53
Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

  • View
    219

  • Download
    1

Embed Size (px)

Citation preview

Page 1: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Managerial Epidemiology

Ty Borders, Ph.D.

Assistant Professor

Department of Health Services Research & Management

Texas Tech School of Medicine

Page 2: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Learning objectives– Define epidemiology– Explain the role of epidemiology in health

care management– Calculate major descriptive epidemiologic

indicators– Understand what are the more prevalent

diseases and disorders in the U.S.

– Calculate and interpret Relative Risk

– Calculate and interpret Odds Ratio

– Understand types and purposes of analytical studies

Page 3: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

What is Epidemiology?

• Study of the distribution and determinants of disease

• The doctrine of what is among or happening to people– Epi: among– Demos: people– Logos: Doctrine

Note: from Charles Lynch, M.D., Iowa College of Public Health

Page 4: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

History of epidemiology1662, John Graunt

a petty merchandiser in London, publishes a report on births and deaths in London.

First to quantify disease patterns.

1839, William Farr a physician, establishes system for routine

compiliation of no. and causes of death in England and Wales

1855, John Snow

a physician, studied whether drinking water in Soutwark and Vauxhall increased risk of cholera

Page 5: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Subfields of Epidemiology

• Clinical epidemiology (patients)

• Social epidemiology (populations)

• Genetic epidemiology (patients/populations)

• Health services epidemiology (populations/patients)

Page 6: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

A broader definition

• Study of the distribution and determinants of health-related events and states

– Utilization of health services

– Health-related quality of life

– Satisfaction with care

Page 7: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Managerial Epidemiology

Epidemiological methods applied to the...

I. Evaluation of community health care needs

II. Study of health services utilization (access)

III. Health outcomes research (study of the impact of health care services on health outcomes)

• Effectiveness

• Patient satisfaction

• Health-related quality of life

Page 8: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

I. Evaluation of community health care needs

• Descriptive morbidity and mortality indicators• Cancer incidence and mortality rates

• Infectious disease rates

• Infant mortality rate

• Descriptive social and demographic indicators• Median income, unemployment rates, etc.

• Market research surveys

Page 9: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

II. Utilization

Health care

system

External environment

Predisposing Enabling Need

Environment

Personal health

practices

Use of health

services

Perceived health status

Evaluated health status

Consumer satisfaction

Population Characteristics Behavior Outcomes

Page 10: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

III. Health outcomes research

• Study of the quality of health services

• This includes the effectiveness of health services

– Results from RCTs may not apply in real world– A number of factors influence who receives a

treatment– A treatment may be more/less effective for particular

subgroups

Page 11: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Descriptive vs. Analytical epi.

• Descriptive epidemiology

– Study of the amount and distribution of disease within a population by person, place, and time

– Provides info. on patterns of disease occurrence by age, sex, race, marital status, etc.

• Analytical epidemiology

– Study of the determinants of disease or reasons for relatively high or low frequency in specific groups

Page 12: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Biologic Concepts

• Agent-Host Environment– An agent interacts with a host in a particular

environment to produce disease (the epidemiologic triangle)

Host

Vector

Agent Environment

Page 13: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Biologic Concepts• Almost all diseases have multiple causes

• Necessary and sufficient– Without the factor, the disease never develops

• Necessary but not sufficient– Requires multiple factors

• Sufficient but not necessary– Factor can produce disease, but so can other factors

• Neither sufficient nor necessary– Probably represents causal relationships in most chronic

diseases

Page 14: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Examples of routes of transmission

Agent Disease

Respiratory Cigarette smoke Lung cancer

Influenza virus Flu

Gastrointestinal Vibrio cholera Cholera

Lead Lead poison.

Sexual transm. Papilloma virus Cervical cancer

Perinatal exposure Rubella virus Cong. Defects

Blood stream exp. Clostridium tetani Tetanus

& skin breakage

Page 15: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Incubation or Induction Period

• The period of time between exposure to a causative agent and the appearance of first clinical manifestations

Infection

Incubation/induction/ latent period

Disease

FatalInapparen

t Mild Moderate Severe

Likely to be seen by doctorLikely to be hospitalized

Page 16: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Study types

• Descriptive studies

– Population level: correlational, ecologic, or aggregate

– Individual level: case reports, case series

• Analytical, observational studies

– Cross sectional survey– Case-control studies– Cohort studies

• Analytical, intervention studies

Page 17: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Measures of disease occurrence

• 3 measures used to assess the frequency of disease or other health events

– Cummulative incidence (CI), also called Risk

– Prevalence

– Incidence density, also called incidence rate

Page 18: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Types of Incidence and Prevalence Measures

Rate Type Numer. Denom.

Morbidity rate Incidence # new nonfatal Total pop.

cases at risk

Mortality rate Incidence # deaths from Total pop.

a disease(s)

Case-fatality rate Incidence # deaths from # of cases

a disease of that disease

Period Prevalence # existing cases Total pop.

plus new cases

diagnosed during

given time period

Page 19: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Risk

• Sometimes also called cumulative incidence

• Proportion of unaffected individuals who, on average, will contract disease of interest over a specified period of time

Page 20: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Calculation of risk

R = New cases

Persons at risk

R = 0 if no new occurrences arise

R = 1 if the entire population becomes infected

Page 21: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Example

We are interested in the risk of acquiring a nosocomial infection. A study was conducted on 5031 patients5031 patients.

596 patients developed infection within 48 hours after admission.

R = 596 / 5031 = 0.12 = 12%

Page 22: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Calculation of prevalence

• Prevalence is a measure of the number of existing cases in a population.

• Specifically, the proportion of a population that has a disease at a particular point in time.

Page 23: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Prevalence

P = Number of cases

Number of persons in population

• Prevalence, like risk, ranges between 0 and 1.

Page 24: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Incidence rate

• Also called incidence density

• Reflects the occurrence of new cases (like risk does)

• But, also measures the rapidity with which event occurs

Page 25: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Calculation of incidence rate

IR = New cases

Person time

Page 26: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Calculation of incidence rate

IR = New cases

Person time

Page 27: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

ExamplePatient A develops a disease 2 years after entry into

study. Thus, the person-time for Patient A is 2 years.

Patients B,C,D,E an F contribute 2,3,7,2 and 6 years, respectively. Thus, the number of person-years is 2+2+3+7+2+6 = 22.

IR = new cases/ PT = 2 / 22

Page 28: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Summary

Characteristic Risk Prev. IR

What is Prob. % of pop. Rapiditymeasured of disease with dis. of dis.

Occurrence

Units None None Cases/person-time

Time of disease Newly Existing Newlydiagnosis diagnosed diagnosed

Synonyms Cumulative - Incidence

Incidence Density

Page 29: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Survival

• Probability of remaining alive for a specific length of time

• For chronic disease, like cancer, 1-year and 5-year survival are important indicators of prognosis and severity.

Page 30: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Calculation of survival

Survival = A - D

A

D = number of deaths observed over a defined period of time

A = number of newly diagnosed patients under observation

Page 31: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Calculation of survival

Survival = A - D

A

D = number of deaths observed over a defined period of time

A = number of newly diagnosed patients under observation

Page 32: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Types of rates

• Crude rates– Rates presented for entire population– e.g. Cancer mortality rate in 1980 (416,481

cancer deaths / midyear U.S. population)

• Category specific rates– Rates presented for individuals in specific

categories– e.g. Cancer deaths among persons 45-54

Page 33: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Adjusted rates

• If we are interested in the magnitude of the health problem, we don’t need adjusted rates

• If we are interested in comparing populations, we need to adjust for differences

Page 34: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Adjustment methods

• Take a weighted average of category-specific rates

• Direct method

• Indirect method

Page 35: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Pros/cons of crude, specific, and adjusted rates

Type Strengths LimitationsCrude Actual summary Difficult to interpret

rates b/c populations may

vary in composition

Specific Homogeneous Cumbersome to compare

subgroups many subgroups of 2

or more populations

Adjusted Summary statistics Fictional rates

Differences in Absolute magnitude

composition removed depends on standard

population chosen

Page 36: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Standardized mortality rate (SMR)

• SMR = observed deaths / expected deaths= indirect adjusted rate / crude rate of

standard pop.

• Usually expressed as a percent

Page 37: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Percentage Uninsured

Page 38: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Person-years of life lost (in 1,000s) from leading causes of cancer, 1991 (from Greenberg, 1996)

2145

845

754

352

342

0 500 1000 1500 2000 2500

Lung

Breast

Colon/rectum

Pancreas

Leukemias

Page 39: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Years of Potential Life Lost before age 65 by cause of death (per 100,000 person years) (from Greenberg,

1996)

935

843

628

395

347

0 200 400 600 800 1000

Injuries

Cancer

Heart disease

Homicide

HIV infection

Page 40: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Leading causes of death, 1996

Cause Frequency

Heart disease 31.6%

Cancer 23.4%

Stroke 6.9%

Chronic lung disease 4.6%

Accidents 4.0%

Pneumonia/influenza 3.6%

Diabetes mellitus 2.7%

HIV/AIDS 1.3%

Page 41: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Observational Studies

• Cross - sectional– Provides estimate of the strength of association

between a factor and outcomes or event– Can not determine timing of exposure– e.g. A telephone survey of rural residents

conducted at one point in time

• Case - control study– Compare the prevalence of exposure between 2

or more groups (i.e. cases and controls)

Page 42: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Observational Studies (cont.)

• Prospective cohort studies

• Retrospective (historical) cohort studies

Page 43: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Cohort Studies

Onset of study

Time

Exposed

Unexposed

Eligible subjects Disease

No Disease

Disease

No Disease

Direction of inquiry

Page 44: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Comparison of prospective and retrospective studies

Attribute Retrospective Prospective

Information Less complete More complete

Discontinued

exposures Useful Not useful

Emerging, new

exposures Not useful Useful

Expense Less costly More costly

Completion

time Shorter Longer

from Greenberg et al.

Page 45: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Adv./Disadv. of cohort studiesAdvantages Disadvantages

Direct calculation Time consuming

of relative risk

May yield info. on incidence Require large sample sizes

Clear temporal relationship Expensive

Can yield info. on multiple Not efficient for study of

exposures rare events

Minimizes bias Losses to follow-up

Strongest observational design

for establishing cause-effect

from Greenberg et al.

Page 46: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Relative Risk

Relative risk (or risk ratio) = ratio of two rates

RR = incidence rate among exposed

incidence rate among unexposed

Page 47: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Relative Risk

Exposure

Yes No

Outcome Death a b a+b

No death c d c+da+c b+d

RR = a/ (a+c)

b/ (b+d)

Page 48: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Example of Relative Risk

Apgar score 0-3 4-6

Outcome Death 42 43 85No death 80 302 382

122 345 467

Risk among exposed = 42 / 122 = 34.4%Risk among unexposed = 43 / 345 = 12.5%RR = 34.4 / 12.5 = 2.8

Page 49: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Observational Studies (cont.)

• Case - control study– Compare the prevalence of exposure between 2

or more groups (i.e. cases and controls)– Pairwise matching

Page 50: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Case-Control Studies

Onset of study

Time

Controls

Direction of Inquiry

CasesExposed

Unexposed

Unexposed

Exposed

Study Onset

Page 51: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Odds Ratio

• Often, we do not have info. about risk

• Therefore, we calculate the OR

Exposure

Yes No

Outcome Yes a b

No c d

Odds of case exposure = (a/a+b) / (b/a+b) = a / b

Odds of control exposure = c / d

Page 52: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Example of Odds Ratio

Exposure

Yes No

Cases 50 15

Controls 30 20

OR = (a/b) / (c/d)

= ad / bc = 50*20 / 30*15 = 2.22

Page 53: Managerial Epidemiology Ty Borders, Ph.D. Assistant Professor Department of Health Services Research & Management Texas Tech School of Medicine

Experimental Studies

• Experimental

– Randomization to an intervention

– Voluntary participation

– Experimental control