EAB 705 Comprehensive Notes on Epidemiological Measures of Morbidity and Mortality wk 2

Foundations of Epidemiological Measurement

• Core Purpose of Morbidity and Mortality Measurement:

  • Identify the etiology (cause) of disease.

  • Monitor disease trends over time.

  • Evaluate public health interventions and healthcare programs.

• Key Parameters for Defining Disease Outcomes:

  • Establish clear diagnostic criteria specifying what counts and what does not count as a case.

  • Specify the exact timing of occurrence and the period of disease risk.

  • Characterize the population of interest (whether individuals are at risk, and the state of the cohort).

Population Dynamics and Cohort Structures

• Closed Population (Cohort):

  • Defined at the start of a study period based on a shared characteristic (typically exposure to a risk factor).

  • No new members are added after initial cohort definition.

  • The population size can only decrease over time due to death or disease onset.

  • The average age of the population increases monotonically over time.

  • Example Cohort: In a closed population of 1,000 individuals followed over time, death eventually reduces the population size to zero. If disease onset is the study endpoint, individuals are followed until they develop the disease or until the follow-up period ends, whichever occurs first.

• Open Population:

  • Members can enter or leave the population during the observation period.

  • Inflow sources: Births, immigration, and new participant recruitment.

  • Outflow sources: Emigration, death, and loss to follow-up.

  • The average age and total population size can increase, decrease, or remain constant.

• Steady (Stationary) State:

  • An open population condition where inflow and outflow balance out, maintaining a constant population size and age distribution.

  • Example Assessment: In a study tracking participants (labeled A through N):

    • 7 participants begin at time point 0.

    • 4 participants (B, D, H, M) terminate follow-up due to disease onset.

    • 1 participant (F) is lost to follow-up without developing the disease.

    • 5 new participants (C, E, G, I, N) are enrolled during the study period.

    • Because participant recruitment replaces departing members, the population remains in a steady state.

Basic Measures of Disease Frequency

• Count:

  • The simplest and most basic measure in epidemiology, referring to the absolute number of cases of a disease or health event.

  • Examples: The number of college dorm residents diagnosed with hepatitis; the number of traffic fatalities in a New York City borough during a 24-hour period.

• Ratio:

  • A measure obtained by dividing one quantity by another ($X / Y$).

  • The numerator is not necessarily included in or related to the denominator.

  • Example: The ratio of male cases to female cases ($2:1$ male to female ratio).

• Proportion:

  • A specific type of ratio describing the fraction of a population affected by a condition.

  • The numerator is strictly included within the denominator.

  • Formula: $\text{Proportion} = \frac{A}{A + B}$

  • Example: The number of male cases divided by the total number of all cases.

• Rate:

  • A measure of how fast an event or disease occurs in a population per unit of time.

  • The numerator contains the number of disease events, while the denominator incorporates a specified period of time.

  • Example: Crude death rate (number of deaths in a calendar year divided by the reference population at the midpoint of the year).

Measures of Morbidity: Incidence

• Definition of Morbidity:

  • Any departure, subjective or objective, from a state of physiological or psychological well-being (sickness, illness, or disability).

• Incidence Overview:

  • Measures the frequency of new cases of disease in a population over a defined period.

  • Mnemonic: I-N-C-I-D-E-N-C-E $\rightarrow$ N for New cases.

  • Directly quantifies individual risk of developing disease and helps establish exposure-outcome relationships.

• Incidence Rate:

  • Describes the rate of disease development in a group over time.

  • Numerator: Number of new cases accruing during the time period (excluding existing cases).

  • Denominator: Population at risk during the time period. Excludes individuals who already have the disease or are biologically incapable of developing it (e.g., women with previous oophorectomy excluded from ovarian cancer risk cohorts).

    • Chronic vs. Infectious Disease Cohorts: Chronic disease rates often use average populations due to widespread susceptibility; infectious disease cohorts must rigorously exclude immune individuals.

  • Time Period: Specified duration (week, month, year, or multi-year).

  • Multiplier: A base factor ($100$, $1,000$, $10,000$, or $100,000$) selected for clarity.

  • Calculation Example: 500 new disease cases in a population of 30,000 over one year using a 1,000 multiplier:

    • $\text{Incidence Rate} = \frac{500}{30,000} \times 1,000 = 16.7\text{ per } 1,000\text{ per year}$

    • Reporting Requirement: The multiplier and time unit must be explicitly included in the units (e.g., "per 1,000 per year").

• Cumulative Incidence:

  • Measures the average risk of an individual developing the disease over a specified time period.

  • Calculated exclusively in a closed population.

  • Formula: $\text{Cumulative Incidence} = \frac{\text{Number of new cases during time period}}{\text{Population at risk at start of time period}}$

  • Calculation Example: 2 new cases developing in a closed cohort of 7 individuals at risk:

    • $\text{Cumulative Incidence} = \frac{2}{7} = 0.29$

• Incidence Density:

  • Used in open populations or dynamic cohorts where individuals contribute varying lengths of follow-up time.

  • Formula: $\text{Incidence Density} = \frac{\text{Number of new cases}}{\text{Total person-time at risk}}$

  • Person-Time: Sum of exact periods of time each individual was at risk ($\text{person-years}$, $\text{person-months}$, or $\text{person-days}$).

  • Cohort Tracking Calculation Example: A 12-participant study where individual follow-up times are measured:

    • Participant A (enrolled 1981, left post-1989): 8.3 years.

    • Participant B (enrolled 1981, disease onset 1992): 11.0 years.

    • Summing all 12 participants yields $107.7\text{ person-years}$ at risk.

    • With 3 total disease cases recorded:

    • $\text{Incidence Density} = \frac{3}{107.7} \times 1,000 = 28\text{ per } 1,000\text{ person-years}$

    • If expressed with a multiplier of 100: $2.8\text{ per } 100\text{ person-years}$.

• Comparative Example (Cumulative Incidence vs. Incidence Density):

  • Scenario Data: 5 new HIV/AIDS cases, total cases = 56, total population = 100,000, population at risk = 20,000. Each person's time at risk is 0.5 years.

  • Cumulative Incidence Calculation:

    • $\text{Cumulative Incidence} = \frac{5}{20,000} = 2.5\text{ per } 10,000\text{ people}$

  • Incidence Density Calculation:

    • $\text{Person-Years} = 20,000 \times 0.5 = 10,000\text{ person-years}$

    • $\text{Incidence Density} = \frac{5}{10,000} = 5\text{ per } 10,000\text{ person-years}$

  • Analytical Significance: Relying on cumulative incidence without accounting for actual follow-up time understates true disease occurrence rate by half.

• Attack Rate:

  • An alternative form of incidence rate (incidence proportion) used for acute disease outbreaks over short time intervals (e.g., foodborne illness, acute viral outbreaks).

  • Spans the timeframe from the primary case to the final outbreak case.

  • Formula: $\text{Attack Rate} = \frac{\text{Number of ill people}}{\text{Total population at risk}} \times 100$

  • Calculation Example: A cholera outbreak in April 2012 yields 490 new cases in a population at risk of 18,600:

    • $\text{Attack Rate} = \frac{490}{18,600} \times 100 = 2.6\text{\%}$

Measures of Morbidity: Prevalence

• Prevalence Overview:

  • The proportion of a population affected by a disease at a specific point or period of time.

  • Includes both new and existing (old) cases.

  • Mnemonic: P-R-E-V-A-L-E-N-C-E $\rightarrow$ E for Existing cases.

  • Measures overall disease burden, not risk.

  • Essential for public health resource allocation, healthcare planning, and evaluating disease etiologies.

  • Formula: $\text{Prevalence} = \frac{\text{Number of existing cases}}{\text{Total population at risk at specified time}}$

  • Calculation Example: In a survey of 1,150 post-partum women in Maryland (2000), 468 reported taking multivitamins $\text{\textgreater}= 4$ times per week in the month prior to pregnancy:

    • $\text{Prevalence} = \frac{468}{1,150} = 40.7\text{\%}$

• Point Prevalence vs. Period Prevalence:

  • Point Prevalence: Proportion of population with disease at a single specific calendar date or timepoint.

  • Period Prevalence: Proportion of population with disease during a defined calendar window (e.g., 12 months).

  • Calculations from a 20-Person Dynamic Cohort (10 individuals tracked over 15 months; 10 healthy controls omitted from graphic representation):

    • Point Prevalence on 04/01/1981: 7 individuals actively ill on April 1. 2 participants (ID #2 and #8) died prior to April 1 (removed from denominator at risk, leaving 18):

      • $\text{Point Prevalence} = \frac{7}{18} = 38.9\text{\%}$

    • Period Prevalence (October 1 to September 30): 10 total individuals experienced disease during the window, out of total population of 20:

      • $\text{Period Prevalence} = \frac{10}{20} = 50.0\text{\%}$

• Special Prevalence Measures (Misnamed Proportions):

  • Autopsy Rate: Proportion of post-mortem examinations displaying a specific finding divided by total autopsies performed.

  • Birth Defect Rate: Calculated as the number of infants with congenital abnormalities divided by total live births (or total live births plus stillbirths).

    • Epidemiological Caveat: Functionally a prevalence measure. True incidence cannot be calculated because the true denominator—total human conceptions at risk—is unobtainable due to unrecognized early pregnancy loss and unrecorded spontaneous abortions.

• Numerator Exercise Comparison (Cohort of 5 Individuals over 2017):

  • Incidence Numerator in 2017: 2 (3 individuals had disease onset prior to January 2017 and are excluded).

  • Point Prevalence Numerator in May 2017: 5 (all 5 individuals active with disease at that point).

  • Period Prevalence Numerator (July to November 2017): 4 (4 individuals active with disease during interval).

• Interrelationship Between Incidence and Prevalence:

  • In a steady state (where incidence rate and disease duration remain stable over time):

  • Formula: $P = I \times d$

    • $P$ = Prevalence

    • $I$ = Incidence Rate

    • $d$ = Average Duration of Disease

  • Analogy: Prevalence is a container holding disease cases. Incidence is the inflow pipe. Outflow occurs via recovery, death, or loss to follow-up.

  • Factors Increasing Prevalence: Increased incidence (via screening or more sensitive diagnostics), longer disease duration, extended survival without cure (improved supportive treatments), influx of cases.

    • Coronary Heart Disease (CHD): Prevalence rises due to earlier diagnosis and life-prolonging care, despite absence of a definitive cure.

    • Obesity: Prevalence rises due to increased new cases combined with prolonged duration.

  • Factors Decreasing Prevalence: Decreased incidence, shorter disease duration, high case-fatality rates, improved cure rates, diagnostic coding changes excluding minor cases.

Measures of Mortality

• Mortality Overview:

  • Mortality is a specialized application of incidence rate measuring death frequency in a defined population during a specific time period.

  • Reflects disease severity and guides resource prioritization, program design, and intervention monitoring.

  • Serves as an accurate surrogate marker for disease incidence when the case fatality rate is extremely high and duration of disease is short.

• Annual All-Cause Mortality Rate:

  • Formula: $\text{Annual All-Cause Mortality Rate} = \frac{\text{Total deaths from all causes in a calendar year}}{\text{Total midyear reference population}} \times 100,000$

  • Midyear population (July 1) is used to approximate dynamic population fluctuations.

• Age-Specific Mortality Rate:

  • Measures all-cause mortality within a restricted age bracket.

  • Formula: $\text{Age-Specific Mortality Rate} = \frac{\text{Total deaths among individuals in age group } X \text{ during year}}{\text{Midyear population in age group } X} \times 100,000$

• Cause-Specific Mortality Rate:

  • Measures deaths attributed to a specific primary disease.

  • Formula: $\text{Cause-Specific Mortality Rate} = \frac{\text{Total deaths from disease } Y \text{ during year}}{\text{Total midyear reference population}} \times 100,000$

• Case Fatality Rate (Percent):

  • Quantifies the intrinsic "killing power" or severity of a disease; measures the proportion of diagnosed cases that result in death.

  • Formula: $\text{Case Fatality Rate} = \frac{\text{Number of deaths from disease } Y}{\text{Total diagnosed cases of disease } Y} \times 100$

  • Calculation Example: Out of 20 diagnosed individuals, 18 die:

    • $\text{Case Fatality Rate} = \frac{18}{20} \times 100 = 90.0\text{\%}$

• Proportionate Mortality (Percent):

  • Measures the proportion of all community deaths caused by a specific condition (not a true rate).

  • Formula: $\text{Proportionate Mortality} = \frac{\text{Number of deaths from disease } Y}{\text{Total deaths from all causes}} \times 100$

• Standardized Mortality Rates (CDC 2002 Data Applications):

  • Contextual Note: CDC data table population numbers represent values $\times 1,000$ (e.g., listed value 288,357 represents $288,357,000$).

  • Unintentional Injury Specific Mortality Rate (Overall):

    • Deaths from unintentional injuries = 106,742; Total Midyear Population = 288,357,000.

    • $\text{Rate} = \frac{106,742}{288,357,000} \times 100,000 = 37.0\text{ deaths per } 100,000\text{ population}$

  • All-Cause Mortality Rate (Ages 25–34):

    • Deaths among 25–34 year olds = 41,355; Midyear population ages 25–34 = 39,928,000.

    • $\text{Rate} = \frac{41,355}{39,928,000} \times 100,000 = 103.6\text{ deaths per } 100,000\text{ population}$

  • All-Cause Mortality Rate (Males):

    • Male deaths = 1,199,264; Male midyear population = 141,656,000.

    • $\text{Rate} = \frac{1,199,264}{141,656,000} \times 100,000 = 846.6\text{ deaths per } 100,000\text{ males}$

  • Unintentional Injury Mortality Rate (Males Ages 25–34):

    • Unintentional injury deaths in group = 9,635; Male population ages 25–34 = 20,203,000.

    • $\text{Rate} = \frac{9,635}{20,203,000} \times 100,000 = 47.7\text{ deaths per } 100,000\text{ males aged 25--34}$

  • Proportionate Mortality for Unintentional Injuries:

    • Unintentional injury deaths = 106,742; Total deaths = 2,443,387 (implied total).

    • $\text{Proportionate Mortality} = \frac{106,742}{2,443,387} \times 100 = 4.4\text{\%}$

Years of Potential Life Loss (YPLL)

• Definition and Application:

  • An epidemiological metric estimating average years of life lost when an individual dies prematurely relative to a standard life expectancy benchmark (commonly 65 years).

  • Quantifies societal, economic, and human loss resulting from early death.

  • Prioritizes public health resources toward conditions causing early mortality.

• Individual Calculation Protocol:

  • $\text{YPLL} = \text{Standard Target Age (65)} - \text{Age at Death}$

  • If an individual dies at or after the target age, their contributed YPLL is zero.

  • Individual 1 (Death at 6 months / 0.5 years):

    • $\text{YPLL} = 65 - 0.5 = 64.5\text{ years}$

  • Individual 4 (Death at 85 years):

    • $\text{YPLL} = 0\text{ years}$

  • Cohort Aggregate Example: Across a sample tracking set, total summed YPLL = 129.5 years.

• Public Health Utilization (MMWR Data):

  • Among youth under 20 years of age, Motor Vehicle Accidents produce a dramatically higher total YPLL than falls, burns, or poisonings combined, demonstrating that injury prevention funds must target motor vehicle safety over other unintentional domestic injuries.

Rate Standardization and Adjustment Methods

• Necessity for Adjustment:

  • Directly comparing crude mortality rates between distinct populations can yield misleading conclusions due to underlying structural demographic differences (e.g., age distribution).

  • Illustrative Scenario: Comparing cancer mortality in a high school vs. a nursing home. Cancer incidence concentrates in older demographics; high school populations exhibit far lower crude cancer mortality strictly due to youthful age structure.

• Direct Standardization:

  • Indication: Used when age-specific mortality rates of all populations under study are explicitly known.

  • Methodology: Applies a single standard external population distribution (e.g., Census reference population) to the age-specific rates of each comparison group to compute "expected deaths."

  • Direct Adjustment Calculation Example (US vs. Mexico):

    • Crude Mortality Data: US = $8.7\text{ per } 1,000$; Mexico = $4.7\text{ per } 1,000$.

    • Step 1: Multiply age-specific mortality of each stratum by the standard census population size ($100,000$ total) for each country.

    • Step 2: Expected total deaths derived: Mexico = 639 deaths; US = 574 deaths.

    • Step 3: Divide expected deaths by standard population ($100,000$) and apply multiplier ($1,000$):

      • $\text{US Age-Adjusted Rate} = \frac{574}{100,000} \times 1,000 = 5.74\text{ per } 1,000$

      • $\text{Mexico Age-Adjusted Rate} = \frac{639}{100,000} \times 1,000 = 6.39\text{ per } 1,000$

    • Conclusion: Crude rates suggested US mortality was double Mexico's rate. Age adjustment reveals Mexico's age-adjusted mortality rate is actually higher ($6.39$ vs. $5.74\text{ per } 1,000$).

• Indirect Standardization and Standardized Mortality Ratio (SMR):

  • Indication: Used when age-specific mortality rates of the study population are unknown, missing, or unstable (common in occupational hazard cohorts or environmental exposure disasters).

  • Methodology: Applies general population age-specific rates to the known age structure of the study cohort to project expected deaths.

  • Formula: $\text{SMR} = \frac{\text{Observed Deaths}}{\text{Expected Deaths}} \times 100$

  • Calculation Example (Coal Miners Cohort):

    • Sum of expected deaths (Miners population strata $\times$ general population rates) = 181.09.

    • Observed deaths in miners cohort = 436.

    • $\text{SMR} = \frac{436}{181.09} \times 100 = 241$

  • Interpretation: An SMR $> 100$ indicates that the study group experiences greater mortality than expected relative to the general population. An SMR of 241 signifies miners have a 141% excess death rate linked to occupational hazards.

Case Ascertainment, Diagnostic Criteria, and Data Sources

• Epidemiological Data Sources:

  • Population Surveys: National Health Interview Survey (NHIS), Health Examination Survey (HES).

  • Disease Registries: Surveillance, Epidemiology, and End Results (SEER) program (population-based cancer tracking).

  • Administrative Data: Hospital records, medical insurance claims, death certificates.

• Mandatory Reporting Laws and Surveillance Systems:

  • Legally Reportable Diseases: Tuberculosis (TB) requires mandatory healthcare provider notification to local public health departments/boards of health within 24 hours.

  • Sexually Transmitted Infections (STIs): Chlamydia, Gonorrhea, and Syphilis are reportable by law; Herpes Simplex Virus (HSV) is not reportable.

  • Unreported Conditions: Common cold is completely unmonitored by public health notification systems.

• Healthcare Utilization Biases in Data Records:

  • Help-Seeking Behavior: Conditions causing severe acute distress (e.g., Asthma) lead patients to hospital care, securing accurate registry entries. Conditions managed with over-the-counter (OTC) medications (e.g., Seasonal Allergies) go unrecorded.

  • Stigma & Non-Disclosure: Mental health conditions (e.g., Depression) suffer from under-reporting in population interviews.

  • Hospital Record Limitations: Post-partum depression recorded in hospital datasets captures only individuals actively seeking medical care, omitting individuals who wait out symptoms.

• Diagnostic Definition Changes and Systemic Shifts:

  • Broadening diagnostic definitions increases reported case numbers; tightening definitions decreases reported case numbers.

  • Historical Example 1 (ICD Coding for Diabetes, 1949):

    • Pre-1949: Any death certificate listing diabetes anywhere on the certificate (including contributing causes) was classified as a diabetes death.

    • Post-1949: Policy shifted to code diabetes deaths only when listed as the primary underlying cause.

    • Impact: An immediate, sharp decline in reported national diabetes mortality occurred, reflecting coding methodology revision rather than clinical improvement.

  • Historical Example 2 (AIDS Case Definition Expansion, 1993):

    • In 1993, expanding the surveillance case definition produced a massive artificial spike in reported annual AIDS cases.

• Natural History and Disease Subtypes:

  • Undiagnosed asymptomatic periods (e.g., Type 2 Diabetes, early-stage cancers) lead to systematic underestimation of true prevalence.

  • Subtypes exhibit variable presentation (e.g., Multiple Sclerosis presents in four distinct patterns: Relapsing-Remitting, Secondary Progressive, Primary Progressive, and Progressive Relapsing).

Errors and Biases in Epidemiological Data Collection

• Subject-Based Response Errors:

  • Asymptomatic Status: Individuals unaware of disease report healthy status falsely.

  • Subclinical Care: Patients undergoing pre-disease treatment (e.g., pre-diabetes lifestyle modification) reporting lack of formal diagnosis.

  • Recall Bias: Differential reporting inaccuracies caused by subjects failing to remember past disease events or exposures.

  • Intentional Withholding: Deliberate suppression of clinical status due to social stigma or privacy concerns.

• Interviewer-Based Errors:

  • Data Entry Errors: Misrecording subject responses on survey instruments.

  • Question Formulation: Asking flawed or invalid questions failing to measure targeted phenomena.

  • Information Bias: Interviewers probing cases deeper than controls because disease status is known.

  • Selection Bias: Non-comparability between eligible study participants and non-responding eligible non-participants.

• Limitations of Hospital Datasets:

  • The baseline population at risk is ill-defined, making selection of appropriate comparison/control cohorts difficult in observational research.