EMS World Magazine is the most authoritative source in the world for clinical and educational material designed to improve the delivery of prehospital emergency medical care.

Issue link: https://emsworld.epubxp.com/i/996071

40 JULY 2018 | EMSWORLD.com ISSUE FOCUS: EVIDENCE-BASED MEDICINE W hen conducting research, it is impractical (and usually impossible) to study every person with a disease or problem—what researchers call the target population. Imagine the resources required to study every heart attack patient in a year or every car crash victim over a decade! Instead, researchers select a sample of patients to represent the larger target population. Although researchers care about the people in their studies, what they really want to know is what those people (the sample) can teach them about the larger (target) population. One statistical tech- nique researchers use to describe what they learn about the target population from a sample is confidence intervals. A confidence inter val takes data for some mea sure obtained from a sample and then calculates what that measure prob- ably looks like in the target population. Typically researchers use and report con- fidence inter vals of 95% (95% CI). For example, if researchers are study- ing the hear t rates of trauma patients, they might find an avera ge hear t rate in their sample of 102 bpm. Using that avera ge, the s tandard dev iation, and the number of people in their s ample (find the formula at w w w.wikihow.com/ C alcu l ate - Co nf i d e n ce - Inte r v al), th ey might calculate a 95% CI of 98–106. That means in this sample of trauma patients, the average heart rate was 102, and in the target population of all trauma patients, the researchers are 95% sure the aver- age heart rate is somewhere between 98 and 106. Ideally the 95% CI is narrow enough that th e re is n o p rac tic al dif fe re n ce between the measure found in the sample and the probable range of that measure in the target population. If the confidence inter val is wide—for example, a sample average of 102 but a 95% CI ranging from 52 to 152—then the sample doesn't pro- vide a ver y clear indication of what's going on in the target population. The width of a 95% CI is driven by the number of subjects in the sample and the natural variation in whatever's being measured. Confidence inter vals can be calculated for almost ever y type of measure (aver- ages, medians, percentages, ratios, etc.). Comparing Subgroups Confidence inter vals can also be used to compare two or more subgroups within a sample. For example, imagine research- ers studying a target population of con- gestive hear t failure patient s selec t a representative sample of patients. They administer nitroglycerin to half the sam- ple (the intervention group) and a placebo to the other half (the control group). If 50% of the patients in the inter ven- tion group and 60% of the patients in the control group require ICU admission, then there is a difference in admission rates of By Lawrence H. Brown, PhD A GUIDE TO STATISTICAL SIGNIFICANCE IN EMS RESEARCH Understand a few key concepts to better comprehend research and its clinical signifi cance Figure 1: Overlapping vs. non-overlapping confidence intervals

- Cover
- EMS_52
- EMS_1
- EMS_2
- EMS_3
- EMS_4
- EMS_5
- EMS_6
- EMS_7
- EMS_8
- EMS_9
- EMS_10
- EMS_11
- EMS_12
- EMS_13
- EMS_14
- EMS_15
- EMS_16
- EMS_17
- EMS_18
- EMS_19
- EMS_20
- EMS_21
- EMS_22
- EMS_23
- EMS_24
- EMS_25
- EMS_26
- EMS_27
- EMS_28
- EMS_29
- EMS_30
- EMS_31
- EMS_32
- EMS_33
- EMS_34
- EMS_35
- EMS_36
- EMS_37
- EMS_38
- EMS_39
- EMS_40
- EMS_41
- EMS_42
- EMS_43
- EMS_44
- EMS_45
- EMS_46
- EMS_47
- EMS_48
- EMS_49
- EMS_50
- EMS_51

- https://www.emsworld.com/
- https://www.wikihow.com/Calculate-Confidence-Interval
- https://www.wikihow.com/Calculate-Confidence-Interval