-
Article
Open AccessDifferences in Symptom Burden in Primary Brain Tumor Patients Based on Sex, Race, and Ethnicity: a Single-Center Retrospective Study
Symptom burden affects quality of life and prognosis in primary brain tumor (PBT) patients. Knowing whether symptom burden varies based on sex, race, or ethnicity may affect the interpretation of the relations...
-
Article
Open AccessThe economic burden of RSV-associated illness in children aged < 5 years, South Africa 2011–2016
Data on the economic burden of RSV-associated illness will inform decisions on the programmatic implementation of maternal vaccines and monoclonal antibodies. We estimated the cost of RSV-associated illness in...
-
Article
Open AccessThe burden of RSV-associated illness in children aged < 5 years, South Africa, 2011 to 2016
Vaccines and monoclonal antibodies to protect the very young infant against the respiratory syncytial virus (RSV)-associated illness are effective for limited time periods. We aimed to estimate age-specific bu...
-
Chapter
Publication Bias
Publication bias occurs when negative studies are either never published or are very delayed in their publication. Publication bias can be visualized using a Funnel plot. Assessing abstracts and databases of r...
-
Chapter
Changing Endpoints
Statistical calculations about power and chances of false positive and negative are based on a plan set before a study starts. If new tests are done after the data is examined, so-called post-hoc analyses, the...
-
Chapter
Hawthorne Effect
The Hawthorne effects happens when people change their behavior because they know they are being observed, which commonly happens in studies that look at behavioral effects before and after an intervention. Of...
-
Chapter
Overpowering
A study is overpowered if there is such a large sample size that small, clinically meaningless differences can be detected statistically. Studies should be planned to detect differences that are clinically mea...
-
Chapter
Lead Time Bias
Lead time bias refers to the apparent increase in survival when a disease is diagnosed earlier and survival is measured from the time of diagnosis to the time of death. In screening studies it is avoided by me...
-
Chapter
Confounding by Severity
Confounding by severity occurs when people with worse disease are treated differently than people with more mild disease, which can make it appear that the intervention causes poor outcomes. It can be addresse...
-
Chapter
Immortal Time Bias
Immortal time bias occurs when one is looking at the risk related to an exposure than can occur over time, so that it appears the exposure causes people to live longer when in fact living longer makes people m...
-
Chapter
Multiple Comparisons
When multiple tests are done on the same data, the chance of finding random false correlations increases. Post hoc analyses of secondary or other endpoints that are not prespecified can also lead to false-posi...
-
Chapter
Ascertainment Bias
Ascertainment bias can mean several things. It occurs when the likelihood of detecting an effect depends on the method of detecting that effect or when different methods are used in different groups. Ascertain...
-
Chapter
Proportional Hazard Violation
Time-dependent endpoints are assessed using Kaplan-Meier survival curves. A hazard ratio is a global assessment of the differences between survival curves that tells the relationship between the average rate o...
-
Chapter
Berkson’s Bias
Berkson’s bias refers to the fact that health conditions may appear associated merely because they are correlated with hospitalization or increased exposure to healthcare. Berkson’s bias can be accounted for b...
-
Chapter
Underpowering
Power refers to the ability of a study to detect a true difference. When a study is too small, it is underpowered and false-negative results become common. A sample size should be chosen that is large enough t...
-
Chapter
Will Rogers Phenomenon
The Will Rogers phenomenon occurs when recategorizing people causes survival to improve in all subgroups while not affecting overall survival. It can be avoided by not looking at subgroups in isolation without...
-
Chapter
Length Time Bias
Length time bias refers to the fact that slower and more indolent diseases are more likely to be detected by screening. In its most extreme form, length time bias can lead to overdiagnosis of diseases that wou...
-
Chapter
Simpson’s Paradox
Simpson’s paradox occurs when the direction of correlation between two variables changes sign (from positive to negative or netagive to positive) when subgroupa are analyzed compared to when the entire populat...
-
Chapter
Selection Bias
Selection bias occurs when there are differences in outcome due to differences in populations being studied rather than the effect of interventions. Selection bias is especially likely to happen in small, sing...
-
Chapter
Validation
Testing a model is called validation. When validation is done on the same dataset that was used to develop the model, it is called internal validation. Because of the risk of overfitting, internal validation m...
