Windsor University School of medicine

Biostatistics

Sensitivity and specificity:

	Disease present	Disease absent
Positive screening	A (90)	B (20)
Negative screening	C (10)	D (80)

Sensitivity =A / A + C = 90 / 100 = 90%

Specificity = D/ B + D = 80 / 1OO = 80%

Positive predictive value = A / A + B = 90 / 110 = 81%

Negative predictive value D / C +D = 80 / 90 = 88%

Sensitivity Persons with the disease and positive screening test divided by number of persons tested with disease (100). i.e sensitivity of test means it gives a positive finding when the person has the disease.

Specificity = Persons without the disease and negative screening test is divided by number of persons tested without diseases (100), i.e.specificity of a test means it gives a negative result when the person does not have disease.

Positive predictive value = Persons with the disease and positive screening test divided by the total number of positive screening test.

Negative predictive value Persons without the disease and negative screening tests divided by the total number of negative screening test. In syphilis, RPR and VDRL have high sensitivity, so the test can give high false positive and few false negative results, but FTA-ABS has high specificity which gives correct diagnosis.

Overall accuracy A + D / A + B +C + D = 170 / 200 = 85%

A = True positive, D True negative.

B False positive, C = False negative.

Incidence of the exposed and non exposed group:

	Disease present	Disease absent
Exposure present	A (80)	B (20)
Exposure absent	C (30)	D (70)

Incidence rate of exposed group = A / A + B = 80 / 80+ 20 = 80%

Incidence rate of non exposed group = C / C + D 30 / 30 + 70 = 30%

incidence rate = Only new disease cases over a period of time is divided by population at risk, so incidence means new cases.

Prevalence rate Total number of cases at a given time is divided by total populations, so prevalence means all eases.

Case control studies.

	Lung failure	Normal Lung
Smoker	A (70)	B (20)
Non-Smoker	C (30)	D (80)

Cases = A / A + C = 70 / 100 = 70%

Controls = B / B + D = 20 / 100 = 20%

Incidence of exposed group = A / A + B = 70 / 90 = 77%

Incidence of non-exposed group = C / C + D = 30 / 110 = 27%

Mean Sum of the numbers associated with the observations is divided by the number of observations.

Median = Middle number.

Mode Most frequently occuring number.

Relative risk Incidence rate among exposed group is divided by incidence rate among non-exposed group 77% / 27% = 2.9% (calculated from Lung failure and Smoking abuse).

The odds ratio: In case control studies or when the proportion of disease is small in cohort studies, odds ratio is used to estimate the relative risk, which cannot be calculated directly.

(A x D) / (B x C) = (70 x 80) / (20 x 30) = 9.33

Attributal risk Incidence rate among exposed group minus incidence rate among non- exposed group = 77% - 27% = 50% (calculated from Lung failure and smoking abuse). Absolute risk Same as incidence.

A / A + B = 70 / 90 = 77% (calculated from Lung failure and smoking abuse from item 3 in case control study).

Probability of a disease condition.

Probability = Total number of times disease occurs is divided by total number of times disease can occur (e.g., total number of AIDS patients = 60, total number of pneumocystis pneumonia = 20, so probability = 20 / 60 = 33%).

Chi-square test: This test is most commonly used for differences between proportions, i.e., comparing effects of two different drugs used in two different groups.

Null hypothesis: When we compare two groups in a study and notice some differences between them, the null hypothesis may suggest that the observed differences are due to random variation in the data. If you accept null hypothesis, you think that observed differences are due to such random variations, but if you reject the null hypothesis, you think that observed differences are not due to random variations. Null hypothesis may he true or false.

Type I (alpha) error means null hypothesis is true, but rejected.

Type II (beta) error means null hypothesis is false, but accepted.

No error means either null hypothesis is true and accepted or is false and rejected.

The P value: P value < 0.01 means the result is statistically significant because the probability of random variation alone is very small.

Back