Chapters 7 & 8: Statistics & Data in Public Health

National Center for Health Statistics (CDC)

• Federal agency

• States report to NCHS

  • Local data on births, deaths, vital statistics

• Surveys: health status, lifestyle, health-related behaviors

Vital Statistics

• Birth certificate

  • Issued by local government

  • Reported to state and NCHS

  • Family information, medical information

• Death certificate

  • Cause of death

• Marriages/divorces

• Spontaneous fetal deaths; abortions

• Infant deaths

  • Before 1st birthday

Census

Numbers in population; necessary to calculate rates

• U.S. Census Bureau

• Ten-year census

  • Geographic distribution

  • Sex, age, ethnic characteristics

  • Education, housing, health insurance

  • Trends in immigration and emigration also tracked by census bureau

Calculated Statistics & Rates

• Birth rates

• Mortality rates

• Age-adjusted mortality rate

• Age specific mortality rate

• Infant mortality rate

  • Death before 1^st birthday/total live births

• Life expectancy: average number of years left to individuals of certain age

• Years Potential Life Lost

  • Death < 75 yrs. premature

Surveys

• National Health Interview Survey

  • 50,000 households

• National Health and Nutrition Examination Survey  (NHANES)

  • 15 counties, 5000 people

  • Comprehensive health assessment

• Behavioral Risk Factor Survey (BRFS)

  • States conduct and report to CDC

  • Risk factors, preventative care

Uses of Data

•Detect health issues in the population

•Development of public health policy

•Plan allocation of resources to address need of community

•Evaluate efficacy of public health interventions

Confidentiality

•Removal of identifying information

•Restricted access to databases

•IRB reviews requests for access to data by researchers

Uncertainty & Probability

Epidemiological studies measure probability

• Degree of probability: p value

  • Could the observed result have occurred by chance?

  • P= 0.05 means that there is a 5 out of 100 probability that the experimental result occurred by chance

    • Statistically significant (p=0.05)

Confidence Interval

Range of values into which the true value falls

• Margins of error

• 50% +  3%:  47 to 53 %

• More narrow CI,  lower likelihood of random error

Power

Probability of finding an effect; large numbers overcome random chance

• Statisticians can calculate size of study necessary

False Positives

A test result which incorrectly indicates that a particular condition or attribute is present.

False Negatives

A test result which incorrectly indicates that a particular condition or attribute is absent.

Screening Tests

May yield false positives or false negatives

• Sensitive: few false negatives

• Specific: few false positives

• Population Screening tests

  • Want to avoid missing any positives

  • Sacrifice specificity for sensitivity

  • Follow-up with more specific test

  • High number of false positives may render test ineffective or too costly

Covid-19 Screening Tests

Tests for virus and Tests for antibody

• Diagnostic test for virus

• Test for exposure and immune response

  • Presence of reactive antibodies in blood

• Pressure on FDA to get test kits deployed

  • Emergency Authorization- lower standards shorter time; “may be effective” 

  • Little data available on performance

  • Companies were asked to do follow up accuracy studies

• Abbot Test, ID Now Rapid Test (15 min)

  • Screens for virus

  • NYU study indicated false negatives in 1/3 to ½ of cases

    • Disputed by Abbot – 90-95% accurate

    • NYU study supported by studies at Stanford, Loyola and Cleveland Clinic

    • If 1 million people tested, 95% accurate, 50,000 positive people will be given negative report

Analytical Sensitivity

Likelihood test will not miss pathogen; minimum concentration of pathogen the test can detect.

Clinical Test Performance

How test accurately diagnoses using clinical specimens

Clinical Sensitivity

Percent of positive tests among patients with confirmed disease

• Requires a reference standard- independent confirmation of disease by symptoms or another test

Covid-19 Testing Challenges

• Diagnostic testing for opening country required tests be highly sensitive

  • Will not miss positive cases in screened population

  • Low false negatives

  • Can tolerate false positives

  • Test must be able to detect asymptomatic individuals who may be contagious

• FDA must ensure clinical sensitivity and specificity before giving authorization to market

• Doctors must be careful with false negative test results when patient presents with clinical symptoms

Cost-Benefit Analysis

Estimated cost of implementing policy vs. estimated benefit

• Costs of immunization program vs. cost of treating disease

• Cannot serve as the sole or primary determinant of a health care decision.”

• Long term vs. short term benefits and costs