The Medical Benefits of Moderate Drinking Revisited: Alcohol Use and Self-Reported Health Status Michael T. French Silvana K. Zavala University of Miami

1
The Medical Benefits of Moderate Drinking
Revisited Alcohol Use and Self-Reported Health
Status Michael T. FrenchSilvana K.
ZavalaUniversity of Miami
Health Economics Research Group
2
Alcohol Use and Health

• Heavy drinking and alcohol abuse negatively
  related to health
• Cardiovascular disease, psychiatric disorders,
  injury, etc.
• Moderate drinking may be protective of adverse
  health outcomes such as hear disease, cancer,
  diabetes, and total mortality risk
• About 2 drinks per day for men
• About 1 drink per day for women
• J-shaped relationship
• Is this effect causal?
• Does moderate drinking display a similar
  relationship with a subjective, self-assessed
  measure of health status
• Few studies in this area and most use data from
  European countries

3
Sample and Data

• 2002 National Health Interview Survey (NHIS)
• Cross-sectional, multistage probability survey
• Representative of the non-institutionalized
  civilian household population of the U.S.
• In 2002, data were collected from 31,044 adults
  for the sample adult core questionnaire
• Response rate 74.4 percent

4
Measures

• How would you describe your general health
  status?
• excellent, very good, good, fair, or poor
• Dichotomous variable equal to 1 if excellent or
  very good and 0 if good, fair, or poor
• Alcohol use
• Total number of drinks consumed in the past year
• Number of days consumed 5 or more drinks (past
  year)
• Lifetime abstainer (fewer than 12 drinks in
  lifetime)
• Former infrequent drinker (12 or more drinks in
  lifetime, but never as many as 12 drinks in a
  single year, and no drinks in the past year)
• Former regular drinker (12 or more drinks in
  lifetime, 12 or more drinks in one or more years,
  but no drinks in the past year)
• Current drinker (12 or more drinks in lifetime
  and at least 1 drink in the past year)
• Current infrequent drinker (12 or more drinks in
  lifetime and 1-11 drinks in past year)
• Current light drinker (12 or more drinks in
  lifetime, 12 or more drinks in past year, and 3
  drinks or fewer per week, on average)
• Current moderate drinker (12 or more drinks in
  lifetime, 12 or more drinks in past year, and
  4-14 drinks per week for men and 4-7 drinks per
  week for women, on average)
• Current heavy drinker (12 or more drinks in
  lifetime, 12 or more drinks in past year, and
  more than 14 drinks per week for men and more
  than 7 drinks per week for women)

5
Theoretical Considerations

• What is the mechanism(s) whereby moderate
  drinkers are more likely than any other group to
  rate their health status as above average
• Moderate drinking is a surrogate measure for a
  host of other, often unobservable factors that
  are significantly related to health status
• If moderate drinkers are more likely to follow a
  moderate lifestyle, then the estimated effect of
  moderate drinking will probably be overstated
• Another possible pathway leads from health status
  to alcohol consumption rather than from alcohol
  consumption to health status
• A heavy drinker reduces consumption to moderate
  drinking after a health shock
• All else equal, moderate alcohol consumption has
  a beneficial effect on the body as well as the
  mind
• This is the mechanism postulated in the present
  paper

6
Data Analysis

• Standard logistic regression model that can be
  expressed as follows
• HS is equal to 1 if the subject reported
  excellent or very good health status and to 0 if
  good, fair, or poor health status
• X is a vector of control variables that influence
  health status, including demographic
  characteristics (e.g., age, race, education,
  marital status), select chronic health conditions
  (e.g., cancer, diabetes, hypertension), and risk
  factors (e.g., smoking, exercise)
• Depending on the specification, A is a
  dichotomous, categorical, or continuous measure
  of alcohol use
• bs are parameters to estimate
• Coefficient estimates for these models are log
  odds ratios
• Separate analysis for men and women

7
Descriptive Statistics

• Among men, 62 percent reported above average
  health status whereas 58 percent of women were in
  this group
• Those in the above average health status group
  were significantly more likely to be younger,
  non-Black, employed, more educated, with lower
  BMI, more people in the household, and more
  income
• Among women, being Hispanic was negatively
  related to above average health status and being
  married was positively related.
• Individuals within all of the current drinking
  groups except heavy were significantly more
  likely to report above average health status
• Former drinkers and lifetime abstainers were more
  likely to report average or below average health
  status
• Annual drinks positively related to above average
  health status and number of days consuming 5 or
  more drinks negatively related
• Former and current smokers were significantly
  more likely to report average or below average
  health status as were those individuals with a
  chronic health condition
• Regular exercisers were significantly more likely
  to report above average health status

8
Regression Results

• Current moderate drinkers significantly more
  likely to self-report excellent or very good
  health status relative to former infrequent
  drinkers and lifetime abstainers
• OR1.27 (plt0.01) among men
• Current light (OR1.40, plt0.01) and current heavy
  (OR1.55, plt0.01) drinkers also had significant
  odds ratios among women, but the magnitudes were
  smaller than the estimate for current moderate
  drinkers (OR2.03, plt0.01)
• Alcohol use has a significant concave
  relationship (i.e., inverted U-shaped) with above
  average health status for women
• Number of days 5 drinks not significant for
  females and significant in the negative direction
  for males
• Implication that heavy or binge drinking is
  detrimental to subjective health status

9
Discussion

• Does moderate drinking lead to improved health
  status or does excellent/very good health lead to
  moderate drinking
• If alcohol use is significantly correlated with
  the error structure of the model then omitted
  variable bias is possible
• Alcohol consumption may be instrumented with
  exogenous variables such as alcohol prices or
  taxes, alcohol laws and restrictions, and alcohol
  treatment availability
• Due to the lack of geographical identifiers in
  the 2002 NHISit was not possible to determine
  whether endogeneity bias was present
• Findings may support the theory of a
  cardio-protective effect of moderate drinking and
  delayed overall mortality within certain
  populations, both of which may lead to better
  health status and lower health care utilization
• Any public or private campaign to promote the
  health benefits of moderate drinking should
  carefully consider the susceptibility to problem
  drinking and alcoholism for a large segment of
  the population