Abstract 187: Electrocardiographic Heart Rate-corrected Qt Interval And Risk Of Stroke In The REasons For Geographic And Racial Differences In Stroke (REGARDS) Study

Stroke ◽  
2012 ◽  
Vol 43 (suppl_1) ◽  
Author(s):  
Elsayed Z Soliman ◽  
George Howard ◽  
George Howard ◽  
Mary Cushman ◽  
Brett Kissela ◽  
...  
Keyword(s):  
Heart Rate ◽  
Racial Differences ◽  
Qt Interval ◽  
Proportional Hazards ◽  
Left Ventricular ◽  
Cox Proportional Hazards ◽  
Corrected Qt Interval ◽  
Cardiovascular Morbidity And Mortality ◽  
Increased Risk ◽  
Regards Study

Background: Prolongation of heart rate-corrected QT interval (QTc) is a well established predictor of cardiovascular morbidity and mortality. Little is known, however, about the relationship between this simple electrocardiographic (ECG) marker and risk of stroke. Methods: A total of 27,411 participants aged > 45 years without prior stroke from the REasons for Geographic and Racial Differences in Stroke (REGARDS) study were included in this analysis. QTc was calculated using Framingham formula (QTcFram). Stroke cases were identified and adjudicated during an up to 7 years of follow-up (median 2.7 years). Cox proportional hazards analysis was used to estimate the hazard ratios for incident stroke associated with prolonged QTcFram interval (vs. normal) and per 1 standard deviation (SD) increase, separately, in a series of incremental models. Results: The risk of incident stroke in the study participants with baseline prolonged QTcFram was almost 3 times the risk in those with normal QTcFram [HR (95% CI): 2.88 (2.12, 3.92), p<0.0001]. After adjustment for age, race, sex, antihypertensive medication use, systolic blood pressure, current smoking, diabetes, left ventricular hypertrophy, atrial fibrillation, prior cardiovascular disease, QRS duration, warfarin use, and QT-prolonging drugs (full model), the risk of stroke remained significantly high [HR (95% CI): 1.67 (1.16, 2.41), p=0.0060)], and was consistent across several subgroups of REGARDS participants. When the risk of stroke was estimated per 1 SD increase in QTcFram, a 24% increased risk was observed [HR (95% CI): 1.24 (1.16, 1.33), p<0.0001)]. This risk remained significant in the fully adjusted model [HR (95% CI): 1.12 (1.03, 1.21), p=0.0055]. Similar results were obtained when other QTc correction formulas including Hodge’s, Bazett’s and Fridericia’s were used. Conclusions: QTc prolongation is associated with a significantly increased risk of incident stroke independently from known stroke risk factors. In light of our results, examining the risk of stroke associated with QT-prolonging drugs may be warranted.

Abstract P316: Multiple Vulnerabilities to Health Disparities and Incident Coronary Heart Disease in the Reasons for Geographic and Racial Differences in Stroke (REGARDS) Study

Circulation ◽  
2018 ◽  
Vol 137 (suppl_1) ◽  
Author(s):  
Monika M Safford ◽  
Laura Pinheiro ◽  
Madeline Sterling ◽  
Joshua Richman ◽  
Paul Muntner ◽  
...  
Keyword(s):  
Health Disparities ◽  
Racial Differences ◽  
Low Income ◽  
Proportional Hazards ◽  
Health Management ◽  
Cox Proportional Hazards ◽  
Southeastern Us ◽  
Cox Proportional Hazards Models ◽  
Increased Risk ◽  
Regards Study

Social determinants contribute to disparities in incident CHD but it is not known if they have an additive effect. We hypothesized that having more socially determined vulnerabilities to health disparities is associated with increased risk of incident CHD in the REGARDS study, a large biracial prospective cohort with physiological and survey measures. Experts adjudicated incident fatal and nonfatal CHD over 10 years of follow-up. Vulnerabilities included black race, low education, low income, and Southeastern US residence. The risks for CHD outcomes associated with 1, 2, and 3+ vs 0 vulnerabilities were calculated with Cox proportional hazards models adjusted for medical conditions, functional status, health behaviors, and physiologic variables. Of the 19,645 participants free of CHD at baseline (mean age 64 years, 57% women), 16% had 0 vulnerabilities, 36% had 1, 29% had 2, and 18% had 3+. Increasing numbers of vulnerabilities were associated with higher incidence (Figure) and risk of CHD that attenuated somewhat after multivariable adjustment (Table). These findings may provide a method of risk stratification useful for population health management.

Abstract 14867: Atrial Fibrillation is Associated With Risk of Incident Heart Failure and Its Subtypes

Circulation ◽  
2020 ◽  
Vol 142 (Suppl_3) ◽  
Author(s):  
Charles D Nicoli ◽  
Wesley T O'Neal ◽  
Emily B Levitan ◽  
Matthew J Singleton ◽  
Suzanne E Judd ◽  
...  
Keyword(s):  
Heart Failure ◽  
Atrial Fibrillation ◽  
Risk Factor ◽  
Racial Differences ◽  
Proportional Hazards ◽  
Left Ventricular ◽  
Cox Proportional Hazards ◽  
Left Ventricular Ejection ◽  
Increased Risk

Background: Heart failure (HF) is an established risk factor for atrial fibrillation (AF). However, the extent to which AF is a risk factor for HF and its subtypes in a racially diverse population is unclear. Methods: This analysis included 25,787 participants free of baseline HF from the REasons for Geographic And Racial Differences in Stroke (REGARDS). AF at baseline was identified by electrocardiogram and self-reported physician diagnosis. HF events during follow-up were ascertained from medical records with subclassification by left ventricular ejection function (EF) at time of diagnosis as HF with reduced EF (HFrEF; EF<40%), HF with preserved EF (HFpEF; EF≥50%), mid-range HF (EF 40-49%) and unclassified. Cox proportional-hazards regression was used to separately examine the association of baseline AF and incident overall HF, HFpEF, and HFrEF. The Lunn-McNeil method was used to test differences in the association of AF by HFrEF & HFpEF. Consistency of the associations of AF with HF and its subtypes was examined in subgroups stratified by sex and race. Results: AF was detected in 1,924 (7.5%) participants at baseline (2003-2007). Over 10.1 years median follow-up, 1,109 HF events occurred (388 HFrEF, 356 HFpEF, 77 mid-range HF, and 288 unclassified). AF was associated with more than 2-fold increased risk of overall HF as well as its subtypes HFpEF and HFrEF in models adjusted for socio-demographics and cardiovascular risk factors. The strength of associations was slightly attenuated after adjustments for Warfarin, aspirin and statin. A stronger association of AF with HFrEF than HFpEF was observed, but the difference was not statistically significant. These associations were consistent among men, women, Black and White subgroups (Table). Conclusions: AF is strongly associated with both HFrEF and HFpEF. While further investigation of the underlying mechanisms is needed, our findings extend the sequelae of AF beyond stroke to include HF regardless of type.

Abstract P316: Prehypertension and Incident Acute Coronary Heart Disease in the REasons for Geographic and Racial Differences in Stroke (REGARDS) Study

Circulation ◽  
2013 ◽  
Vol 127 (suppl_12) ◽  
Author(s):  
Stephen P Glasser ◽  
Yulia Khodneva ◽  
Daniel Lackland ◽  
Ronald Prineas ◽  
Monika Safford
Keyword(s):  
Coronary Heart Disease ◽  
Heart Disease ◽  
Racial Differences ◽  
Proportional Hazards ◽  
Cox Proportional Hazards ◽  
Community Dwelling ◽  
Cox Proportional Hazards Models ◽  
Chd Risk ◽  
Regards Study ◽  
Chd Risk Factors

Objective: The independent prognostic value of prehypertension (preHTN) for incident coronary heart disease (CHD) remains unsettled. Using the REasons for Geographic And Racial Differences in Stroke (REGARDS) cohort study, we examined associations between preHTN and incident acute CHD and CVD death. Methods: REGARDS includes 30,239 black and white community-dwelling adults age 45 and older at baseline. Recruitment occurred from 2003-7, with baseline interviews and in-home data collection for physiologic measures. Follow-up is conducted by telephone every 6 months to detect events and deaths, which are adjudicated by experts. Systolic BP was categorized into <120 mmHg (n=4385), 120-129 mmHg (n=4000), 130-139 (n=2066), and hypertension was categorized into controlled (<140/90 mmHg on treatment) (n=8378), and uncontrolled (>140/90 mmHg) (n=5364). Incident acute CHD was defined as definite or probable myocardial infarction (MI) or acute CHD death. CVD death was defined as acute CHD, stroke, heart failure or other cardiovascular disease related. Cox proportional hazards models estimated the hazard ratios (HR) for incident CHD by BP categories, adjusting for sociodemographics and CHD risk factors. Results: The 23,393 participants free of CHD at baseline were followed for a median of 4.4 years. Mean age was 64.1, 58% were women and 42% were black. There was a significant interaction between sex and BP categories, therefore analyses were stratified by sex. There were 252 non-fatal and fatal acute CHD events among women and 407 among men. Among women, compared with SBP<120 mmHg, BP categories above SBP 120 mmHg were associated with incident CHD (adjusted HR for SBP120-129 mmHg=1.94 {95% CI 1.04-3.62]; SBP 130-139 mmHg=1.92 {0.95-3.87}; controlled HTN=2.16 {1.25-3.75}; uncontrolled HTN=3.25 {1.87-5.65}) in fully adjusted models. Among men, only uncontrolled HTN was associated with incident CHD (HR=1.55 {1.11-2.17}). Conclusion: In this sample, preHTN may be associated with incident CHD among women but not men.

scholarly journals Premature Ventricular Complexes on Screening Electrocardiogram and Risk of Ischemic Stroke

Stroke ◽  
2015 ◽  
Vol 46 (5) ◽  
pp. 1365-1367 ◽  
Author(s):  
Sunil K. Agarwal ◽  
Jennifer Chao ◽  
Frederick Peace ◽  
Suzanne E. Judd ◽  
Brett Kissela ◽  
...  
Keyword(s):  
Blood Pressure ◽  
Ischemic Stroke ◽  
Racial Differences ◽  
Proportional Hazards Model ◽  
Geographic Region ◽  
Left Ventricular ◽  
Cox Proportional Hazards ◽  
Cox Proportional Hazards Model ◽  
Premature Ventricular Complexes ◽  
Increased Risk

Background and Purpose— Premature ventricular complexes (PVCs) detected from long-term ECG recordings have been associated with an increased risk of ischemic stroke. Whether PVCs seen on routine ECG, commonly used in clinical practice, are associated with an increased risk of ischemic stroke remains unstudied. Methods— This analysis included 24 460 participants (aged, 64.5+9.3 years; 55.1% women; 40.0% blacks) from the Reasons for Geographic and Racial Differences in Stroke (REGARDS) study who were free of stroke at the time of enrollment. PVCs were ascertained from baseline ECG (2003–2007), and incident stroke cases through 2011 were confirmed by an adjudication committee. Results— A total of 1415 (5.8%) participants had at least 1 PVC at baseline, and 591 developed incident ischemic stroke during an average (SD) follow-up of 6.0 (2.0) years. In a cox proportional hazards model adjusted for age, sex, race, geographic region, education, previous heart disease, systolic blood pressure, blood pressure–lowering medications, current smoking, diabetes mellitus, left ventricular hypertrophy by ECG, and aspirin use and warfarin use, the presence of PVCs was associated with 38% increased risk of ischemic stroke (hazard ratio [95% confidence interval], 1.38 [1.05–1.81]). Conclusions— PVCs are common on routine screening ECGs and are associated with an increased risk of ischemic stroke.

Abstract MP50: Neighborhood Social And Economic Environment And Heart Failure Risk Among Adults With & Without Diabetes: The Reasons For Geographic And Racial Differences In Stroke (regards) Study

Circulation ◽  
2021 ◽  
Vol 143 (Suppl_1) ◽  
Author(s):  
Gargya Malla ◽  
Andrea Cherrington ◽  
Monika M Safford ◽  
Parag Goyal ◽  
Doyle M Cummings ◽  
...  
Keyword(s):  
Heart Failure ◽  
Racial Differences ◽  
Proportional Hazards ◽  
Signs And Symptoms ◽  
Cox Proportional Hazards ◽  
Economic Environment ◽  
Individual Level ◽  
Black And White ◽  
Regards Study ◽  
Year 2000

Background: Heart failure (HF) mortality rates have been increasing since 2011. Individual-level education and occupation have been inversely associated with HF mortality among those with diabetes mellitus (DM) but not among those without DM. However, less is known about the association between neighborhood social and economic environment (NSEE) and HF risk and whether this association varies by DM status. Methods: This study included 21,244 Black and White adults age >=45 years at baseline (2003-07) from the REGARDS Study. NSEE quartiles were created using z-scores based on 6 census tract variables from year 2000 (% <high school education, % unemployed, % household with <$30,000, % living in poverty, % on public assistance, % without car). Incident HF events (fatal or non-fatal) were adjudicated based on hospitalization with HF signs and symptoms, supportive imaging or biomarkers. Diabetes was defined as fasting glucose >=126 mg/dL or random glucose >=200 mg/dL or use of diabetes medications. Cox proportional hazards regression was used to obtain hazard ratios (95% CI) with HF follow-up through 2016. Results: Mean age was 65 years, 54% were women, 61% were White and 18% had prevalent DM at baseline. During a median 10.1 years, 829 incident HF events occurred. Among adults with DM, neighborhood disadvantage was associated with an increased HF risk , but this association was not statistically significant (Table). Among adults without DM, the risk of HF was higher for participants living in any neighborhood that was not the most advantaged, and the magnitude of association was smiliar across NSEE quartiles. Conclusion: Adults living in disadvantaged neighborhoods had a higher risk of HF, particularly among those without DM. Addressing neighborhood social and economic conditions may be important for HF prevention.

scholarly journals Electrocardiographic ST-T Abnormities Are Associated With Stroke Risk in the REGARDS Study

Stroke ◽  
2020 ◽  
Vol 51 (4) ◽  
pp. 1100-1106
Author(s):  
Mitsuaki Sawano ◽  
Ya Yuan ◽  
Shun Kohsaka ◽  
Taku Inohara ◽  
Takeki Suzuki ◽  
...  
Keyword(s):  
Ischemic Stroke ◽  
Racial Differences ◽  
Stroke Risk ◽  
Proportional Hazards ◽  
The United States ◽  
Incidence Rates ◽  
Cox Proportional Hazards ◽  
Hazard Ratio ◽  
Common Finding ◽  
Increased Risk

Background and Purpose— In previous studies, isolated nonspecific ST-segment and T-wave abnormalities (NSSTTAs), a common finding on ECGs, were associated with greater risk for incident coronary artery disease. Their association with incident stroke remains unclear. Methods— The REGARDS (Reasons for Geographic and Racial Differences in Stroke) study is a population-based, longitudinal study of 30 239 white and black adults enrolled from 2003 to 2007 in the United States. NSSTTAs were defined from baseline ECG using the standards of Minnesota ECG Classification (Minnesota codes 4-3, 4-4, 5-3, or 5-4). Participants with prior stroke, coronary heart disease, and major and minor ECG abnormalities other than NSSTTAs were excluded from analysis. Multivariable Cox proportional hazards regression was used to examine calculate hazard ratios of incident ischemic stroke by presence of baseline NSSTTAs. Results— Among 14 077 participants, 3111 (22.1%) had NSSTTAs at baseline. With a median of 9.6 years follow-up, 106 (3.4%) with NSSTTAs had ischemic stroke compared with 258 (2.4%) without NSSTTAs. The age-adjusted incidence rates (per 1000 person-years) of stroke were 2.93 in those with NSSTTAs and 2.19 in those without them. Adjusting for baseline age, sex, race, geographic location, and education level, isolated NSSTTAs were associated with a 32% higher risk of ischemic stroke (hazard ratio, 1.32 [95% CI, 1.05–1.67]). With additional adjustment for stroke risk factors, the risk of stroke was increased 27% (hazard ratio, 1.27 [95% CI, 1.00–1.62]) and did not differ by age, race, or sex. Conclusions— Presence of NSSTTAs in persons with an otherwise normal ECG was associated with a 27% increased risk of future ischemic stroke.

Left ventricular geometry and risk of incident hypertension

Heart ◽  
2019 ◽  
Vol 105 (18) ◽  
pp. 1402-1407 ◽  
Author(s):  
Sung Keun Park ◽  
Ju Young Jung ◽  
Jeong Gyu Kang ◽  
Pil-Wook Chung ◽  
Chang-Mo Oh
Keyword(s):  
Wall Thickness ◽  
Proportional Hazards ◽  
Proportional Hazards Model ◽  
Left Ventricular ◽  
Cox Proportional Hazards ◽  
Cox Proportional Hazards Model ◽  
Left Ventricular Geometry ◽  
Incident Hypertension ◽  
Increased Risk ◽  
Echocardiographic Parameters

ObjectiveLeft ventricular (LV) geometry change is an independent predictor for cardiovascular disease. However, data are equivocal on the association of echocardiographic parameters of LV geometry with incident hypertension. Thus, we were to investigate the risk of hypertension according to the baseline echocardiographic parameters of LV geometry.MethodsStudy participants were 12 562 Koreans without hypertension who received echocardiography as an item of health check-up. They were divided into normotensive or prehypertensive group according to baseline blood pressure. In each group, study subjects were classified by quintiles of baseline echocardiographic parameters including left ventricular mass index (LVMI), relative wall thickness (RWT), interventricular septal thickness (IVST), posterior wall thickness (PWT) and IVST plus PWT and followed up for 5 years. Cox proportional hazards model was used in calculating adjusted HRs and their 95% CI for hypertension according to each quintile group. Area under the curve (AUC) analysis (AUC [95% CI]) was performed to compare the predictability of LVMI, RWT, IVST, PWT, IVST plus PWT for hypertension.ResultsPrehypertensive group had the worse clinical and echocardiographic parameters in baseline analysis than normotensive group. The risk of hypertension significantly increased proportionally to baseline LVMI, RWT, IVST, PWI and IVST plus PWT above specific quintile levels, which was identified in both normotensive and prehypertensive group. In AUC analysis, IVST, PWT and IVST plus PWT showed a significantly increased AUC, compared with LVMI.ConclusionLV geometry change was significantly associated with the increased risk for hypertension in non-hypertensive individuals.

scholarly journals High sodium:potassium intake ratio increases the risk for all-cause mortality: the REasons for Geographic And Racial Differences in Stroke (REGARDS) study

2013 ◽  
Vol 2 ◽  
Author(s):  
Suzanne E. Judd ◽  
Kristal J. Aaron ◽  
Abraham J. Letter ◽  
Paul Muntner ◽  
Nancy S. Jenny ◽  
...  
Keyword(s):  
Racial Differences ◽  
Proportional Hazards ◽  
Reference Group ◽  
Cox Proportional Hazards ◽  
Stroke Incidence ◽  
Black And White ◽  
Cox Proportional Hazards Models ◽  
Group Data ◽  
All Cause Mortality ◽  
Regards Study

AbstractIncreased dietary Na intake and decreased dietary K intake are associated with higher blood pressure. It is not known whether the dietary Na:K ratio is associated with all-cause mortality or stroke incidence and whether this relationship varies according to race. Between 2003 and 2007, the REasons for Geographic And Racial Differences in Stroke (REGARDS) cohort enrolled 30 239 black and white Americans aged 45 years or older. Diet was assessed using the Block 98 FFQ and was available on 21 374 participants. The Na:K ratio was modelled in race- and sex-specific quintiles for all analyses, with the lowest quintile (Q1) as the reference group. Data on other covariates were collected using both an in-home assessment and telephone interviews. We identified 1779 deaths and 363 strokes over a mean of 4·9 years. We used Cox proportional hazards models to obtain multivariable-adjusted hazard ratios (HR). In the highest quintile (Q5), a high Na:K ratio was associated with all-cause mortality (Q5 v. Q1 for whites: HR 1·22; 95 % CI 1·00, 1·47, P for trend = 0·084; for blacks: HR 1·36; 95 % CI 1·04, 1·77, P for trend = 0·028). A high Na:K ratio was not significantly associated with stroke in whites (HR 1·29; 95 % CI 0·88, 1·90) or blacks (HR 1·39; 95 % CI 0·78, 2·48), partly because of the low number of stroke events. In the REGARDS study, a high Na:K ratio was associated with all-cause mortality and there was a suggestive association between the Na:K ratio and stroke. These data support the policies targeted at reduction of Na from the food supply and recommendations to increase K intake.

Abstract 19554: Lipoprotein(a) and Risk of Ischemic Stroke in the Reasons for Geographic and Racial Differences in Stroke (REGARDS) Study

Circulation ◽  
2015 ◽  
Vol 132 (suppl_3) ◽  
Author(s):  
Kristine S Alexander ◽  
Neil A Zakai ◽  
Fred Unverzagt ◽  
Virginia Wadley ◽  
Brett M Kissela ◽  
...  
Keyword(s):  
Ischemic Stroke ◽  
Racial Differences ◽  
Stroke Risk ◽  
Proportional Hazards ◽  
Continuous Variable ◽  
Cox Proportional Hazards ◽  
Stroke Incidence ◽  
Lipoprotein A ◽  
Cox Proportional Hazards Models ◽  
Regards Study

Background: Increased lipoprotein (a) (Lp(a)) is associated with coronary risk, but links with stroke have been less consistent. Blacks have 2-4-fold higher Lp(a) levels than whites, and have higher stroke incidence than whites, but have been under-represented in studies of Lp(a) and stroke to date. Hypothesis: Lp(a) is a risk factor for ischemic stroke, and this risk differs by race. Methods: REGARDS recruited 30,239 black and white U.S. men and women in 2003-7 to study regional and racial differences in stroke mortality. We measured Lp(a) by immunonepholometric assay in 572 cases of incident ischemic stroke and a 1,104-person cohort random sample. The hazard ratio of stroke by baseline Lp(a) was calculated using Cox proportional hazards models, stratified by race. Lp(a) was modeled both as a continuous variable (per sex- and race-specific SD) and in sex- and race-specific quartiles, given known differences in distributions by race and sex. Results: As shown in the Figure, being in the 4 th vs 1 st Lp(a) quartile was associated with ischemic stroke in black but not white participants, adjusted for age and sex (Model 1). The HRs were essentially unchanged with added adjustment for stroke risk factors (Model 2). There was no significant association between Lp(a) as a continuous variable and stroke, though race-specific patterns were similar. There remained no association between Lp(a) and stroke in whites when we used the sex- and race-specific 90 th percentile as a cut-off (HR: 0.91 95% CI: 0.52, 1.60). Discussion: Lp(a) was associated with ischemic stroke risk in black but not white REGARDS participants, this might partly explain the black/white disparity in stroke. Further studies in racially diverse groups are necessary to confirm these findings. Figure 1. Hazard ratios for Lp(a) and stroke in blacks and whites, per quartile (compared with first quartile) and SD.

Abstract P089: Adiposity and risk of stroke: body shape matters more than body weight

Circulation ◽  
2014 ◽  
Vol 129 (suppl_1) ◽  
Author(s):  
Suzanne E Judd ◽  
Virginia J Howard ◽  
George Howard ◽  
Susan Lakoski ◽  
Mary Cushman ◽  
...  
Keyword(s):  
Racial Differences ◽  
Stroke Risk ◽  
Proportional Hazards ◽  
Obesity Paradox ◽  
The United States ◽  
Cox Proportional Hazards ◽  
Stroke Risk Factor ◽  
Medical Risk ◽  
Increased Risk ◽  
History Of

Background: Higher body mass index (BMI) and larger waist circumference (WC), characteristics of increased adiposity, are associated with greater risk of cardiovascular disease (CVD). However, an obesity paradox (improved survival in those with history of stroke) has been observed in some populations. We hypothesized that BMI would be associated with decreased risk of stroke but that this association would be attenuated after accounting for WC. Methods: The REasons for Geographic and Racial Differences in Stroke (REGARDS) study recruited 30,239 black and white participants age 45 years and older from across the United States between 2003 and 2007. WC, height and weight were measured during a baseline in home visit and participants were followed every 6 months. Strokes were adjudicated by physicians after review of medical records. Cox proportional hazards models were used to assess stroke risk. BMI and WC were modeled in quintiles. Results: We observed 1047 incident and recurrent strokes over a mean of 6.3 years. Higher BMI and lower WC were associated with decreased risk of stroke after adjustment for demographic and medical risk factors (Table). Having both BMI and WC in the model dramatically increased the strength of association for the other variable. Collinearity was not present. The observed association was consistent across race, sex, age, WC, and history of stroke (all p for interaction >0.20). Conclusion: Increased BMI, the main measure used to define obesity, was not a stroke risk factor. Increased risk of stroke was observed in WC less than the current recommendations of 88 cm for women and 102 cm for men. Smaller waist sizes and maintenance of lean mass are important targets for stroke prevention.<br

Sign in / Sign up

Export Citation Format

Share Document