Notice that they mention '95 Percent Confidence Interval = % Reduction +/- 1.96 x Standard Error' in the article. Where did the value 1.96 come from?

U.S.News and John Hopkins Medicine's Armstrong Institute for Patient Safety and Quality wanted to calculate the number of lives that could be saved each year if all congenital heart surgery patients had their surgeries at high-volume hospitals. Most of the patients in the study were children. This article uses confidence intervals as a part of their research. Confidence interval, as I understand it, measures the degree of uncertainty in a sampling method. This article does not include the sample size, but it does have a level of the confidence interval. This study uses data from 61 hospitals downloaded from The Society of Thoracic Surgeons website, where the most recent data at the time of this study displayed 30-day mortality for congenital heart surgery patients. In this study, the researchers placed the patients into five categories, with category 1 being the lowest risk and category five as the highest risk. The analysis was restricted to categories four and five since the first three categories were found to have a weak volume outcome based on prior research. However, categories four and five were assigned a rating of one to three stars. O/E Ratio = Observed Mortality Rate/ Expected Mortality Rate 1 star: 95% confidence interval that the O/E ratio is above 1.0 2 stars: 95% confidence interval that the O/E ratio is above 1.0

3 stars: 95% confidence interval that the O/E ratio is above 1.0 This analysis used information from the most recent study time, from the Society of Thoracic Surgeons website. The Society thresholds previously identified it for categorizing hospitals as high volume (250 or more procedures per year), medium-volume (100 to 249 procedures per year) and low-volume (fewer than 100 procedures per year). It was found that 26 hospitals were categorized as high volume. The society calculated the expected mortality rates in account with severity of the conditions, and complications in each hospital's patient population. For the purpose of this study, only category four and five were counted by the researchers. The researchers counted the number of deaths and number of expected deaths. This article does not discuss the sample size number used in the study. It also does not discuss the results of the study. This article only discusses the method used in the actual study.