Correlation between the utilization of sodium-glucose cotransporter-2 (SGLT2) inhibitors and the risk of dementia: a nationwide population-based study in Taiwan

Abstract

The impact of sodium-glucose cotransporter 2 inhibitors (SLGT2i) usage on reducing the risk of dementia remains uncertain. Our research seeks to establish the association between dementia risk and SLGT2 inhibitors among individuals with diabetes. This study relied on data from the Taiwan National Health Insurance Database (NHIRD), which was established in 1995 coinciding with the launch of the National Health Insurance (NHI) program by the Taiwanese government. The NHI program was implemented to enhance the healthcare system and public health in Taiwan. Patients with type 2 diabetes mellitus (T2DM) administered SGLT2i between 2016 and 2019 were included in the SGLT2i cohort. The comparison cohort consisted of patients who did not receive SGLT2i, propensity score matching by sex, age (in 5-y intervals), index date year, insurance fee, urbanization, comorbidities, and medications, with a 1:1 ratio of the exposure group. SGLT2i users had a significantly lower risk of dementia than non-SGLT2i users after adjusting for age, sex, insurance fees, urbanization, comorbidities, and medications (adjusted HR = 0.53, 95%CI: 0.50–0.57). The results revealed that patients treated with SGLT2i have a lower risk of dementia in Taiwan.

Introduction

The number of aging people worldwide is increasing. Dementia is one of the main neurological disorders leading to morbidity and mortality in aging people(Livingston et al. 2020). Alzheimer’s disease and vascular dementia are the most prevalent forms of dementia (Lobo et al. 2000; Jhoo et al. 2008; Chan et al. 2013). Although the exact pathological mechanism of dementia is not yet clear, physical inactivity, hypertension, obesity, diabetes, hearing impairment, depression, traumatic brain injury, smoking, alcohol consumption, low education, low social contact, and air pollution are modifiable risk factors for dementia(Kivimäki et al. 2019; Tanaka et al. 2023; Smith et al. 2023; Sheng et al. 2023; Kim et al. 2023; Zhao et al. 2023; Hackett et al. 2023; Kalam et al. 2023; Mollayeva et al. 2023; Arora and Bhagianadh 2021; Maccora et al. 2020; Wilker et al. 2023; Sommerlad and Liu 2023). In some countries, the prevalence of dementia has been decreased by decreasing these modifiable risk factors(Livingston et al. 2020). Diabetes is associated with a higher risk of long-term microvascular and cardiovascular complications, and systematic research has shown that individuals with diabetes have a higher risk of dementia(Chatterjee et al. 2016). Hypertension is also a risk factor for dementia, and the risk of vascular dementia is higher in patients with higher systolic blood pressure (Lee et al. 2022; Nagai et al. 2010). Within the category of oral anti-diabetic medications, sodium-glucose cotransporter 2 inhibitors (SGLT2i) have demonstrated the capacity to reduce blood glucose levels, lower blood pressure, and facilitate weight loss. In addition, it also provides beneficial advantages for heart failure and chronic renal failure protection(Solomon et al. 2022; Group et al. 2022). Nonetheless, it remains uncertain whether the use of SGLT2 inhibitors will lead to a reduced risk of dementia. Our research sought to elucidate the long-term dementia risk in individuals with diabetes who received treatment with SGLT2 inhibitors.

Methods

Data source

The data source for this study was the Taiwan National Health Insurance Database (NHIRD), which was established in 1995 in conjunction with the launch of the National Health Insurance (NHI) program by the Taiwanese government to enhance healthcare and public health. The NHIRD contains comprehensive information on patient demographics, inpatient and outpatient records, medications, and treatments. The disease codes in this database were derived from the International Classification of Diseases, 9th and 10th editions, Clinical Modification (ICD-9-CM and ICD-10-CM). Approval for this study was granted by the Research Ethics Committee of China Medical University Hospital (Institutional Review Board), under the reference CMUH110-REC3-133.

Ethics statement

To protect patient confidentiality, the NHIRD employs encryption measures, safeguarding personal information. Researchers are assigned anonymous identification numbers that are linked to essential claims data, including gender, date of birth, medical services, and prescriptions. As a result, accessing the NHIRD does not require patient consent. This research project obtained an exemption approval from the Institutional Review Board (IRB) at China Medical University (Approval Number CMUH109-REC2-031 (CR2)), and the IRB explicitly waived the need for consent.

Study population

Patients diagnosed with type 2 diabetes mellitus (T2DM) who received SGLT2 inhibitors (SGLT2i) between 2016 and 2019 constituted the SGLT2i cohort. The comparison cohort was comprised of patients who did not receive SGLT2i. Propensity score matching was conducted based on factors including sex, age (grouped in 5-year intervals), index date, insurance fee, urbanization, comorbidities, and medications, with a 1:1 ratio between the two groups.

For SGLT2i users, the index date was defined as the date of their first prescription of SGLT2i following their T2DM diagnosis. Non-users of SGLT2i had their index date randomly assigned after their T2DM diagnosis date. Exclusion criteria for this study encompassed individuals under 20 years of age, those previously diagnosed with dementia before the index date, those with missing data on sex and age, and individuals taking SGLT2i for less than 90 days. All participants were followed up until the occurrence of newly diagnosed dementia, withdrawal from NHIRD, death, or the conclusion of the year 2019.

Main outcome and comorbidities

Dementia (ICD-9:290, 294.1, 331.2 and ICD-10: F01, F02, F03, F05, G30, and G31.1) was the main outcome of our study. Furthermore, we considered sex, age, insurance fees (low, medium, and high), urbanization (low, medium, and high), comorbidities, and related medications as confounding factors. In addition to continuous age values, age was categorized into three groups: 20–44, 45–69, and ≥ 70 years old. The comorbidities encompassed hypertension (HT, ICD-9: 401–405, ICD-10: I10-I16), dyslipidemia (DL, ICD-9: 272, ICD-10: E71.30, E75.21, E75.22, E75.24, E75.3, E75.5, E75.6, E77, E78.0, E78.1, E78.2, E78.3, E78.4, E78.5, E78.6, E78.70, E78.79, E78.8, E78.9), coronary heart disease (CAD, ICD-9: 410–414, ICD-10: I20-I25), alcoholism (ICD-9: 291, 303, 305.0, 571.0–571.3, 790.3, V11.3, V79.1, ICD-10: F10, K70, R78.0, Z65.8), smoking (ICD-9: 305.1, V15.82, ICD-10: F17.2, Z87.891), and obesity (ICD-9: 278, 783.1, ICD-10: E66.09, E66.1, E66.8, E66.9, E66.01, E66.2, E65, E67.0, E67.1, E67.3, E67.2, E67.8, E68 R63.5). The relevant medications included metformin, sulfonylureas, meglitinides, alpha-glucosidase inhibitor (AGI), thiazolidinedione, dipeptidyl peptidase-4 (DPP-4), glucagon-like peptide-1 receptor agonists (GLP-1 RA), and insulin.

Statistical analysis

Categorical data and continuous variables were presented as numbers (percentages) and mean [standard deviation (SD)], respectively. Differences in all variables between the SGLT2i cohort and the control group were assessed using the standardized mean difference (SMD)(Austin 2009). A SMD value less than 0.1 indicated a negligible difference between the two cohorts. We employed Cox proportional hazard regression models to compute the adjusted hazard ratio (aHR), along with their corresponding 95% confidence intervals (CIs), to evaluate the risk of dementia. The aHR was derived from a multivariate model incorporating variables such as sex, mean age, comorbidities, and medications. All statistical analyses were conducted using SAS, version 9.4, with a significance level set at p < 0.05.

Results

A total of 305,810 patients were included in this study (152905 SGLT2i group and 152905 control group). The study population’s demographic and baseline clinical characteristics are presented in Table 1. We can’t find any significant differences in the distribution of demographic variables between the two groups; however, male participants were the majority in both the SGLT2i (58.13%) and control (57.85%) cohorts. The mean age was 58.70 ± 11.76 years in the SGLT2i cohort and 58.24 ± 11.63 years in the comparison cohort. Both groups mainly comprised patients with medium insurance fees and high urbanization rates. Compared to the controls, there were more comorbidities in SGLT2i patients, except for HT (non-SGLT2i: 76.05% vs. SGLT2i: 74.89%) and CAD (non-SGLT2i: 30.88% vs. SGLT2i: 30.22%). The most common medications in both cohorts were metformin (non-SGLT2i: 97.59% vs. SGLT2i: 97.39%), sulfonylureas (non-SGLT2i: 84.43% vs. SGLT2i: 83.82%), and DPP-4 inhibitors (non-SGLT2i: 72.21% vs. SGLT2i: 72.31%). The follow-up time for SGLT2i users (2.25 ± 0.78) was longer than that of SGLT2i non-users (2.22 ± 0.98).

Table 1.

Comparison of demographic characteristics and comorbidities between patients with SGLT2i and controls

Covariates		SGLT2i (N = 152905)		Control (N = 152905)		SMD
		n	%	n	%
Male		88,456	57.85	88,887	58.13	0.006
Age	20–44	78,598	51.40	76,205	49.84	0.031
	45–69	70,241	45.94	72,252	47.25	0.026
	≥ 70	4066	2.66	4448	2.91	0.015
	mean ± SD	58.24 ± 11.63		58.70 ± 11.76		0.039
Insurance fee	Low	31,481	20.59	33,157	21.68	0.027
	Medium	82,205	53.76	81,476	53.29	0.010
	High	39,219	25.65	38,272	25.03	0.014
Urbanization	Low	13,128	8.59	12,994	8.50	0.003
	Medium	59,737	39.07	60,000	39.24	0.004
	High	80,040	52.35	79,911	52.26	0.002
Comorbidities	HT	114,513	74.89	116,283	76.05	0.027
	DL	128,394	83.97	128,310	83.91	0.001
	CAD	46,213	30.22	47,224	30.88	0.014
	Alcoholism	3879	2.54	3752	2.45	0.005
	Smoking	7519	4.92	7345	4.80	0.005
	Obesity	7675	5.02	7307	4.78	0.011
Medication	Metformin	148,913	97.39	149,220	97.59	0.013
	Sulphonylurea	128,162	83.82	129,093	84.43	0.017
	Meglitinides	27,727	18.13	28,272	18.49	0.009
	AGI	56,723	37.10	55,473	36.28	0.017
	Thiazolidinedione	60,888	39.82	59,149	38.68	0.023
	DPP-4 inhibitors	110,558	72.31	110,406	72.21	0.002
	GLP-1 RA	3129	2.05	2971	1.94	0.007
	Insulin	62,820	41.08	64,278	42.04	0.019
Follow up time (years)	Mean ± SD	2.25 ± 0.78		2.22 ± 0.98		0.035

HT hypertension; DL dyslipidemia; CAD coronary artery disease; AGI alpha-glucosidase inhibitor; DPP-4 dipeptidyl peptidase-4; GLP-1 RA glucagon-like peptide-1 receptor agonists (GLP-1 RA)

As shown in Table 2, the numbers of patients with dementia were 1243 and 2428 in SGLT2i users and the control group, respectively. SGLT2i users had a significantly lower risk of dementia than those without dementia after adjusting for age, sex, insurance fees, urbanization, comorbidities, and medications (aHR = 0.53, 95%CI = 0.50–0.57). The HR of dementia for male patients was lower (adjusted HR = 0.79; 95%CI = 0.74, 0.85) than that of female patients. Furthermore, participants aged 45–69 had a higher risk of dementia (adjusted HR = 38.6; 95%CI = 30.24, 49.41) than those aged 20–44 and those aged ≥ 70 (adjusted HR = 133; 95%CI = 103.5, 173.0). Compared to participants with low insurance fees, the hazard ratio of dementia for participants with medium or high insurance fees was lower (medium: adjusted HR = 0.91; 95%CI = 0.84, 0.98 / high: adjusted HR = 0.83; 95%CI = 0.75, 0.91). With regard to comorbidities, the risk of dementia in patients with HT (adjusted HR = 1.37; 95%CI = 1.22, 1.54), CAD (adjusted HR = 1.18; 95%CI = 1.10, 1.26), and smoking (adjusted HR = 1.25; 95%CI = 1.03, 1.52) was higher than that in those without. Similarly, participants taking meglitinides (adjusted HR = 1.20; 95%CI = 1.11, 1.29), thiazolidinedione (adjusted HR = 1.11; 95%CI = 1.04, 1.19), and insulin (adjusted HR = 1.51; 95%CI = 1.41, 1.62) had a higher hazard ratio for dementia than those without. In contrast, the risk of dementia in patients using GLP-1 RA was 0.65 times higher than in those who did not use GLP-1 RA (95%CI = 0.45, 0.92).

Table 2.

Hazard ratios and 95% confidence intervals of dementia development

Covariates		n	PY	IR	aHR	95%CI	P
SGLT2i
	No	2428	338,891	7.16	1.00
	Yes	1243	343,585	3.62	0.53	0.50, 0.57	< 0.001
Sex
	female	2030	291,592	6.96	1.00
	male	1641	390,884	4.20	0.79	0.74, 0.85	< 0.001
Age
	20–44	66	348,085	0.19	1.00
	45–69	2967	317,604	9.34	38.60	30.24, 49.41	< 0.001
	≥ 70	638	16,788	38.00	133.0	103.5, 173.0	< 0.001
Insurance fee
	Low	1094	143,247	7.64	1.00
	Medium	1920	366,141	5.24	0.91	0.84, 0.98	0.013
	High	657	173,089	3.80	0.83	0.75, 0.91	< 0.001
Urbanization
	Low	319	58,205	5.48	1.00
	Medium	1522	267,127	5.70	1.07	0.94, 1.20	0.306
	High	1830	357,144	5.12	1.04	0.92, 1.17	0.512
Comorbidities
	HT	3352	513,252	6.53	1.37	1.22, 1.54	< 0.001
	DL	3036	573,243	5.30	0.93	0.85, 1.01	0.094
	CAD	1848	204,666	9.03	1.18	1.10, 1.26	< 0.001
	Alcoholism	49	16,184	3.03	1.21	0.91, 1.61	0.192
	Smoking	107	30,770	3.48	1.25	1.03, 1.52	0.026
	Obesity	80	32,600	2.45	0.93	0.74, 1.16	0.499
Medication
	Metformin	3603	666,857	5.40	1.01	0.79, 1.30	0.921
	Sulphonylurea	3335	580,824	5.74	1.05	0.93, 1.19	0.406
	Meglitinides	1062	125,317	8.47	1.20	1.11, 1.29	< 0.001
	AGI	1831	254,334	7.20	1.02	0.95, 1.09	0.548
	Thiazolidinedione	1938	272,428	7.11	1.11	1.04, 1.19	0.002
	DPP-4 inhibitors	3021	499,186	6.05	1.04	0.95, 1.14	0.361
	GLP-1 RA	31	12,061	2.57	0.65	0.45, 0.92	0.017
	Insulin	2191	283,456	7.73	1.51	1.41, 1.62	< 0.001

HT hypertension; DL dyslipidemia; CAD coronary artery disease; n numbers of dementia development; PY person-years; IR incidence rate (per 1000 person-years); aHR adjusted hazard ratio; 95%CI, 95% confidence interval; AGI alpha-glucosidase inhibitor; DPP-4 dipeptidyl peptidase-4; GLP-1 RA glucagon-like peptide-1 receptor agonists (GLP-1 RA)

In multivariate analysis, the risk of dementia in the SGLT2i cohort was significantly lower than that in the controls, except in patients with obesity and GLP-1 RA (Table 3). Compared with SGLT2i non-users, the HR of dementia in SGLT2i users was 0.57 (95%CI: 0.52, 0.62) in female patients and 0.50 (95%CI: 0.45, 0.56) in male patients. In patients with 20–44, 45–69, and ≥ 70 years old, the risks of dementia in comparison with the control group were 0.33 (95%CI: 0.19, 0.58), 0.52 (95%CI: 0.48, 0.56), and 0.57 (95%CI: 0.48, 0.67), respectively. For insurance fees, the adjusted HRs of dementia in SGLT2i users were 0.52 (95%CI = 0.46, 0.59) in low insurance fee patients, 0.53 (95%CI = 0.48, 0.58) in medium insurance fee patients, and 0.57 (95%CI = 0.49, 0.67) in high insurance fee patients compared to the controls. Besides, the risk of dementia in participants with SGLT2i was 0.56 (95%CI = 0.44, 0.70) among the low urbanization group, 0.56 (95%CI = 0.50, 0.62) among the medium urbanization group, and 0.50 (95%CI = 0.46, 0.56) among the high urbanization group than those without SGLT2i. In comparison with the control cohort, the risk of developing dementia in the SGLT2i cohort was 0.52 (95% CI:0.48, 0.55) in patients with HT, 0.52 (95% CI:0.48, 0.56) in patients with DL, 0.51 (95% CI:0.46, 0.56) in patients with CAD, 0.56 (95% CI:0.31, 1.00) in patients with alcoholism, and 0.40 (95% CI:0.27, 0.62) in patients with smoking. As for medications, the adjusted HR of dementia in participants taking SGLT2i were 0.53 (95%CI = 0.49, 0.57) in metformin users, 0.54 (95%CI = 0.50, 0.58) in sulphonylurea users, 0.45 (95%CI = 0.39, 0.51) in meglitinides users, 0.52 (95%CI = 0.48, 0.58) in AGI users, 0.54 (95%CI = 0.49, 0.59) in thiazolidinedione users, 0.53 (95%CI = 0.49, 0.57) in DPP-4 inhibitors users, and 0.50 (95%CI = 0.46, 0.55) in insulin users compared with the comparison group.

Table 3.

Hazard ratios and 95% confidence intervals of dementia development with and without SGLT2i stratified by age, comorbidities and medication

Covariates		Control			SGLT2i			aHR	95%CI
		n	PY	IR	n	PY	IR
Sex
	female	1330	144,559	9.20	700	147,033	4.76	0.57	0.52, 0.62
	male	1098	194,332	5.65	543	196,552	2.76	0.50	0.45, 0.56
Age
	20–44	49	170,394	0.29	17	177,691	0.10	0.33	0.19, 0.58
	45–69	1973	160,093	12.32	994	157,511	6.31	0.52	0.48, 0.56
	≥ 70	406	8404	48.31	232	8384	27.67	0.57	0.48, 0.67
Insurance fee
	Low	716	72,324	9.90	378	70,923	5.33	0.52	0.46, 0.59
	Medium	1278	181,488	7.04	642	184,653	3.48	0.53	0.48, 0.58
	High	434	85,079	5.10	223	88,009	2.53	0.57	0.49, 0.67
Urbanization
	Low	213	28,930	7.36	106	29,274	3.62	0.56	0.44, 0.70
	Medium	987	132,765	7.43	535	134,362	3.98	0.56	0.50, 0.62
	High	1228	177,196	6.93	602	179,948	3.35	0.50	0.46, 0.56
Comorbidities
	HT	2249	256,428	8.77	1103	256,824	4.29	0.52	0.48, 0.55
	DL	2034	285,031	7.14	1002	288,213	3.48	0.52	0.48, 0.56
	CAD	1246	102,232	12.19	602	102,433	5.88	0.51	0.46, 0.56
	Alcoholism	30	7679	3.91	19	8505	2.23	0.56	0.31, 1.00
	Smoking	76	14,878	5.11	31	15,891	1.95	0.40	0.27, 0.62
	Obesity	45	15,414	2.92	35	17,186	2.04	0.76	0.49, 1.20
Medication
	Metformin	2387	331,448	7.20	1216	335,409	3.63	0.53	0.49, 0.57
	Sulphonylurea	2207	289,342	7.63	1128	291,483	3.87	0.54	0.50, 0.58
	Meglitinides	756	61,882	12.22	306	63,435	4.82	0.45	0.39, 0.51
	AGI	1219	123,786	9.85	612	130,548	4.69	0.52	0.48, 0.58
	Thiazolidinedione	1281	131,935	9.71	657	140,493	4.68	0.54	0.49, 0.59
	DPP-4 inhibitors	2003	246,359	8.13	1018	252,827	4.03	0.53	0.49, 0.57
	GLP-1 RA	17	5091	3.34	14	6970	2.01	0.62	0.29, 1.29
	Insulin	1487	139,789	10.64	704	143,668	4.90	0.50	0.46, 0.55

HT hypertension; DL dyslipidemia; CAD coronary artery disease; n numbers of dementia development; PY person-years; IR incidence rate (1000 person-years); aHR, adjusted hazard ratio; 95%CI, 95% confidence interval; AGI alpha-glucosidase inhibitor; DPP-4 dipeptidyl peptidase-4; GLP-1 RA glucagon-like peptide-1 receptor agonists (GLP-1 RA)

Figure 1 show the cumulative incidence comparison of dementia for SGLT2i cohort (dashed line) or control group (solid line).

Fig. 1.

Cumulative Incidence of Dementia in SGLT2i cohort and non-SGLT2i cohort

Discussion

In our study, we identified an inverse relationship between the use of SGLT2 inhibitors and the onset of dementia. A prior study conducted in Taiwan reported an 8.08% prevalence of dementia among older individuals (Sun et al. 2014). In our study, the prevalence of type 2 diabetes in patients aged ≥ 70 years old was 38%, 9.34% in those aged 45–69 years, and 0.19% in those aged 20–44 years. In numerous studies, the dementia prevalence rate in patients with type 2 diabetes was higher than in patients without type 2 diabetes.

In a cohort study based on the population, the use of SGLT2 inhibitors was linked to a reduced risk of dementia compared to dipeptidyl peptidase 4 (DPP-4) inhibitors (Wu et al. 2023). One case-control study showed a lower risk of dementia in people who used metformin, DPP-4 inhibitors, GLP1-RA, or SGLT2i(Wium-Andersen et al. 2019). In our study, there was a lower incidence rate in patients who use SGLT2i than in non-user, and the adjusted incidence rate in SGL2i users was lower than that in non-users. In addition, the cumulative incidence of dementia graph shows that the cumulative incidence of dementia is lower in the SGLT2i cohort over time.

Anemia is associated with dementia in elder people (Hong et al. 2013; Weiss et al. 2022). In individuals with type 2 diabetes, anemia is correlated with an increased risk of dementia, as evidenced in a nationwide population-based cohort study (Choi et al. 2020). Beyond the glucose-lowering effect, SGLT2i may increase hematocrit and prevent anemia. In the DAPA-CKD trial, in patients with anemia and CKD, anemia was corrected in 53.3% of patients in the SGLT2i group and 29.4% in the placebo group (Koshino et al. 2023). Our study is limited because hematocrit change and difference were not recorded in NHIRD.

Dementia development is linked to cardiovascular risk factors such as hypertension, elevated cholesterol levels, and diabetes mellitus (Whitmer et al. 2005). Angiotensin receptor blockers work by blocking angiotensin II receptors, resulting in a reduction of blood pressure. When used for hypertension treatment, angiotensin receptor blockers are connected with a decreased risk (HR) of dementia and enhanced neurocognitive function (Forette et al. 2002; Li et al. 2010; Hajjar et al. 2020). Our findings further demonstrated that hypertension constitutes a risk factor for dementia. Numerous studies have provided evidence that SGLT2 inhibitors have the capacity to reduce blood pressure in individuals with diabetes (Kario et al. 2018a, b; Kinguchi et al. 2019). Use of SGLT2i can help reduce blood pressure, and this may be a possible mechanism that explains the neuroprotective effect of SGLT2i. In different blood pressure levels, the risk of dementia may be different. In one study, vascular dementia is higher in those with high blood pressure not using antihypertensive drugs; however, the association is U-shaped in people taking antihypertensive drugs(Lee et al. 2022). Our study is limited in that the baseline blood pressure and blood pressure changes after SGL2Ti treatment were unknown.

Limitations and strengths

Considering the homogeneity between the SGLT2i and control groups, we conducted propensity score matching based on the demographic data and comorbidities. Consequently, the control group did not accurately reflect the true prevalence of comorbidities or demographic characteristics, potentially resulting in an underestimation of dementia risk. It’s essential to recognize that dementia is a condition that develops over the long term. Although this was a nationwide study, the mean follow-up year was approximately 2 years due to the participant enrollment period. Therefore, an additional clinical or experimental study is needed to explore the underlying mechanisms linking SGLT2i and dementia. We could not enroll patients diagnosed with dementia after the follow-up period. Therefore, the risk of dementia may have been underestimated in this study. In addition, most participants came from high and medium levels of urbanization, and potential surveillance bias may have occurred because urban and rural areas were quite different in terms of the convenience of healthcare. Nonetheless, a remarkable 99% or more of the residents were enrolled in the NHI program. Previous research has indicated that NHI programs offer cost-free healthcare services both within the country and in remote regions (Cheng and Chiang 1997; Huang et al. 2006). This may eliminate barriers and minimize surveillance bias. Furthermore, one of the major limitations of NHIRD is the lack of hematology data. Thus, we can’t consider the HbA1c and duration of diabetes as the covariates in this study.

Conclusion

SGLT2i initially was used for diabetes mellitus and then at current, it is widely used in heart failure or chronic kidney disease patients without diabetes. Diabetic patients had higher risk of all types of dementia. There’s not yet to clarify the priority of choosing glucose-lowering agents in diabetic patients with dementia or high risk of dementia. Our study shows that SGLT2i may be beneficial in reducing dementia risk in diabetic patients and implys early initiation of SGLT2i should be considered in individuals with diabetes for further dementia risk reduction. However, it’s not clear yet that whether if SGL2i can provide brain cognition benefit for non-diabetic patients. If it’s proved in the future, SGLT2i for dementia may change standard of care. However, this study has some limitations. Further researches are needed to evaluate the mechanism of neuroprotective effect of SGLT2i.

Abbreviations

NHIRD

National Health Insurance Database

T2DM

type 2 diabetes mellitus

SGLT2

sodium-glucose cotransporter-2

NHI

National Health Insurance

ICD-9-CM and ICD-10-CM

International Classification of Diseases,9th and 10th editions, Clinical Modification

IRB

Institutional Review Board

HT

hypertension

DL

dyslipidemia

CAD

coronary heart disease

AGI

alpha-glucosidase inhibitor

DPP-4

dipeptidyl peptidase-4

GLP-1 RA

glucagon-like peptide-1 receptor agonists (GLP-1 RA)

SD

standard deviation

SMD

standardized mean difference

cHR

crude hazard ratio

aHR

adjusted hazard ratio

Author contributions

Conceptualization, K.-H.C., and W.-L. C.; methodology, K.-H.C. and H.-J. L.; software, K.-H.C. and H.-J. L.; validation, K.-H.C. and W.-L. C.; formal analysis, K.-H.C. and H.-J. L.; investigation, W.-L. C., S.C.-S.T., Y.-C. Y., T.-L. H., H.-J. L.Y.-H. H., F.-J. T., and K.-H.C.; resources, F.-J. T., and K.-H.C.; data curation, K.-H.C. Y.-H. H., and H.-J. L.; writing—original draft preparation, K.-H.C., and W.-L. C.; writing—review and editing, K.-H.C., and W.-L. C.; visualization, W.-L. C., S.C.-S.T., Y.-C. Y., T.-L. H., H.-J. L., Y.-H. H., F.-J. T., and K.-H.C.; supervision, K.-H.C.; project administration, F.-J. T., and K.-H.C; funding acquisition, F.-J. T., and K.-H.C. All authors have read and agreed to the published version of the manuscript. The authors declare that all data were generated in-house and that no paper mill was used.

Funding

This study is supported in part by Taiwan Ministry of Health and Welfare Clinical Trial Center (MOHW113-TDU-B-212-114009), China Medical University Hospital (DMR-111-105; DMR-112-087; DMR-113-009; DMR-113-156), and Tungs’ Taichung MetroHarbor Hospital (TTMHH-R1140003; TTMHH-R1140083; TTMHH-R1140085; TTMHH-R1140089). We are grateful to Health Data Science Center, China Medical University Hospital for providing administrative, technical and funding support. No additional external funding was received for this study.

Data availability

All source data for this work (or generated in this study) are available upon reasonable request.

Declarations

Ethics approval

The NHIRD secures patient personal data by employing encryption to safeguard privacy. It also provides researchers with anonymous identification numbers that are associated with relevant claims data, including details such as gender, date of birth, medical services, and prescriptions. As a result, patient consent is not required to access the NHIRD for this research project. This research project received approval for exemption from the Institutional Review Board (IRB) at China Medical University (Approval Number CMUH109-REC2-031 (CR2)), and the IRB explicitly waived the requirement for informed consent. All the procedures were followed by the relevant guidelines and regulations.

Consent for publication

All authors agree to publish the article.

Competing interests

The authors declare no competing interests.