YPLL Overrides Age Selection?
Introduction:
Historically, Vitalnet overrode age selection when YPLL (or YPLL Rate) was selected as the main statistic. In 2020, based on suggestion / request / discussion from Maria Cooper at Texas Department of State Health Services, we decided to change the algorithm to NOT override age selection, as the best decision. The purpose of this page is to discuss the change, and to verify that Vitalnet is working correctly with the change.
Why Override?
1) The great majority of published YPLL results analyze all ages. Reinforcing that YPLL normally uses all ages, the Wikipedia YPLL article makes no mention of not using all ages (for example, calculating YPLL among those age 0-19 years).
2) In an attempt to protect less experienced users, who might unintentionally select an age group such as 0-19 years, and then mistakenly think the YPLL output covered all years, Vitalnet previously overrode the age selection, setting age to 0-99+ years for YPLL output.
Why NOT Override?
1) It is also possible that a user might intentionally select an age group such as 0-19 years, make the output (with age selection overridden), and then mistakenly think the YPLL output only covered 0-19 years.
2) Assuming no override, the output clearly says that an age group (such as 0-19 years) is selected, so there really should not be misinterpretation.
3) It is useful to run YPLL for a specific age group, such as 0-19 years, so removing the override makes Vitalnet more useful.
4) Retaining the override would require an explanation in the documentation, leading to the reader getting bogged down in extraneous details.
Our Decision
Vitalnet prevents certain things that make zero sense, such as [age rows and age columns in one table], or [table with no years selected]. In contrast, making YPLL for an age groups such as 0-19 makes sense, so it should be allowed. Finally, users want the override removed. Therefore, the override is best removed.
Methods:
The override was removed by editing the Vitalnet source code. To verify that the software still works correctly related to this issue, a series of YPLL (and YPLL rate) tables were made, using 2010 Texas ICD-10 underlying cause mortality, with either 0-19 selected, or with 4 groups [0-19, 20-39, 40-59, 60-99+] selected, first with age override turned on, then with age override turned off.
The tables were compared to make sure that Vitalnet is still working correctly with the change, and to look for any unintended consequences. Some of the tables (C-F below) should not differ with the change. Other tables should differ, and need to be checked by hand to make sure the numbers are correct.
Results:
Test | Rows | Cols | Ages Used |
Multi Tables? |
Should Differ? |
---|---|---|---|---|---|
A | Race | Sex | 0-19 | No | Yes |
B | Sex | Race | 0-19 | No | Yes |
C | Age | Sex | 4 groups | No | No |
D | Sex | Age | 4 groups | No | No |
E | Race | Age | 4 groups | No | No |
F | Age | Race | 4 groups | No | No |
G | Race | Sex | 4 groups | By Age | Yes |
Test | YPLL | YPLL Rate |
Results Correct? |
||
---|---|---|---|---|---|
Override On (Old) |
Override Off (New) |
Override On (Old) |
Override Off (New) |
||
A | 0227mpee | 0227bjrr | 0227duax | 0227frsz | Yes |
B | 0227bspw | 0227vjbm | 0227nchw | 0227atea | Yes |
C | 0227uuhz | 0227uiaf | 0227dkmi | 0227xiyb | Yes |
D | 0227wuxg | 0227pfeh | 0227mxsd | 0227qmak | Yes |
E | 0227hfsf | 0227fzia | 0227metu | 0227vpfn | Yes |
F | 0227aupj | 0227wpwb | 0227rueu | 0227jybu | Yes |
G | 0227fdrs | 0227cqpq | 0227sxbi | 0227skfz | Yes |
A few key facts to help check tables (YPLL cutoff 65):
• YPLL for 0-19 = 250,384.0
• YPLL for 0-64 = 898,205.0
• YPLL rate for 0-19 = 3,285.1 / 100,000
• YPLL rate for 0-64 = 3,984.3 / 100,000
• Population for 0-19 = 7,621,714
• Population for 0-64 = 22,543,675
Discussion:
It was decided to remove the YPLL age override, and the reasoning given. To make sure that Vitalnet is still working correctly after removing the YPLL override, a series of YPLL tables were made. The tables check out. The change was successful.