What is the mean survival time? What is the median survival time? What is the mode survival time?
The mean survival time in the above data set is 9 months, the median survival time is 10.5 months, and the mode survival time is 11 months. A group of data can be described by their central distribution and degree of variation (scale statistics). Measures of central distribution include mean, median, and mode. Scale statistics describe the variability or spread of the sample data (i.e., how scattered or clustered the data are about the center of the distribution) and include variance, standard deviation, standard error of the mean, coefficient of variation, range, and interquartile range.
Mean
•Arithmetic average
•Appropriate for data that is normally distributed
•Equal to the sum of all the sample values divided by the sample size
•Strengths:
•Calculated from all the sample values so makes maximum use of all available data
•Weaknesses:
•Can be influenced by any extreme value (outlier)
•Possible solutions include weighting or trimming the data set
Median
•Midpoint of a series of ordered values
•Appropriate for non-normally distributed data
•Determination involves arranging the numbers in sequence from smallest to largest, then finding the midpoint or calculating the average of two midpoints
•Most common way to report survival data
•Strengths:
•Insensitive to outliers; may be preferred to the mean when dealing with skewed (non-normal, asymmetric) data
•Weaknesses:
•Does not account for all data values; a mean generally is the preferred estimation of central tendency for symmetric distributions
•Relationship to mean
•Symmetric distributions: mean = median
•Left-skewed data: mean < median
•Right-skewed data: mean > median
Mode
•Most frequently occurring value within the data set
•There may be more than one mode, or none at all
•Strength:
•Less sensitive to skewed data than either the sample mean or median
•Weaknesses:
•More subject to sample variation than either the sample mean or median
•Limited applicability (samples may not have any repeated data values)
If the distribution is unimodal and symmetric, then the sample mean, median, and mode are all estimates of the same value, the population mean.
Standard Deviation
•Measure of spread, scatter, or dispersion of a sample
•When data are distributed normally, 1, 2, and 3 standard deviations from the mean encompass 68%, 95%, and 99% of the population respectively
Standard Error of the Mean
•Measure of the precision with which the mean is known
•Calculated by dividing the standard deviation by the square root of the sample size; it is always smaller than the standard deviation
•The choice between using standard deviation vs. standard error of the mean is controversial
Confidence Interval (CI)
•A range within which 95% (or other computed value, 1-α) of the population values being estimated would be expected to fall
•Reflects the strength (precision) of the evidence:
•Wide confidence intervals indicate less precise estimates
•Narrow confidence intervals indicate more precise estimates
•The larger the sample size, the narrower the confidence interval, and the greater the confidence that the true value is close to the stated value
•In a positive finding, the lower boundary of the confidence interval should remain important or clinically significant if the results are to be accepted
•In a negative finding, the upper boundary of the confidence interval is not clinically important for acceptance of the result
Validity
•Validity is the extent to which you measured what you intended to measure (accuracy)
•Internal validity is the integrity of a study’s experimental design
•External validity refers to the ability to generalize the study’s results to non-study patients or populations
Reliability
•Reliability is the reproducibility of your results (precision)
•Different assessors making the same conclusions with the same data
•One assessor making the same conclusions with the same data, on different occasions
A recently published paper describes the circumstances, characteristics, and outcomes of one trauma group’s clinical management of 15 cases of penetrating neck injury. What type of study is this?
A case series is a descriptive, observational study of a series of cases, which typically portrays the manifestations, clinical course, and prognosis of a condition or single intervention without a control group.
Study Designs
•Listed in the order of least to most robust study design
•Case report (no control group)
•Case series (no control group)
•Cross-sectional study
•Case–control study (select outcomes first and then look for exposures)
•Cohort (select exposure first and then look for outcomes; can be retrospective or prospective)
•Randomized controlled trial (lowest likelihood of bias)
•Meta-analysis (pooled data)
Case Report
•Description of a single case, typically describing the manifestations, clinical course, and prognosis of that novel disease or intervention
•Disseminates new information quickly
•Good for rare conditions
•Weaknesses:
•Anecdotal evidence
•Provides little empirical evidence to the clinician
Case Series
•Most common type of study in the surgical literature
•Represents a series of case reports
•Strengths:
•Useful as a source of hypotheses for investigation by other study designs
•Weaknesses:
•Provides weak empirical evidence because of the lack of comparability
A study evaluated the long-term health status and quality of life of heart transplant recipients. 293 patients were asked to come in five years after their transplant, whereupon a full clinical examination and written survey were conducted. What type of study is this?
A cross-sectional study is a descriptive study of the relationship between diseases and other factors, within a defined population, at a specified point in time.
Cross-Sectional Study
•Also referred to as a prevalence study
•A prospective comparison of subjects with similar exposure at a certain point in time
•Strengths:
•Good estimate of prevalence
•Good generalizability
•Good feasibility
•Weaknesses:
•Cannot estimate incidence or temporal sequence
•Cannot provide information on causality