Thyroid Disease as a Risk Factor for Cerebrovascular Disease Meng-Han Yang, PhD,* Fu-Yi Yang, MD,† and Ding-Dar Lee, MD, PhD‡x
Background: Thyroid disease is the medical condition impairing function of the thyroid. Among this disorder category, hyperthyroidism is that the thyroid gland produces excessive amounts of thyroid hormones whereas hypothyroidism is that the thyroid gland does not produce enough thyroid hormone. Various studies have supported the comorbid association between thyroid disease and cardiovascular disorder. However, there is insufficient evidence to prove the relationship between cerebrovascular disease (CVD) and thyroid disease. Methods: In this study, we tried to verify that thyroid disease increases the risk of CVD development employing a population-based database, National Health Insurance Research Database of Taiwan. A total of 16,808 hyperthyroidism cases and 5793 hypothyroidism patients with corresponding control subjects were studied, respectively. Hazard ratio (HR) by the Cox regression was used to quantify risk of CVD in different groups of subjects, that is, case patients versus matched controls. Further stratification studies for risk factors of CVD were performed to evaluate the comorbid association between CVD and hyperthyroidism/hypothyroidism. Results: Evaluation results have shown that hyperthyroidism increased 38% of the hazard of developing follow-up CVD (adjusted HR, 1.38) whereas hypothyroidism increased even higher the risk (adjusted HR, 1.89). Further stratification studies for risk factors of CVD suggested that the comorbid association between hypothyroidism and CVD was comparable to those influences from cardiac risk factors, such as diabetes mellitus, hyperlipidemia, hypertension, or renal failure and so forth. Conclusions: Thyroid disease may predispose to onset of CVD. Advanced analysis is required to investigate the pathologic mechanism underlying the association between CVD and thyroid disease. Key Words: Hyperthyroidism—hypothyroidism—cerebrovascular disease—hazard ratio—NHIRD. Ó 2015 by National Stroke Association
Thyroid disease is the medical condition impairing function of the thyroid. Among this disorder category, hyperthyroidism is a condition in which the thyroid gland produces and secretes excessive amounts of the free thyroid hormones: triiodothyronine (T3) and/or thyroxine (T4). Symptoms of hyperthyroidism may be
nervousness, irritability, heart racing, hand tremors, anxiety, muscular weakness, and so forth. Thyroid hormone is also critical to normal function of cells, therefore in excess, it both overstimulates metabolism and exacerbates the effect of sympathetic nervous system. Graves’ disease is the most common cause of hyperthyroidism,
From the *Department of Computer Science and Information Engineering, National Kaohsiung University of Applied Sciences, Kaohsiung, Taiwan; †Department of Neurology, Taipei Tzu Chi General Hospital, New Taipei, Taiwan; ‡Department of Dermatology, TaipeiVeterans General Hospital, Taipei City, Taiwan; and xFaculty of Medicine, Department of Dermatology, National Yang-Ming University, Taipei City, Taiwan, ROC. Received October 11, 2014; revision received November 25, 2014; accepted November 28, 2014. This research has been supported by National Kaohsiung University of Applied Sciences and National Taiwan University. The authors declare that they have no competing interest.
Analysis procedures were designed by M.-H.Y. and F.-Y.Y. Data collection and analysis was performed by M.-H.Y. and supervised by both D.-D.L. and F.-Y.Y. Article was prepared by M.-H.Y. and D.-D.L. All authors read and approved the final article. Address correspondence to Fu-Yi Yang, MD, The Department of Neurology, Taipei Tzu Chi General Hospital, No. 289, Jianguo Rd., Xindian District, New Taipei 23142, Taiwan, ROC. E-mail: [email protected]; [email protected]. 1052-3057/$ - see front matter Ó 2015 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2014.11.032
912
Journal of Stroke and Cerebrovascular Diseases, Vol. 24, No. 5 (May), 2015: pp 912-920
THYROID DISEASE AS A RISK FACTOR FOR CVD
and the generally accepted treatment of hyperthyroidism involves temporary use of suppressive thyrostatic medication (antithyroid drugs).1,2 Contrarily, hypothyroidism is a common endocrine disorder in which the thyroid gland does not produce enough thyroid hormone. Numerous symptoms are associated with hypothyroidism, including fatigue, poor memory and concentration, swelling of the limbs, heavy periods, abnormal sensation, and so forth. In children, hypothyroidism leads to delays in growth and intelligence development.3 Hypothyroidism can be well treated with manufactured levothyroxine, which is the hormone replacement management.4 Thyroid function has a profound effect on the heart, and cardiovascular mortality rates are increased in hyperthyroidism.5 A recent nationwide study has shown that subjects had a higher risk of developing cardiovascular diseases (hazard ratio [HR], 1.34) after the diagnosis of hyperthyroidism.6 On the other hand, it has been observed that adults with hypothyroidism had a high prevalence of coronary heart disease,7 and long-lasting untreated hypothyroidism in children may be associated with proatherogenic abnormalities.8 Therefore, it has been suggested that diagnosis and treatment of cardiac disease may benefit from including analysis of thyroid hormone status.9 However, although thyroid autoantibodies may be associated with the presence of intracranial artery stenosis in young stroke patients,10 thyroid hormone cannot be used to estimate the functional stroke outcome.11 There is insufficient evidence to prove that hyperthyroidism predisposes to cerebral ischemia.12 Similarly, there is no evidence to suggest an association between hypothyroidism and ischemic stroke among healthy postmenopausal women.13 A population-based cohort study also cannot validate the comorbid relationship between hypothyroidism and mild cognitive impairment.14 On the other hand, a study reviewed records of patients admitted to the University of Louisville Stroke Center with a diagnosis of acute ischemic stroke or transient ischemic attack and found that 12% of patients with acute ischemic stroke or transient ischemic attack had hypothyroidism.15 Moreover, a cohort study has found a slightly increased risk of stroke (adjusted HR, 1.10; 95% confidence interval [CI], 1.01-1.20) in patients with autoimmune thyroiditis.16 Because the association between cerebrovascular disease (CVD) and thyroid disease remains unclear, in this study, we tried to verify that thyroid disease increases the risk of CVD development using a nationwide population-based data source, National Health Insurance Research Database (NHIRD) of Taiwan. Validating the comorbid relationship between CVD and thyroid disease may benefit from good demographic diversity of this database because insufficient supports for this association concluded from previous studies might be because of selection bias.
913
Materials and Methods National Health Insurance Research Database The National Health Insurance (NHI) program in Taiwan was initiated in 1995 and currently covers over 99% of the Taiwanese population. Nearly all hospitals and clinics in Taiwan are contracted to support the program. In 2000, the National Health Research Institutes (NHRI) began being authorized by Bureau of NHI to construct the NHIRD to make release claims data available for academic research. Identities of subjects were encrypted by Bureau of NHI before sending claims data to NHRI, and the encrypted identities would be encrypted again by NHRI before the database being released for applications from academic institutes. This is now one of the largest nationwide medical data sets in the world and has been used in many epidemiologic studies. This study used the NHIRD through application from Bureau of NHI; the version was updated in 2010, which contains 1,000,000 random samples of insurants with no significant difference in age, sex, or insurance cost between the sample and the overall population. Insurance claim data of these samples over the period from 1996 to 2010, including diagnosis records and medication orders for outpatient and inpatient visits, were provided. In this database, disease diagnosis has been coded by The International Classification of Disease, ninth Revision, Clinical Modification (ICD-9 CM).
Case Definition and Control Selection In this study, we would like to validate, respectively, the comorbid relationship between CVD and hyperthyroidism as well as hypothyroidism. The first group of cases in this study were adults (age $18 years) initially diagnosed with hyperthyroidism (ICD-9 CM codes: 242.xx and 775.3) and the second group with hypothyroidism (codes: 243 and 244.x), between 2002 and 2008, according to outpatient and/or inpatient records. For each case, the date when the diagnosis of hyperthyroidism or hypothyroidism was first made was defined as the index date. In this case–control study, each case was matched against 3 insurants, selected randomly from the NHIRD without any diagnosis record for hyperthyroidism, hypothyroidism, and thyroid goiter between 1996 and 2010. As a control, he or she must have at least 1 outpatient/inpatient visit for any condition during that year and month of the index date for his/her corresponding matched case. The date of the first visit during that year and month was assigned as the control’s index date. The control was also matched with the case in terms of age and sex. In our selection criteria, controls do not need to be healthy because exclusion of ill people as controls might distort the results.17 Controls should have the same risk of CVD as the cases, whereas healthy
M.-H. YANG ET AL.
914
Figure 1. The study period was defined from 24 months before corresponding index date until 24 months after it. The medical histories of arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure within the study period for each cohort subject were recorded. The first visit for cerebrovascular disease within 24 months after corresponding index date was recorded.
individuals might have lower possibilities for CVD development. Any subjects who had a record of diagnosis of CVD (codes: 430438.xx) before the index date were removed. For each cohort subject, as shown in Figure 1, the study period was defined from 24 months before corresponding index date until 24 months after it. The medical histories of arrhythmias (codes: 426.0, 426.1x, 426.7, 426.9, 427.xx, 785.0, 996.01, 996.04, V45.0x, and V53.3x), diabetes mellitus (codes: 250, 250.0x, 250.1x, 250.2x, 250.3x, and 250.7x), hyperlipidemia (codes: 272.x), hypertension (codes: 401.x, 402.x0, 404.x0, and 404.x2), and renal failure (codes: 403.x1, 404.x2, 404.x3, 585, 586, V42.0, V45.1, and V56.xx) within the study period for each cohort subject were recorded.
Potential Comorbid Event and Risk Factors In this study, we tried to evaluate, respectively, the comorbid relationship between CVD and hyperthyroidism as well as hypothyroidism. For each cohort subject, all outpatient/inpatient claims within the 24 months after corresponding index date were examined and the first visit for CVD was recorded. For each cohort subject, the event date for CVD was defined as date of onset for this illness, or 2010/12/31, which is censoring date of the source NHIRD if this subject had no diagnosis records for CVD. To validate the comorbid association between CVD and hyperthyroidism/hypothyroidism, respectively, we would like to take into account several risk factors for CVD as adjustment. Separate lists of patients suffering four frequent risk factors for CVD during the period from 1996 to 2010 would be extracted from the source NHIRD, respectively: diabetes mellitus, hyperlipidemia, hypertension, and renal failure. Case patients and control subjects would be screened according to existence in distinct intersections of these lists for studying variance of hazards for CVD onset from subgroups of subjects with different kinds of risk factors. Demographics and clinical variables are compared between case patients and the control group using the chi-square test or student t test when appropriate. We will use the HR by multivariate Cox regression, to
quantify risk of CVD in different groups of subjects, that is, hyperthyroidism patients versus corresponding control subjects or hypothyroidism patients versus corresponding control subjects. All HR values are multivariate adjusted by factors of age, sex, and medical histories within corresponding study periods for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure. All tests were 2-tailed and P values less than .05 were considered significant.
Results Table 1 shows the demographics of the data set, which includes 16,808 hyperthyroidism cases and 51,654 matched controls. As expected, the distribution of gender (case versus control 5 22.9% versus 23.0% for male distribution) and the mean age (case versus control 5 40.9 versus 41.3 years) were similar among hyperthyroidism cases and controls. Furthermore, there were significant differences in proportions of medical histories for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure among case and control groups. It has been noted that 35.7% (n 5 6002) of hyperthyroidism cases and 21.0% (n 5 10,862) of matched controls had visiting records for any one of diabetes mellitus, hyperlipidemia, or hypertension within corresponding study periods. To evaluate whether hyperthyroidism increases the hazard of developing CVD, we counted the number of patients whose first visit for CVD has occurred within the 2-year follow-up from the index date. The cohort subjects whose onset of CVD occurred before the index date were removed; that is why numbers of subjects between case and control groups could not follow strictly the ratio of 1 versus 3. The value of HR with 95% CI, after adjusting factors of age, sex, and medical histories for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure, was represented in Table 1. It could be observed that hyperthyroidism increased the hazard of developing CVD (multivariate adjusted HR, 1.38). After removing cohort subjects whose onset of CVD occurred before the index date, Table 2 shows the demographics of the data set, which includes 5793 hypothyroidism cases and 18,239 matched controls. Again the
THYROID DISEASE AS A RISK FACTOR FOR CVD
915
Table 1. Demographics of the data set and relative hazard of developing CVD (within the 24-month follow-up from corresponding index date) among hyperthyroidism cases versus matched controls
Variable
Hyperthyroidism (N 5 16,808)
Male (%) 3842 (22.9) Mean age, y (SD) 40.9 (14.34) No. with medical histories within study period for Arrhythmias (%) 2884 (17.2) Diabetes mellitus (%) 2269 (13.5) Hyperlipidemia (%) 3239 (19.3) Hypertension (%) 3349 (19.9) Renal failure (%) 235 (1.4) No. of developing follow-up CVD (%) 462 (2.7)
Matched control (N 5 51,654) 11,870 (23.0) 41.3 (14.51) 1893 (3.7) 3476 (6.7) 5460 (10.6) 7258 (14.1) 448 (.9) 917 (1.8)
P value
HR (95% CI)*
P value
1.38 (1.23-1.56)
,.001
.745 .012 ,.001 ,.001 ,.001 ,.001 ,.001 ,.001
Abbreviations: CI, confidence interval; CVD, cerebrovascular disease; HR, hazard ratio; SD, standard deviation. *The hazard ratio values are adjusted by factors of age, sex, and medical histories for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure.
distribution of gender (case versus control 5 16.5% versus 16.9% for male distribution) and the mean age (case versus control 5 45.1 versus 45.8 years) were similar among hypothyroidism cases and controls. But there were significant differences in proportions of medical histories for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure among case and control groups. It has been shown that 44.8% (n 5 2593) of hypothyroidism cases and 27.7% (n 5 5045) of matched controls had diagnosis records for any one of diabetes mellitus, hyperlipidemia, or hypertension within corresponding study periods. Furthermore, as shown in Table 2, hypothyroidism also increased the hazard of developing follow-up CVD (multivariate adjusted HR, 1.89). As mentioned in subsection ‘‘potential comorbid event and risk factors’’, to further evaluate the comorbid relationships between CVD and hyperthyroidism/ hypothyroidism, separate lists of patients suffering 4
frequent risk factors for CVD during the period from 1996 to 2010 have been extracted from the source NHIRD: diabetes mellitus (list 1), hyperlipidemia (list 2), hypertension (list 3), and renal failure (list 4). Then another list named as ‘‘three_hypers’’ was produced by unifying list 1, 2, and 3. Either hyperthyroidism or hypothyroidism cases who did not exist in the list ‘‘three_hypers’’ would be removed. Corresponding control subjects, who must also be presented in list ‘‘three_hypers’’ were selected according to criteria mentioned in subsection ‘‘case patient definition and control selection’’. According to results presented in Table 3, the data set consisted of 8698 hyperthyroidism cases and 26,621 matched controls, after removing cohort subjects whose onset of CVD occurring before the index date. In all, 69.0% (n 5 6002) of case patients and 54.0% (n 5 14,371) of control subjects had diagnosis records for any one of diabetes mellitus, hyperlipidemia, or hypertension within corresponding study periods. Compared with
Table 2. Demographics of the data set and relative hazard of developing CVD (within the 24-month follow-up from corresponding index date) among hypothyroidism cases versus matched controls
Variable
Hypothyroidism (N 5 5793)
Male (%) 956 (16.5) Mean age, y (SD) 45.1 (15.37) No. with medical histories within study period for Arrhythmias (%) 691 (11.9) Diabetes mellitus (%) 912 (15.7) Hyperlipidemia (%) 1597 (27.6) Hypertension (%) 1423 (24.6) Renal failure (%) 187 (3.2) No. of developing follow-up CVD (%) 257 (4.4)
Matched control (N 5 18,239) 3,088 (16.9) 45.8 (15.73) 958 (5.3) 1637 (9.0) 2483 (13.6) 3546 (19.4) 235 (1.3) 435 (2.4)
P value
HR (95% CI)*
P value
1.89 (1.61-2.22)
,.001
.448 .002 ,.001 ,.001 ,.001 ,.001 ,.001 ,.001
Abbreviations: CI, confidence interval; CVD, cerebrovascular disease; HR, hazard ratio; SD, standard deviation. *The hazard ratio values are adjusted by factors of age, sex, and medical histories for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure.
M.-H. YANG ET AL.
916
Table 3. Demographics of the data set and relative hazard of developing CVD among hyperthyroidism cases versus matched controls
Variable
Hyperthyroidism (N 5 8698)
Male (%) 2325 (26.7) Mean age, y (SD) 47.8 (14.08) No. with medical histories within study period for Arrhythmias (%) 1903 (21.9) Diabetes mellitus (%) 2269 (26.1) Hyperlipidemia (%) 3239 (37.2) Hypertension (%) 3349 (38.5) Renal failure (%) 213 (2.4) No. of developing follow-up CVD (%) 401 (4.6)
Matched control (N 5 26,621) 7148 (26.9) 48.0 (14.12) 1739 (6.5) 4482 (16.8) 7401 (27.8) 9024 (33.9) 516 (1.9) 829 (3.1)
P value
HR (95% CI)*
P value
1.34 (1.18-1.52)
,.001
.825 .233 ,.001 ,.001 ,.001 ,.001 .004 ,.001
Abbreviations: CI, confidence interval; CVD, cerebrovascular disease; HR, hazard ratio; SD, standard deviation. All cohort subjects must have diagnosis records for one of diabetes mellitus, hyperlipidemia, or hypertension in the source NHIRD. *The hazard ratio values are adjusted by factors of age, sex, and medical histories for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure.
outcomes in Table 1, this data set were older in age (Table 1 versus Table 3 5 41.2 versus 48.0 years) and had higher proportions for male gender (Table 1 versus Table 3 5 22.9% versus 26.8%) as well as visiting records for arrhythmias (Table 1 versus Table 3 5 7.0% versus 10.3%), diabetes mellitus (8.4% versus 19.1%), hyperlipidemia (12.7% versus 30.1%), hypertension (15.5% versus 35.0%), and renal failure (1.0% versus 2.1%). The comorbid association between hyperthyroidism and CVD still remained (multivariate adjusted HR, 1.34). On the other hand, after removing cohort subjects whose onset of CVD occurring before the index date, the data set whose outcomes were listed in Table 4 contained 3601 hypothyroidism cases and 11,359 matched controls. In all, 72.0% (n 5 2593) of case patients and 57.8% (n 5 6560) of control subjects had visiting records for any one of diabetes mellitus, hyperlipidemia, or hypertension within corresponding study periods. Compared with results in Table 2,
again this data set were older in age (Table 2 versus Table 4 5 45.6 versus 51.3 years) and had higher proportions for male gender (Table 2 versus Table 4 5 16.8% versus 19.3%) as well as diagnosis records for arrhythmias (Table 2 versus Table 4 5 6.9% versus 9.4%), diabetes mellitus (10.6% versus 20.0%), hyperlipidemia (17.0% versus 32.8%), hypertension (20.7% versus 38.6%), and renal failure (1.8% versus 3.0%). The comorbid relationship between hypothyroidism and CVD was significant too (multivariate adjusted HR, 1.80). Finally, we would like to know whether the comorbid association between hyperthyroidism/hypothyroidism and CVD was comparable to influences from risk factors of CVD. Based on the patient lists described in last paragraph for diabetes mellitus (list 1), hypertension (list 3), and renal failure (list 4), 3 further lists were constructed. The first one named as ‘‘HT_woDM_woRF’’ was produced by copying list 3 but excluding entries from list 1
Table 4. Demographics of the data set and relative hazard of developing CVD among hypothyroidism cases versus matched controls
Variable
Hypothyroidism (N 5 3601)
Male (%) 690 (19.2) Mean age, y (SD) 50.8 (14.61) No. with medical histories within study period for Arrhythmias (%) 548 (15.2) Diabetes mellitus (%) 912 (25.3) Hyperlipidemia (%) 1597 (44.3) Hypertension (%) 1423 (39.5) Renal failure (%) 170 (4.7) No. of developing follow-up CVD (%) 230 (6.4)
Matched control (N 5 11,359) 2199 (19.4) 51.5 (14.96) 864 (7.6) 2075 (18.3) 3315 (29.2) 4349 (38.3) 274 (2.4) 423 (3.7)
P value
HR (95% CI)*
P value
1.80 (1.53-2.13)
,.001
.793 .012 ,.001 ,.001 ,.001 .186 ,.001 ,.001
Abbreviations: CI, confidence interval; CVD, cerebrovascular disease; HR, hazard ratio; SD, standard deviation. All cohort subjects must have diagnosis records for one of diabetes mellitus, hyperlipidemia, or hypertension in the source NHIRD. *The hazard ratio values are adjusted by factors of age, sex, and medical histories for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure.
THYROID DISEASE AS A RISK FACTOR FOR CVD
917
Table 5. Demographics of the data set and relative hazard of developing CVD among hyperthyroidism cases accompanied with matched controls
Variable
Hyperthyroidism (N 5 9840)
Male (%) 1884 (19.1) Mean age, y (SD) 34.9 (10.95) No. with medical histories within study period for Arrhythmias (%) 1262 (12.8) Diabetes mellitus (%) 0 (0) Hyperlipidemia (%) 863 (8.8) Hypertension (%) 0 (0) Renal failure (%) 0 (0) No. of developing follow-up CVD (%) 95 (1.0)
Matched control (N 5 29,849) 5726 (19.2) 35.0 (11.00) 566 (1.9) 0 (0) 992 (3.3) 0 (0) 0 (0) 141 (.5)
P value
HR (95% CI)*
P value
1.88 (1.43-2.48)
,.001
.936 .417 ,.001 NA ,.001 NA NA ,.001
Abbreviations: CI, confidence interval; CVD, cerebrovascular disease; HR, hazard ratio; NA, not applicable; SD, standard deviation. All cohort subjects were without any diagnosis records for all diabetes mellitus, hypertension, and renal failure in the source NHIRD. *The hazard ratio values are adjusted by factors of age, sex, and medical histories for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure.
and/or 4. The second named as ‘‘HT_DM_noRF’’ was created by intersecting lists 1 and 3 but excluding entries from list 4. The last one named as ‘‘HT_DM_RF’’ was generated by intersecting lists 1, 3, and 4. It should be noted that subjects from these 3 lists were exclusive. Accordingly, hyperthyroidism patients were selected to constitute 4 subgroups: cases existed in list ‘‘HT_woDM_ woRF’’, cases existed in list ‘‘HT_DM_noRF’’, cases existed in list ‘‘HT_DM_RF’’, and cases without any diagnosis records for diabetes mellitus, hypertension, and renal failure. Control subjects, who must also have corresponding set of risk factors for matched case subgroup, were sampled according to criteria mentioned in subsection ‘‘case patient definition and control selection’’. Likewise, hypothyroidism patients were also selected to constitute 4 subgroups: cases existed in list ‘‘HT_woDM_ woRF’’, cases existed in list ‘‘HT_DM_noRF’’, cases existed in list ‘‘HT_DM_RF’’, and cases without any diagnosis records for diabetes mellitus, hypertension, and renal failure. Again control subjects, who must also be with corresponding set of risk factors for matched case subgroup, were sampled according to criteria mentioned previously. All cohort subjects, either case or control samples, whose onset of CVD occurring before the index date would be removed. Observing the results presented in Table 5, for cohort subjects without any diagnosis records for all diabetes mellitus, hypertension, and renal failure in the source NHIRD, there was significant difference between hyperthyroidism cases and matched controls for the risk of developing follow-up CVD (multivariate adjusted HR, 1.88). On the other hand, for cohort subjects with visiting records for risk factors of CVD in the source NHIRD, the relative hazards of developing CVD between hyperthyroidism cases and matched controls did not reach statistical significance (for cohort of Table S1 in Appendix with diagnosis records for all diabetes mellitus, hypertension,
and renal failure in the source NHIRD, P value 5 .178; for cohort of Table S3 in Appendix with diagnosis records for hypertension, but without any visiting records for both diabetes mellitus and renal failure in the source NHIRD, P value 5 .196). The exception was cohort subjects with diagnosis records for both diabetes mellitus and hypertension, but without any visiting records for renal failure in the source NHIRD; the case group was 1.30 times higher than corresponding control group to develop follow-up CVD (multivariate adjusted HR, 1.30; 95% CI, 1.06-1.60; P value 5 .012; Table S2 in Appendix). However, for various subgroups of subjects with different combinations of risk factors for CVD, the hypothyroidism patients always had a significantly higher hazard for follow-up CVD development than matched controls (Table 6 and Tables S4-S6 in Appendix). For cohort subjects with diagnosis records for all diabetes mellitus, hypertension, and renal failure in the source NHIRD, the multivariate adjusted HR was 1.76 (Table S4 in Appendix). For subjects with diagnosis records for both diabetes mellitus and hypertension, but without any visiting records for renal failure in the source NHIRD, the multivariate adjusted HR value was 1.77 (Table S5 in Appendix). For subjects with diagnosis records for hypertension, but without any visiting records for both diabetes mellitus and renal failure in the source NHIRD, the multivariate adjusted HR value was 2.23 (Table S6 in Appendix). Finally, for subjects without any diagnosis records for all diabetes mellitus, hypertension, and renal failure in the source NHIRD, the multivariate adjusted HR value was 2.18 (Table 6).
Discussion In this study, we have found that hyperthyroidism increased 38% the hazard of developing follow-up CVD (multivariate adjusted HR, 1.38; Table 1). However,
M.-H. YANG ET AL.
918
Table 6. Demographics of the data set and relative hazard of developing CVD among hypothyroidism cases accompanied with matched controls
Variable
Hypothyroidism (N 5 2893)
Male (%) 372 (12.9) Mean age, y (SD) 37.4 (11.98) No. with medical histories within study period for Arrhythmias (%) 199 (6.9) Diabetes mellitus (%) 0 (0) Hyperlipidemia (%) 422 (14.6) Hypertension (%) 0 (0) Renal failure (%) 0 (0) No. of developing follow-up CVD (%) 37 (1.3)
Matched control (N 5 8811) 1159 (13.2) 37.5 (12.04) 213 (2.4) 0 (0) 382 (4.3) 0 (0) 0 (0) 46 (.5)
P value
HR (95% CI)*
P value
2.18 (1.39-3.42)
.001
.683 .610 ,.001 NA ,.001 NA NA ,.001
Abbreviations: CI, confidence interval; CVD, cerebrovascular disease; HR, hazard ratio; NA, not applicable; SD, standard deviation. All cohort subjects were without any diagnosis records for all diabetes mellitus, hypertension, and renal failure in the source NHIRD. *The hazard ratio values are adjusted by factors of age, sex, and medical histories for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure.
focused on cohort subjects with diagnosis records for some risk factors of CVD in the source NHIRD, the comorbid association between hyperthyroidism and CVD would be reduced. Four frequent risk factors for CVD were taken into account: diabetes mellitus, hyperlipidemia, hypertension, and renal failure. For instance, concentrated on subjects with visiting records for any one of diabetes mellitus, hyperlipidemia, or hypertension in the source NHIRD, hyperthyroidism increased 34% the hazard of developing follow-up CVD (multivariate adjusted HR, 1.34; Table 3). Moreover, for cohort subjects with diagnosis records for both diabetes mellitus and hypertension, but without any visiting records for renal failure in the source NHIRD, hyperthyroidism increased 30% the hazard of developing follow-up CVD (multivariate adjusted HR, 1.30; Table S2 in Appendix). Finally, for cohort subjects with diagnosis records for all diabetes mellitus, hypertension, and renal failure in the source NHIRD, the comorbid relationship between hyperthyroidism and CVD did not reach statistical significance (P value 5 .178; Table S1 in Appendix); neither did the sample group with diagnosis records for hypertension, but without any visiting records for both diabetes mellitus and renal failure in the source NHIRD (P value 5 .196; Table S3 in Appendix). The results suggested that the comorbid association between hyperthyroidism and CVD was weaker than those influences from risk factors such as diabetes mellitus, hyperlipidemia, hypertension, and/or renal failure, and so forth. The observation for which cluster of hyperthyroidism cases was 1.88 times higher than corresponding control cluster to develop follow-up CVD in the sample group without any diagnosis records for all diabetes mellitus, hypertension, and renal failure in the source NHIRD (multivariate adjusted HR, 1.88; Table 5), further validated this assumption. However, it still should be
cautious for such conclusion because in diabetic hemodialysis patients, sudden cardiac death may be influenced by hyperthyroidism and regular assessment of thyroid status may help estimate the cardiac risk.18 We have also found that hypothyroidism increased even higher the hazard of developing follow-up CVD (multivariate adjusted HR, 1.89; Table 2). The comorbid association between hypothyroidism and CVD always remained even focused on cohort subjects with diagnosis records for some risk factors of CVD in the source NHIRD. For instance, concentrated on subjects with visiting records for any one of diabetes mellitus, hyperlipidemia, or hypertension in the source NHIRD, hypothyroidism increased 80% the hazard of developing follow-up CVD (multivariate adjusted HR, 1.80; Table 4). Moreover, for cohort subjects with diagnosis records for all diabetes mellitus, hypertension, and renal failure in the source NHIRD, the case group was 76% higher than corresponding group for the risk of CVD development (multivariate adjusted HR, 1.76; Table S4 in Appendix). Similarly, the multivariate adjusted HR value was 1.77 (Table S5 in Appendix) in the sample group with diagnosis records for both diabetes mellitus and hypertension, but without any visiting records for renal failure in the source NHIRD. Finally, the multivariate adjusted HR value was 2.23 (Table S6 in Appendix) in cohort subjects with diagnosis records for hypertension, but without any visiting records for both diabetes mellitus and renal failure in the source NHIRD. The results suggested that the comorbid association between hypothyroidism and CVD was comparable to those influences from risk factors such as diabetes mellitus, hyperlipidemia, hypertension, and/or renal failure, and so forth. The observation for which cluster of hypothyroidism cases was 2.18 times higher than corresponding control cluster to develop follow-up CVD in the sample group without any
THYROID DISEASE AS A RISK FACTOR FOR CVD
diagnosis records for all diabetes mellitus, hypertension, and renal failure in the source NHIRD (multivariate adjusted HR 5 2.18; Table 6), further validated this assumption. One recent study has reported that 28% of ischemic stroke patients had thyroid hormone concentrations outside the reference range.11 Hyperthyroidism is associated with an enhanced metabolic state and an increased production of reactive oxygen species (ROS). ROS generation overrides the ability of endogenous antioxidant enzymes as hyperthyroidism promotes the depletion of antioxidant molecules. Consequently, the deleterious effects of thyroid hormones on heart tissue involve oxidative stress. Moreover, hyperthyroidism sensitizes the rat heart to reperfusion injury, which is characterized by severe arrhythmias and tissue damage.19 It is also known that hyperthyroidism is associated with increased left ventricular mass of the heart,5 hemodynamic changes, endothelial dysfunction, and coagulopathy.6 Genetic confounding and living habit might contribute to the relationship between cardiovascular disease and hyperthyroidism, as both disorders show familial aggregation and have the risk factor of smoking.6 Cerebral venous thrombosis and arterial ischemic events can happen concomitantly in patients with hyperthyroidism,12 whereas thyroid autoantibodies may be associated with the presence of intracranial artery stenosis in stroke patients.10 Thyroid hormone can regulate blood lipid levels, then influence the occurrence and progress of coronary heart disease. The promoter of the low-density lipoprotein (LDL) receptor gene contains a thyroid hormone responsive element, which can promote free triiodothyronine to increase the expression of LDL receptors, therefore, increasing the clearance of LDL.7 It has been confirmed that serum free triiodothyronine levels were inversely correlated with carotid atherosclerosis.20 Moreover, hypothyroidism can increase the levels of several cardiovascular risk factors such as total cholesterol, triglyceride, and LDL cholesterol, and accelerate the development of atherosclerosis and coronary artery disease.7,21 Low thyroid hormones alter the arterial muscle structure, then cause thickening of the vessel wall and increased stiffness of the heart.5,7 A record review study suggested that elevated homocysteine levels associated with hypothyroidism may represent a risk factor of stroke.15 However, higher effect size was found in the first year after diagnosis of autoimmune thyroiditis (HR, 1.33; 95% CI, 1.14-1.56), but not in the long term, consistent with a residual effect of hypothyroidism.16 Consequently, it is believed that L-thyroxine treatment and not low thyroid hormones per se explained this increased risk because L-thyroxine elevates coagulation factor levels and inhibits fibrinolysis in a dose-dependent manner.16,22 It has been observed in our analysis results that cohort subjects, who had diagnosis records in the source NHIRD
919
for more kinds of risk factors for CVD, also were older in age, and had higher proportions for CVD development, male gender, as well as visiting records within study periods for arrhythmias. It is true that older people have a higher risk for stroke than the general population and the risk of stroke doubles for every decade after the age of 55 years.23,24 Besides, men are 25% more likely to suffer strokes than women.23,24 Finally, other important risk factors that contribute to stroke onset include hypertension, heart disease, blood cholesterol level, diabetes, and so forth.23 All outcomes of our analysis coincide with these medical theories. We acknowledge that this study had several limitations. First, the disease diagnoses were coded according to the ICD-9 CM and obtained from administrative claims reported by hospitals or clinics, which may be considered less accurate than clinical diagnoses by standard criteria. Despite this potential drawback, the data remain representative, as the Bureau of NHI routinely and randomly sampled a fixed percentage of claims each year from every contracted medical institution. Moreover, an independent group of physicians reviewed the charts and validated the diagnoses. Consequently, hundreds of studies based on the NHIRD can be found in the US National Library of Medicine (PubMed). Second, the administrative claims data from the NHIRD did not include detailed personal information (e.g., body mass index, living habits, or laboratory test results), which may add confounding factors to thyroid disease or CVD. Finally, enhanced metabolic state induced by hyperthyroidism will increase the production of ROS, and estrogen receptors make the hyperthyroid heart more susceptible to ischemia/reperfusion. Because tamoxifen is well known as an antioxidant and is a selective antagonist of estrogen receptors, a recent study has reported that tamoxifen inhibits the hypersensitivity of hyperthyroid rat myocardium to reperfusion damage.19 Medicine treatments for thyroid disorder should be taken into account to verify the comorbid relationship between thyroid disease and CVD.
Conclusion As far as we know, the association between CVD and thyroid disease remains uncertain. Because insufficient evidence for this association supported by previous studies might be because of selection bias, in this article we used a nationwide population-based database with good demographic diversity to validate the comorbid relationship between CVD and thyroid disease. According to the evaluation results, hyperthyroidism increased 38% the hazard of developing follow-up CVD (multivariate adjusted HR, 1.38) whereas hypothyroidism increased even higher the risk (multivariate adjusted HR, 1.89). Further stratification studies for risk factors of CVD suggested that the comorbid association between
M.-H. YANG ET AL.
920
hypothyroidism and CVD was comparable to those influences from cardiac risk factors. Advanced analysis is required to investigate the pathologic mechanism underlying the association between CVD and thyroid disease. Acknowledgments: The authors would like to thank Wei-Zen Sun, MD, at the Department of Anesthesiology, National Taiwan University Hospital, for provision and utilization of the source NHIRD.
Supplementary Data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.jstrokecerebrovas dis.2014.11.032.
References 1. Siegenthaler W. Differential diagnosis in internal medicine: from symptom to diagnosis. Stuttgart, Germany: Thieme Medical Publishers 2007. 2. Torre Dario M, Lamb Geoffrey C, Ruiswyk Jerome Van, et al. Kochar’s clinical medicine for students. Philadelphia, PA: Lippincott Williams & Wilkins 2008. 3. Dan L, Fauci A, Kasper D, et al. Harrison’s principles of internal medicine. New York, NY: McGraw-Hill Professional 2011. 4. Khandelwal D, Tandon N. Overt and subclinical hypothyroidism: who to treat and how. Drugs 2012;72:17-33. 5. Faber J, Selmer C. Cardiovascular disease and thyroid function. Front Horm Res 2014;43:45-56. 6. Brandt F, Thvilum M, Almind D, et al. Morbidity before and after the diagnosis of hyperthyroidism: a nationwide register-based study. PLoS One 2013;8:e66711. 7. Bai MF, Gao CY, Yang CK, et al. Effects of thyroid dysfunction on the severity of coronary artery lesions and its prognosis. J Cardiol 2014;64:496-500. 8. Cerbone M, Capalbo D, Wasniewska M, et al. Cardiovascular risk factors in children with long-standing untreated idiopathic subclinical hypothyroidism. J Clin Endocrinol Metab 2014;99:2697-2703. 9. Danzi S, Klein I. Thyroid disease and the cardiovascular system. Endocrinol Metab Clin North Am 2014;43: 517-528. 10. Shi Z, Zhang X, Chen Z, et al. Elevated thyroid autoantibodies and intracranial stenosis in stroke at an early age. Int J Stroke 2014;9:735-740.
11. Leonards CO, Schneider HJ, Liman TG, et al. Thyroidstimulating hormone, white matter hyperintensities, and functional outcome in acute ischemic stroke patients. Cerebrovasc Dis Extra 2014;4:61-68. 12. Janovsky CC, Fukuda TG, Silva GS, et al. An unusual association between acute ischaemic stroke and cerebral venous thrombosis with thyrotoxic state. BMJ Case Rep 2013;2013. 13. Giri A, Edwards TL, LeGrys VA, et al. Subclinical hypothyroidism and risk for incident ischemic stroke among postmenopausal women. Thyroid 2014;24:1210-1217. 14. Parsaik AK, Singh B, Roberts RO, et al. Hypothyroidism and risk of mild cognitive impairment in elderly persons: a population-based study. JAMA Neurol 2014; 71:201-207. 15. Remmel KS, Wanahita A, Moore K, et al. Acute ischemic stroke and hypothyroidism. J Ky Med Assoc 2006; 104:191-193. 16. Karch A, Thomas SL. Autoimmune thyroiditis as a risk factor for stroke: a historical cohort study. Neurology 2014;82:1643-1652. 17. Grimes DA, Schulz KF. Compared to what? Finding controls for case-control studies. Lancet 2005;365:1429-1433. 18. Drechsler C, Schneider A, Gutjahr-Lengsfeld L, et al. Thyroid function, cardiovascular events, and mortality in diabetic hemodialysis patients. Am J Kidney Dis 2014; 63:988-996. 19. Pavon N, Hernandez-Esquivel L, Buelna-Chontal M, et al. Antiarrhythmic effect of tamoxifen on the vulnerability induced by hyperthyroidism to heart ischemia/ reperfusion damage. J Steroid Biochem Mol Biol 2014; 143:416-423. 20. Tatar E, Kircelli F, Asci G, et al. Associations of triiodothyronine levels with carotid atherosclerosis and arterial stiffness in hemodialysis patients. Clin J Am Soc Nephrol 2011;6:2240-2246. 21. Pei YJ, Wang AM, Zhao Y, et al. Studies of cardiovascular risk factors in polycystic ovary syndrome patients combined with subclinical hypothyroidism. Gynecol Endocrinol 2014;30:553-556. 22. Van Zaane B, Squizzato A, Debeij J, et al. Alterations in coagulation and fibrinolysis after levothyroxine exposure in healthy volunteers: a controlled randomized crossover study. J Thromb Haemost 2011;9:1816-1824. 23. Stroke: hope through research. National Institute of Neurological Disorders and Stroke (NINDS). http:// www.ninds.nih.gov/disorders/stroke/detail_stroke.htm. July 25, 2014. 24. Haast RA, Gustafson DR, Kiliaan AJ. Sex differences in stroke. Journal of cerebral blood flow and metabolism. J Cereb Blood Flow Metab 2012;32:2100-2107.
Background: Thyroid disease is the medical condition impairing function of the thyroid. Among this disorder category, hyperthyroidism is that the thyroid gland produces excessive amounts of thyroid hormones whereas hypothyroidism is that the thyroid gland does not produce enough thyroid hormone. Various studies have supported the comorbid association between thyroid disease and cardiovascular disorder. However, there is insufficient evidence to prove the relationship between cerebrovascular disease (CVD) and thyroid disease. Methods: In this study, we tried to verify that thyroid disease increases the risk of CVD development employing a population-based database, National Health Insurance Research Database of Taiwan. A total of 16,808 hyperthyroidism cases and 5793 hypothyroidism patients with corresponding control subjects were studied, respectively. Hazard ratio (HR) by the Cox regression was used to quantify risk of CVD in different groups of subjects, that is, case patients versus matched controls. Further stratification studies for risk factors of CVD were performed to evaluate the comorbid association between CVD and hyperthyroidism/hypothyroidism. Results: Evaluation results have shown that hyperthyroidism increased 38% of the hazard of developing follow-up CVD (adjusted HR, 1.38) whereas hypothyroidism increased even higher the risk (adjusted HR, 1.89). Further stratification studies for risk factors of CVD suggested that the comorbid association between hypothyroidism and CVD was comparable to those influences from cardiac risk factors, such as diabetes mellitus, hyperlipidemia, hypertension, or renal failure and so forth. Conclusions: Thyroid disease may predispose to onset of CVD. Advanced analysis is required to investigate the pathologic mechanism underlying the association between CVD and thyroid disease. Key Words: Hyperthyroidism—hypothyroidism—cerebrovascular disease—hazard ratio—NHIRD. Ó 2015 by National Stroke Association
Thyroid disease is the medical condition impairing function of the thyroid. Among this disorder category, hyperthyroidism is a condition in which the thyroid gland produces and secretes excessive amounts of the free thyroid hormones: triiodothyronine (T3) and/or thyroxine (T4). Symptoms of hyperthyroidism may be
nervousness, irritability, heart racing, hand tremors, anxiety, muscular weakness, and so forth. Thyroid hormone is also critical to normal function of cells, therefore in excess, it both overstimulates metabolism and exacerbates the effect of sympathetic nervous system. Graves’ disease is the most common cause of hyperthyroidism,
From the *Department of Computer Science and Information Engineering, National Kaohsiung University of Applied Sciences, Kaohsiung, Taiwan; †Department of Neurology, Taipei Tzu Chi General Hospital, New Taipei, Taiwan; ‡Department of Dermatology, TaipeiVeterans General Hospital, Taipei City, Taiwan; and xFaculty of Medicine, Department of Dermatology, National Yang-Ming University, Taipei City, Taiwan, ROC. Received October 11, 2014; revision received November 25, 2014; accepted November 28, 2014. This research has been supported by National Kaohsiung University of Applied Sciences and National Taiwan University. The authors declare that they have no competing interest.
Analysis procedures were designed by M.-H.Y. and F.-Y.Y. Data collection and analysis was performed by M.-H.Y. and supervised by both D.-D.L. and F.-Y.Y. Article was prepared by M.-H.Y. and D.-D.L. All authors read and approved the final article. Address correspondence to Fu-Yi Yang, MD, The Department of Neurology, Taipei Tzu Chi General Hospital, No. 289, Jianguo Rd., Xindian District, New Taipei 23142, Taiwan, ROC. E-mail: [email protected]; [email protected]. 1052-3057/$ - see front matter Ó 2015 by National Stroke Association http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2014.11.032
912
Journal of Stroke and Cerebrovascular Diseases, Vol. 24, No. 5 (May), 2015: pp 912-920
THYROID DISEASE AS A RISK FACTOR FOR CVD
and the generally accepted treatment of hyperthyroidism involves temporary use of suppressive thyrostatic medication (antithyroid drugs).1,2 Contrarily, hypothyroidism is a common endocrine disorder in which the thyroid gland does not produce enough thyroid hormone. Numerous symptoms are associated with hypothyroidism, including fatigue, poor memory and concentration, swelling of the limbs, heavy periods, abnormal sensation, and so forth. In children, hypothyroidism leads to delays in growth and intelligence development.3 Hypothyroidism can be well treated with manufactured levothyroxine, which is the hormone replacement management.4 Thyroid function has a profound effect on the heart, and cardiovascular mortality rates are increased in hyperthyroidism.5 A recent nationwide study has shown that subjects had a higher risk of developing cardiovascular diseases (hazard ratio [HR], 1.34) after the diagnosis of hyperthyroidism.6 On the other hand, it has been observed that adults with hypothyroidism had a high prevalence of coronary heart disease,7 and long-lasting untreated hypothyroidism in children may be associated with proatherogenic abnormalities.8 Therefore, it has been suggested that diagnosis and treatment of cardiac disease may benefit from including analysis of thyroid hormone status.9 However, although thyroid autoantibodies may be associated with the presence of intracranial artery stenosis in young stroke patients,10 thyroid hormone cannot be used to estimate the functional stroke outcome.11 There is insufficient evidence to prove that hyperthyroidism predisposes to cerebral ischemia.12 Similarly, there is no evidence to suggest an association between hypothyroidism and ischemic stroke among healthy postmenopausal women.13 A population-based cohort study also cannot validate the comorbid relationship between hypothyroidism and mild cognitive impairment.14 On the other hand, a study reviewed records of patients admitted to the University of Louisville Stroke Center with a diagnosis of acute ischemic stroke or transient ischemic attack and found that 12% of patients with acute ischemic stroke or transient ischemic attack had hypothyroidism.15 Moreover, a cohort study has found a slightly increased risk of stroke (adjusted HR, 1.10; 95% confidence interval [CI], 1.01-1.20) in patients with autoimmune thyroiditis.16 Because the association between cerebrovascular disease (CVD) and thyroid disease remains unclear, in this study, we tried to verify that thyroid disease increases the risk of CVD development using a nationwide population-based data source, National Health Insurance Research Database (NHIRD) of Taiwan. Validating the comorbid relationship between CVD and thyroid disease may benefit from good demographic diversity of this database because insufficient supports for this association concluded from previous studies might be because of selection bias.
913
Materials and Methods National Health Insurance Research Database The National Health Insurance (NHI) program in Taiwan was initiated in 1995 and currently covers over 99% of the Taiwanese population. Nearly all hospitals and clinics in Taiwan are contracted to support the program. In 2000, the National Health Research Institutes (NHRI) began being authorized by Bureau of NHI to construct the NHIRD to make release claims data available for academic research. Identities of subjects were encrypted by Bureau of NHI before sending claims data to NHRI, and the encrypted identities would be encrypted again by NHRI before the database being released for applications from academic institutes. This is now one of the largest nationwide medical data sets in the world and has been used in many epidemiologic studies. This study used the NHIRD through application from Bureau of NHI; the version was updated in 2010, which contains 1,000,000 random samples of insurants with no significant difference in age, sex, or insurance cost between the sample and the overall population. Insurance claim data of these samples over the period from 1996 to 2010, including diagnosis records and medication orders for outpatient and inpatient visits, were provided. In this database, disease diagnosis has been coded by The International Classification of Disease, ninth Revision, Clinical Modification (ICD-9 CM).
Case Definition and Control Selection In this study, we would like to validate, respectively, the comorbid relationship between CVD and hyperthyroidism as well as hypothyroidism. The first group of cases in this study were adults (age $18 years) initially diagnosed with hyperthyroidism (ICD-9 CM codes: 242.xx and 775.3) and the second group with hypothyroidism (codes: 243 and 244.x), between 2002 and 2008, according to outpatient and/or inpatient records. For each case, the date when the diagnosis of hyperthyroidism or hypothyroidism was first made was defined as the index date. In this case–control study, each case was matched against 3 insurants, selected randomly from the NHIRD without any diagnosis record for hyperthyroidism, hypothyroidism, and thyroid goiter between 1996 and 2010. As a control, he or she must have at least 1 outpatient/inpatient visit for any condition during that year and month of the index date for his/her corresponding matched case. The date of the first visit during that year and month was assigned as the control’s index date. The control was also matched with the case in terms of age and sex. In our selection criteria, controls do not need to be healthy because exclusion of ill people as controls might distort the results.17 Controls should have the same risk of CVD as the cases, whereas healthy
M.-H. YANG ET AL.
914
Figure 1. The study period was defined from 24 months before corresponding index date until 24 months after it. The medical histories of arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure within the study period for each cohort subject were recorded. The first visit for cerebrovascular disease within 24 months after corresponding index date was recorded.
individuals might have lower possibilities for CVD development. Any subjects who had a record of diagnosis of CVD (codes: 430438.xx) before the index date were removed. For each cohort subject, as shown in Figure 1, the study period was defined from 24 months before corresponding index date until 24 months after it. The medical histories of arrhythmias (codes: 426.0, 426.1x, 426.7, 426.9, 427.xx, 785.0, 996.01, 996.04, V45.0x, and V53.3x), diabetes mellitus (codes: 250, 250.0x, 250.1x, 250.2x, 250.3x, and 250.7x), hyperlipidemia (codes: 272.x), hypertension (codes: 401.x, 402.x0, 404.x0, and 404.x2), and renal failure (codes: 403.x1, 404.x2, 404.x3, 585, 586, V42.0, V45.1, and V56.xx) within the study period for each cohort subject were recorded.
Potential Comorbid Event and Risk Factors In this study, we tried to evaluate, respectively, the comorbid relationship between CVD and hyperthyroidism as well as hypothyroidism. For each cohort subject, all outpatient/inpatient claims within the 24 months after corresponding index date were examined and the first visit for CVD was recorded. For each cohort subject, the event date for CVD was defined as date of onset for this illness, or 2010/12/31, which is censoring date of the source NHIRD if this subject had no diagnosis records for CVD. To validate the comorbid association between CVD and hyperthyroidism/hypothyroidism, respectively, we would like to take into account several risk factors for CVD as adjustment. Separate lists of patients suffering four frequent risk factors for CVD during the period from 1996 to 2010 would be extracted from the source NHIRD, respectively: diabetes mellitus, hyperlipidemia, hypertension, and renal failure. Case patients and control subjects would be screened according to existence in distinct intersections of these lists for studying variance of hazards for CVD onset from subgroups of subjects with different kinds of risk factors. Demographics and clinical variables are compared between case patients and the control group using the chi-square test or student t test when appropriate. We will use the HR by multivariate Cox regression, to
quantify risk of CVD in different groups of subjects, that is, hyperthyroidism patients versus corresponding control subjects or hypothyroidism patients versus corresponding control subjects. All HR values are multivariate adjusted by factors of age, sex, and medical histories within corresponding study periods for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure. All tests were 2-tailed and P values less than .05 were considered significant.
Results Table 1 shows the demographics of the data set, which includes 16,808 hyperthyroidism cases and 51,654 matched controls. As expected, the distribution of gender (case versus control 5 22.9% versus 23.0% for male distribution) and the mean age (case versus control 5 40.9 versus 41.3 years) were similar among hyperthyroidism cases and controls. Furthermore, there were significant differences in proportions of medical histories for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure among case and control groups. It has been noted that 35.7% (n 5 6002) of hyperthyroidism cases and 21.0% (n 5 10,862) of matched controls had visiting records for any one of diabetes mellitus, hyperlipidemia, or hypertension within corresponding study periods. To evaluate whether hyperthyroidism increases the hazard of developing CVD, we counted the number of patients whose first visit for CVD has occurred within the 2-year follow-up from the index date. The cohort subjects whose onset of CVD occurred before the index date were removed; that is why numbers of subjects between case and control groups could not follow strictly the ratio of 1 versus 3. The value of HR with 95% CI, after adjusting factors of age, sex, and medical histories for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure, was represented in Table 1. It could be observed that hyperthyroidism increased the hazard of developing CVD (multivariate adjusted HR, 1.38). After removing cohort subjects whose onset of CVD occurred before the index date, Table 2 shows the demographics of the data set, which includes 5793 hypothyroidism cases and 18,239 matched controls. Again the
THYROID DISEASE AS A RISK FACTOR FOR CVD
915
Table 1. Demographics of the data set and relative hazard of developing CVD (within the 24-month follow-up from corresponding index date) among hyperthyroidism cases versus matched controls
Variable
Hyperthyroidism (N 5 16,808)
Male (%) 3842 (22.9) Mean age, y (SD) 40.9 (14.34) No. with medical histories within study period for Arrhythmias (%) 2884 (17.2) Diabetes mellitus (%) 2269 (13.5) Hyperlipidemia (%) 3239 (19.3) Hypertension (%) 3349 (19.9) Renal failure (%) 235 (1.4) No. of developing follow-up CVD (%) 462 (2.7)
Matched control (N 5 51,654) 11,870 (23.0) 41.3 (14.51) 1893 (3.7) 3476 (6.7) 5460 (10.6) 7258 (14.1) 448 (.9) 917 (1.8)
P value
HR (95% CI)*
P value
1.38 (1.23-1.56)
,.001
.745 .012 ,.001 ,.001 ,.001 ,.001 ,.001 ,.001
Abbreviations: CI, confidence interval; CVD, cerebrovascular disease; HR, hazard ratio; SD, standard deviation. *The hazard ratio values are adjusted by factors of age, sex, and medical histories for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure.
distribution of gender (case versus control 5 16.5% versus 16.9% for male distribution) and the mean age (case versus control 5 45.1 versus 45.8 years) were similar among hypothyroidism cases and controls. But there were significant differences in proportions of medical histories for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure among case and control groups. It has been shown that 44.8% (n 5 2593) of hypothyroidism cases and 27.7% (n 5 5045) of matched controls had diagnosis records for any one of diabetes mellitus, hyperlipidemia, or hypertension within corresponding study periods. Furthermore, as shown in Table 2, hypothyroidism also increased the hazard of developing follow-up CVD (multivariate adjusted HR, 1.89). As mentioned in subsection ‘‘potential comorbid event and risk factors’’, to further evaluate the comorbid relationships between CVD and hyperthyroidism/ hypothyroidism, separate lists of patients suffering 4
frequent risk factors for CVD during the period from 1996 to 2010 have been extracted from the source NHIRD: diabetes mellitus (list 1), hyperlipidemia (list 2), hypertension (list 3), and renal failure (list 4). Then another list named as ‘‘three_hypers’’ was produced by unifying list 1, 2, and 3. Either hyperthyroidism or hypothyroidism cases who did not exist in the list ‘‘three_hypers’’ would be removed. Corresponding control subjects, who must also be presented in list ‘‘three_hypers’’ were selected according to criteria mentioned in subsection ‘‘case patient definition and control selection’’. According to results presented in Table 3, the data set consisted of 8698 hyperthyroidism cases and 26,621 matched controls, after removing cohort subjects whose onset of CVD occurring before the index date. In all, 69.0% (n 5 6002) of case patients and 54.0% (n 5 14,371) of control subjects had diagnosis records for any one of diabetes mellitus, hyperlipidemia, or hypertension within corresponding study periods. Compared with
Table 2. Demographics of the data set and relative hazard of developing CVD (within the 24-month follow-up from corresponding index date) among hypothyroidism cases versus matched controls
Variable
Hypothyroidism (N 5 5793)
Male (%) 956 (16.5) Mean age, y (SD) 45.1 (15.37) No. with medical histories within study period for Arrhythmias (%) 691 (11.9) Diabetes mellitus (%) 912 (15.7) Hyperlipidemia (%) 1597 (27.6) Hypertension (%) 1423 (24.6) Renal failure (%) 187 (3.2) No. of developing follow-up CVD (%) 257 (4.4)
Matched control (N 5 18,239) 3,088 (16.9) 45.8 (15.73) 958 (5.3) 1637 (9.0) 2483 (13.6) 3546 (19.4) 235 (1.3) 435 (2.4)
P value
HR (95% CI)*
P value
1.89 (1.61-2.22)
,.001
.448 .002 ,.001 ,.001 ,.001 ,.001 ,.001 ,.001
Abbreviations: CI, confidence interval; CVD, cerebrovascular disease; HR, hazard ratio; SD, standard deviation. *The hazard ratio values are adjusted by factors of age, sex, and medical histories for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure.
M.-H. YANG ET AL.
916
Table 3. Demographics of the data set and relative hazard of developing CVD among hyperthyroidism cases versus matched controls
Variable
Hyperthyroidism (N 5 8698)
Male (%) 2325 (26.7) Mean age, y (SD) 47.8 (14.08) No. with medical histories within study period for Arrhythmias (%) 1903 (21.9) Diabetes mellitus (%) 2269 (26.1) Hyperlipidemia (%) 3239 (37.2) Hypertension (%) 3349 (38.5) Renal failure (%) 213 (2.4) No. of developing follow-up CVD (%) 401 (4.6)
Matched control (N 5 26,621) 7148 (26.9) 48.0 (14.12) 1739 (6.5) 4482 (16.8) 7401 (27.8) 9024 (33.9) 516 (1.9) 829 (3.1)
P value
HR (95% CI)*
P value
1.34 (1.18-1.52)
,.001
.825 .233 ,.001 ,.001 ,.001 ,.001 .004 ,.001
Abbreviations: CI, confidence interval; CVD, cerebrovascular disease; HR, hazard ratio; SD, standard deviation. All cohort subjects must have diagnosis records for one of diabetes mellitus, hyperlipidemia, or hypertension in the source NHIRD. *The hazard ratio values are adjusted by factors of age, sex, and medical histories for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure.
outcomes in Table 1, this data set were older in age (Table 1 versus Table 3 5 41.2 versus 48.0 years) and had higher proportions for male gender (Table 1 versus Table 3 5 22.9% versus 26.8%) as well as visiting records for arrhythmias (Table 1 versus Table 3 5 7.0% versus 10.3%), diabetes mellitus (8.4% versus 19.1%), hyperlipidemia (12.7% versus 30.1%), hypertension (15.5% versus 35.0%), and renal failure (1.0% versus 2.1%). The comorbid association between hyperthyroidism and CVD still remained (multivariate adjusted HR, 1.34). On the other hand, after removing cohort subjects whose onset of CVD occurring before the index date, the data set whose outcomes were listed in Table 4 contained 3601 hypothyroidism cases and 11,359 matched controls. In all, 72.0% (n 5 2593) of case patients and 57.8% (n 5 6560) of control subjects had visiting records for any one of diabetes mellitus, hyperlipidemia, or hypertension within corresponding study periods. Compared with results in Table 2,
again this data set were older in age (Table 2 versus Table 4 5 45.6 versus 51.3 years) and had higher proportions for male gender (Table 2 versus Table 4 5 16.8% versus 19.3%) as well as diagnosis records for arrhythmias (Table 2 versus Table 4 5 6.9% versus 9.4%), diabetes mellitus (10.6% versus 20.0%), hyperlipidemia (17.0% versus 32.8%), hypertension (20.7% versus 38.6%), and renal failure (1.8% versus 3.0%). The comorbid relationship between hypothyroidism and CVD was significant too (multivariate adjusted HR, 1.80). Finally, we would like to know whether the comorbid association between hyperthyroidism/hypothyroidism and CVD was comparable to influences from risk factors of CVD. Based on the patient lists described in last paragraph for diabetes mellitus (list 1), hypertension (list 3), and renal failure (list 4), 3 further lists were constructed. The first one named as ‘‘HT_woDM_woRF’’ was produced by copying list 3 but excluding entries from list 1
Table 4. Demographics of the data set and relative hazard of developing CVD among hypothyroidism cases versus matched controls
Variable
Hypothyroidism (N 5 3601)
Male (%) 690 (19.2) Mean age, y (SD) 50.8 (14.61) No. with medical histories within study period for Arrhythmias (%) 548 (15.2) Diabetes mellitus (%) 912 (25.3) Hyperlipidemia (%) 1597 (44.3) Hypertension (%) 1423 (39.5) Renal failure (%) 170 (4.7) No. of developing follow-up CVD (%) 230 (6.4)
Matched control (N 5 11,359) 2199 (19.4) 51.5 (14.96) 864 (7.6) 2075 (18.3) 3315 (29.2) 4349 (38.3) 274 (2.4) 423 (3.7)
P value
HR (95% CI)*
P value
1.80 (1.53-2.13)
,.001
.793 .012 ,.001 ,.001 ,.001 .186 ,.001 ,.001
Abbreviations: CI, confidence interval; CVD, cerebrovascular disease; HR, hazard ratio; SD, standard deviation. All cohort subjects must have diagnosis records for one of diabetes mellitus, hyperlipidemia, or hypertension in the source NHIRD. *The hazard ratio values are adjusted by factors of age, sex, and medical histories for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure.
THYROID DISEASE AS A RISK FACTOR FOR CVD
917
Table 5. Demographics of the data set and relative hazard of developing CVD among hyperthyroidism cases accompanied with matched controls
Variable
Hyperthyroidism (N 5 9840)
Male (%) 1884 (19.1) Mean age, y (SD) 34.9 (10.95) No. with medical histories within study period for Arrhythmias (%) 1262 (12.8) Diabetes mellitus (%) 0 (0) Hyperlipidemia (%) 863 (8.8) Hypertension (%) 0 (0) Renal failure (%) 0 (0) No. of developing follow-up CVD (%) 95 (1.0)
Matched control (N 5 29,849) 5726 (19.2) 35.0 (11.00) 566 (1.9) 0 (0) 992 (3.3) 0 (0) 0 (0) 141 (.5)
P value
HR (95% CI)*
P value
1.88 (1.43-2.48)
,.001
.936 .417 ,.001 NA ,.001 NA NA ,.001
Abbreviations: CI, confidence interval; CVD, cerebrovascular disease; HR, hazard ratio; NA, not applicable; SD, standard deviation. All cohort subjects were without any diagnosis records for all diabetes mellitus, hypertension, and renal failure in the source NHIRD. *The hazard ratio values are adjusted by factors of age, sex, and medical histories for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure.
and/or 4. The second named as ‘‘HT_DM_noRF’’ was created by intersecting lists 1 and 3 but excluding entries from list 4. The last one named as ‘‘HT_DM_RF’’ was generated by intersecting lists 1, 3, and 4. It should be noted that subjects from these 3 lists were exclusive. Accordingly, hyperthyroidism patients were selected to constitute 4 subgroups: cases existed in list ‘‘HT_woDM_ woRF’’, cases existed in list ‘‘HT_DM_noRF’’, cases existed in list ‘‘HT_DM_RF’’, and cases without any diagnosis records for diabetes mellitus, hypertension, and renal failure. Control subjects, who must also have corresponding set of risk factors for matched case subgroup, were sampled according to criteria mentioned in subsection ‘‘case patient definition and control selection’’. Likewise, hypothyroidism patients were also selected to constitute 4 subgroups: cases existed in list ‘‘HT_woDM_ woRF’’, cases existed in list ‘‘HT_DM_noRF’’, cases existed in list ‘‘HT_DM_RF’’, and cases without any diagnosis records for diabetes mellitus, hypertension, and renal failure. Again control subjects, who must also be with corresponding set of risk factors for matched case subgroup, were sampled according to criteria mentioned previously. All cohort subjects, either case or control samples, whose onset of CVD occurring before the index date would be removed. Observing the results presented in Table 5, for cohort subjects without any diagnosis records for all diabetes mellitus, hypertension, and renal failure in the source NHIRD, there was significant difference between hyperthyroidism cases and matched controls for the risk of developing follow-up CVD (multivariate adjusted HR, 1.88). On the other hand, for cohort subjects with visiting records for risk factors of CVD in the source NHIRD, the relative hazards of developing CVD between hyperthyroidism cases and matched controls did not reach statistical significance (for cohort of Table S1 in Appendix with diagnosis records for all diabetes mellitus, hypertension,
and renal failure in the source NHIRD, P value 5 .178; for cohort of Table S3 in Appendix with diagnosis records for hypertension, but without any visiting records for both diabetes mellitus and renal failure in the source NHIRD, P value 5 .196). The exception was cohort subjects with diagnosis records for both diabetes mellitus and hypertension, but without any visiting records for renal failure in the source NHIRD; the case group was 1.30 times higher than corresponding control group to develop follow-up CVD (multivariate adjusted HR, 1.30; 95% CI, 1.06-1.60; P value 5 .012; Table S2 in Appendix). However, for various subgroups of subjects with different combinations of risk factors for CVD, the hypothyroidism patients always had a significantly higher hazard for follow-up CVD development than matched controls (Table 6 and Tables S4-S6 in Appendix). For cohort subjects with diagnosis records for all diabetes mellitus, hypertension, and renal failure in the source NHIRD, the multivariate adjusted HR was 1.76 (Table S4 in Appendix). For subjects with diagnosis records for both diabetes mellitus and hypertension, but without any visiting records for renal failure in the source NHIRD, the multivariate adjusted HR value was 1.77 (Table S5 in Appendix). For subjects with diagnosis records for hypertension, but without any visiting records for both diabetes mellitus and renal failure in the source NHIRD, the multivariate adjusted HR value was 2.23 (Table S6 in Appendix). Finally, for subjects without any diagnosis records for all diabetes mellitus, hypertension, and renal failure in the source NHIRD, the multivariate adjusted HR value was 2.18 (Table 6).
Discussion In this study, we have found that hyperthyroidism increased 38% the hazard of developing follow-up CVD (multivariate adjusted HR, 1.38; Table 1). However,
M.-H. YANG ET AL.
918
Table 6. Demographics of the data set and relative hazard of developing CVD among hypothyroidism cases accompanied with matched controls
Variable
Hypothyroidism (N 5 2893)
Male (%) 372 (12.9) Mean age, y (SD) 37.4 (11.98) No. with medical histories within study period for Arrhythmias (%) 199 (6.9) Diabetes mellitus (%) 0 (0) Hyperlipidemia (%) 422 (14.6) Hypertension (%) 0 (0) Renal failure (%) 0 (0) No. of developing follow-up CVD (%) 37 (1.3)
Matched control (N 5 8811) 1159 (13.2) 37.5 (12.04) 213 (2.4) 0 (0) 382 (4.3) 0 (0) 0 (0) 46 (.5)
P value
HR (95% CI)*
P value
2.18 (1.39-3.42)
.001
.683 .610 ,.001 NA ,.001 NA NA ,.001
Abbreviations: CI, confidence interval; CVD, cerebrovascular disease; HR, hazard ratio; NA, not applicable; SD, standard deviation. All cohort subjects were without any diagnosis records for all diabetes mellitus, hypertension, and renal failure in the source NHIRD. *The hazard ratio values are adjusted by factors of age, sex, and medical histories for arrhythmias, diabetes mellitus, hyperlipidemia, hypertension, and renal failure.
focused on cohort subjects with diagnosis records for some risk factors of CVD in the source NHIRD, the comorbid association between hyperthyroidism and CVD would be reduced. Four frequent risk factors for CVD were taken into account: diabetes mellitus, hyperlipidemia, hypertension, and renal failure. For instance, concentrated on subjects with visiting records for any one of diabetes mellitus, hyperlipidemia, or hypertension in the source NHIRD, hyperthyroidism increased 34% the hazard of developing follow-up CVD (multivariate adjusted HR, 1.34; Table 3). Moreover, for cohort subjects with diagnosis records for both diabetes mellitus and hypertension, but without any visiting records for renal failure in the source NHIRD, hyperthyroidism increased 30% the hazard of developing follow-up CVD (multivariate adjusted HR, 1.30; Table S2 in Appendix). Finally, for cohort subjects with diagnosis records for all diabetes mellitus, hypertension, and renal failure in the source NHIRD, the comorbid relationship between hyperthyroidism and CVD did not reach statistical significance (P value 5 .178; Table S1 in Appendix); neither did the sample group with diagnosis records for hypertension, but without any visiting records for both diabetes mellitus and renal failure in the source NHIRD (P value 5 .196; Table S3 in Appendix). The results suggested that the comorbid association between hyperthyroidism and CVD was weaker than those influences from risk factors such as diabetes mellitus, hyperlipidemia, hypertension, and/or renal failure, and so forth. The observation for which cluster of hyperthyroidism cases was 1.88 times higher than corresponding control cluster to develop follow-up CVD in the sample group without any diagnosis records for all diabetes mellitus, hypertension, and renal failure in the source NHIRD (multivariate adjusted HR, 1.88; Table 5), further validated this assumption. However, it still should be
cautious for such conclusion because in diabetic hemodialysis patients, sudden cardiac death may be influenced by hyperthyroidism and regular assessment of thyroid status may help estimate the cardiac risk.18 We have also found that hypothyroidism increased even higher the hazard of developing follow-up CVD (multivariate adjusted HR, 1.89; Table 2). The comorbid association between hypothyroidism and CVD always remained even focused on cohort subjects with diagnosis records for some risk factors of CVD in the source NHIRD. For instance, concentrated on subjects with visiting records for any one of diabetes mellitus, hyperlipidemia, or hypertension in the source NHIRD, hypothyroidism increased 80% the hazard of developing follow-up CVD (multivariate adjusted HR, 1.80; Table 4). Moreover, for cohort subjects with diagnosis records for all diabetes mellitus, hypertension, and renal failure in the source NHIRD, the case group was 76% higher than corresponding group for the risk of CVD development (multivariate adjusted HR, 1.76; Table S4 in Appendix). Similarly, the multivariate adjusted HR value was 1.77 (Table S5 in Appendix) in the sample group with diagnosis records for both diabetes mellitus and hypertension, but without any visiting records for renal failure in the source NHIRD. Finally, the multivariate adjusted HR value was 2.23 (Table S6 in Appendix) in cohort subjects with diagnosis records for hypertension, but without any visiting records for both diabetes mellitus and renal failure in the source NHIRD. The results suggested that the comorbid association between hypothyroidism and CVD was comparable to those influences from risk factors such as diabetes mellitus, hyperlipidemia, hypertension, and/or renal failure, and so forth. The observation for which cluster of hypothyroidism cases was 2.18 times higher than corresponding control cluster to develop follow-up CVD in the sample group without any
THYROID DISEASE AS A RISK FACTOR FOR CVD
diagnosis records for all diabetes mellitus, hypertension, and renal failure in the source NHIRD (multivariate adjusted HR 5 2.18; Table 6), further validated this assumption. One recent study has reported that 28% of ischemic stroke patients had thyroid hormone concentrations outside the reference range.11 Hyperthyroidism is associated with an enhanced metabolic state and an increased production of reactive oxygen species (ROS). ROS generation overrides the ability of endogenous antioxidant enzymes as hyperthyroidism promotes the depletion of antioxidant molecules. Consequently, the deleterious effects of thyroid hormones on heart tissue involve oxidative stress. Moreover, hyperthyroidism sensitizes the rat heart to reperfusion injury, which is characterized by severe arrhythmias and tissue damage.19 It is also known that hyperthyroidism is associated with increased left ventricular mass of the heart,5 hemodynamic changes, endothelial dysfunction, and coagulopathy.6 Genetic confounding and living habit might contribute to the relationship between cardiovascular disease and hyperthyroidism, as both disorders show familial aggregation and have the risk factor of smoking.6 Cerebral venous thrombosis and arterial ischemic events can happen concomitantly in patients with hyperthyroidism,12 whereas thyroid autoantibodies may be associated with the presence of intracranial artery stenosis in stroke patients.10 Thyroid hormone can regulate blood lipid levels, then influence the occurrence and progress of coronary heart disease. The promoter of the low-density lipoprotein (LDL) receptor gene contains a thyroid hormone responsive element, which can promote free triiodothyronine to increase the expression of LDL receptors, therefore, increasing the clearance of LDL.7 It has been confirmed that serum free triiodothyronine levels were inversely correlated with carotid atherosclerosis.20 Moreover, hypothyroidism can increase the levels of several cardiovascular risk factors such as total cholesterol, triglyceride, and LDL cholesterol, and accelerate the development of atherosclerosis and coronary artery disease.7,21 Low thyroid hormones alter the arterial muscle structure, then cause thickening of the vessel wall and increased stiffness of the heart.5,7 A record review study suggested that elevated homocysteine levels associated with hypothyroidism may represent a risk factor of stroke.15 However, higher effect size was found in the first year after diagnosis of autoimmune thyroiditis (HR, 1.33; 95% CI, 1.14-1.56), but not in the long term, consistent with a residual effect of hypothyroidism.16 Consequently, it is believed that L-thyroxine treatment and not low thyroid hormones per se explained this increased risk because L-thyroxine elevates coagulation factor levels and inhibits fibrinolysis in a dose-dependent manner.16,22 It has been observed in our analysis results that cohort subjects, who had diagnosis records in the source NHIRD
919
for more kinds of risk factors for CVD, also were older in age, and had higher proportions for CVD development, male gender, as well as visiting records within study periods for arrhythmias. It is true that older people have a higher risk for stroke than the general population and the risk of stroke doubles for every decade after the age of 55 years.23,24 Besides, men are 25% more likely to suffer strokes than women.23,24 Finally, other important risk factors that contribute to stroke onset include hypertension, heart disease, blood cholesterol level, diabetes, and so forth.23 All outcomes of our analysis coincide with these medical theories. We acknowledge that this study had several limitations. First, the disease diagnoses were coded according to the ICD-9 CM and obtained from administrative claims reported by hospitals or clinics, which may be considered less accurate than clinical diagnoses by standard criteria. Despite this potential drawback, the data remain representative, as the Bureau of NHI routinely and randomly sampled a fixed percentage of claims each year from every contracted medical institution. Moreover, an independent group of physicians reviewed the charts and validated the diagnoses. Consequently, hundreds of studies based on the NHIRD can be found in the US National Library of Medicine (PubMed). Second, the administrative claims data from the NHIRD did not include detailed personal information (e.g., body mass index, living habits, or laboratory test results), which may add confounding factors to thyroid disease or CVD. Finally, enhanced metabolic state induced by hyperthyroidism will increase the production of ROS, and estrogen receptors make the hyperthyroid heart more susceptible to ischemia/reperfusion. Because tamoxifen is well known as an antioxidant and is a selective antagonist of estrogen receptors, a recent study has reported that tamoxifen inhibits the hypersensitivity of hyperthyroid rat myocardium to reperfusion damage.19 Medicine treatments for thyroid disorder should be taken into account to verify the comorbid relationship between thyroid disease and CVD.
Conclusion As far as we know, the association between CVD and thyroid disease remains uncertain. Because insufficient evidence for this association supported by previous studies might be because of selection bias, in this article we used a nationwide population-based database with good demographic diversity to validate the comorbid relationship between CVD and thyroid disease. According to the evaluation results, hyperthyroidism increased 38% the hazard of developing follow-up CVD (multivariate adjusted HR, 1.38) whereas hypothyroidism increased even higher the risk (multivariate adjusted HR, 1.89). Further stratification studies for risk factors of CVD suggested that the comorbid association between
M.-H. YANG ET AL.
920
hypothyroidism and CVD was comparable to those influences from cardiac risk factors. Advanced analysis is required to investigate the pathologic mechanism underlying the association between CVD and thyroid disease. Acknowledgments: The authors would like to thank Wei-Zen Sun, MD, at the Department of Anesthesiology, National Taiwan University Hospital, for provision and utilization of the source NHIRD.
Supplementary Data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.jstrokecerebrovas dis.2014.11.032.
References 1. Siegenthaler W. Differential diagnosis in internal medicine: from symptom to diagnosis. Stuttgart, Germany: Thieme Medical Publishers 2007. 2. Torre Dario M, Lamb Geoffrey C, Ruiswyk Jerome Van, et al. Kochar’s clinical medicine for students. Philadelphia, PA: Lippincott Williams & Wilkins 2008. 3. Dan L, Fauci A, Kasper D, et al. Harrison’s principles of internal medicine. New York, NY: McGraw-Hill Professional 2011. 4. Khandelwal D, Tandon N. Overt and subclinical hypothyroidism: who to treat and how. Drugs 2012;72:17-33. 5. Faber J, Selmer C. Cardiovascular disease and thyroid function. Front Horm Res 2014;43:45-56. 6. Brandt F, Thvilum M, Almind D, et al. Morbidity before and after the diagnosis of hyperthyroidism: a nationwide register-based study. PLoS One 2013;8:e66711. 7. Bai MF, Gao CY, Yang CK, et al. Effects of thyroid dysfunction on the severity of coronary artery lesions and its prognosis. J Cardiol 2014;64:496-500. 8. Cerbone M, Capalbo D, Wasniewska M, et al. Cardiovascular risk factors in children with long-standing untreated idiopathic subclinical hypothyroidism. J Clin Endocrinol Metab 2014;99:2697-2703. 9. Danzi S, Klein I. Thyroid disease and the cardiovascular system. Endocrinol Metab Clin North Am 2014;43: 517-528. 10. Shi Z, Zhang X, Chen Z, et al. Elevated thyroid autoantibodies and intracranial stenosis in stroke at an early age. Int J Stroke 2014;9:735-740.
11. Leonards CO, Schneider HJ, Liman TG, et al. Thyroidstimulating hormone, white matter hyperintensities, and functional outcome in acute ischemic stroke patients. Cerebrovasc Dis Extra 2014;4:61-68. 12. Janovsky CC, Fukuda TG, Silva GS, et al. An unusual association between acute ischaemic stroke and cerebral venous thrombosis with thyrotoxic state. BMJ Case Rep 2013;2013. 13. Giri A, Edwards TL, LeGrys VA, et al. Subclinical hypothyroidism and risk for incident ischemic stroke among postmenopausal women. Thyroid 2014;24:1210-1217. 14. Parsaik AK, Singh B, Roberts RO, et al. Hypothyroidism and risk of mild cognitive impairment in elderly persons: a population-based study. JAMA Neurol 2014; 71:201-207. 15. Remmel KS, Wanahita A, Moore K, et al. Acute ischemic stroke and hypothyroidism. J Ky Med Assoc 2006; 104:191-193. 16. Karch A, Thomas SL. Autoimmune thyroiditis as a risk factor for stroke: a historical cohort study. Neurology 2014;82:1643-1652. 17. Grimes DA, Schulz KF. Compared to what? Finding controls for case-control studies. Lancet 2005;365:1429-1433. 18. Drechsler C, Schneider A, Gutjahr-Lengsfeld L, et al. Thyroid function, cardiovascular events, and mortality in diabetic hemodialysis patients. Am J Kidney Dis 2014; 63:988-996. 19. Pavon N, Hernandez-Esquivel L, Buelna-Chontal M, et al. Antiarrhythmic effect of tamoxifen on the vulnerability induced by hyperthyroidism to heart ischemia/ reperfusion damage. J Steroid Biochem Mol Biol 2014; 143:416-423. 20. Tatar E, Kircelli F, Asci G, et al. Associations of triiodothyronine levels with carotid atherosclerosis and arterial stiffness in hemodialysis patients. Clin J Am Soc Nephrol 2011;6:2240-2246. 21. Pei YJ, Wang AM, Zhao Y, et al. Studies of cardiovascular risk factors in polycystic ovary syndrome patients combined with subclinical hypothyroidism. Gynecol Endocrinol 2014;30:553-556. 22. Van Zaane B, Squizzato A, Debeij J, et al. Alterations in coagulation and fibrinolysis after levothyroxine exposure in healthy volunteers: a controlled randomized crossover study. J Thromb Haemost 2011;9:1816-1824. 23. Stroke: hope through research. National Institute of Neurological Disorders and Stroke (NINDS). http:// www.ninds.nih.gov/disorders/stroke/detail_stroke.htm. July 25, 2014. 24. Haast RA, Gustafson DR, Kiliaan AJ. Sex differences in stroke. Journal of cerebral blood flow and metabolism. J Cereb Blood Flow Metab 2012;32:2100-2107.
