1732 Diabetes Care Volume 43, August 2020

Hsuan-Ying Chen,1 Shihchen Kuo,2

Health Care Costs Associated With Pei-Fang Su,3 Jin-Shang Wu,4,5,6 and
Huang-Tz Ou1,7,8
Macrovascular, Microvascular, and
Metabolic Complications of Type 2
Diabetes Across Time: Estimates
From a Population-Based Cohort
of More Than 0.8 Million
EPIDEMIOLOGY/HEALTH SERVICES RESEARCH

Individuals With Up to 15 Years

of Follow-up
Diabetes Care 2020;43:1732–1740 | https://doi.org/10.2337/dc20-0072

OBJECTIVE
Developing country-specific unit-cost catalogs is a key area for advancing economic
research to improve medical and policy decisions. However, little is known about 1
Institute of Clinical Pharmacy and Pharmaceutical
how health care costs vary by type 2 diabetes (T2D) complications across time in Sciences, College of Medicine, National Cheng
Asian countries. We sought to quantify the economic burden of various T2D Kung University, Tainan, Taiwan
2
Division of Metabolism, Endocrinology and Di-
complications in Taiwan. abetes, Department of Internal Medicine, Uni-
versity of Michigan Medical School, Ann Arbor,
RESEARCH DESIGN AND METHODS MI
3
A nationwide, population-based, longitudinal study was conducted to analyze Department of Statistics, College of Manage-
802,429 adults with newly diagnosed T2D identified during 1999–2010 and followed ment, National Cheng Kung University, Tainan,
Taiwan
up until death or 31 December 2013. Annual health care costs associated with T2D 4
Department of Family Medicine, College of Med-
complications were estimated, with multivariable generalized estimating equation icine, National Cheng Kung University, Tainan,
models adjusted for individual characteristics. Taiwan
5
Department of Family Medicine, National Cheng
RESULTS Kung University Hospital, Tainan, Taiwan
6
Division of Family Medicine, National Cheng
The mean annual health care cost was $281 and $298 (2017 U.S. dollars) for a male Kung University Hospital, Dou-Liu Branch, Douliu,
and female, respectively, diagnosed with T2D at age <50 years, with diabetes Taiwan
7
duration of <5 years, and without comorbidities, antidiabetic treatments, and Department of Pharmacy, College of Medicine,
National Cheng Kung University, Tainan, Taiwan
complications. Depression was the costliest comorbidity, increasing costs by 64– 8
Department of Pharmacy, National Cheng Kung
82%. Antidiabetic treatments increased costs by 72–126%. For nonfatal compli- University Hospital, Tainan, Taiwan
cations, costs increased from 36% (retinopathy) to 202% (stroke) in the event year Corresponding author: Huang-Tz Ou, huangtz@
and from 13% (retinopathy or neuropathy) to 49% (heart failure) in subsequent mail.ncku.edu.tw
years. Costs for the five leading costly nonfatal subtype complications increased by Received 9 January 2020 and accepted 20 April
201–599% (end-stage renal disease with dialysis), 37–376% (hemorrhagic/ischemic 2020
stroke), and 13–279% (upper-/lower-extremity amputation). For fatal complica- This article contains supplementary material online
tions, costs increased by 1,784–2,001% and 1,285–1,584% for cardiovascular and at https://doi.org/10.2337/figshare.12173988.
other-cause deaths, respectively. H.-Y.C. and S.K. have equal first authorship.
© 2020 by the American Diabetes Association.
CONCLUSIONS Readers may use this article as long as the work is
properly cited, the use is educational and not for
The cost estimates from this study are crucial for parameterizing diabetes economic
profit, and the work is not altered. More infor-
simulation models to quantify the economic impact of clinical outcomes and mation is available at https://www.diabetesjournals
determine cost-effective interventions. .org/content/license.
care.diabetesjournals.org Chen and Associates 1733

Diabetes is a global issue that not only such as comorbidities and antidiabetic Supplementary Fig. 1. A total of 802,429
causes a serious health burden but also treatments (12,13,15,16,18,19). patients were included in this study.
imposes a significant economic impact Against this background, the current
for individuals, health care systems, and study used Taiwan’s NHI claims data Identification of Study Variables
countries (1). In 2017, ;2 million people from a nationwide, population-based co- Diabetes-Related Complications,
in Taiwan, or 10.9% of adults aged 20–79 hort of .0.8 million adults with T2D for Comorbidities, and Death

years, had diabetes, which is higher than up to 15 years of follow-up to compre- Each patient was followed up from T2D
the global average of 8.8% (1). Diabetes is hensively and systematically estimate diagnosis to death or the end of 2013,
among the top five causes of death in the health care costs associated with a whichever came first. Diabetes-related
Taiwan (2). In 2017, diabetes was the wide range of diabetes-related macro- complications of interest included macro-
third most expensive disease, with an vascular, microvascular, and metabolic and microvascular complications and
annual health care expenditure of more complications in the event year and sub- metabolic complications. We also identi-
than US$850 million, accounting for ;4% sequent follow-up years, with adjustment fied the subtype of diseases within each
of the total national health care expen- for various patient characteristics. These major complication (e.g., stroke included
diture reimbursed by the Taiwan Na- estimates are useful not only for decision ischemic or hemorrhagic stroke or tran-
tional Health Insurance (NHI) program makers to quantify and differentiate the sient ischemic attack). Comorbidities in-
(3). Type 2 diabetes (T2D), which accounts economic burden of T2D but also for cluded hypertension, hyperlipidemia, liver
for .99% of the diabetes population in simulation modelers to parameterize disease, cancer, and depression. Compli-
Taiwan, contributed to the majority of the the cost data in diabetes economic sim- cations and comorbidities were identi-
health and economic burden attributable ulation models (5–8). fied using ICD-9-CM codes. Macrovascular
to diabetes (4). and metabolic complications were iden-
Considering the substantial economic RESEARCH DESIGN AND METHODS tified using the emergency department/
burden of T2D, the parameterization of inpatient files of the NHIRD, and micro-
This study was approved by the institu-
cost components associated with diabe- vascular complications and comorbidities
tional review board of National Cheng
tes progression and management is im- were based on the outpatient and emer-
Kung University Hospital (B-EX-103-015).
portant for quantifying the cost impact of gency department/inpatient files of the
T2D and serving as the cost data input NHIRD. The coding details for complica-
Data Source
in economic simulation modeling studies tions and comorbidities are provided in
This was a retrospective study that used
(5–8). The results of cost-effectiveness Supplementary Tables 1 and 2. The accu-
the Longitudinal Cohort of Diabetes Pa-
studies vary with respect to the sources racy of identifying disease conditions on
tients (LHDB) data set for 1996–2013
from which unit costs are derived, which the basis of ICD-9-CM coding in the NHIRD
from Taiwan’s NHI Research Database
compromises comparability. The devel- has been validated and published else-
(NHIRD). The NHIRD is a population-
opment of country-specific unit-cost cat- where (22). For example, the positive
based database derived from health
alogs has been recognized by the Second predictive value for the diagnosis of myo-
claims data of Taiwan’s NHI program,
Panel on Cost-Effectiveness in Health and cardial infarction, ischemic stroke, and
which is a mandatory-enrollment, single-
heart failure is as high as 93%, 94%,
Medicine (9–11) as a key area for future payment system that covers .99% of
exploration in cost-effectiveness research and 98%, respectively. In addition, death
Taiwan’s population (20). The LHDB is a
because it is valuable to standardize cost events comprised a composite of fatal
nationally representative data set for the
inputs, reduce the cost data collection cardiovascular diseases (CVDs) (including
population with diabetes in Taiwan and
burden, and facilitate the comparison of death as a result of stroke, myocardial
has been validated for research purposes
studies within and across countries. A few infarction, ischemic heart disease, or heart
(21). This data set consists of longitudinal
studies have analyzed the medical costs failure) and other-cause death. The op-
health data from a random sample of
of diabetes-related complications in Asian erational definitions of confirming mor-
120,000 deidentified patients with in-
countries (12–19), with only one study in tality status using the inpatient files of the
cident diabetes from each calendar year
Taiwan (12). Nevertheless, none of these NHIRD have been validated (23). We
since 1999 who can be tracked back to
studies fulfill all the requirements for the further used the disenrollment records
1 January 1996 and followed up to
complete cost parameterization in a dia- from the NHIRD registration files of ben-
31 December 2013.
betes economic simulation model. eficiaries to confirm mortality (24).
Specifically, they used restrictive and Study Population Health Care Costs
underrepresentative study cohorts Study patients extracted from the LHDB This economic analysis was conducted
(12–17), considered a limited number were newly diagnosed with T2D (ICD-9, from the perspective of the health care
of complications or just a single com- Clinical Modification [ICD-9-CM], codes sector. The health care costs comprised
plication (13,14,16–19), excluded fatal 250.x0 and 250.x2, where x 5 0–9) at two aspects of medical costs from the
complications (15–17,19), had short- age $18 years during 1999–2010. We formal health care sector: costs paid by the
term follow-up or did not even ac- excluded those with any macrovascular third-party payer (reimbursement costs by
count for the temporal distribution or microvascular diseases in the year Taiwan’s NHI program) and out-of-pocket
and change of costs (12,13,15–17,19), before or at T2D diagnosis, undefined costs paid by individual beneficiaries (co-
did not distinguish between type 1 di- sex, or zero medical expenses through- payments by patients). Cost components
abetes (T1D) and T2D (13,15), or did not out the entire study period. Selection of in the NHIRD claims reimbursed by the NHI
consider other potential cost drivers, the study cohort is described in detail in program included the fees of diagnosis,
1734 Health Care Costs of Type 2 Diabetes in Taiwan Diabetes Care Volume 43, August 2020

treatments (i.e., examinations, proce- Cost Multipliers of Annual Health Care Costs subtypes of complications (sensitivity
dures, special materials), pharmaceuti- Associated With Patient Demographics, analysis 1). Second, we considered the
cal services, and medications. Costs were Comorbidities, Complications, Antidiabetic status of comorbidities measured in the
adjusted on the basis of the medical con- Treatments, and Death year before or at T2D diagnosis and
sumer price index and are expressed in Our primary interest was to quantify treated them as the time-invariant var-
2017 U.S. dollars. the impact associated with a wide range iables in the analysis (sensitivity analysis
of patient demographic and clinical char- 2). Third, we modeled the annual MPR of
Exposure to Antidiabetic Drugs
acteristics (i.e., age, sex, diabetes duration, antidiabetic drugs as a time-varying vari-
Medication utilization was identified us-
comorbidities, complications, antidia- able in the analysis (sensitivity analysis 3).
ing drug identification numbers, linked to
betic treatments) on health care costs. The annual MPR during the follow-up
the World Health Organization Anatom-
We applied the generalized estimating was calculated by summing days supply
ical Therapeutic Chemical Classification
equation (GEE) model with a log-link from the first to the last antidiabetic pre-
System, to define active ingredients of
function to estimate the log-transformed scription in a year divided by 365 days (25).
antidiabetic drugs. Patients were classi-
annual health care costs as a function of A cutoff point for MPR of 0.8 (25) was
fied into four groups on the basis of the
status of their antidiabetic drug expo-
patient demographics, comorbidities, applied to determine low (MPR ,0.8) and
sure: 1) none (diet, exercise control), 2)
complications, and antidiabetic treat- high (MPR $0.8) exposure/adherence to
ments. The coefficients and 95% CIs antidiabetic drugs. SAS 9.4 software (SAS
only use of oral drugs, 3) only use of
from the GEE model were then back- Institute, Cary, NC) was used for all statis-
injectable drugs, or 4) combined use of
transformed to the ordinal scale using tical analyses.
oral and injectable drugs. Further, the
an exponential function to form the cost
annual medication possession ratio (MPR)
multipliers. The GEE accounts for the RESULTS
for antidiabetic treatments was measured
nonindependence of yearly repeated- Table 1 describes the baseline patient
to quantify the amount of antidiabetic
measured cost data within each subject characteristics. The study population of
drug exposure.
during the follow-up. In the multivari- 802,429 individuals had a mean age of
able modeling analysis, unless indicated 54.2 years and 46.3% females. Prevalent
Statistical Analyses otherwise, patient demographics (age at comorbidities were present from 2.8 to
Statistical analyses were initiated in early T2D diagnosis and sex) and death status 37.0% of the study population; hyper-
2018 and finalized in mid-2019. Descrip- were the time-invariant variables, whereas tension, hyperlipidemia, cancer, and de-
tive analyses were performed for patient diabetes duration, comorbidities, diabetes- pression were more frequent in females.
characteristics. Means and SDs were related complications, and antidiabetic Most individuals (55.5%) were treated
used for continuous variables, and per- drug exposure were treated as time- with oral antidiabetic drugs only.
centages and frequencies were applied varying variables, which were measured Supplementary Table 3 shows the
for dichotomized and categorical varia- each year during the follow-up. We event rates of comorbidities, incidence
bles. The event rates for diabetes-related examined the collinearity among the rates of diabetes-related complications,
complications and comorbidities during variables in the GEE model by using a antidiabetic drug exposure rates, and
the follow-up are presented as the num- variance inflation factor; no significant death rates during the follow-up. The
ber of events per 100 person-years. The collinearity problems (supported by var- average follow-up duration was 6.9 years.
antidiabetic treatments during the fol- iance inflation factor ,10) were found. Hyperlipidemia and hypertension were
low-up are presented as the number of In the GEE models, the base case was the two most common comorbid con-
antidiabetic drug exposures per 100 per- determined as the annual health care ditions. Nonfatal microvascular compli-
son-years. The death rate is presented as costs for a male diagnosed with T2D at cations were the most frequent, followed
the number of events per 100 people. age ,50 years, with a diabetes duration by nonfatal macrovascular and metabolic
Crude Health Care Costs for Diabetes- of ,5 years, and without comorbidities, complications. About 57% of patients
Related Complications complications, and antidiabetic treat- had exposure to antidiabetic drugs dur-
The costs of diabetes-related complica- ments. The GEE models were multipli- ing the follow-up. A total of 55,193 (6.9%)
tions were estimated as event-year costs cative. That is, to determine the relative patients died during the follow-up pe-
and annual state-year costs. Event-year impact on health care costs for a given riod, with 0.9% as a result of fatal CVD and
costs were the costs associated with patient with characteristics other than 6.0% as a result of other causes.
medical management of an acute care those of the base case patient, we can Table 2 presents the crude health care
episode (initial management in emer- multiply the health care costs for the base costs of diabetes-related complications
gency department, inpatient, or out- case patient (baseline cost) by the prod- in the event year and subsequent years.
patient care settings) and any subsequent uct of the cost multipliers (or multiplicative The mean time from T2D diagnosis to the
care provided within the first year fol- factors) calculated for each demographic development of the first complication
lowing the acute episode. State-year and clinical condition applied to that event ranged from 4.0 to 7.3 years. The
costs reflected the annual resource patient. average event-year cost of complications
use required beyond the first year for In the primary analysis, we considered was from $1,772 for proteinuria to
the ongoing medical management 14 fatal or nonfatal complications in the $17,652 for fatal ischemic heart disease.
while a given health state/chronic event multivariable model. A series of sensi- The average annual state-year cost
is present for the remainder of the tivity analyses were conducted. First, we of complications ranged from $1,798
patient’s life. examined a total of 29 fatal or nonfatal for other retinopathy to $16,404 for
care.diabetesjournals.org Chen and Associates 1735

Table 1—Patient characteristics in the year before or at T2D diagnosis stratified by results when we used the MPR to
sex examine the cost impact of adherence
Characteristic Total population Male Female to antidiabetic treatments. Higher ad-
herence to oral or injectable treat-
Cohort size, n (%) 802,429 (100.0) 430,799 (53.7) 371,630 (46.3)
ments was associated with increased
Age at T2D diagnosis (years), mean (SD) 54.16 (13.69) 53.28 (13.46) 55.18 (13.89)
costs.
Comorbidity, %
Hypertension 36.97 34.92 39.35
Hyperlipidemia 30.85 29.75 32.13
Liver disease 19.47 21.58 17.03
CONCLUSIONS
Cancer 10.64 8.05 13.64 This study provides a comprehensive
Depression 2.80 2.15 3.56 view of the real-world economic impact
Antidiabetic drug exposure, % associated with each of the 29 incident
None 39.95 36.60 43.84 fatal or nonfatal diabetes-related com-
Only oral 55.45 58.06 52.44 plications, independently of the effects
Only injectable 0.73 0.77 0.69
of patient demographics, comorbidities,
Oral 1 injectable 3.86 4.58 3.09
antidiabetic treatments, and other-cause
Injectable antidiabetic drugs included insulin and glucagon-like peptide 1 receptor agonists. death among patients with T2D in Tai-
wan. We found that the health care costs
of T2D varied by patient demograph-
end-stage renal disease (ESRD) with di- product of the cost multipliers for ics, clinical characteristics, and cause of
alysis. Supplementary Fig. 2 shows the each characteristic of that patient. For death. Fatal complications generally had
annual health care costs before, during, instance, the mean annual health care a higher economic impact than nonfatal
and after the occurrence of 10 nonfatal cost for a patient aged 70 years is 1.68 complications. Mortality as a result of
major macrovascular, microvascular, and times that of a patient aged ,50 years, a composite of CVD events (fatal CVD)
metabolic complications. The event- that for a patient with T2D duration of was the costliest event, which increased
year costs were relatively high for mac- 5 years is 0.85 times that of a patient the event-year health care costs by
rovascular complications, and the cost with a T2D duration of ,5 years, that 1,784–2,001%, followed by other-cause
decline in the subsequent years was for a female is 1.06 times that of a male, death (1,285–1,584%), nonfatal stroke
relatively slow for microvascular com- that for a patient with a comorbidity is (195–202%), and myocardial infarction
plications. Supplementary Figs. 3–14 0.98–1.81 times that of a patient without (145–154%). When the subtypes of
illustrate the crude annual event- and comorbidities, and that for a patient complications were assessed, among
state-year health care costs of diabetes- treated with oral antidiabetic drugs is fatal complications, other-cause death
related complications (as a composite 1.78 times that of a patient without oral was the costliest event, which increased
outcome or by subtype of diseases) and treatments. Having nonfatal complica- the event-year health care costs by
other-cause death. For nonfatal compli- tions increases the cost by 1.37–2.95 1,285%, followed by ischemic or hem-
cations, the event-year cost was the times in the event year and 1.13–1.49 orrhagic stroke (613–728%) and heart
highest for lower-extremity amputa- times in subsequent years. Patients who failure (447%). Among nonfatal compli-
tion ($12,989) followed by hemorrhagic died as a result of CVD have an increased cations, ischemic or hemorrhagic stroke,
stroke ($12,304) and myocardial infarc- cost in the year of death by 20.49 times. ESRD with dialysis, upper- or lower-extremity
tion ($11,103). The annual state-year For a female patient aged 50 years, with a amputation, and coronary heart disease
cost for most of the complications in- T2D duration of 5 years, treated with (e.g., ischemic heart disease, myocardial
creased gradually over time. ESRD with oral antidiabetic drugs, and developing infarction, heart failure) were among the
chronic dialysis or kidney transplant was stroke, the estimated mean annual most significant cost drivers, increasing
the costliest followed by lower-extremity health care cost in the event year is the event-year and annual state-year
amputation and heart failure. For death $1,583 ($281.21 3 1.06 3 1.19 3 health care costs by 103–376% and
cases, the event-year cost was the high- 0.85 3 1.78 3 2.95). 13–599%, respectively.
est for deaths associated with ischemic Table 4 shows the results when we Some of cost estimates in our study
heart disease. considered subtypes of complications are different from those reported in
Table 3 shows the cost multipliers from in the GEE analysis. Hemorrhagic or ische- the four most relevant and recently
the primary analysis. The baseline cost of mic stroke, upper- or lower-extremity published studies from Germany (26),
$281 represents the mean annual health amputation, ESRD with dialysis, ischemic Taiwan (12), Hong Kong (15), and the
care cost for the base case patient, who heart disease, and proliferative reti- U.K. (27). The discrepancy is likely due
is a male diagnosed with T2D at age ,50 nopathy are the leading cost drivers to differences in the health care set-
years, with diabetes duration of ,5 for the event- and state-year costs. ting or system (e.g., a mixture of public
years, and without complications, co- Supplementary Table 4 shows the results and private providers [15] vs. universal
morbidities, and antidiabetic treatments. when we considered the baseline comor- coverage [12,26]), definition of com-
If a patient has any characteristics other bidity status in the GEE analysis. Com- plication events (e.g., mixed prevalent
than those for the base case, the mean pared with the primary analysis results, and incident events [12,15,26,27]), in-
annual health care cost is estimated by depression is still the costliest comorbid- clusion of cost components for estimat-
multiplying the baseline cost by the ity. Supplementary Table 5 presents the ing health care costs (e.g., health care
1736 Health Care Costs of Type 2 Diabetes in Taiwan Diabetes Care Volume 43, August 2020

Table 2—Crude event-year and annual state-year health care costs of diabetes-related complications
Patients with Time to the first event (years), Event-year cost (US$), Average annual state-year cost
Event event, n mean (SD) mean (SD) (US$), mean (SD)
Stroke 59,543 5.21 (3.55) 7,166.06 (12,868.20) 3,572.20 (7,789.74)
Ischemic stroke 43,663 5.36 (3.59) 6,790.06 (12,032.49) 3,654.47 (7,738.86)
Hemorrhagic stroke 12,162 5.10 (3.56) 12,304.31 (18,946.12) 4,181.91 (10,159.10)
Transient ischemic attack 13,137 5.59 (3.59) 4,559.47 (7,391.18) 3,215.62 (6,644.44)
Myocardial infarction 14,942 5.76 (3.63) 11,103.10 (12,788.41) 3,968.37 (7,724.08)
Ischemic heart disease 62,218 5.35 (3.53) 7,186.14 (10,786.39) 3,490.74 (7,054.18)
Heart failure 29,077 5.94 (3.73) 10,070.83 (15,839.95) 5,691.68 (10,222.44)
Arteriosclerotic CVD 388 5.22 (3.45) 7,750.24 (12,037.88) 3,972.51 (8,470.16)
Arrhythmia 32,617 5.57 (3.66) 8,101.12 (15,216.45) 3,586.78 (8,284.03)
Nephropathy 228,983 4.73 (3.49) 3,158.42 (7,987.84) 2,544.21 (5,623.36)
CKD with or without short-
term dialysis 47,441 6.31 (3.90) 3,894.18 (8,336.73) 2,349.70 (4,940.97)
ESRD with dialysis 8,990 6.52 (3.56) 9,846.21 (12,071.96) 16,403.68 (11,295.76)
Proteinuria 36,146 5.16 (3.56) 1,772.43 (4,274.87) 2,193.31 (5,188.79)
Kidney transplant 95 7.32 (3.11) 10,866.27 (4,427.81) 11,123.23 (6,402.39)
Other nephropathy 142,307 4.78 (3.53) 3,159.59 (8,395.77) 1,868.30 (4,276.29)
Retinopathy 296,940 4.29 (3.22) 1,972.96 (3,763.25) 2,042.24 (4,343.47)
Proliferative retinopathy 26,569 6.22 (3.82) 3,548.15 (5,082.46) 4,687.43 (7,841.06)
Blindness 1,342 5.01 (3.44) 4,394.43 (7,143.49) 3,025.47 (6,211.11)
Other retinopathy 269,144 4.27 (3.20) 1,885.07 (3,696.92) 1,797.51 (3,809.84)
Neuropathy 249,533 4.10 (3.10) 2,212.41 (4,341.18) 2,072.20 (4,510.40)
Peripheral vascular disease 130,957 4.65 (3.36) 3,444.45 (7,594.15) 2,490.97 (5,295.28)
Diabetes foot ulcer 21,354 5.32 (3.60) 4,701.95 (8,150.95) 3,284.20 (6,700.11)
Upper-extremity amputation 601 4.93 (3.44) 8,335.78 (12,434.98) 3,067.11 (6,019.06)
Lower-extremity amputation 4,441 6.03 (3.81) 12,988.73 (13,761.50) 5,777.34 (10,361.16)
Other peripheral vascular
disease 106,991 4.66 (3.36) 3,119.31 (7,299.63) 2,262.14 (4,796.40)
Hospitalized diabetic
ketoacidosis 17,049 4.01 (3.48) 5,401.15 (9,323.16) NA
Hospitalized HHS 13,388 4.51 (3.81) 7,557.12 (13,550.72) NA
Hospitalized hypoglycemia 44,808 6.14 (3.91) 6,881.47 (11,491.43) NA
All-cause death 55,193 5.82 (3.82) 13,079.68 (13,938.41) NA
Fatal CVD 7,488 5.49 (3.71) 14,542.10 (15,581.51) NA
Fatal stroke 4,034 5.94 (3.74) 14,329.33 (15,251.66) NA
Fatal myocardial infarction 2,220 6.79 (3.80) 14,963.03 (16,070.27) NA
Fatal ischemic heart disease 2,255 6.04 (3.83) 17,651.55 (17,798.79) NA
Fatal heart failure 3,722 6.26 (3.81) 16,883.87 (17,182.75) NA
Other-cause death 47,705 5.87 (3.83) 12,850.14 (13,648.53) NA
CKD, chronic kidney disease; HHS, hyperosmolar hyperglycemic syndrome; NA, not applicable.

sector costs [12,26] vs. mixed public and ESRD with dialysis). Despite these differ- 82%. This finding supports the recent
private medical costs [15,27]), inclusion ences, however, our study and previous guidelines emphasizing the importance
of covariates in the modeling analy- studies (12,15,26,27) consistently revealed of assessing, preventing, and managing
sis (e.g., without considering comorbid- that fatal CVD as well as nonfatal stroke, coexistent mental illnesses in patients with
ities [12,15,26,27]), and study design ESRD, amputation, and coronary heart diabetes (28). Furthermore, consistent with
and analytical approaches (e.g., cross- disease have a high impact on the health previous studies (29,30), we found that the
sectional [12,15] vs. longitudinal [15,26,27]). care costs of T2D. These results highlight use of antidiabetic treatments contrib-
Even with similar health care systems, the importance of early effective preven- utes to the cost burden, increasing health
studies from different countries may yield tion of vascular complications and their care costs by 76–86%. Beyond this, we
different results. For example, in a pre- associated deaths to restrain the consider- further showed that higher use/exposure
vious study from Germany (26), which has able economic burden incurred in the T2D of antidiabetic treatments (MPR $0.8)
the statutory national health insurance population. yielded a higher cost impact compared
program as Taiwan, the top three costly Moreover, we found that the eco- with that for lower use/exposure (MPR
complications are ESRD, fatal ischemic nomic impact of depression in the T2D ,0.8). This may suggest a trade-off
heart disease, and amputation, which population cannot be neglected because when enhancing an antidiabetic ther-
are different from what is found in this it was the costliest comorbidity, increas- apy; more exposure/adherence to an-
study (i.e., other-cause death, fatal stroke, ing the annual health care costs by 64– tidiabetic treatments increases upfront
care.diabetesjournals.org Chen and Associates 1737

Table 3—Cost multipliers and associated 95% CIs for patient demographics, Compared with existing studies on
comorbidities, diabetes-related complications, antidiabetic drug exposure, and the economic burden of T2D, this study
death: primary analysis has several strengths. First, by using
Variable Multiplier 95% CI the individual-level data from the NHI
Baseline annual health care cost (2017 U.S. $), mean (95% CI) 281.21 279.25 283.21
claims covering all medical utilization and
Age at T2D diagnosis (years) (ref. ,50)
costs, this study used the largest nation-
50–59 1.19 1.19 1.20 ally representative T2D study cohort with
60–69 1.43 1.42 1.44 the longest follow-up period to estimate
$70 1.68 1.67 1.69 the health care costs associated with a
Diabetes duration (years) (ref. 1–4) wide range of diabetes-related compli-
$5 0.85 0.84 0.85 cations. This study addresses several lim-
Female (ref. male) 1.06 1.06 1.06 itations in previous studies, including
Comorbidity (ref. none) that the analyses were based on restric-
Hypertension 1.21 1.21 1.22 tive/underrepresentative or ancient study
Hyperlipidemia 1.07 1.06 1.07 cohorts (13–19,27,29,30,33); included
Liver disease 0.98 0.98 0.98
only a limited number of study patients
Cancer 1.04 1.04 1.05
Depression 1.81 1.80 1.82
for severe/chronic complication events
Oral antidiabetic drug exposure (ref. none) 1.78 1.78 1.79
(12,19,26); assessed only a few diabetes-
related complications, only nonfatal com-
Injectable antidiabetic drug exposure (ref. none) 1.83 1.82 1.84
plications, or just a single complication
Complication (event year) (ref. none)
Stroke 2.95 2.92 2.98
(12–19,27,29,30,33); or were cross-sectional
Myocardial infarction 2.53 2.48 2.58 or conducted with only a short-term
Ischemic heart disease 2.51 2.49 2.53 follow-up (12,15–19,29,30). Importantly,
Heart failure 2.24 2.21 2.27 the large sample size in our study ensured
Arteriosclerotic CVD 1.80 1.62 2.01 statistical power for assessing relatively
Arrhythmia 2.04 2.01 2.06 rare complication events and differenti-
Nephropathy 1.49 1.48 1.50
ating them into subtypes (e.g., hemor-
Retinopathy 1.37 1.36 1.37
Neuropathy 1.41 1.40 1.41 rhagic stroke, blindness, amputation,
Peripheral vascular disease 1.55 1.54 1.56 deaths). The long-term follow-up al-
Hospitalized diabetic ketoacidosis 1.90 1.87 1.93 lowed us to measure chronic events
Hospitalized HHS 1.72 1.69 1.76 (e.g., ESRD with dialysis or kidney trans-
Hospitalized hypoglycemia 1.90 1.88 1.92 plant, death), and the longitudinal nature
Complication (state year) (ref. none) of the data allowed us to explore the
Stroke 1.44 1.43 1.45 costs in both the event year and sub-
Myocardial infarction 1.18 1.16 1.20
sequent years. The differences between
Ischemic heart disease 1.32 1.31 1.33
Heart failure 1.49 1.47 1.51 costs in the event year and those in
Arteriosclerotic CVD 1.15 1.04 1.26 subsequent years can be substantial.
Arrhythmia 1.14 1.13 1.15 This is also observed in previous studies
Nephropathy 1.18 1.18 1.19 (12,15,26,27). In addition, with the large
Retinopathy 1.13 1.13 1.13 study cohort with a long-term follow-up,
Neuropathy 1.13 1.13 1.13 this study included sufficient numbers of
Peripheral vascular disease 1.14 1.14 1.15
individuals with T2D with diverse de-
Death (ref. none)
mographic and clinical characteristics to
Fatal CVD 20.49 19.93 21.05
Other-cause death 16.42 16.20 16.64 allow a comprehensive assessment of the
independent cost impact associated with
All variables in the model were statistically significant at P , 0.05. Injectable antidiabetic drugs
included insulin and glucagon-like peptide 1 receptor agonists. Fatal CVD was a composite end
diabetes-related complications.
point of death as a result of stroke, myocardial infarction, ischemic heart disease, or heart failure. Second, this study was based on NHI
All variables were treated as time-dependent variables in the model, except age and sex. claims data, which are recognized as the
Hypertension, hyperlipidemia, liver disease, and cancer were considered as chronic diseases and most suitable source because of the large
assumed to be irreversible once they occurred, so the status of these comorbidities remained in the
subsequent follow-up years. Depression status could be in and out, depending on whether patients sample size, wide coverage, detailed
had a diagnosis in a given year. HHS, hyperosmolar hyperglycemic syndrome; ref., reference. longitudinal cost data, and elimination
of self-report bias (i.e., recall and social
desirability biases), unlike previous stud-
ies, which used data from either patient
costs that are due to high drug acquisition complication costs and overall health surveys or a chart review (29,30,34).
costs, but these costs could be offset care costs (31,32). Further studies are Third, to avoid misclassification and en-
given that intensive antidiabetic treat- warranted to examine the association of sure result applicability to patients with
ments may decrease the risks of devel- intensive medication therapies/adherence T2D, we applied valid procedures to
oping costly vascular complications, which with the complication and all-cause med- define the study cohort and assess clin-
might thus lead to lower downstream ical costs in diabetes. ical characteristics and events; these
1738 Health Care Costs of Type 2 Diabetes in Taiwan Diabetes Care Volume 43, August 2020

Table 4—Cost multipliers and associated 95% CIs for patient demographics, procedures are documented elsewhere
comorbidities, diabetes-related complications, antidiabetic drug exposure, and (18,35–38). In contrast, previous studies
death: sensitivity analysis 1 (subtypes of diseases) did not differentiate T1D from T2D
Variable Multiplier 95% CI (13,15,39,40) or did not document the
Baseline annual health care cost (2017 U.S. $), mean (95% CI) 293.04 290.99 295.10
validity of the approaches used for out-
Age at T2D diagnosis (years) (ref. ,50)
come identification (13,15,40–42). In
50–59 1.20 1.19 1.20 addition, we analyzed incident diabetes-
60–69 1.45 1.44 1.45 related complications on the basis of the
$70 1.73 1.72 1.74 study cohort without any history of vas-
Diabetes duration, years (ref. 1–4) cular complications by T2D diagnosis, un-
$5 0.86 0.86 0.86 like previous studies, which either did not
Female (ref. male) 1.06 1.06 1.07 distinguish between prevalent and incident
Comorbidity (ref. none) complication events (29,30) or simply ap-
Hypertension 1.19 1.19 1.20 plied multivariable regression analyses to
Hyperlipidemia 1.06 1.05 1.06 adjust for the history of complication
Liver disease 0.97 0.96 0.97
events (12,15,26,27). Using a complica-
Cancer 1.04 1.03 1.04
Depression 1.82 1.81 1.83
tion-free cohort design allows for the
Oral antidiabetic drug exposure (ref. none) 1.80 1.79 1.80
exclusion of potential influence of prior
complication events on the future
Injectable antidiabetic drug exposure (ref. none) 1.76 1.75 1.77
events of interest, and thus, one could
Complication (event year) (ref. none)
Stroke
argue that analyses that are based on a
Ischemic stroke 2.57 2.54 2.59 complication-free cohort may provide
Hemorrhagic stroke 4.76 4.66 4.87 more reliable and valid cost estimates of
Transient ischemic attack 1.46 1.44 1.49 complication events that can be attrib-
Myocardial infarction 2.45 2.41 2.50 uted to diabetes itself.
Ischemic heart disease 2.50 2.48 2.52 Fourth, this study used a wide spec-
Heart failure 2.03 2.00 2.06
trum of analyses. We assessed a wide
Arteriosclerotic CVD 1.87 1.68 2.07
Arrhythmia 2.03 2.01 2.06 range of diabetes-related complications
Nephropathy considered as the composite outcome or
CKD with or without short-term dialysis 1.54 1.53 1.56 by subtype, considered a number of
ESRD with dialysis 3.01 2.94 3.09 important comorbidities as the time-
Proteinuria 1.20 1.19 1.21 varying covariates in all analyses and
Kidney transplant 2.29 1.94 2.72 further performed a sensitivity analysis
Other nephropathy 1.48 1.47 1.50
Retinopathy
to assess their effects as the baseline
Proliferative retinopathy 1.80 1.78 1.82 time-invariant covariates, and consid-
Blindness 1.66 1.57 1.75 ered antidiabetic treatments as the
Other retinopathy 1.34 1.34 1.35 time-varying covariates in all analyses
Neuropathy 1.41 1.41 1.42 and conducted a sensitivity analysis to
Peripheral vascular disease further assess the effects by the level of
Diabetes foot ulcer 1.67 1.65 1.69
medication exposure/adherence. In con-
Upper-extremity amputation 3.79 3.49 4.11
Lower-extremity amputation 3.22 3.11 3.33
trast, previous studies focused on either
Other peripheral vascular disease 1.46 1.45 1.46 aggregate-level complications (i.e., incre-
Hospitalized diabetic ketoacidosis 1.94 1.91 1.98 mental cost burden of people with di-
Hospitalized HHS 1.73 1.69 1.76 abetes relative to those without diabetes
Hospitalized hypoglycemia 1.73 1.71 1.74 [43–46]) or only a single complication
Complication (state year) (ref. none) (e.g., death [13], severe hypoglycemia
Stroke [14]). In addition, the potential economic
Ischemic stroke 1.37 1.36 1.38
impact of comorbidities was neglected in
Hemorrhagic stroke 1.48 1.45 1.51
Transient ischemic attack 1.03 1.02 1.05
previous studies (12,26,27,33). The cost
Myocardial infarction 1.16 1.14 1.18 impact of antidiabetic treatments was
Ischemic heart disease 1.31 1.30 1.32 either not considered (12,26,27,33) or
Heart failure 1.24 1.22 1.25 only analyzed at an aggregated level as
Arteriosclerotic CVD 1.17 1.08 1.28 exposure versus nonexposure to antidi-
Arrhythmia 1.17 1.16 1.18 abetic treatments (29,30,34,40). Finally,
Nephropathy
to ensure the validity of study results, we
CKD with or without short-term dialysis 1.09 1.08 1.10
ESRD with dialysis 6.99 6.82 7.17 adopted rigorous analyses to systemat-
Proteinuria 1.03 1.02 1.04 ically quantify the crude cost burden of
Kidney transplant 1.04 0.89 1.21 individual diabetes-related complications
at their initial occurrences (event-year costs)
Continued on p. 1739
and in subsequent years (state-year
care.diabetesjournals.org Chen and Associates 1739

Table 4—Continued 15 years of follow-up in Taiwan, the

Variable Multiplier 95% CI
economic burden of incident vascular
complications and death is found to be
Other nephropathy 1.07 1.07 1.08 compelling over time. The cost impact
Retinopathy attributable to depression and inten-
Proliferative retinopathy 1.27 1.26 1.29
Blindness 1.08 1.03 1.13
sive antidiabetic treatments cannot be
Other retinopathy 1.10 1.09 1.10 ignored. The cost estimates provide a
Neuropathy 1.13 1.13 1.14 Taiwan-specific cost catalog of T2D and
Peripheral vascular disease are useful for facilitating the parame-
Diabetes foot ulcer 1.10 1.09 1.12 terization of diabetes economic simu-
Upper-extremity amputation 1.16 1.08 1.25 lation models to quantify the economic
Lower-extremity amputation 1.13 1.09 1.16
impact of clinical outcomes, determine
Other peripheral vascular disease 1.11 1.10 1.11
cost-effective interventions, and in-
Death (ref. none)
Fatal CVD
form clinical and policy efforts of im-
Fatal ischemic stroke 8.28 7.75 8.84 proving diabetes care.
Fatal hemorrhagic stroke 7.13 6.63 7.68
Fatal myocardial infarction 2.99 2.77 3.22
Fatal ischemic heart disease 4.21 3.90 4.55 Funding. This work was supported by the
Fatal heart failure 5.47 5.17 5.79 Ministry of Science and Technology, Taiwan,
Other-cause death 13.85 13.66 14.04 under grant MOST 108-2320-B-006-048. S.K.’s
In sensitivity analysis 1, stroke, nephropathy, retinopathy, peripheral vascular disease, and fatal research is in part supported by National Institute
CVD were divided into subtypes of diseases. All variables in the model were statistically significant of Diabetes and Digestive and Kidney Diseases
at P , 0.05, except kidney transplant (state year) (P 5 0.521). Injectable antidiabetic drugs grant P30-DK-092926 (Michigan Center for Di-
included insulin and glucagon-like peptide 1 receptor agonists. Fatal CVD was a composite end abetes Translational Research [MCDTR] and
point of death as a result of stroke, myocardial infarction, ischemic heart disease, or heart failure. MCDTR Methods and Measurement Core).
All variables were treated as time-dependent variables in the model, except age and sex. The sponsors had no role in the design and
Hypertension, hyperlipidemia, liver disease, and cancer were considered as chronic diseases and conduct of the study; collection, management,
assumed to be irreversible once they occurred, so the status of these comorbidities remained in the analysis, and interpretation of the data; prepa-
subsequent follow-up years. Depression status could be in and out, depending on whether patients ration, review, or approval of the manuscript; and
had a diagnosis in a given year. CKD, chronic kidney disease; HHS, hyperosmolar hyperglycemic decision to submit the manuscript for publication.
syndrome; ref., reference. Duality of Interest. No potential conflicts of
interest relevant to this article were reported.
Author Contributions. H.-Y.C.,S.K.,P.-F.S.,J-S.W.,
and H.-T.O. contributed to data acquisition, anal-
costs) and model their cost impacts with analyses to minimize the potential un-
ysis, and interpretation. H.-Y.C., S.K., and H.-T.O
adjustment for a wide range of potential measured confounding effects. Second, drafted the manuscript. H.-Y.C. and P.-F.S. pro-
confounders, including patient demo- socioeconomic data (e.g., income, edu- vided statistical analysis and expertise. H.-Y.C.
graphics, comorbidities, and antidiabetic cation) are not collected in the health and H.-T.O. provided administrative, technical,
treatments. Some previous studies only claims database. However, these varia- material, or logistical support. S.K., P.-F.S.,
J.-S.W., and H.-T.O. critically reviewed the man-
analyzed the crude costs of diabetes- bles have been found to have no signif- uscript for intellectual content. S.K. and H.-T.O
related complication events (42,47). More- icant impact and were not considered contributed to the study concept and design and
over, the cost estimates in this study are in previous studies (12,15,26,27,29,30). supervised the study. H.-Y.C. and H.-T.O. are the
presented as the adjusted cost multipliers, Third, our cost measures considered med- guarantors of this work and, as such, had full
which represent the independent cost ical costs from the formal health care access to all the data in the study and take
responsibility for the integrity of the data and
impact of specific variables in the regres- sector. The costs from the informal health the accuracy of the data analysis.
sion analyses. The multipliers would be care sector (e.g., patient time costs, un-
easier to interpret from different perspec- paid caregiver time costs, transportation References
tives and by different stakeholders and are costs) and non–health care sectors (e.g., 1. International Diabetes Federation. IDF Dia-
not tied to actual monetary cost values that cost of unpaid lost productivity as a result betes Atlas. 8th ed. Brussels, Belgium, Interna-
would vary by time. of illness [9]) were not included because of tional Diabetes Federation, 2017
2. Taiwan Ministry of Health and Welfare.
Several limitations should be acknowl- data unavailability. Finally, the absolute
Cause of death statistics [Internet], 2017.
edged. First, we used health claims data crude medical costs may be limited to Available from https://www.mohw.gov.tw/
as the primary study data source, where country-specific cost estimates (i.e., a np-128-2.html. Accessed 21 November 2019
clinical biomarkers (e.g., HbA1c for de- Taiwanese population with T2D under a 3. Taiwan Ministry of Health and Welfare. Na-
termining the severity and control level universal health care coverage setting). tional annual health statistics from the National
Health Insurance program [Internet], 2017. Avail-
of diabetes) are typically lacking, which However, the adjusted cost multipliers, able from https://dep.mohw.gov.tw/DOS/np-
might affect clinical events and costs. instead of absolute dollar values, reflect 1918-113.html. Accessed 21 November 2019
However, we carefully measured diabe- the magnitude of the economic impact 4. Jiang Y-D, Chang C-H, Tai T-Y, Chen J-F, Chuang
tes duration, diabetes-related complica- according to the given individual charac- L-M. Incidence and prevalence rates of diabetes
tions, comorbidities, and antidiabetic teristics and would be more appropriate mellitus in Taiwan: analysis of the 2000-2009
Nationwide Health Insurance database. J Formos
treatments, which could be surrogate for international comparisons. Med Assoc 2012;111:599–604
proxies for diabetes severity and control. In summary, among a T2D population 5. American Diabetes Association Consensus
We adjusted all these variables in the of .0.8 million individuals with up to Panel. Guidelines for computer modeling of
1740 Health Care Costs of Type 2 Diabetes in Taiwan Diabetes Care Volume 43, August 2020

diabetes and its complications. Diabetes Care 19. Duan X, Li Y, Liu Q, Liu L, Li C. Epidemiological 4 inhibitors with other second- and third-line
2004;27:2262–2265 characteristics, medical costs and healthcare antidiabetic drugs in patients with type 2 di-
6. McEwen LN, Herman WH. Health care utili- resource utilization of diabetes-related compli- abetes. Br J Clin Pharmacol 2017;83:1556–1570
zation and costs of diabetes. In Chapter 40 in cations among Chinese patients with type 2 36. Lu C-H, Yang C-Y, Li C-Y, Hsieh C-Y, Ou H-T.
Diabetes in America. 3rd ed. Cowie CC, diabetes mellitus. Expert Rev Pharmacoecon Lower risk of dementia with pioglitazone, com-
Casagrande SS, Menke A, et al., Eds. Bethesda, Outcomes Res 2019;3:1–9 pared with other second-line treatments, in
MD, National Institutes of Health, 2018, p. 20. Cheng T-M. Taiwan’s new national health metformin-based dual therapy: a population-
40.1–40.78 (NIH Publ. no. 17-1468) insurance program: genesis and experience so based longitudinal study. Diabetologia 2018;61:
7. Govan L, Wu O, Lindsay R, Briggs A. How do far. Health Aff (Millwood) 2003;22:61–76 562–573
diabetes models measure up? A review of di- 21. Lin C-C, Lai M-S, Syu C-Y, Chang S-C, Tseng F- 37. Hou W-H, Chang K-C, Li C-Y, Ou H-T.
abetes economic models and ADA guidelines. J Y. Accuracy of diabetes diagnosis in health in- Dipeptidyl peptidase-4 inhibitor use is not
Health Econ Outcomes Res 2015;3:132–152 surance claims data in Taiwan. J Formos Med associated with elevated risk of severe joint
8. Charokopou M, Sabater FJ, Townsend R, Assoc 2005;104:157–163 pain in patients with type 2 diabetes: a pop-
Roudaut M, McEwan P, Verheggen BG. Meth- 22. Hsieh C-Y, Su C-C, Shao S-C, et al. Taiwan’s ulation-based cohort study. Pain 2016;157:
ods applied in cost-effectiveness models for National Health Insurance Research Database: 1954–1959
treatment strategies in type 2 diabetes mellitus past and future. Clin Epidemiol 2019;11:349–358 38. Hou WH, Chang KC, Li CY, Ou HT. Dipeptidyl
and their use in Health Technology Assess- 23. Cheng C-L, Chien H-C, Lee C-H, Lin S-J, Yang Y- peptidase-4 inhibitor use is associated with de-
ments: a systematic review of the literature HK. Validity of in-hospital mortality data among creased risk of fracture in patients with type 2
from 2008 to 2013. Curr Med Res Opin 2016;32: patients with acute myocardial infarction or diabetes: a population-based cohort study. Br J
207–218 stroke in National Health Insurance Research Clin Pharmacol 2018;84:2029–2039
9. Sanders GD, Neumann PJ, Basu A, et al. Database in Taiwan. Int J Cardiol 2015;201:96– 39. Hoffmann F, Claessen H, Morbach S, Waldeyer
Recommendations for conduct, methodologi- 101 R, Glaeske G, Icks A. Impact of diabetes on costs
cal practices, and reporting of cost-effectiveness 24. Lien H-M, Chou S-Y, Liu J-T. Hospital owner- before and after major lower extremity ampu-
analyses: second panel on cost-effectiveness in ship and performance: evidence from stroke and tations in Germany. J Diabetes Complications
health and medicine. JAMA 2016;316:1093–1103 cardiac treatment in Taiwan. J Health Econ 2008; 2013;27:467–472
10. Neumann PJ, Kim DD, Trikalinos TA, et al. 27:1208–1223 40. von Ferber L, Köster I, Hauner H. Medical
Future directions for cost-effectiveness analyses 25. Lam WY, Fresco P. Medication adherence costs of diabetic complications total costs and
in health and medicine. Med Decis Making 2018; measures: an overview. BioMed Res Int 2015; excess costs by age and type of treatment results
38:767–777 2015:217047 of the German CoDiM Study. Exp Clin Endocrinol
11. Predmore Z. Cost-effectiveness analysis 26. Kähm K, Laxy M, Schneider U, Rogowski WH, Diabetes 2007;115:97–104
and geographic variation in health care costs Lhachimi SK, Holle R. Health care costs associated 41. Martin S, Schramm W, Schneider B, et al.
in the United States. Med Decis Making 2019; with incident complications in patients with Epidemiology of complications and total treat-
39:3–4 type 2 diabetes in Germany. Diabetes Care ment costs from diagnosis of Type 2 diabetes in
12. Cheng S-W, Wang C-Y, Chen J-H, Ko Y. 2018;41:971–978 Germany (ROSSO 4). Exp Clin Endocrinol Diabe-
Healthcare costs and utilization of diabetes-re- 27. Alva ML, Gray A, Mihaylova B, Leal J, Holman tes 2007;115:495–501
lated complications in Taiwan: a claims database RR. The impact of diabetes-related complications 42. Ward A, Alvarez P, Vo L, Martin S. Direct
analysis. Medicine (Baltimore) 2018;97:e11602 on healthcare costs: new results from the UKPDS medical costs of complications of diabetes in the
13. Wong CKH, Jiao F, Tang EHM, Tong T, Thokala (UKPDS 84). Diabet Med 2015;32:459–466 United States: estimates for event-year and
P, Lam CLK. Direct medical costs of diabetes 28. American Diabetes Association. 5. Lifestyle annual state costs (USD 2012). J Med Econ
mellitus in the year of mortality and year pre- management: standards of medical care in 2014;17:176–183
ceding the year of mortality. Diabetes Obes diabetesd2019. Diabetes Care 2019;42(Suppl. 43. American Diabetes Association. Economic
Metab 2018;20:1470–1478 1):S46–S60 costs of diabetes in the U.S. in 2017. Diabetes
14. Wong CKH, Tong T, Cheng GHL, et al. Direct 29. Brandle M, Zhou H, Smith BR, et al. The direct Care 2018;41:917–928
medical costs in the preceding, event and sub- medical cost of type 2 diabetes. Diabetes Care 44. Brown JB, Nichols GA, Glauber HS, Bakst AW.
sequent years of a first severe hypoglycaemia 2003;26:2300–2304 Type 2 diabetes: incremental medical care costs
episode requiring hospitalization: a population- 30. Li R, Bilik D, Brown MB, et al. Medical costs during the first 8 years after diagnosis. Diabetes
based cohort study. Diabetes Obes Metab 2019; associated with type 2 diabetes complications Care 1999;22:1116–1124
21:1330–1339 and comorbidities. Am J Manag Care 2013;19: 45. Donnan PT, Leese GP, Morris AD; Diabetes
15. Jiao F, Wong CKH, Tang SCW, et al. Annual 421–430 Audit and Research in Tayside, Scotland/
direct medical costs associated with diabetes- 31. Hirsch JD, Morello CM. Economic impact of Medicine Monitoring Unit Collaboration. Hospi-
related complications in the event year and in and treatment options for type 2 diabetes. Am J talizations for people with type 1 and type 2
subsequent years in Hong Kong. Diabet Med Manag Care 2017;23(Suppl.):S231–S240 diabetes compared with the nondiabetic pop-
2017;34:1276–1283 32. Morello CM, Hirsch JD. Strategies for ad- ulation of Tayside, Scotland: a retrospective co-
16. Kim TH, Chun KH, Kim HJ, et al. Direct medical dressing the cost of nonadherence in diabetes. hort study of resource use. Diabetes Care 2000;
costs for patients with type 2 diabetes and related Am J Manag Care 2017;23(Suppl.):S247–S252 23:1774–1779
complications: a prospective cohort study based 33. Clarke P, Gray A, Legood R, Briggs A, Holman 46. de Lagasnerie G, Aguadé A-S, Denis P,
on the Korean National Diabetes Program. J R. The impact of diabetes-related complications Fagot-Campagna A, Gastaldi-Menager C. The
Korean Med Sci 2012;27:876–882 on healthcare costs: results from the United economic burden of diabetes to French national
17. Shuyu Ng C, Toh MPHS, Ko Y, Yu-Chia Lee J. Kingdom Prospective Diabetes Study (UKPDS health insurance: a new cost-of-illness method
Direct medical cost of type 2 diabetes in Singa- Study No. 65). Diabet Med 2003;20:442–450 based on a combined medicalized and incre-
pore. PLoS One 2015;10:e0122795 34. Henriksson F, Agardh CD, Berne C, et al. mental approach. Eur J Health Econ 2018;19:
18. Rhee SY, Hong SM, Chon S, et al. Hypogly- Direct medical costs for patients with type 2 189–201
cemia and medical expenses in patients with diabetes in Sweden. J Intern Med 2000;248:387– 47. Ray JA, Valentine WJ, Secnik K, et al. Review
type 2 diabetes mellitus: an analysis based on the 396 of the cost of diabetes complications in Australia,
Korea National Diabetes Program Cohort. PLoS 35. Ou HT, Chang KC, Li CY, Wu JS. Comparative Canada, France, Germany, Italy and Spain. Curr
One 2016;11:e0148630 cardiovascular risks of dipeptidyl peptidase Med Res Opin 2005;21:1617–1629