Prognosis and Therapy

prognosis

the likely outcome of a disease

prognostic factors

• demographic (age, gender, socioeconomic factors)

• individual patient behaviors (fear-avoidance, copers)

• disease/injury-specific factors

• comorbid considerations

do all future risk studies measure prognosis

no → not every study showing future risk is a prognostic prediction study

prediction studies

use multiple variables (referred to as “predictors”) to estimate the probability of an outcome

two types of prediction studies

1. diagnostic prediction models

2. prognostic prediction models

diagnostic prediction models

determine the presence of a particular disease or injury

prognostic prediction models

estimate whether a particular event/injury will occur in the future

what is the point of prognostic studies

• predict future outcomes

• develop/validate prediction models

• incorporate predictive performance

• What is likely to happen?

what are not prognostic studies

• not descriptive

• not causal

• not evaluating interventions

• does not answer what caused this or what is the effect b

a good prognostic model should answer 4 questions

• Discrimination: can it distinguish high-risk from low-risk patients?

• Calibration: are the predicted risks accurate?

• Validation: does it work in other patients?

• Clinical Utility: does it improve clinical decisions?

discrimination

can it distinguish high-risk from low-risk patients?

calibration

are the predicted risks accurate?

validation

does it work in other patients?

clinical utility

does it improve clinical decisions?

building an evidence table: column 1

condition → what condition is targeted by the therapy

building an evidence table: column 2

study design → report the type of study found

building an evidence table: column 3

author, year

building an evidence table: column 4

“n = total number of subjects used in the study

building an evidence table: column 5

statistically significant → not if results are or aren’t

building an evidence table: column 6

quality of study → numerical score assigned

• PEDro scale

• QUADAS Tool

building an evidence table: column 7

magnitude of benefit → how strong is the benefit (small, medium, large, none)

building an evidence table: column 8

absolute risk reduction = ARR = (event rate in the intervention group - event rate in control rate)

• correlates with risk difference

building an evidence table: column 9

Number Needed to treat = NNT = 1/ARR

building an evidence table: column 10

comments → on study design flaws, notable elements, dosing, specifics, etc