Notes on How we Choose Papers

Our Preferences:

· Cause specific relative risks preferred over all-cause risks

· Moderate/average of studies figures preferred over extreme numbers

· More adjustment for toher risk factors preffered over less adjustments

· More detailed divisions of population prefered over less divisions

· More stratification of risk factors prefered over less stratification

· More recent data prefered over outdated data

Why some papers are rejected:

· Irrelevant data or data not suited for our purpose (noted as no specific relative risks given)

· Some other paper gave better data in accordance to the list of preferences above

· Non-positive/non-statistically significant data

· Trend especially monotonic trends, which should be there; isn't

Miscellaneous:

· Income variable is mainly absorbed by education and occupational class, and thus, is not used as a primary variable. Furthermore, education stabilizes after mid-20's but income fluctuates throughout the career.

· Serum cholesterol variable is mainly absorbed by BMI, nutrition, and fat intake; and is therefore left out.

· Good data for existing cancer and heart diseases; and family history of asthma cannot be found. It is also suspected that with a positive diagnosis of cancer or heart diseases, the mortality rate becomes difficult to predict even to a reasonable range of accuracy.

· Data on how gun possession affect homicide/suicide and how salt intake affect mortality is often contradictory/inconclusive. Thus, in the interest of preventing Type I errors, they are excluded from our calculations.

· Synnergistic effects among some risks factors when combined together are not taken into consideration. This factor may be included in future versions of the calculator. However, the correlation of the distribution of the population among the characteristics are taken into consideration.