Abstract
Background
This study aimed to determine the relationship between hematocrit (HCT) levels and cardiovascular risk factors in a community‐based population of middle‐aged adults.
Methods
From April 2011 to February 2012, a total of 1,884 middle‐aged adults were selected from a community‐based population in China. Blood and urine samples were collected for routine blood and urine tests, and measurement of plasma glucose and lipid levels. Baseline information including traditional cardiovascular risk factors was obtained by standard questionnaire to analyze. We evaluated the distribution of the HCT values for middle‐aged adults with or without cardiovascular risk factors. There were 548 males and 1,336 females in this study. The mean age of all subjects was 54.7 ± 6.7 years. There were 1,209 subjects with risk factors and 675 without risk factors.
Results
The HCT levels in subjects with risk factors were higher than those without risk factors (P = 0.005). According to a simplified tool for evaluation of the 10‐year risk of ischemic cardiovascular diseases (CVDs) in Chinese populations, all subjects were divided into four groups: the ultralow‐risk group (1,367, 72.6%), low‐risk group (232, 12.3%), intermediate‐risk group (201, 10.7%), and high‐risk/ultrahigh‐risk group (84, 4.4%). Compared with HCT levels in the ultralow‐risk group, significant differences were found in the low‐risk, intermediate‐risk, and high‐risk/ultrahigh‐risk groups (all P < 0.05).
Conclusion
Our results indicate that elevated HCT levels may be positively associated with cardiovascular risk factors. Thus, the combination of HCT values and cardiovascular risk factors may enable early diagnosis of CVDs.
Keywords: hematocrit, cardiovascular diseases, risk factors, community‐based population, middle‐aged adults
INTRODUCTION
Cardiovascular disease (CVD), also identical to circulatory diseases, is a series of diseases closely related to the circulatory system 1, 2. CVD represents the leading cause of death and morbidity worldwide. Due to the high mortality rate, severe damage to physical and mental health, long rehabilitation periods, and a high rate of disability, the burden of CVD is a highly significant social issue 3, 4. Generally, CVD is recognized as a multifactorial disease induced by complex interactions between environmental and genetic factors 5. Converging evidence indicates that traditional risk factors, such as smoking habit, hypertension, dyslipidemia, and diabetes, occupy an important position in the development and progression of CVD 6, 7, 8. Novel markers of cardiovascular risk, such as an altered hemorheologic profile may also influence cardiovascular mortality in these patients 6. An increasing evidence revealed that abnormalities in blood rheology, including elevations of blood and plasma viscosity, are related to various CVDs 9.
Although the multifactorial mechanisms by which abnormalities of blood viscosity lead to an increased incidence of clinical CVD are not yet fully understood, there are clearly potential mechanisms whereby the elevated blood viscosity abnormalities may cause the disorders of blood circulation, alter microcirculatory perfusion, increase atherogenesis, promote plaque rupture, and thereby contributing to ischemia 10, 11. Clinical studies provided evidence that abnormalities in blood flow properties may worse microcirculatory blood flow and influences CVD 12, 13, 14, 15, 16. More specifically, numerous epidemiologic studies have shown that main factors of blood rheology, including hematocrit (HCT), erythrocyte sedimentation rate, and fibrinogen, are independent risk elements of cardiovascular morbidity and mortality 17, 18. Previous studies demonstrated that the elevated HCT was significantly correlated to the incidence of coronary atherosclerosis, unstable angina, myocardial infarction, etc. 18, 19, 20. HCT is strongly beneficial for the flow properties of blood in the microcirculation, and helps to provide a more thorough understanding of the red blood cells variation that plays a role in predicting health conditions, which can be interpreted that HCT may have a corresponding increase when the absolute value of red blood cells is elevating due to various reasons 21. Therefore, we performed a pilot study to determine the relationship between HCT levels and cardiovascular risk factors in a community‐based population of middle‐aged adults, which may provide a more comprehensive and reasonable conclusion for further diagnostic evaluation of CVD.
MATERIALS AND METHODS
Ethics Statement
The study was approved by the Ethics Committee of the Fourth Affiliated Hospital of China Medical University. Informed consent was obtained in written from all participants using procedures approved by institutional review boards.
Study Design and Subjects
From April 2011 to February 2012, a total of 1,884 middle‐aged adults were selected from a community‐based population in China. Subjects must meet certain criteria in order to be enrolled in this study: (1) healthy middle‐aged adults (age ≥ 30 years); (2) subjects with no recent trauma, surgery, severe diarrhea, vomiting, heavy perspiration, and large area burn; (3) subjects have no history of infectious diseases, malignant tumors, hematopathy, multiple organ dysfunction, and autoimmune diseases. Subjects who could not meet the inclusion criteria would be excluded.
Evaluation of Cardiovascular Risk Factors
Height, weight, waist circumstance, blood pressure, blood glucose, and blood lipid were measured. Cardiovascular risk factors included hypertension, diabetes, dyslipidemia, smoking, and obesity. We evaluated the risk of CVD in a community‐based population according to a simplified tool for evaluation of the 10‐year risk of ischemic CVDs in Chinese populations 22. This criteria was designed based on age (0∼4), systolic blood pressure (−2∼8), total cholesterol (0∼1), body mass index (0∼2), diabetes (0∼1), and smoking (0∼2). The score ranged from −2 to 18. According to the scores, the 10‐year risks of CVD were stratified into five grades: ultralow‐risk (≤5%), low‐risk (>5%), intermediate‐risk (>10%), high‐risk (>20%), and ultrahigh‐risk (>40%).
A cross‐sectional questionnaire survey of cardiovascular risk factors was conducted in all subjects. The following information was collected: demographic characteristics, personal and family history, physical activity, and medication history, etc. Physical examination included anthropometry index, blood pressure, and 12‐lead ECG. Blood and urine samples were collected for routine blood and urine tests, and measurement of plasma glucose and lipid levels according to International Committee for Standardization in Hematology. Measurement of HCT levels was performed using the Sysmex XE‐2100 automated blood cell analyzer (Sysmex, Kobe, Japan).
Statistical Analysis
Data were presented as mean ± standard deviation (SD), median with interquartile ranges (IQR), or frequencies. A χ2 test was used to compare frequencies. One‐way analysis of variance (ANOVA) and Student's t‐test were used for normally distributed variables, whereas the Mann–Whitney's U‐test was used for nonnormal distributed variables. Comparisons between two groups for nominal variables were made by the Fisher's exact test. All tests of statistical significance were two‐sided with a P‐value less than 0.05 considered statistically significant. All statistical analyses were performed using the SPSS 18.0 software (SPSS, Inc., Chicago, IL).
RESULTS
Baseline Characteristics of Subjects
In this study, a total of 1,884 healthy middle‐aged adults were involved, including 548 males (29.1%) and 1,336 females (70.9%). The mean age of all subjects was 54.7 ± 6.7 years (range, 35∼64 years). There were statistically significant differences in HCT, body mass index, waist circumference, systolic blood pressure, diastolic blood pressure, fasting blood glucose, cholesterol, triglycerides, HDL‐C, LDL‐C, creatinine, uric acid, and serum sodium between males and females (all P < 0.05). The baseline clinical characteristics of all subjects are presented in Table 1.
Table 1.
Comparison of Baseline Characteristics and Clinical Features Between Males and Females
| Male | Female | ||
|---|---|---|---|
| (n = 548) | (n = 1,336) | P‐value | |
| Mean age (years) | 54.9 ± 7.2 | 54.6 ± 6.5 | 0.132 |
| Family history | |||
| Hypertension (%) | 205 (37.4) | 573 (42.9) | 0.028 |
| Coronary heart disease (%) | 114 (20.8) | 342 (25.6) | 0.027 |
| Cerebrovascular disease (%) | 136 (24.8) | 379 (28.4) | 0.116 |
| Diabetes (%) | 109 (19.9) | 292 (21.9) | 0.344 |
| Pulmonary disease (%) | 55 (10.0) | 161 (12.1) | 0.356 |
| Malignant tumor (%) | 95 (17.3) | 227 (17.0) | 0.887 |
| Smoking (%) | 261 (47.6) | 61 (4.6) | <0.001 |
| BMI (kg/m2) | 25.3 ± 3.2 | 24.9 ± 3.3 | 0.027 |
| Waist circumference (cm) | 90 (85, 95) | 84 (78, 90) | <0.001 |
| SBP (mmHg) | 128.6 ± 19.5 | 125.7 ± 2.5 | 0.003 |
| DBP (mmHg) | 78.7 ± 11.6 | 75.6 ± 11.6 | <0.001 |
| RBC (×1012) | 5.0 ± 0.4 | 4.6 ± 0.4 | <0.001 |
| Hb (g/l) | 145.5 ± 10.7 | 126.7 ± 10.7 | <0.001 |
| HCT | 0.44 (0.43, 0.46) | 0.40 (0.38, 0.41) | 0.009 |
| FPG (mmol/l) | 5.23 (4.84, 5.84) | 5.04 (4.77, 5.51) | 0.006 |
| TC (mmol/l) | 5.0 ± 1.0 | 5.3 ± 1.0 | <0.001 |
| TG (mmol/l) | 1.44 (0.96, 2.28) | 1.27 (0.88, 1.91) | 0.02 |
| HDL‐C (mmol/l) | 1.17 (1.02, 1.37) | 1.35 (1.16, 1.55) | <0.001 |
| LDL‐C (mmol/l) | 2.88 ± 0.74 | 3.08 ± 0.80 | <0.001 |
| Creatinine (umol/l) | 88.7 ± 13.5 | 72.5 ± 13.5 | <0.001 |
| Uric acid (umol/l) | 343.3 (297.1, 398.2) | 263.5 (225.9, 307.2) | <0.001 |
| Albumin (g/l) | 44.1 ± 2.6 | 44.0 ± 2.2 | 0.383 |
| Na+ (mmol/) | 140.9 ± 2.0 | 141.3 ± 2.5 | 0.043 |
| Microalbuminuria (mg/l) | 24.6 ± 37.8 | 20.3 ± 32.6 | 0.388 |
CHD, coronary heart disease; BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; RBC, red blood cell; Hb, hemoglobin; HCT, hematocrit; FPG, fasting plasma glucose; TC, total cholesterol; TG, triglycerides; HDL‐C, high‐density lipoprotein cholesterol; LDL‐C, low‐density lipoprotein cholesterol.
HCT Levels of Subjects With Different Cardiovascular Risk Factors
Table 2 showed HCT levels of subjects with different cardiovascular risk factor. There were 491 (35.8%) subjects with one risk factor, 263 (14.0%) with two risk factors, 455 (24.1%) with more than or equal to three risk factors, and 675 (35.8%) without risk factors. The HCT levels in subjects with risk factors were higher than those without risk factors (P = 0.005). There also existed obvious difference in HCT levels between subjects with one risk factor and subjects with two risk factors (P = 0.007). Furthermore, we observed significant difference in HTC levels between subjects with two risk factors and subjects with greater than or equal to three risk factors (P = 0.001).
Table 2.
The HCT Levels of Subjects With Different Cardiovascular Risk Factor
| n | Percentage | HCT levels | |
|---|---|---|---|
| Subjects without risk factors | 675 | 35.9% | 0.40 (0.38, 0.41) |
| Subjects with risk factors | 1209 | 64.1% | 0.42 (0.40, 0.44)* |
| 1 | 491 | 26.1% | 0.40 (0.38, 0.42)* |
| 2 | 263 | 13.9% | 0.41 (0.39, 0.43)* |
| ≥3 | 455 | 24.1% | 0.43 (0.41, 0.45)* |
*P < 0.05 compared with HCT levels of subjects without risk factors.
HCT Levels of Subjects With Different CVD Risk Grade
The HCT levels of middle‐aged adults with different CVD risk grade are summarized in Table 3. There were 1,367 (72.6%) in the ultralow‐risk group, 232 (12.3%) in the low‐risk group, 201 (10.7%) in the intermediate‐risk group, and 84 (4.4%) in the high‐risk/ultrahigh‐risk group. Compared with HCT levels in the ultralow‐risk group, significant differences were found in the low‐risk, intermediate‐risk, and high‐risk/ultrahigh‐risk groups (all P < 0.05).
Table 3.
The HCT Levels of Middle‐Aged Adults With Different CVD Risk Grade
| Risk grade | N | Percentage | HCT levels |
|---|---|---|---|
| Ultra‐low risk | 1367 | 72.6% | 0.40 (0.38, 0.42) |
| Low risk | 232 | 12.3% | 0.41 (0.39, 0.43)* |
| Intermediate risk | 201 | 10.7% | 0.42 (0.40, 0.45)* |
| High‐risk/Ultra‐high risk | 84 | 4.4% | 0.42 (0.40, 0.45)* |
*P < 0.05 compared with HCT levels of subjects in the ultralow‐risk group.
DISCUSSION
The HCT, also known as packed cell volume or erythrocyte volume fraction, is the volume percentage (%) of red blood cells in blood and functions as a major determinant of blood viscosity, blood pressure, venous return, cardiac output, and platelet adhesiveness 23, 24. Abundant epidemiologic studies have shown that HCT level is an important index of blood rheology and an independent risk of cardiovascular morbidity and mortality 6, 25. Skretteberg et al. have demonstrated that HCT levels, as a major determinant of whole blood viscosity, might have a potential adverse role in the risk of CVD 17. Elevated HCT levels might function significantly in increasing Hb levels and promote the increase of blood viscosity; thus dynamically altering the blood rheological parameters, followed by weakening of microvascular perfusion and the acceleration of the thrombus formation. Therefore, it was biologically plausible that HCT might be one of the major pathological mediators of CVD 17. Recently, a growing number of prospective studies have demonstrated that increased levels of HCT may be closely related to cardiovascular risk factors 18, 26, 27, 28, 29.
In this study, an independent cross‐sectional population study was conducted to explore the relationship between HCT levels and cardiovascular risk factors in a community‐based population of 1,884 middle‐aged adults. Our results revealed that the HCT levels in subjects with cardiovascular risk factors (hypertension, diabetes, dyslipidemia, smoking, or obesity) were higher than those without risk factors, suggesting that elevated HCT levels were correlated with cardiovascular risk factors. In addition, we also found that subjects with greater than or equal to three risk factors had higher levels of HCT than those with one or two risk factors, indicating that the total number of cardiovascular risk factors was positively associated with the level of HCT. Although the exact mechanism of HCT in the association of cardiovascular risk factors is not clear yet, a potential explanation might be that high HCT levels would alter its function as a major determinant of blood viscosity, whose abnormalities may cause the disorders of blood circulation, alter microcirculatory perfusion, increase atherogenesis, promote plaque rupture, and thereby contributing to ischemia 10. In our study, the results also demonstrated that the HCT levels in the ultralow‐risk group were significantly lower than in the low‐risk, intermediate‐risk, and high‐risk/ultrahigh‐risk groups, revealing that the HCT levels may be an important determinant of CVD risk grade. Therefore, we speculated that the combination of HCT levels and cardiovascular risk factors may additionally contribute to the diagnosis of CVD. Paul et al. suggested that HCT levels should be taken into consideration as an important risk predictor in the assessment and management of newly diagnosed hypertensive patients 18. Tamariz et al. confirmed that elevated estimated whole blood viscosity and HCT levels would be cross‐sectionally associated with emerging risk factors for insulin resistance and type 2 diabetes mellitus 30. A recent study has showed that continuous cigarette smoking has severe adverse affects on hematological parameters including HCT, and might be associated with a greater risk for developing CVDs 31. Our findings are consistent with previous studies that HCT levels may be associated with cardiovascular risk factors; thereby estimating the combination of HCT levels and cardiovascular risk factors may improve diagnostic accuracy of CVD.
Our study has several limitations that should be acknowledged. The first limitation is that the gender ratio and mean age of subjects in this study were fairly unrepresentative, which may have effects in estimating the relationships between HCT levels and cardiovascular risk factors. On the other hand, this cross‐sectional study could not establish the cause‐and‐effect relationship between HCT levels and cardiovascular risk factors; thereby possibly influencing the reliability of our results 32. Importantly, we failed to confirm the effects of HCT levels on prognosis of CVD patients.
In conclusion, our study indicates that elevated HCT levels may be positively associated with cardiovascular risk factors. Thus, the combination of HCT values and cardiovascular risk factors may enable early diagnosis of CVDs. Due to the limitation of study mentioned above, more detailed studies are needed to confirm our findings.
CONFLICT OF INTEREST
No competing financial interests exist.
ACKNOWLEDGMENTS
We thank our colleagues at the Department of Cardiology, Fourth Affiliated Hospital of China Medical University.
REFERENCES
- 1. Chen Y, Graziano JH, Parvez F, et al. Arsenic exposure from drinking water and mortality from cardiovascular disease in Bangladesh: Prospective cohort study. BMJ 2011;342:d2431. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Steinberger J, Daniels SR, Eckel RH, et al. Progress and challenges in metabolic syndrome in children and adolescents: A scientific statement from the American Heart Association Atherosclerosis, Hypertension, and Obesity in the Young Committee of the Council on Cardiovascular Disease in the Young; Council on Cardiovascular Nursing; and Council on Nutrition, Physical Activity, and Metabolism. Circulation 2009;119:628–647. [DOI] [PubMed] [Google Scholar]
- 3. Wang Z, Klipfell E, Bennett BJ, et al. Gut flora metabolism of phosphatidylcholine promotes cardiovascular disease. Nature 2011;472:57–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Deaton C, Froelicher ES, Wu LH, et al. The global burden of cardiovascular disease. Eur J Cardiovasc Nurs 2011;10:S5–S13. [DOI] [PubMed] [Google Scholar]
- 5. May AL, Kuklina EV, Yoon PW. Prevalence of cardiovascular disease risk factors among US adolescents, 1999–2008. Pediatrics 2012;129:1035–1041. [DOI] [PubMed] [Google Scholar]
- 6. Zanazzi M, Fatini C, Farsetti S, et al. Blood rheology and renal transplantation: An intriguing relationship for assessing cardiovascular risk. Transplant Proc 2010;42:1383–1384. [DOI] [PubMed] [Google Scholar]
- 7.Mur Marti T, Villaro Gabarros M, Porta Martinez N, Jaen Manzanera A. Prevalence of chronic renal insufficiency in diabetic type 2 diabetes patients based on the estimated glomerular filtration rate and relation with cardiovascular risk. Med Clin (Barc) 2013;140:395–400. [DOI] [PubMed] [Google Scholar]
- 8. Nishiyama M, Kimijima M, Muto T, Kimura K. Presence of an interaction between smoking and being overweight increases risks of hypertension, diabetes, and cardiovascular disease in outpatients with mood disorders. Environ Health Prev Med 2012;17:285–291. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Salazar Vazquez BY, Martini J, Chavez Negrete A, et al. Cardiovascular benefits in moderate increases of blood and plasma viscosity surpass those associated with lowering viscosity: Experimental and clinical evidence. Clin Hemorheol Microcirc 2010;44:75–85. [DOI] [PubMed] [Google Scholar]
- 10. Tikhomirova IA, Oslyakova AO, Mikhailova SG. Microcirculation and blood rheology in patients with cerebrovascular disorders. Clin Hemorheol Microcirc 2011;49:295–305. [DOI] [PubMed] [Google Scholar]
- 11. Church TS, Lavie CJ, Milani RV, Kirby GS. Improvements in blood rheology after cardiac rehabilitation and exercise training in patients with coronary heart disease. Am Heart J 2002;143:349–355. [DOI] [PubMed] [Google Scholar]
- 12. Lowe GD. Can haematological tests predict cardiovascular risk? The 2005 Kettle Lecture. Br J Haematol 2006;133:232–250. [DOI] [PubMed] [Google Scholar]
- 13. Pope JV, Jones AE, Gaieski DF, et al. Multicenter study of central venous oxygen saturation (ScvO2) as a predictor of mortality in patients with sepsis. Ann Emerg Med 2010;55:40–46.e41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Resch H, Garhofer G, Fuchsjäger‐Mayrl G, Hommer A, Schmetterer L. Endothelial dysfunction in glaucoma. Acta Ophthalmol 2009;87:4–12. [DOI] [PubMed] [Google Scholar]
- 15. Dominguez‐Rodriguez A, Abreu‐Gonzalez P, Sanchez‐Sanchez JJ, Kaski JC, Reiter RJ. Melatonin and circadian biology in human cardiovascular disease. J Pineal Res 2010;49:14–22. [DOI] [PubMed] [Google Scholar]
- 16. Pérez‐López FR, Larrad‐Mur L, Kallen A, Chedraui P, Taylor HS. Review: Gender differences in cardiovascular disease: Hormonal and biochemical influences. Reprod Sci 2010;17:511–531. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Skretteberg PT, Bodegard J, Kjeldsen SE, et al. Interaction between inflammation and blood viscosity predicts cardiovascular mortality. Scand Cardiovasc J 2010;44:107–112. [DOI] [PubMed] [Google Scholar]
- 18. Paul L, Jeemon P, Hewitt J, et al. Hematocrit predicts long‐term mortality in a nonlinear and sex‐specific manner in hypertensive adults. Hypertension 2012;60:631–638. [DOI] [PubMed] [Google Scholar]
- 19. Surampudi PN, Wang C, Swerdloff R. Hypogonadism in the aging male diagnosis, potential benefits, and risks of testosterone replacement therapy. Int J Endocrinol 2012;2012:625434. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Li G, Hu H, Shi W, et al. Elevated hematocrit in nonalcoholic fatty liver disease: A potential cause for the increased risk of cardiovascular disease? Clin Hemorheol Microcirc 2012;51:59–68. [DOI] [PubMed] [Google Scholar]
- 21. Pasipoularides A. Optimal hematocrit: A Procrustean bed for maximum oxygen transport rate? J Appl Physiol (1985) 2012;113:353–354. [DOI] [PubMed] [Google Scholar]
- 22. Puska P, Nissinen A, Tuomilehto J, et al. The community‐based strategy to prevent coronary heart disease: Conclusions from the ten years of the North Karelia project. Annu Rev Public Health 1985;6:147–193. [DOI] [PubMed] [Google Scholar]
- 23. Brooks AC, Guillaumin J, Cooper ES, Couto CG. Effects of hematocrit and red blood cell‐independent viscosity on canine thromboelastographic tracings. Transfusion 2014;54:727–734. [DOI] [PubMed] [Google Scholar]
- 24. Nemeth N, Alexy T, Furka A, et al. Inter‐species differences in hematocrit to blood viscosity ratio. Biorheology 2009;46:155–165. [DOI] [PubMed] [Google Scholar]
- 25. Romagnoli M, Alis‐Pozo R, Martinez‐Bello V, et al. Blood rheology effect of submaximal exercise on young subjects. Clin Hemorheol Microcirc 2014;56:111–117. [DOI] [PubMed] [Google Scholar]
- 26. Toss F, Nordstrom A, Nordstrom P. Association between hematocrit in late adolescence and subsequent myocardial infarction in Swedish men. Int J Cardiol 2013;168:3588–3593. [DOI] [PubMed] [Google Scholar]
- 27. Brækkan SK, Mathiesen EB, Njølstad I, Wilsgaard T, Hansen J‐B. Hematocrit and risk of venous thromboembolism in a general population. The Tromsø Study. Haematologica 2010;95:270–275. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. van Straten AH, Kats S, Bekker MW, et al. Risk factors for red blood cell transfusion after coronary artery bypass graft surgery. J Cardiothorac Vasc Anesth 2010;24:413–417. [DOI] [PubMed] [Google Scholar]
- 29. Solomon SD, Uno H, Lewis EF, et al. Erythropoietic response and outcomes in kidney disease and type 2 diabetes. New Engl J Med 2010;363:1146–1155. [DOI] [PubMed] [Google Scholar]
- 30. Tamariz LJ, Young JH, Pankow JS, et al. Blood viscosity and hematocrit as risk factors for type 2 diabetes mellitus: The atherosclerosis risk in communities (ARIC) study. Am J Epidemiol 2008;168:1153–1160. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Pamukcu B, Oflaz H, Onur I, Cimen A, Nisanci Y. Effect of cigarette smoking on platelet aggregation. Clin Appl Thromb Hemost 2011;17:E175–E180. [DOI] [PubMed] [Google Scholar]
- 32. Castro VM, Clements CC, Murphy SN, et al. QT interval and antidepressant use: A cross sectional study of electronic health records. BMJ 2013;346:f288. [DOI] [PMC free article] [PubMed] [Google Scholar]