RESEARCH DESIGNS:

OBSERVATIONAL STUDIES

LABORATORY STUDIES OF
DISEASE MECHANISMS AND INTERVENTION/TREATMENT MECHANISM

RANDOMIZED CONTROLLED INTERVENTION TRIALS

ISSUES: WHICH RESEARCH REPORTS CAN BE BELIEVED?
WHAT STUDIES, AND TYPES OF STUDIES, NEED TO BE CONDUCTED?

OBSERVATIONAL STUDIES:

RANDOMIZED CONTROLLED TRIALS:

1. CIGARETTE SMOKING AND LUNG CANCER

• OBSERVATIONAL STUDY DATA EVENTUALLY PREVAILED BECAUSE:
• STRONG ASSOCIATION/APPARENT RISK REDUCTION FOLLOWING CESSATION
• SUBSTANTIAL FRACTION OF NEVER SMOKERS
• CONSISTENT FINDINGS ACROSS DIVERSE POPULATIONS
  (BUT LOWER RR IN JAPAN VS. USA)

  LABORATORY STUDIES:

  BIOLOGICAL MEASURES OF EXPOSURE

  BIOLOGICAL MECHANISMS/PATHWAYS

  LITTLE USE OF RCT'S

  CERTAINTY OF RESULTS FOR PASSIVE SMOKING AND LUNG CANCER?

• MANY OBSERVATIONAL STUDIES AND META ANALYSES REPORTING A 40-50% REDUCTION IN CORONARY HEART DISEASE; A DURATION-DEPENDENT 20-30% INCREASE IN BREAST CANCER; REDUCTIONS IN FRACTURE RATES; AN INCREASE IN VENOUS THROMBOEMBOLIC; AND A REDUCTION IN TOTAL MORTALITY

  E.G., Grodstein et al (1997, NEJM)
  'Background. Postmenopausal hormone therapy has both benefits and hazards, including decreased risk of osteoporosis and cardiovascular disease and an increased risk of breast cancer'
  'After adjustment for confounding variables, current hormone users had a lower risk of death [RR=0.63 (0.56,0.70)]'
  'Current hormone users with coronary risk factors had the largest reduction in mortality [RR=0.51 (0.45,0.57)]'

• POSTMENOPAUSAL ESTROGEN/PROGESTIN INTERVENTION TRIAL (1995, JAMA) INCREASE IN HDL-C AND DECREASE IN LDL-C FOR EACH OF SEVERAL HRT PREPARATIONS VERSUS PLACEBO (also coagulation factors, blood pressure, insulin, endometrial histology, bone mineral density,…)
  HEART AND ESTROGEN/PROGESTIN REPLACEMENT STUDY (HERS, JAMA 1998)
  
  2763 postmenopausal women with established coronary disease
  625mg/day CEE plus 2.5mg/day MPA

  11% lower LDL-C and 10% higher HDL-C in active versus placebo groups

  No difference in CHD events over 4.1 year average follow-up

• 'EXPERIMENTATION TRUMPS OBSERVATION' (Diana Petitti, JAMA, 1998)
  'These findings are a sobering reminder of the limitations of observational research, the incompleteness of current understanding of the mechanisms of vascular disease and the dangers of extrapolation'.

  'Compliance bias is large enough to explain entirely reductions in the relative risk of CHD between users and non-users of ERT and HRT of the magnitude found in observational studies'.

  'The lipid hypothesis has dominated thinking about CHD for at least 4 decades. There is a growing recognition that thrombotic phenomena play an important role in acute coronary syndromes'.

  'When an exposure can be assigned at random, it should be assigned randomly. Commitment to randomized trials as the standard of proof must be especially strong when the public health implications are so great'.

  EDITORIALS FOLLOWING HERS

  'HERS - A MISSED OPPORTUNITY'
  (Malcolm Whitehead and Meir Stampfer 1998, Climacteric)

  • One preparation only. Only among women with established coronary heart disease

  • Failure of HERS to achieve the planned period of observation (4.1 vs. 4.75 years) is a 'major flaw'.

  • 'There are other problems with HERS'. (Adherence rates vs. projected)

  • 'Preliminary data from NHS support a pattern of early transient increase followed by a substantial decline in risk as the duration of therapy is extended. Thus, the results of HERS do not contradict the directly relevant observational data'.

  • 'Unless prospective, randomized trials possess sufficient power then definitive conclusions cannot be drawn and more questions than answers will be the result. This, regrettably, is the HERS legacy'.

  EDITORIALS FOLLOWING HERS

  'HRT AND THE HERS FINDINGS - HAS THE GROUND SHIFTED?
  

  Wulf Utian (1998, Menopause Management)
  

  • 'Unfortunately HERS was poorly conceived and designed, taking an unnecessary gamble that has now come back to haunt all parties concerned, and confuse consumers and providers alike'.

  • '…already indications that progestins might attenuate some of the estrogen-induced cardiac-benefit effect. HERS would therefore have best included an estrogen-only arm, or have been designed as an estrogen versus placebo comparative study'.

  • 'The initial negative effect is almost certainly due to the attenuation of estrogen-induced increase in coronary flow'.

  LESSONS/QUESTIONS FROM AVAILABLE HRT AND CHD DATA

  • OBSERVATIONAL STUDIES TO BE REGARDED AS HYPOTHESIS GENERATING UNLESS MAGNITUDE OF POTENTIAL BIASES CAN BE QUANTIFIED AND ARE SMALL ENOUGH TO HAVE LITTLE EFFECT ON RESULTS. CAN STUDIES BE DESIGNED TO ASSESS

  • PRESCRIPTION (CONFOUNDING) BIAS?

  • COMPLIANCE BIAS?

  • CONSIDERATION OF POTENTIAL SURROGATE OUTCOMES FOR A DISEASE (E.G., LIPID LOWERING FOR CHD) TYPICALLY NEED TO BE CARRIED OUT AFRESH FOR EACH TREATMENT.

  • RCT'S NEED TO BE LARGE ENOUGH, AND LONG ENOUGH TO ANSWER THE IMPORTANT PUBLIC HEALTH QUESTIONS.

  • CAN JOINT CONDUCT AND ANALYSIS OF RANDOMIZED TRIALS AND OBSERVATIONAL STUDIES EXTEND THE IMPLICATIONS OF RCT'S?

  2. BETA CAROTENE SUPPLEMENTATION AND CANCER, CORONARY HEART DISEASE AND ALL-CAUSE MORTALITY

  • MANY OBSERVATIONAL STUDIES REPORT THAT PERSONS SELF-REPORTING A RELATIVELY LARGE CONSUMPTION OF FOODS RICH IN BETA CAROTENE (AND OTHER CAROTENOIDS) HAVE A LOWER RISK OF CANCER AT SEVERAL SITES. OBSERVATIONAL STUDIES ARE PARTICULARLY CONSISTENT FOR LUNG CANCER (ZIEGLER ET AL, CANCER CAUSES AND CONTROL, 1996)

  • AT LEAST 10 PROSPECTIVE STUDIES REPORTING HIGHER RISK OF LUNG CANCER, HEART DISEASE, OTHER CANCER OR ALL-CAUSE MORTALITY AMONG PERSONS HAVING RELATIVELY LOW BLOOD LEVELS OF BETA CAROTENE E.G., RR=0.62 (0.44,0.87) FOR TOTAL MORTALITY OVER 8.2 YEARS AVERAGE FOLLOW-UP FOR PERSONS HAVING BC BLOOD CONCENTRATION IN HIGHEST VERSUS LOWEST QUARTILE (GREENBERG ET AL, 1996, JAMA)

  INTERVENTION TRIALS OF BETA CAROTENE SUPPLEMENTATION

  ALPHA TOCOPHEROL, BETA CAROTENE CANCER PREVENTION STUDY GROUP (NEJM, 1994)

  • 29133 male smokers 50-69 years of age;
  • 20mg/day beta carotene, 5-8 years follow-up

  • Lung cancer RR=1.18 (1.03,1.36) &nbps;    (876 cases)
  • No significant difference for other cancers
  • Total mortality RR=1.08 (1.01,1.16)

  CAROTENE AND RETINOL EFFICACY TRIAL (Omenn et al, 1996, NEJM)

  • 18,314 smokers, former smokers and asbestos exposed persons
  • 30mg BC plus 25000 IU retinol/day
  • Trial stopped early after 4 years average follow-up

  • Lung cancer RR=1.28 (1.04,1.57) &nbps;    (388 cases)
  • No significant difference for other cancers
  • cardiovascular disease mortality RR=1.26 (0.99,1.61)
  • Total mortality RR=1.17 (1.03,1.33)

  PHYSICIANS HEALTH STUDY
  (Hennekens et al, NEJM, 1996)

  • 22,071 male U.S. physicians, ages 40-84; 12 years follow-up
  • 50mg BC supplementation on alternate days
  • 11% current smokers, 39% current smokers
  • No difference in cancer incidence, cardiovascular disease incidence or total mortality
  • Lung cancer (82 - BC group, 88 - placebo group)

  QUESTIONS/LESSONS

  • Reliability of observational data
  • diet a complex mixture of foods and nutrients
  • measurement error in dietary self reports, and also in blood measurements
  • ability to control for distinguishing characteristics of high fruit/vegetable consumers and supplement users
  • Supplements versus foods - could BC be an antioxidant at low levels and a pro-oxidant at higher levels?
  • Methods of assessing unexpected CT results?
  • Developmental process needed to justify full-scale intervention trial?

  2. DIETARY FAT AND POSTMENOPAUSAL BREAST CANCER

  NUTRIENT CONSUMPTION ASSESSMENT TOOLS
  

  • Food records
  • Dietary recalls
  • Food frequencies
  • Biological measures (e.g., from blood or urine)

  ISSUES/CHALLENGES
  

  • Human diet a complex mixture of foods and nutrients with many highly correlated elements
  • Nutrient intakes may not be highly variable within populations available for study
  • Ability to assess short and long term nutrient consumption may be limited by substantial random and systematic biases?

  RESEARCH DESIGNS
  

  • Aggregate data (ecologic) studies
  • Cohort and case-control studies
  • Randomized controlled intervention trials

  ISSUES/LESSONS
  

  • NEED STRONGER RESEARCH TOOLS
  • OBJECTIVE MEASURES (BIOMARKERS) OF NUTRIENT INTAKE ON APPROPRIATE SAMPLES AS A BASIC COMPONENT OF COHORT AND CASE-CONTROL STUDIES
  • MORE APPROPRIATE MEASUREMENT MODELS IN ANALYSIS AND REPORTING
  • NEED A MORE VARIED RESEARCH STRATEGY
  • IS THERE A ROLE FOR GOOD QUALITY POPULATION COMPARISONS (E.G., PRENTICE AND SHEPPARD, 1995, BIOMETRIKA)?
  • METHODS TO EFFECTIVELY USE BIOMARKER ENDPOINTS IN SMALL SCALE HUMAN FEEDING TRIALS TO IDENTIFY PROMISING INTERVENTIONS?
    

  OPPORTUNITIES AND CHALLENGES ASSOCIATED WITH FORTHCOMING GENETICS/GENOMICS ARRAY DATA

  
  

  
  

  
  

  
  Risk Evolution with Time

  ISSUES/LESSONS

  • IDENTIFICATION OF DISEASE-RELATED GENES VIA LINKAGE/LINKAGE-DISEQUILIBRIUM ANALYSIS
  • EFFICIENT DESIGN AND ANALYSIS OF FAMILY STUDIES FOR IDENTIFICATION AND COURSE MAPPING OF DISEASE GENES
  • NEED TO MODEL 'ENVIRONMENTAL' RISK FACTORS?
  • METHODS TO STUDY GENE-ENVIRONMENT INTERACTIONS?
  • ROLE OF MEASURES OF GENE EXPRESSION, OR OF ACQUIRED GENETIC CHANGES IN INTERVENTION DEVELOPMENT; OR AS BATTERY OF INTERMEDIATE OUTCOMES IN PREVENTION OR TREATMENT TRIALS

  SUMMARY: EMERGING RESEARCH AREAS

  • RESEARCH TOWARD AN INTEGRATED AGENDA FOR CHRONIC DISEASE POPULATION RESEARCH
  • STUDIES TO QUANTIFY VARIOUS BIASES
  • INNOVATIVE APPROACHES (E.G., DATA AND MODELS) FOR REDUCING OR ACCOMMODATING SUCH BIASES
  • METHODS TO EFFECTIVELY COMMUNICATE STUDY RELIABILITY
  • CONSENSUS AS TO CIRCUMSTANCES (STRENGTH OF HYPOTHESIS; FEASIBILITY STUDY; PUBLIC HEALTH IMPORTANCE) THAT JUSTIFY RCT
  • RESEARCH INTO ROLE/USE OF BIOMARKERS FOR INTERVENTION DEVELOPMENT/TESTING
  • SURROGATE OUTCOME IDENTIFICATION
  • METHODS FOR COMBINING HIGH DIMENSIONAL RESPONSE DATA TO MEANINGFULLY COMPARE TREATMENTS
  • RESEARCH INTO METHODS NEEDED FOR DISEASE GENE DISCOVERY AND FOR STUDY OF GENE-ENVIRONMENT INTERACTIONS