaching hospitals. The average annual total health-care cost per person was 19,299 baht or $551 (SD 64,754 baht or $1,850). The median annual total health-care cost per person was 5,658 baht or $162 (Interquartile Range, IQR = 14,209 baht or $406). The annual LOS per person was 2.52 (SD 9.10) days. The average annual number of hospitalizations per person was 0.35 (SD 0.89), and the average annual number of outpatient visits per person was 7.39 (SD 6.20). In this analysis, there were 77% of patients who had only outpatient visits. Only 21% of patients were admitted to the hospitals and the average annual number of hospitalizations per person of these patients was 1.63 (SD 1.26), which was higher than that of total patients (0.35 [SD 0.89]). In addition, 12% of diabetic patients took insulin.

Table 2.  Descriptive statistics of the sample
VariablesStatistical values (N = 24,051)
Demographics:
Average age (years)59 (SD = 13.14)
 Female sex66%
 Type II diabetes99%
Payment method:
 Fee-for-service
  Civil servant medical benefit scheme19%
 Capitation34%
  Universal coverage28%
  Social security scheme6%
 Out-of-pocket47%
Hospital characteristics:
 Number of patients in teaching hospitals61%
 Number of patients in nonteaching hospitals39%
Health-care costs and utilization:
 Average annual cost per person (baht)19,299 baht or $551 (SD = 64,754 baht or $1850)
 Median annual cost per person (baht)5658 baht or $162 (IQR = 14,209 baht or $406)
 Average annual length of stay per person (day)2.52 (SD = 9.10)
 Average annual number of hospitalizations per person0.35 (SD = 0.89)
 Average annual number of outpatient visits per person7.39 (SD = 6.20)
 Number of patients with only outpatient visits77%
 Number of patients admitted to hospitals21%
 Average annual number of hospitalizations per person1.63 (SD = 1.26)
Insulin utilization:
 Number of diabetic patients taking insulin12%
Comorbidity:
 Number of diabetic patients with coronary artery diseases6.15%
 Number of diabetic patients with cardiovascular diseases1.46%
 Number of diabetic patients with peripheral vascular diseases0.59%
 Number of diabetic patients with hyperlipidemia12.79%
 Number of diabetic patients with hypertension33.33%
 Number of diabetic patients with cancer4.10%
Complication:
 Number of diabetic patients with nephropathy1.77%
 Number of diabetic patients with neuropathy3.95%
 Number of diabetic patients with retinopathy8.67%

Tables 3 and 4 shows the results of OLS and logistic regression analyses, respectively. Age (parameter estimates [PE] = 0.006, P < 0.001) or male sex (PE = −0.019, P < 0.002) had a significant impact on an increase in health-care costs. Payment methods (e.g., capitation [PE = 0.083, P < 0.001], FFS [PE = 0.211, P < 0.001], and out-of-pocket [PE = 0.057, P < 0.001]) had a significant positive effect on an increase in health-care costs. In addition, diabetic patients under capitation (PE = 2.163, odds ratio [OR] = 8.69, P < 0.001), FFS (PE = 2.365, OR = 10.64, P < 0.001) were more likely to have higher hospitalizations compared to those paid by out-of-pocket (PE = 1.502, OR = 4.49, P < 0.001). Diabetic patients receiving treatment from a teaching hospital had significantly higher health-care costs (PE = 0.359, P < 0.001), but they were less likely to have hospitalizations (PE = −1.600, OR = 0.20, P < 0.001).

Table 3.  Results of multiple linear regression analysis
Dependent variable = log of total health-care costs
Independent variablesParameter estimatesP-value
  • *

    Statistically significant at P < 0.05.

  • Model significant at P < 0.001; Adjusted R-square = 0.54.

Age0.006<0.001*
Female−0.0190.002*
Capitation0.083<0.001*
Fee-for-service0.211<0.001*
Out-of-pocket0.057<0.001*
Teaching hospital0.359<0.001*
Hospitalization0.615<0.001*
Outpatient visit0.041<0.001*
Insulin utilization0.344<0.001*
Hypertension0.096<0.001*
Hyperlipidemia0.0290.002*
Cancer0.154<0.001*
Nephropathy0.016<0.001*
Neuropathy0.064<0.001*
Retinopathy0.035<0.001*
Coronary artery disease0.141<0.001*
Cardiovascular disease0.0580.024*
Peripheral vascular disease0.213<0.001*
Table 4.  Results of logistic regression analysis
Dependent variable = occurrence of hospitalization (yes = 1)
Independent variablesParameter estimatesOdds ratioP-value
  • *

    Statistically significant at P < 0.05; Model significant at P < 0.001.

  • CAD, coronary artery disease; CVD, cardiovascular disease; FFS, fee-for-service; PVD, peripheral vascular disease.

Female−0.1160.890.006*
Capitation2.1638.69<0.001*
FFS2.36510.64<0.001*
Out-of-pocket1.5024.49<0.001*
Teaching hospital−1.6000.20<0.001*
Outpatient visit−0.0290.97<0.001*
Insulin utilization1.3083.70<0.001*
Hypertension0.7512.12<0.001*
Hyperlipidemia0.0271.030.687
Cancer1.5254.60<0.001*
Nephropathy2.84517.21<0.001*
Neuropathy1.7905.99<0.001*
Retinopathy0.5231.69<0.001*
CAD1.9647.13<0.001*
CVD0.3241.380.024
PVD1.2333.43<0.001*

Patients admitted to hospital (PE = 0.615, P < 0.001) were significantly associated with an increase in health-care costs. Patients with more outpatient visits significantly consumed higher health-care costs (PE = 0.041, P < 0.001). Insulin users significantly had higher health-care costs (PE = 0.344, P < 0.001) and were about four times more likely to have hospitalizations compared to noninsulin users (PE = 1.308, OR = 3.70, P < 0.001).

Diabetic patients with comorbidities (e.g., hypertension [PE = 0.096, P < 0.001]), hyperlipidemia (PE = 0.029, P = 0.002), and cancer (PE = 0.154, P < 0.001)] had significantly higher health-care costs than those without comorbidities. In addition, diabetic patients with hypertension (PE = 0.751, OR = 2.12, P < 0.001) or cancer (PE = 1.525, OR = 4.60, P < 0.001) also were about two or four times more likely to hospitalize compared to those without hypertension or cancer, respectively. Nevertheless, there was no statistical significant association between an increase in risk of hospitalization and having hyperlipidemia (PE = 0.027, OR = 1.03, P < 0.687). Furthermore, patients with microvascular complications (e.g., nephropathy [PE = 0.016, P < 0.001]), neuropathy (PE = 0.064, P < 0.001), and retinopathy (PE = 0.035, P = 0.001)] had a positive impact on health-care costs. Especially, diabetic patients with nephropathy (PE = 2.845, OR = 17.21, P < 0.001), neuropathy (PE = 1.790, OR = 5.99, P < 0.001), or retinopathy (PE = 0.524, OR = 1.69, P < 0.001) were about 18, 6, or 2 times more likely to have hospitalizations than those without microvascular complications, respectively. Diabetic patients with macrovascular complications (e.g., CAD [PE = 0.141, P < 0.001]), CVD (PE = 0.058, P = 0024), and PVD (PE = 0.213, P < 0.001)] were positively associated with higher health-care costs. In addition, diabetic patients with CAD (PE = 1.964, OR = 7.13, P < 0.001), CVD (PE = 0.325, OR = 1.38, P < 0.001), or PVD (PE = 1.223, OR = 3.39, P < 0.001) were 7, 1, or 3 times more likely to hospitalize compared to those without CAD, CVD, or PVD, respectively. Multiple linear and logistic regression models were significant (P < 0.001) and the adjusted R2 was 54%, meaning that all significant factors in the model were able to explain 54% of the variation in total health-care costs.

Discussion

The results of this study suggested that demographic factors, payment methods, hospital characteristics, health-care utilizations, comorbidities, and complications were significantly associated with higher health-care costs and hospitalizations. All previous studies supported the finding that older age patients had higher health-care costs and hospitalizations [8–15]. Moreover, male patients were more likely to have higher costs and hospitalizations than female patients. Krop et al. [8] found the same result, whereas the study of Bhattacharyya [11] showed that female patients were more likely to consume higher health-care costs and utilization.

Regarding the payment methods, particularly patients under FFS (i.e., CSMBS) or capitation (i.e., SSS or UC) significantly had higher health-care costs and hospitalizations compared to those paid by out-of-pocket. In contrast to previous studies, there was no impact of payment method factor on health-care costs. Whether patients were enrolled in the FFS or capitation systems did not have any significant effect on the total direct costs of diabetes [10,12]. In addition, there was no statistically significant difference in patients under FFS plan on hospitalization use [11]. In this study, it could be explained that because all health-care costs of patients under FFS were covered by the government and patients under capitation would pay only some amount of copay for their health-care costs, these patients could easily acquire their treatments as much as they needed and would not be worried about the affordability of health-care expenses. Therefore, they tended to consume higher health-care costs and hospitalizations. This could suggest that patient's eligible benefits could be an important indicator of health-care cost drivers in patients with diabetes in Thailand. Most patients under SSS were the working-age adults who were likely to be healthier than the patients under UC, so that they tended to consume less health-care costs and hospitalizations.

For hospital characteristic factor (e.g., teaching hospitals), patients receiving treatment from teaching hospitals significantly consumed higher health-care costs but had less probability of hospitalization. This could explain that most patients with higher disease severity from nonteaching hospitals were usually referred to a teaching hospital. These patients mostly had only outpatient visits and might not be able to admit to a teaching hospital due to the lack of space. The results reveal that more patients receiving treatment at a teaching hospital had only outpatient visits (86%) compared to those receiving treatment at nonteaching hospitals (61%). In addition, there were fewer patients admitted to teaching hospital (11%) compared to those admitted to nonteaching hospitals (37%).

The results show that the more health-care utilizations (e.g., hospitalization, outpatient visit, and insulin utilization), the higher health-care costs the patients significantly had. Moreover, diabetic patients taking insulin had significantly higher risk of hospitalization. Similar results were also found in the studies of Bhattacharyya [11] and Guo et al. [14].

Furthermore, diabetic patients with comorbidities (e.g., hypertension, hyperlipidemia, and cancer) had significantly higher health-care costs, and diabetic patients with hypertension or cancer tended to have higher hospitalizations. Patients with diabetes and microvascular complications (e.g., nephropathy, neuropathy, and retinopathy) had significantly higher health-care costs and hospitalizations. Diabetic patients with macrovascular complications (e.g., CAD, CVD, and PVD) had significantly higher health-care costs and hospitalizations. Similar to the studies of Bhattacharyya [11] and Bhattacharyya and Else [12], diabetic complications (e.g., retinopathy, nephropathy, and neuropathy) and comorbidities (e.g., hypertension, hyperlipidemia, CAD, and CVD) also had a significant positive impact on health-care costs and hospitalizations.

Two limitations have been addressed in this study. First, the administrative claims data used might be limited. In Thailand, there has been no standardized claims data collection system and standardized data coding excluding ICD-10 codes across hospitals yet, so that different hospitals have different types of claims data collected and data coding. This study used the claims data obtained from four public hospitals and combined into one data set, therefore, unmatched variables were not able to be used for the analysis. Some different coding of administrative claims data would not allow us to identify which type of health-care cost was either diabetic or nondiabetic-related treatment. Thus, in this study, all health-care costs consumed by patients with diabetes were used instead of the costs related to diabetic-related treatment only. Last, like any other retrospective claims data analysis, clinical information such as blood glucose level and other laboratory values would have been highly associated with health-care costs and hospitalizations. Without these clinical measures, assessing the perfect association between factors and health-care costs and hospitalizations might not be possible. Nevertheless, the finding may still be useful information for health-care providers and health policymakers because significant factors in this analysis were able to explain 54% of the variation in total health-care costs.

Based on the results of this study, it is suggested that health-care providers and health policymakers may need to focus on the factors associated with an increase in health-care costs and hospitalizations, such as patients with older age, male sex, comorbidities, complications, patients under capitation or FFS system, and patients taking insulin. Health-care providers may set up the interventions such as diabetic patient counseling, pharmaceutical care, or disease management to delay the progression of comorbidities or complications that diabetic patients may possibly have in the future [17,18]. Although patients under capitation or FFS system have significantly higher health-care costs and hospitalization, these patients may not be at risk. This factor signals the eligible benefits rather than the potential prognostic factors of health-care costs and utilizations. This may relatively indicate the issue of inequity in health care rather than disease severity. It may be used as the information for health policymakers to solve the inequity problem. An investigation of factors associated with health-care costs and hospitalizations may help health-care providers and administrators intervene to improve patient management and possibly reduce health-care costs in the future.

Acknowledgment

This study is supported by a grant from the Thailand Research Fund. We would like to give particular thanks to the Department of Pharmacy, Mahidol University and the Health Intervention and Technology Assessment Program (HITAP) supported by the Thai Health Foundation, the National Health System Research Institute (HSRI) and the Bureau of Health Policy and Strategy, Ministry of Public Health.

Ancillary