B5

The Epidemiologic Study Cycle

The epidemiologic study cycle is a four-step process used to investigate health patterns and test hypotheses within populations.

• Descriptive Studies to Aggregate and Analyze Data: Descriptive studies are used to identify groups with high and low rates of a specific health pattern. This analysis can occur within a single population or between different populations.
• Model Building and Formulation of Hypothesis: Hypotheses are generated based on the results obtained from the descriptive study stage.
• Analytic Studies to Test Hypotheses: Studies are conducted to test the generated hypotheses. This stage usually begins with a retrospective study design. If the results of the retrospective study warrant further investigation, prospective and experimental studies may follow.
• Analysis of Result: Data from analytic studies are analyzed to confirm or reject hypotheses. This stage also serves to suggest further descriptive studies and the formulation of new hypotheses.

Errors in Epidemiologic Studies

Errors in epidemiological studies can be categorized into random errors and systematic errors (bias).

Random Error

Random error is the result of fluctuations around a true value caused by sampling variability. It is characterized as being random and can occur during data collection, coding, transfer, or analysis.

• Characteristics: Random error affects measurements in a transient and inconsistent manner. It is impossible to fully correct for random error.
• Sampling Error: This is the random error present in all sampling procedures.
• Examples: Poorly worded questions, misunderstanding an individual answer from a specific respondent, or typographical errors during coding.
• Precision: In epidemiological variables, precision is a measure of random error. Precision is inversely related to random error; therefore, reducing random error increases precision.
• Confidence Intervals: These are computed to demonstrate the precision of relative risk estimates. A narrower confidence interval indicates a more precise relative risk estimate.
• Methods to Reduce Random Error:
  1. Increase the sample size of the study.
  2. Reduce variability in the measurement.

Systematic Error (Bias)

Systematic error, or bias, occurs when there is a difference between the true value in a population and the observed value in a study from any cause other than sampling variability.

• Bias: Defined as any systematic error that results in an incorrect estimate of the association between exposure and the risk of disease. Observational studies are particularly susceptible to chance, bias, and confounding, which must be addressed during both the design and analysis stages.
• Selection Bias: This occurs when study subjects are selected or become part of the study due to a third, unmeasured variable that is associated with both the exposure and the outcome of interest.
• Information Bias: This arises from a systematic error in the assessment of a variable. A common example is recall bias.
• Confounding: Traditionally defined as bias arising from the co-occurrence or mixing of effects of extraneous factors (called confounders) with the main effects of interest.

Measures of Diagnostic Efficacy

Parameters of diagnostic efficiency quantify how useful a specific test is for a given disease or condition.

• Diagnostic Sensitivity: The proportion of individuals with a disease who test positive. It is usually expressed as a percentage.
  • $\text{Sensitivity (\%)} = \frac{100 \times \text{the number of diseased individuals with positive test}}{\text{total number of diseased individuals tested}}$
  • $\text{Sensitivity (\%)} = \frac{\text{Total Positive (TP)}}{\text{True Positive (TP)} + \text{False Negative (FN)}} \times 100$
• Diagnostic Specificity: The proportion of individuals without the disease who test negative for the disease.
  • $\text{Specificity (\%)} = \frac{100 \times \text{the number of individuals without the disease with a negative test}}{\text{total number of individuals tested without the disease}}$
  • $\text{Specificity (\%)} = \frac{\text{Total Negative (TN)}}{\text{True Negative (TN)} + \text{False Positive (FP)}} \times 100$
• Positive Predictive Value (PPV): The probability that subjects with a positive screening test truly have the disease.
  • $\text{PPV} = \frac{\text{Number of diseased individuals with positive test}}{\text{Number with positive test}} \times 100$
  • $\text{PPV} = \frac{\text{Total Positive (TP)}}{\text{True Positive (TP)} + \text{False Positive (FP)}} \times 100$
• Negative Predictive Value (NPV): The probability that subjects with a negative screening test truly do not have the disease.
  • $\text{NPV} = \frac{\text{Number of individuals without the disease with a negative test}}{\text{Number with negative test}} \times 100$
  • $\text{NPV} = \frac{\text{Total Negative (TN)}}{\text{True Negative (TN)} + \text{False Negative (FN)}} \times 100$

Measures of Disease Spectrum

These measures define the ability of an agent to infect, cause disease, or cause death.

• Infectivity (Ability to infect):
  • $\text{Infectivity} = \frac{\text{Number of infected}}{\text{Number of susceptible}} \times 100$
• Pathogenicity (Ability to cause disease):
  • $\text{Pathogenicity} = \frac{\text{Number with clinical disease}}{\text{Number of infected}} \times 100$
• Virulence (Ability to cause death):
  • $\text{Virulence} = \frac{\text{Number of deaths}}{\text{Number with disease}} \times 100$

Case Study: Breast Cancer and Oral Contraceptive Use

A cross-sectional study was conducted to determine the association between breast cancer (identified by the presence of a cyst) and the lifetime use of oral contraceptives. The objective is to calculate the prevalence odds and prevalence ratio.

• Task: Calculate the prevalence odds and prevalence ratio of patients with breast cancer that use oral contraceptives relative to breast cancer patients who never used oral contraceptives.