Presented at the XIV International Symposium on Atherosclerosis

1
Prevalence of Dyslipidemia Estimated from a
Self-Reported Survey Versus Clinical and
Laboratory Evaluation Comparison of SHIELD and
NHANES
Harold E. Bays, MD,1 Richard H. Chapman, PhD,2
Susan Grandy, PhD,3 and the SHIELD Study
Group1L-MARC Research Center, Louisville,
Kentucky, USA 2ValueMedics Research, Falls
Church, Virginia, USA 3AstraZeneca, Wilmington,
Delaware, USA

• Analyses
• The weighted data from each study were used to
  calculate the self-reported, national prevalence
  of diabetes, hypertension, and dyslipidemia and
  its components.
• For SHIELD, the returned sample (N211,097) was
  weighted to match 2003 US census data6 on age,
  gender, and household size.
• For NHANES, overall prevalence estimates
  (self-reported plus laboratory test confirmed)
  were calculated using NHANES sampling weights
  based on age, income, and race/ethnicity to
  represent the US adult population.
• Standard errors were estimated using SUDAAN to
  account for both the complex sample design and
  the use of both interview and morning examination
  sample data in combination.
• SHIELD self-reported prevalence estimates were
  then compared with estimates from NHANES
  19992002 using both self-reported and laboratory
  values.

Objectives Optimal treatment of dyslipidemia,
hypertension and diabetes requires that patients
know they have the diseases. Awareness of these
CHD risk factors was assessed by comparing a
self-reported survey, Study to Help Improve Early
evaluation and management of risk factors Leading
to Diabetes (SHIELD), to objective data from
National Health and Nutrition Examination Survey
(NHANES 1999?2002). Methods A 12-item SHIELD
questionnaire was sent to 200,000 households
representative of US adult population (127,420
returned 64 response). SHIELD included data on
recipient and family members (N211,097), such as
if ever diagnosed with diabetes, high blood
pressure or cholesterol problems. In NHANES,
dyslipidemia was defined as any one of TC 240
mg/dL or diagnosis of high cholesterol TG gt200
LDL-C 160 and HDL-C lt40 mg/dL. Results SHIELD
and NHANES reported prevalences of diabetes (8
vs 9) and hypertension (23 vs 29), which were
not clinically significant differences. However,
dyslipidemia was reported only approximately half
as much in SHIELD (26) as NHANES (53).
Occurrence of individual components of
dyslipidemia was uniformly less in SHIELD than
NHANES high TC17 vs 35, high LDL-C10 vs
14. Even greater differences were seen for high
TG7 vs 17 and low HDL-C5 vs 24. (All plt.01,
Chi-square.) Conclusions In contrast to diabetes
and hypertension, dyslipidemia prevalence was
less in self-reported SHIELD than objectively
assessed NHANES, with especially low self-report
of high TG and low HDL-C.

• Results for diabetes, hypertension, and
  dyslipidemia are shown in Figure 1.
• Baseline Analysis As shown in Figure 2, the
  occurrence of individual components of
  dyslipidemia was uniformly less in SHIELD than in
  NHANES (all p lt0.01 in ?2 tests), using the
  following thresholds TC 240 mg/dL (6.2
  mmol/L), LDL-C 160 mg/dL (4.1 mmol/L), HDL-C
  lt40 mg/dL (lt1.0 mmol/L) or TG gt200 mg/dL (gt2.3
  mmol/L).

• Diabetes mellitus, hypertension, and dyslipidemia
  are important CVD risk factors that have other
  associated comorbidities.1
• Among developed nations, these metabolic diseases
  have a high worldwide prevalence. In the US
• Approximately 18.2 million have diabetes
  mellitus,2 which is projected to increase to 29
  million by the year 2050.3
• Nearly one in three, or 65 million adults, has
  hypertension4 and about 28 (59 million) have
  pre-hypertension.5
• Nearly 100 million adults have total blood
  cholesterol values of 200 mg/dL (5.2 mmol/L) and
  higher, while 34.5 million have levels of 240
  mg/dL (6.2 mmol/L) or above.1
• Optimal treatment of diabetes, hypertension, and
  dyslipidemia requires that patients know they
  have the diseases.
• SHIELD (the Study to Help Improve Early
  evaluation and management of risk factors Leading
  to Diabetes) is a large, ongoing longitudinal
  study in the US adult population that was
  designed to assess the unmet medical need and
  burden of illness in patients with or at risk for
  diabetes, including self-reported rates of
  comorbid conditions.
• In this analysis, self-reported awareness of
  metabolic diseases from the SHIELD survey was
  compared with the prevalence of the same
  metabolic diseases as determined by objective
  data obtained from the National Health and
  Nutrition Examination Surveys (NHANES 1999?2002).5

Figure 2. Comparison of the prevalence of
specific dyslipidemia problems from SHIELD
self-report and NHANES laboratory findings (p
lt0.01 in ?2 tests, for all comparisons)
Figure 1. Prevalence of diabetes, hypertension,
and dyslipidemia in the US adult population, as
estimated from SHIELD and NHANES data
Analysis Using More Aggressive Thresholds As
shown in Figure 3, the occurrence of total
dyslipidemia, and individual components of
dyslipidemia was even less in SHIELD than in
NHANES (all p lt0.01 in ?2 tests with more a more
aggressive definition of dyslipidemia TC 200
mg/dL (5.2 mmol/L), LDL-C 130 mg/dL (3.4
mmol/L), HDL-C lt50 mg/dL (lt1.3 mmol/L) or TG gt150
mg/dL (gt1.7 mmol/L). When subset by gender, the
results similarly demonstrated smaller
differences between SHIELD and NHANES regarding
diabetes and hypertension prevalences, whereas
substantial differences were seen in dyslipidemia
and its components (Figure 4).

• To identify prevalence of diabetes, hypertension,
  and dyslipidemia in the US adult population
• To quantify the level of awareness of each of
  these conditions

• SHIELD
• SHIELD is a 5-year, national, longitudinal study
  of diabetes, CVD, and metabolic disease risks in
  US adults.
• As the first phase of the longitudinal SHIELD
  study, a screener questionnaire was developed by
  a panel of diabetes healthcare experts (the
  SHIELD Study Group) and, in 2004, was mailed to a
  stratified random sample of 200,000 US households
  who were part of the TNS NFO household panel.
• TNS NFO maintains a survey panel of more than
  600,000 households throughout the US, constructed
  to represent the US population in terms of
  geographic residence, age of head of household,
  and household size and income.
• The screener questionnaire consisted of 12
  questions and was completed by the head of
  household, who answered for up to 4 adult
  household members (?18 years of age).
• Respondents were asked if they had ever been
  diagnosed as having each of several conditions,
  including diabetes, high BP, or cholesterol
  problems. (We expected that respondents to a
  self-administered questionnaire would be unlikely
  to recall their actual FPG, BP, or lipid levels.)
• SHIELD data on adults (N211,097) from 127,420
  households (63.7 response rate) were analyzed to
  determine self-reported prevalence of diabetes,
  hypertension, and various types of dyslipidemia.
  
• NHANES
• Data from SHIELD were compared with similar data
  from NHANES 19992002.
• NHANES produces nationally representative data
  about the health and nutritional status of the US
  civilian non-institutionalized population.
• These data have the added value of including
  self-reported risk factors as well as clinical
  evaluation and laboratory testing to confirm
  diagnoses and to identify undiagnosed risk
  factors.
• Because the NHANES data include laboratory values
  along with diagnoses and treatments, they can be
  used with a weighting system to estimate actual
  national prevalence of various conditions.
• NHANES data on adults 18 years old (N4257) were
  analyzed to determine the prevalence of diabetes,
  hypertension, and dyslipidemia.
• Identifying conditions

Figure 3. Comparison of the prevalence of
specific dyslipidemia problems from SHIELD
self-report and NHANES laboratory findings, using
more aggressive thresholds (p lt0.01 in ?2 tests,
for all comparisons)
Figure 4. Percent of population estimated to
have each condition in SHIELD and NHANES, by
gender

• Limitations of SHIELD
• The households participating in the TNS NFO panel
  had voluntarily elected to do so, leading to the
  possibility of bias due to self-selection.
• Household panels also tend to under-represent the
  very wealthy and very poor segments of the
  population and do not include military or
  institutionalized individuals.
• Differences in results from these studies could
  be due to factors other than or in addition to
  SHIELD being self-report only, such as the
  different sampling frames and time periods used
  for the surveys.
• For example, there has been a trend toward lower
  levels of total cholesterol and LDL-C in US
  adults over time.7 However, it is unlikely that
  the time period from the NHANES surveys
  (1999?2002) and the SHIELD survey (2004) was
  sufficient to account for the large differences
  observed here.

• From a clinical significance standpoint, the
  prevalence of diabetes mellitus and hypertension
  were generally similar in the self-reported
  SHIELD compared with NHANES, which was confirmed
  by objective testing.
• In contrast to diabetes and hypertension, the
  estimated prevalence of dyslipidemia from
  self-reported SHIELD data was less than half that
  observed in the more objectively assessed NHANES.
• The especially low self-reporting of high TG and
  low HDL-C led to the greatest discrepancies
  between SHIELD and NHANES.
• This level of difference was accentuated by a
  more aggressive definition of dyslipidemia.
• The lower proportion of dyslipidemia reported in
  SHIELD relative to NHANES highlights the need for
  improved education regarding lipid parameters
  that may be important in assessing CVD risk and
  in treating CVD lipid risk factors.
• Annual results from the SHIELD longitudinal
  surveys may provide greater insight into the
  interplay of diabetes and CVD risk factors.

• 1. American Heart Association. Heart
  Disease and Stroke Statistics 2006 Update.
  Dallas, Texas American Heart Association 2006.
• 2. Centers for Disease Control and
  Prevention. National diabetes fact sheet
  general information and national estimates on
  diabetes
• in the United States, 2003. Rev ed.
  Atlanta, GA U.S. Department of Health and
  Human Services, Centers for Disease Control and
  Prevention, 2004.
• 3. Boyle JP, Honeycutt AA, Narayan
  KMV, et al. Projection of diabetes burden through
  2050. Impact of changing demography and disease
  prevalence in the
• U.S. Diabetes Care
  2001241936-1940.
• 4. Fields LE Burt VL Cutler JA, et al.
  The Burden of Adult Hypertension in the United
  States 1999 to 2000. A Rising Tide. Hypertension
  200444398-404.
• 5. NHANES 1999-2002, CDC/NCHS, NHLBI
• 6. US Census Bureau. 2003 Annual
  Supplement to the Current Population Survey
  Census Bureau Resident Population Estimates of
  the United States.
• Washington, DC US Census Bureau
  2003.
• 7. Carroll MD, Lacher DA, Sorlie PD, et
  al. Trends in serum lipids and lipoproteins of
  adults, 1960-2002. JAMA 20052941773-1781
• BP blood pressure CVD
  cardiovascular disease FPG fasting plasma
  glucose HDL-C high-density lipoprotein
  cholesterol
• LDL-C low-density lipoprotein
  cholesterol NHANES National Health and
  Nutrition Examination Survey SHIELD Study to
  Help Improve Early evaluation and
  management of risk factors Leading to Diabetes
  TC total cholesterol TG triglycerides TNS
  NFO Taylor Nelson Sofres National Family
  Opinion

References
Abbreviations
Presented at the XIV International Symposium on
Atherosclerosis Rome, Italy June 1822,
2006