Increasing Risk of Dementia Among Patients with Subsequent Epilepsy Within 2 Years Post-Traumatic Brain Injury: A Population‐Based Case-Control Study

Figures & data

Table 1 TBI Demographic Features with and without Epilepsy (n = 3740)

Variable	Case Group		Control Group		p value
	TBIs Patients with Epilepsy (n =748)		TBIs Patients without Epilepsy (n =2992)
	n	%	n	%
Age (mean ±SD)	47.8±19.2
18–64	582	77.81	2328	77.81	1.0000
65–74	81	10.83	324	10.83
> 75	85	11.36	340	11.36
Sex
Male	522	69.79	2088	69.79	1.0000
Female	V 226	30.21	904	30.21
Monthly Income^a
NT$ 1~19,199	271	36.23	954	31.89	0.7034
NT$ ≥19,200	476	63.64	2032	67.91
Missing	1	0.13	6	0.20
Urbanicity^b
I (1)	180	24.06	702	23.46	0.7225
II (2+3)	329	43.98	1382	46.19
III (4+5)	137	18.32	545	18.22
IV (6+7)	78	10.43	269	8.99
Missing	24	3.21	94	3.14
Comorbidities
Stroke	209	27.94	219	7.32	<0.0001
Diabetes	88	11.76	261	8.72	0.0105
Hyperlipidemia	13	1.74	76	2.54	0.1980
Hypertension	120	16.04	450	15.04	0.4950
CAD	41	5.48	150	5.01	0.6031
HF	15	2.01	34	1.14	0.0616
AF	25	3.34	61	2.04	0.0334

Notes: Stroke, ICD9: 430–438. Diabetes, ICD9: 250. Hyperlipidemia, ICD9: 272. Hypertension, ICD9: 401–405. CAD, coronary heart disease, ICD9: 410–414. HF, heart failure, ICD9: 428. AF, atrial fibrillation, ICD9: 427. ^aIncome: in 2011, the average exchange rate was US$ 1 = NT$ 29. ^bUrbanization <AREA_NO_I>: 1(most urbanization), 2+3, 4+5, 6+7.

Abbreviations: CAD, coronary heart disease; HF, heart failure; AF, atrial fibrillation.

Table 2 Dementia Incidence Over 5-Year Follow-Up, HRs Next to Crude and Covariate-Adjusted, Indexed by Medical Utilization Initiation

Presence of Dementia in 5-Year Follow-up Period	Total		Male		Female
	TBIs Patients with Epilepsy (n=748)	TBIs Patients Without Epilepsy (n=2992)	TBIs Patients with Epilepsy (n=522)	TBIs Patients Without Epilepsy (n=2088)	TBIs Patients with Epilepsy (n=226)	TBIs Patients Without Epilepsy (n=904)
Yes (dementia, average)	1.32	2.59	1.20	2.57	1.79	2.61
No (dementia, average)	4.32	4.78	4.26	4.76	4.46	4.80
Crude HR (95% CI)	3.16 (2.10–4.76)*	1	5.35 (3.17–9.01)*	1	1.19 (0.55–2.61)	1
Covariate-adjusted HR (95% CI)^a	3.03 (1.96–4.69)*	1	5.32 (3.06–9.26)*	1	1.02 (0.44–2.41)	1

Notes: ^aAdjustments are made for age, sex, patients’ monthly income, geographical location and comorbidities (Stroke, Diabetes, Hyperlipidemia, Hypertension, CAD, HF and AF). *p<0.05.

Abbreviations: HR, Hazard ratio; TBI, traumatic brain injury; CI, confidence interval.

Figure 1 Presents the dementia-free survival curves using the Kaplan–Meier method. TBI Patients with epilepsy had significantly lower 5-year dementia-free survival rates than patients without epilepsy (Log rank test: 33.6462, p=<0.0001).

Table 3 Crude and Adjusted HR for Dementia in Sampled Individuals of Atypical Age Groups Over a 5-Year Follow-Up Periods from the Index Date of Healthcare Utilization (n = 3740)

Age Group (Years)	Total		Male		Female
	TBIs Patients with Epilepsy (n=748)	TBIs Patients Without Epilepsy (n=2992)	TBIs Patients with Epilepsy (n=522)	TBIs Patients Without Epilepsy (n=2088)	TBIs Patients with Epilepsy (n=226)	TBIs Patients Without Epilepsy (n=904)
18–64 (n=2910)
Crude HR (95% CI)	21.14 (7.22–61.86)*	1	28.46 (8.45–95.80)*	1	–	1
Adjusted HR (95% CI)^a	14.37 (4.64–44.51)*	1	18.19 (5.09–64.93)*	1	–	1
65–74 (n=405)
Crude HR (95% CI)	3.69 (1.71–7.96)*	1	4.84 (1.75–13.39)*	1	2.53 (0.76–8.42)	1
Adjusted HR (95% CI)^a	3.41 (1.44–8.08)*	1	6.00 (1.92–18.70)*	1	1.97 (0.46–8.47)	1
> 75 (n=425)
Crude HR (95% CI)	1.23 (0.57–2.66)	1	1.51 (0.50–4.56)	1	0.98 (0.34–2.88)	1
Adjusted HR (95% CI)^a	0.98 (0.43–2.26)	1	2.00 (0.62–6.42)	1	0.57 (0.17–1.91)	1

Notes: *p<0.05. HR was calculated by using the stratified Cox proportional regression method (stratified by sex, age group) during the 19year follow-up period. ^aAdjustments are made for patients’ monthly income, geographical location and comorbidities (Stroke, Diabetes, Hyperlipidemia, Hypertension, CAD, HF and AF).

Data Sharing Statement

In this population-based case-control study, sample patients were retrieved from the Taiwan “Longitudinal Health Insurance Database 2000” (LHID2000) were retrieved). This dataset provides a tremendous opportunity to track enrollees’ healthcare utilization since 1995, the year of the Administration of Health Insurance operations in Taiwan. The Health and Welfare Data Science Center, Ministry of Health and Welfare, collects the LHID2000, which attributes the fundamental medical claims dataset of 2,000,000 patient registration files to a random sample retrieved from over 23,000,000 enrollees of the Administration Health Insurance Research Claim Dataset in Taiwan. The Health and Welfare Data Science Center and other researchers validated the LHID2000 dataset. Researchers can approach the data for research purposes and employ the LHID2000 for publication in peer-reviewed journals.