Comprehensive Study Guide to Epidemiologic Measures of Morbidity and Mortality

Foundations of Morbidity and Mortality Measurement

Epidemiology fundamentally relies on the precise, objective measurement of health outcomes within populations.
To accurately measure any health outcome, four core parameters must be established:
- Outcome Definition: Precise diagnostic criteria establishing what constitutes a case.
- Timing of Occurrence: Determination of whether and exactly when the outcome occurred, accounting for biological mechanisms:
  - Induction Period: The interval between causal exposure and disease initiation.
  - Latent Period: The time interval between disease initiation and its clinical detection or diagnosis.
- Time at Risk: The exact quantification of the duration during which individuals in the cohort are susceptible to the disease.
- Population Specification: Delineation of the target group in which the disease occurs (e.g., at-risk, closed, open, or steady-state populations).

Mathematical Structures in Disease Frequency

Count:
- Represents the absolute frequency of disease occurrence.
- Defined as the simple tally of individuals with a confirmed diagnosis or event.
- Example: The absolute number of college dormitory residents diagnosed with hepatitis.
Ratio:
- A mathematical fraction where the numerator is completely distinct and mutually exclusive from the denominator.
- Expressed mathematically as: $\text{Ratio} = \frac{X}{Y}$
- Example: The ratio of male cases to female cases: $\frac{\text{Number of Male Cases}}{\text{Number of Female Cases}}$
Proportion:
- A fraction where the numerator is a subset of, and included within, the denominator.
- Expressed mathematically as: $\text{Proportion} = \frac{a}{a + b}$
- Has no temporal units and is bounded between $0$ and $1$ (or $0\%$ and $100\%$ ).
- Example: The proportion of cases occurring among males: $\frac{\text{Number of Male Cases}}{\text{Total Cases}}$
Rate:
- Quantifies the speed at which a health status change occurs per unit of time.
- Expressed mathematically as: $\text{Rate} = \frac{\text{Number of Disease Events}}{\text{Period of Time}}$
- Example: The crude mortality rate expressed as: $\frac{\text{Number of Deaths in a Given Year}}{\text{Reference Population at Mid-Year}}$

Classifications of Study Populations

Closed Population (Cohort):
- A cohort is defined as a group of individuals followed over a period of time who share a common characteristic or exposure.
- Defined strictly at the inception of follow-up ( $t_0$ ), with no entry of new members permitted after the study begins.
- Outflow from the population occurs exclusively through death or the onset of the specific disease being investigated.
- Loss to follow-up (censoring) is accounted for using quantitative methods such as the life-table adjustment technique.

Closed cohort population size declining over time

Open Population:
- A dynamic group that freely gains new members over time (inflow) or loses existing members who remain alive (outflow via migration, loss of eligibility).
- Average age and overall population size may increase, decrease, or remain constant over time.
Steady-State (Stationary) Population:
- A specialized subtype of an open population in which member inflow exactly equals outflow.
- Maintains a stable, constant population size and constant demographic age distribution over time.

Core Measures of Morbidity: Incidence and Prevalence

Incidence:
- Quantifies the frequency of development of new cases of disease in a specified population over a designated period.
- Serves as the primary epidemiological metric for identifying causal etiologies and establishing exposure-outcome associations.
- Represents a direct measure of disease risk (the likelihood of an individual developing the disease over time).
Prevalence:
- Quantifies the total proportion of existing cases (encompassing both new and long-standing historical cases) in a population at a single point or interval in time.
- Serves as a measure of total disease burden; indispensable for healthcare service planning, resource allocation, and identifying potential etiologic associations.
- Does not measure disease risk; prevalence is a composite function influenced by both new disease development and the clinical duration of existing cases.

Detailed Measures of Incidence

Incidence Rate:
- Defined as the number of new disease events occurring in a defined population at risk during a specified observation window.
- Numerator: Number of newly diagnosed cases (individuals with a prior history of the disease are excluded).
- Denominator: Population at risk (individuals with existing disease or who are biologically incapable of contracting the disease are excluded).
- Formula: $\text{Incidence Rate} = \frac{\text{Number of New Cases of Disease During Given Period}}{\text{Population at Risk During That Period}} \times \text{Multiplier}$
- Worked Example: If $500$ new cases arise within a population at risk of $30{,}000$ over one year: $\text{Incidence Rate} = \frac{500}{30{,}000} \times 1{,}000 = 16.7\,\text{per 1,000 per year}$
Cumulative Incidence (Incidence Proportion):
- Defined as the proportion of at-risk individuals who develop disease over a specified time horizon.
- Always evaluated over a fixed follow-up window, bounded theoretically between $0$ and $1$ .
- Predicts individual average risk of developing disease in a closed cohort.
- Formula: $\text{Cumulative Incidence} = \frac{\text{Number of New Cases in a Given Time Period}}{\text{Number at Risk at the Start of the Time Period}}$

Closed cohort individual follow-up lines over time for cumulative incidence

* *Worked Example from Cohort Diagram*:
    * Baseline enrollment:  $7$  individuals (A, B, C, D, E, F, G) at risk at time  $t_0$ .
    * Follow-up through time  $t_1$  reveals  $2$  individuals (C and E) develop disease.
    * Calculation:

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

Incidence Density (Person-Time Rate):
- Used when study subjects are followed for differing lengths of time (variable entry and exit).
- Replaces the static population count with total person-time at risk in the denominator.
- Person-time represents the sum of individual observation units contributed by each subject while disease-free (e.g., person-years, person-months, or person-days).
- Formula: $\text{Incidence Density} = \frac{\text{Number of New Cases During a Given Period}}{\text{Total Person-Time at Risk}} \times \text{Multiplier}$

Individual observation lines and person-time accrual

* *Worked Example 1 (Cohort Study with Dynamic Censoring)*:
    * Study tracking 12 subjects (A through L) between years 1981 and 1995:
        * Subject A:  $8.3\,\text{years}$ 
        * Subject B:  $14.0\,\text{years}$  (disease event at 1992)
        * Subject C:  $14.0\,\text{years}$ 
        * Subject D:  $10.2\,\text{years}$ 
        * Subject E:  $3.0\,\text{years}$ 
        * Subject F:  $12.0\,\text{years}$  (disease event at 1986)
        * Subject G:  $7.0\,\text{years}$ 
        * Subject H:  $10.0\,\text{years}$ 
        * Subject I:  $3.0\,\text{years}$ 
        * Subject J:  $9.0\,\text{years}$ 
        * Subject K:  $6.2\,\text{years}$ 
        * Subject L: disease event at 1992
    * Total accumulated person-time =  $107.7\,\text{person-years}$ ; Total new cases =  $3$ .
    * Calculation:

$\text{Incidence Density} = \frac{3}{107.7} \approx 0.028$ $\text{Incidence Density} = 0.028 \times 1{,}000 = 28\,\text{per 1,000 person-years}$

* *Worked Example 2 (HIV/AIDS Population Comparative Scenario)*:
    * Scenario parameters:  $5$  new HIV/AIDS cases reported; total existing HIV/AIDS cases =  $56$ ; total population =  $100{,}000$ ; population at risk =  $20{,}000$ .
    * Observation adjustment: The actual average observation time at risk per person is  $183\,\text{days per year}$  ( $0.5\,\text{years}$  per individual).
    * Standard Incidence Rate:

$\text{Incidence Rate} = \frac{5}{20{,}000} = 0.00025$ $\text{Incidence Rate} = 0.00025 \times 10{,}000 = 2.5\,\text{per 10,000 population}$ * Person-Time Calculation: $\text{Total Person-Time} = 20{,}000 \times 0.5 = 10{,}000\,\text{person-years}$ $\text{Incidence Density} = \frac{5}{10{,}000} = 0.0005$ $\text{Incidence Density} = 0.0005 \times 10{,}000 = 5\,\text{per 10,000 person-years}$

Attack Rate:
- A variant of cumulative incidence utilized during acute disease outbreaks or epidemics characterized by short incubation windows and defined etiologic exposures (e.g., foodborne outbreaks, chemical exposures, infectious clusters).
- Calculated over the entire duration of the epidemic, spanning the time from index to terminal case.
- Formula: $\text{Attack Rate} = \frac{\text{Ill}}{\text{Ill} + \text{Well}} \times 100$
- Worked Example: A cholera outbreak in Country X during April 2012 yields $490$ new cases within an at-risk population of $18{,}600$ : $\text{Attack Rate} = \frac{490}{18{,}600} \times 100 \approx 2.6\%$

Detailed Measures of Prevalence

General Prevalence Formula: $\text{Prevalence} = \frac{\text{Number of Existing Cases of Disease During Given Period}}{\text{Total Population During That Period}} \times \text{Multiplier}$
Survey Worked Example: In a Maryland study (year 2000) surveying $1{,}150$ women following childbirth, $468$ reported multivitamin use at least $4$ times weekly in the month prior to conception: $\text{Prevalence} = \frac{468}{1{,}150} \times 100 \approx 40.7\%$
Point Prevalence:
- The proportion of a population with a disease condition at an exact single point in calendar time.
- Formula: $\text{Point Prevalence} = \frac{\text{Number of Existing Cases at a Specified Point in Time}}{\text{Total Population at That Specific Point in Time}} \times \text{Multiplier}$
Period Prevalence:
- The proportion of a population possessing the disease condition at any time throughout a defined window or interval.
- Formula: $\text{Period Prevalence} = \frac{\text{Number of Existing Cases During a Specified Time Period}}{\text{Total Population During Specified Period}} \times \text{Multiplier}$

Timeline tracking 10 subjects through illness onset, death, and recovery

* *Worked Example from 10-Subject Cohort Diagram (Total Base Population = 20)*:
    * **Point Prevalence on April 1, 1981**:
        * On April 1, 1981, subjects 1, 4, 5, 6, 7, 9, and 10 are actively ill ( $7$  active cases).
        * Two individuals (subjects 2 and 8) died prior to April 1, 1981, leaving an active population of  $18$ .
        * Calculation:

$\text{Point Prevalence} = \frac{7}{18} \times 100 \approx 38.9\%$ * Period Prevalence (October 1, 1980 to September 30, 1981): * A total of $10$ unique individuals were ill at some point during the interval (subjects 1 through 10). * Initial baseline population was $20$ . * Calculation: $\text{Period Prevalence} = \frac{10}{20} \times 100 = 50.0\%$

Timeline diagram of disease episodes over 2017

* *Worked Example from Calendar Year 2017 Diagram*:
    * **Incidence Numerator for 2017**:  $2$  (identifying only subjects whose disease onset initiated within the 2017 calendar window).
    * **Point Prevalence Numerator for May 2017**:  $5$  (identifying all active cases ongoing in May 2017).
    * **Period Prevalence Numerator from July to November 2017**:  $4$  (identifying any case active at any point between July and November).

Special Prevalence Subtypes:
- Prevalence Disease Rate at Autopsy:
  - Necessary for disorders that cannot be reliably diagnosed premortem (e.g., definitive confirmation of Alzheimer's disease pathology).
  - Formula: $\text{Autopsy Disease Rate} = \frac{\text{Number of Cases of Disease}}{\text{Number of Persons Autopsied}} \times \text{Multiplier}$
- Prevalence Birth Defect Rate:
  - True incidence cannot be estimated because the total denominator of conceptions is unknown due to spontaneous pregnancy loss and unrecognized fetal death.
  - Formula: $\text{Birth Defect Rate} = \frac{\text{Number of Babies with a Given Abnormality}}{\text{Number of Live Births (Total Births)}} \times \text{Multiplier}$

Dynamic Equilibrium: The Relationship Between Incidence and Prevalence

Dynamic tub model showing inflow of incidence and outflow of deaths and cures affecting prevalence

Theoretical Steady-State Relationship:
- Under steady-state assumptions where population parameters remain stationary over time: $\text{Prevalence} \approx \text{Incidence} \times \text{Duration of Disease}$
Factors Causing a Decrease in Disease Prevalence:
- Decreased incidence (fewer individuals develop the condition).
- High case-fatality rates (rapid mortality removes prevalent cases from the pool).
- High cure rates (effective therapies resolve disease).
- Introduction of diagnostic criteria that are less sensitive (contracting the case definition).
- Changes in International Classification of Diseases (ICD) coding rules that exclude previous disease designations.
Factors Causing an Increase in Disease Prevalence:
- Increased incidence (accelerated inflow of new cases).
- Improved detection sensitivity and widespread diagnostic screening programs (e.g., PSA testing, mammography).
- Decreased mortality rates without cure (medical interventions that prolong patient survival, lengthening disease duration).
Clinical Case Examples:
- Coronary Heart Disease (CHD): Overall prevalence increased despite declining age-specific incidence, driven by early detection and therapeutic advances that reduce mortality, thereby extending patient survival.
- Obesity: Prevalence continues to surge due to the combination of rising incidence and a long, non-fatal chronic disease duration.

Comparative Matrix of Morbidity Measures

Point Prevalence:
- Numerator: Existing cases (both old and new).
- Denominator: Population at that specific point in time.
- Time Dimension: Single point in time.
- Units: None (dimensionless proportion).
- Mathematical Type: Proportion.
- Survey / Assessment Question: "Do you currently have asthma?"
- Primary Study Design: Cross-sectional studies.
Period Prevalence:
- Numerator: Existing cases during the defined interval.
- Denominator: Mid-point population of the interval.
- Time Dimension: Defined period of calendar time.
- Units: None (dimensionless proportion).
- Mathematical Type: Proportion.
- Survey / Assessment Question: "Have you had asthma during the last year?"
- Primary Study Design: Public health surveillance.
Cumulative Incidence:
- Numerator: New cases developing during the interval.
- Denominator: Initial population at risk at baseline ( $t_0$ ).
- Time Dimension: Defined period of time.
- Units: None (dimensionless proportion).
- Mathematical Type: Proportion (estimate of risk).
- Survey / Assessment Question: "Have you developed asthma over the last year?"
- Primary Study Design: Closed cohort studies.
Incidence Density:
- Numerator: New cases developing during observation.
- Denominator: Total person-time accumulated by susceptible individuals.
- Time Dimension: Continuous duration of observation.
- Units: $\text{Time}^{-1}$ (e.g., person-years, person-months).
- Mathematical Type: True rate.
- Survey / Assessment Question: Evaluated continuously via longitudinal medical records.
- Primary Study Design: Open cohort studies.

Foundations of Mortality Measurement

Utility of Mortality Metrics:
- Measures the frequency of death occurrences within specified populations across defined intervals of time and geography.
- Provides an objective index of clinical disease severity.
- Guides public health resource allocation, interventions, and surveillance.
Mortality as a Proxy for Disease Incidence:
- Mortality rates accurately approximate disease incidence when two clinical criteria are met:
  - The case-fatality rate is high (a high proportion of diagnosed patients die from the illness).
  - The duration of survival following disease onset is short.

Core Formulas for Mortality Measures

Annual Mortality Rate (Crude Death Rate): $\text{Annual Mortality Rate} = \frac{\text{Number of Deaths from All Causes in 1 Year}}{\text{Total Persons in Population at Midyear}} \times \text{Multiplier}$
Age-Specific Mortality Rate: $\text{Age-Specific Mortality Rate} = \frac{\text{Number of Deaths from All Causes in 1 Year in Specified Age Group}}{\text{Total Persons in Specified Age Group at Midyear}} \times \text{Multiplier}$
Cause-Specific Mortality Rate: $\text{Cause-Specific Mortality Rate} = \frac{\text{Number of Deaths from a Specific Cause in 1 Year}}{\text{Total Persons in Population at Midyear}} \times \text{Multiplier}$
Sex-Specific Mortality Rate: $\text{Sex-Specific Mortality Rate} = \frac{\text{Number of Deaths from All Causes in 1 Year for Specified Sex}}{\text{Total Persons of Specified Sex at Midyear}} \times \text{Multiplier}$
Cause-, Age-, and Sex-Specific Mortality Rate: $\text{Specific Mortality Rate} = \frac{\text{Number of Deaths from Specific Cause in 1 Year Among Specified Sex and Age}}{\text{Total Persons in Specified Sex and Age Range at Midyear}} \times \text{Multiplier}$
Case-Fatality Rate (CFR): $\text{Case-Fatality Rate} = \frac{\text{Number of Deaths from Specific Disease During a Defined Time Period}}{\text{Total Persons with Diagnosed Disease}} \times 100$
Proportionate Mortality: $\text{Proportionate Mortality} = \frac{\text{Number of Deaths from a Specific Disease in a Given Year}}{\text{Total Deaths Due to All Causes in That Year}} \times 100$

Worked Mortality Calculations: United States Vital Statistics (2002 Data)

Baseline Vital Statistics (CDC, 2012):
- Total US Population (Both Sexes): $288{,}357{,}000$
- Total Deaths (All Causes, Both Sexes): $2{,}443{,}387$
- Total Unintentional Injury Deaths (Both Sexes): $106{,}742$
- Total Male Population: $141{,}656{,}000$
- Total Male Deaths (All Causes): $1{,}199{,}264$
- Total Male Unintentional Injury Deaths: $69{,}257$
- Population Aged 25–34 (Both Sexes): $39{,}928{,}000$
- Deaths Aged 25–34 (All Causes, Both Sexes): $41{,}355$
- Unintentional Injury Deaths Aged 25–34 (Both Sexes): $12{,}569$
- Male Population Aged 25–34: $20{,}203{,}000$
- Male Deaths Aged 25–34 (All Causes): $28{,}736$
- Male Unintentional Injury Deaths Aged 25–34: $9{,}635$
Calculation 1: Unintentional Injury Cause-Specific Mortality Rate (US, 2002): $\text{Rate} = \frac{106{,}742}{288{,}357{,}000} \times 100{,}000 = 37.0\,\text{per 100,000 population}$
Calculation 2: Age-Specific All-Cause Mortality Rate (25–34 Year Olds): $\text{Rate} = \frac{41{,}355}{39{,}928{,}000} \times 100{,}000 = 103.6\,\text{deaths per 100,000 25–34 year olds}$
Calculation 3: Sex-Specific All-Cause Mortality Rate (Males): $\text{Rate} = \frac{1{,}199{,}264}{141{,}656{,}000} \times 100{,}000 = 846.6\,\text{deaths per 100,000 males}$
Calculation 4: Cause-, Age-, and Sex-Specific Mortality Rate (Unintentional Injuries, 25–34 Year Old Males): $\text{Rate} = \frac{9{,}635}{20{,}203{,}000} \times 100{,}000 = 47.7\,\text{per 100,000 25–34 year old males}$
Calculation 5: Proportionate Mortality for Unintentional Injuries (US, 2002): $\text{Proportionate Mortality} = \frac{106{,}742}{2{,}443{,}387} \times 100 \approx 4.4\%$ (Meaning: $4.4\%$ of all deaths occurring in the United States in 2002 were attributable to unintentional injuries.)

Years of Potential Life Lost (YPLL)

Definition and Function:
- An index quantifying premature, preventable mortality by measuring lost lifespan before reaching an established threshold age (typically age 65).
- Shifts public health prioritization toward conditions causing early death in younger age cohorts.
Calculation Procedure:
- For each individual dying before age 65, subtract their exact age at death from 65.
- Individuals surviving to or past age 65 contribute $0\,\text{YPLL}$ .
- Sum individual values to determine the aggregate loss: $\text{YPLL} = \sum (65 - \text{Age at Death})$
Worked Example Table:
- Individual 1 (died at 6 months / 0.5 years): $65 - 0.5 = 64.5\,\text{YPLL}$
- Individual 2 (died at 55 years): $65 - 55 = 10.0\,\text{YPLL}$
- Individual 3 (died at 15 years): $65 - 15 = 50.0\,\text{YPLL}$
- Individual 4 (died at 85 years): contributes $0.0\,\text{YPLL}$
- Individual 5 (died at 60 years): $65 - 60 = 5.0\,\text{YPLL}$
- Aggregate impact: $\text{Total YPLL} = 64.5 + 10.0 + 50.0 + 0.0 + 5.0 = 129.5\,\text{years}$

YPLL per year per 100,000 for unintentional injuries among individuals under 20 years old

Public Health Application for Individuals Under 20 Years of Age:
- Motor vehicle traffic-related accidents represent the single largest contributor to premature years lost in individuals under age 20 (approaching $500\,\text{YPLL per 100,000}$ ), substantially exceeding suffocation, drowning, poisoning, falls, and fire/burns.

Methods of Mortality Rate Standardization and Adjustment

The Necessity of Rate Adjustment:
- Comparing crude mortality rates directly between two distinct populations can be misleading due to confounding by demographic variables such as age, sex, and race.
- Cancer incidence, for instance, increases sharply with age. A nursing home cohort will naturally display a much higher crude cancer mortality rate than a high school population solely due to its age structure.
- Overall population mortality is the mathematical product of the age-specific risk of death and the relative demographic weight of each age stratum in that population.
Direct Adjustment:
- Evaluates the question: "If the age composition of the two populations were identical, would there be differences in overall mortality between them?"
- Commonly employed for official vital statistics reporting and disease incidence surveillance.
- Methodological Requirement: Requires known age-specific mortality rates for both target populations, which are then applied to a chosen Standard Population (e.g., standard census reference of $100{,}000$ individuals).
- Mathematical procedure:
  1. Compute expected deaths for each age stratum: $\text{Expected Deaths} = \text{Age-Specific Mortality Rate} \times \text{Standard Population Strata Size}$
  2. Sum expected deaths across all strata.
  3. Compute directly adjusted rate: $\text{Adjusted Rate} = \frac{\sum \text{Expected Deaths}}{\text{Total Standard Population}} \times \text{Multiplier}$
Worked Example: Direct Standardization of Mexico vs. United States:
- Reference Standard Population = $100{,}000$
- Crude Mortality Rates: Mexico = $4.7\,\text{per 1,000}$ ; United States = $8.7\,\text{per 1,000}$
- Stratified Data and Expected Deaths:
  - Age 0–1 (Standard: $2{,}400$ ):
    - Mexico rate: $1{,}693.2\,\text{per 100,000}$ $\rightarrow$ Expected: $41$
    - US rate: $737.8\,\text{per 100,000}$ $\rightarrow$ Expected: $18$
  - Age 1–4 (Standard: $9{,}600$ ):
    - Mexico rate: $112.5\,\text{per 100,000}$ $\rightarrow$ Expected: $11$
    - US rate: $38.5\,\text{per 100,000}$ $\rightarrow$ Expected: $4$
  - Age 5–14 (Standard: $19{,}000$ ):
    - Mexico rate: $36.2\,\text{per 100,000}$ $\rightarrow$ Expected: $7$
    - US rate: $21.7\,\text{per 100,000}$ $\rightarrow$ Expected: $4$
  - Age 15–24 (Standard: $17{,}000$ ):
    - Mexico rate: $102.9\,\text{per 100,000}$ $\rightarrow$ Expected: $17$
    - US rate: $90.3\,\text{per 100,000}$ $\rightarrow$ Expected: $15$
  - Age 25–44 (Standard: $26{,}000$ ):
    - Mexico rate: $209.6\,\text{per 100,000}$ $\rightarrow$ Expected: $55$
    - US rate: $176.4\,\text{per 100,000}$ $\rightarrow$ Expected: $46$
  - Age 45–64 (Standard: $19{,}000$ ):
    - Mexico rate: $841.1\,\text{per 100,000}$ $\rightarrow$ Expected: $160$
    - US rate: $702.3\,\text{per 100,000}$ $\rightarrow$ Expected: $133$
  - Age 65+ (Standard: $7{,}000$ ):
    - Mexico rate: $4{,}967.4\,\text{per 100,000}$ $\rightarrow$ Expected: $348$
    - US rate: $5{,}062.6\,\text{per 100,000}$ $\rightarrow$ Expected: $354$
- Sum of Expected Deaths:
  - Mexico = $41 + 11 + 7 + 17 + 55 + 160 + 348 = 639$
  - United States = $18 + 4 + 4 + 15 + 46 + 133 + 354 = 574$
- Adjusted Mortality Rate Calculations:
  - Mexico: $\text{Rate} = \frac{639}{100{,}000} \times 1{,}000 = 6.39\,\text{per 1,000}$
  - United States: $\text{Rate} = \frac{574}{100{,}000} \times 1{,}000 = 5.74\,\text{per 1,000}$
- Epidemiological Insight: While the US had a higher crude death rate due to its older population ( $8.7$ vs. $4.7\,\text{per 1,000}$ ), age adjustment demonstrates that mortality risk was higher in Mexico ( $6.39$ vs. $5.74\,\text{per 1,000}$ ).
Indirect Adjustment and the Standardized Mortality Ratio (SMR):
- Evaluates the question: "Do individuals in a specialized cohort (e.g., an occupational industry) have higher mortality than individuals of the same age in the general reference population?"
- Utilized when age-specific mortality rates are unavailable or unstable due to small case numbers in the index cohort.
- Formula: $\text{SMR} = \frac{\text{Observed Number of Deaths}}{\text{Expected Number of Deaths}} \times 100$
- Interpretation:
  - $\text{SMR} = 100$ : Observed mortality equals expected mortality.
  - $\text{SMR} > 100$ : Observed mortality exceeds expected mortality.
  - $\text{SMR} < 100$ : Observed mortality is less than expected mortality.
Worked Example: Occupational Mortality in Underground Miners:
- Total Miner Population = $534{,}533$ ; Total Observed Deaths = $436$ .
- Age Stratum Calculations using General Population Death Rates:
  - Age 20–24: Population = $74{,}598$ ; General Rate = $12.26\,\text{per 100,000}$ $\rightarrow$ Expected = $9.14$
  - Age 25–29: Population = $85{,}077$ ; General Rate = $16.12\,\text{per 100,000}$ $\rightarrow$ Expected = $13.71$
  - Age 30–34: Population = $80{,}845$ ; General Rate = $21.54\,\text{per 100,000}$ $\rightarrow$ Expected = $17.41$
  - Age 35–44: Population = $148{,}870$ ; General Rate = $33.96\,\text{per 100,000}$ $\rightarrow$ Expected = $50.55$
  - Age 45–54: Population = $102{,}649$ ; General Rate = $56.82\,\text{per 100,000}$ $\rightarrow$ Expected = $58.32$
  - Age 55–59: Population = $42{,}494$ ; General Rate = $75.23\,\text{per 100,000}$ $\rightarrow$ Expected = $31.96$
- Total Expected Deaths: $\sum \text{Expected Deaths} = 9.14 + 13.71 + 17.41 + 50.55 + 58.32 + 31.96 = 181.09$
- Standardized Mortality Ratio Calculation: $\text{SMR} = \frac{436}{181.09} \times 100 \approx 241$
- Epidemiological Conclusion: Underground miners experience $2.41$ times (or $241\%$ of) the mortality expected from the general population.

Data Sources, Quality, and Surveillance Artifacts

Primary Data Sources for Morbidity and Mortality:
- Population interview surveys.
- Morbidity surveys.
- Hospital and clinical discharge records.
- Disease registries (e.g., SEER cancer registries).
- Life and health insurance company databases.
Numerator and Case Ascertainment Challenges:
- Case Definition: Shifts in diagnostic criteria over time alter case counts without changes in underlying biological incidence.
- Case Ascertainment: Dependent on reporting compliance (e.g., complete tuberculosis reporting vs. underreported common cold cases).
- Diagnostic Access: Stigmatized conditions or illnesses in underserved areas frequently go undiagnosed and unreported.

Standard death certificate Cause of Death Part I and Part II format

Death Certificate Structure and Coding:
- Deaths are systematically coded based on the Underlying Cause of Death.
- Part I: Delineates the causal chain of events leading to death:
  - Line a: Immediate cause (final disease/condition resulting in death, e.g., Rupture of myocardium, onset minutes).
  - Line b: Intermediate condition (due to, e.g., Acute myocardial infarction, onset 6 days).
  - Line c: Underlying cause (initiating disease/injury leading to the chain, e.g., Chronic ischemic heart disease, onset 5 years).
- Part II: Documents significant contributing comorbid conditions that did not directly initiate the underlying cause (e.g., Diabetes, Chronic obstructive pulmonary disease, smoking history).
- Cancer diagnoses on death certificates demonstrate higher diagnostic validity than chronic cardiovascular or respiratory conditions.

Diabetes mortality drop between 1948 and 1950 due to sixth revision of ICD coding

Impact of ICD Coding Transitions:
- Revisions to ICD coding guidelines can create artificial shifts in vital statistics.
- The transition to the Sixth Revision of ICD in 1949–1950 caused an artificial drop in reported diabetes mortality in the United States, as diabetes was no longer prioritized as the underlying cause when listed alongside cardiovascular disease.

Surveillance curve of AIDS cases in the US showing the 1993 spike due to case definition expansion

Surveillance Case Definition Expansion (1993 AIDS Spike):
- In the first quarter of 1993, the CDC expanded the surveillance case definition of AIDS to include CD4+ T-lymphocyte counts under $200\,\mu\text{L}$ or a CD4+ percentage under $14\%$ , as well as pulmonary tuberculosis, recurrent pneumonia, and invasive cervical cancer in HIV-infected persons.
- This triggered an artificial spike to over $100{,}000$ reported cases in 1993, followed by a decline as prevalent cases were absorbed into the reporting system.

Systematic Biases and Methodological Limitations

Natural History and Clinical Heterogeneity:
- Clinical Spectrum: Diseases present across a range of stages (e.g., early vs. advanced diabetes).
- Disease Subtypes: Pathologies encompass distinct sub-classifications with differing prognoses (e.g., progressive vs. relapsing-remitting multiple sclerosis).
- Timing of Detection: Identification may occur at subclinical, clinical, recurrent, or relapsing stages (e.g., major depressive disorder).
Systematic Errors in Interview Data:
- Subject-Level Biases:
  - Asymptomatic infection/disease.
  - Symptomatic disease without medical evaluation.
  - Medical care received without formal diagnosis.
  - Recall bias (differential memory between affected and unaffected individuals).
  - Non-disclosure or intentional reporting suppression.
- Interviewer-Level Biases:
  - Incorrect recording of participant responses.
  - Inconsistent question phrasing or administration.
  - Differential probing directed toward suspected exposure groups.
  - Selection bias related to sampling strategy (e.g., reliance on landline telephones).
Limitations Inherent to Hospital Data:
- Hospital admissions are selective based on disease severity, socioeconomic status, and institutional admission policies.
- Medical records are designed for patient care rather than research, leading to incomplete or variable diagnostic data.
- The true catchment population at risk (the denominator) is rarely definable.