| Replicating BDHS Table on History of Diabetes [message #13617] |
Thu, 30 November 2017 06:39  |
jilans
Messages: 5
Registered: November 2017
Location: London
|
Member |
|
|
Hi
I am trying to replicate tables (Table 15.4 of page 243) on history of diabetes for Bangladesh DHS 2011 report.
1. I am finding it difficult to arrive at the same denominator values for numbers of respondents. I seem to have a far greater number of women in the cohort than in the report.
2. I also realised a number of respondents have more than 1 blood glucose measurements (sh284_1 to sh284_8). The report doesn't say which of the measurements or whether an average of the measureents was used in determining cut-off points for classifying DM 'diagnosis' (FPG 7mmol/L and over).
3. On page 242, it says to convert the blood glucose measurements from mg/dl to mmol/L, values were multiplied by 0.0551. Which works out well when I tried to convert sh284 to mmol/L. However, it then said necessary adjustments were made to glucose measurements in 'whole blood' to 'plasma glucose' equivalent by multiplying each value by 1.11. Does it mean that I would need to do sh284 * 0.0551 * 1.11 to arrive at other corresponding national data?
Please can anyone help answer these questions and include Stata commands where possible?
Many thanks.
|
|
|
|
|
|
| Re: Replicating BDHS Table on History of Diabetes [message #13621 is a reply to message #13619] |
Sat, 02 December 2017 05:38  |
jilans
Messages: 5
Registered: November 2017
Location: London
|
Member |
|
|
Dear Bridgette
Thank you for your email.
I am afraid I did not find the CSPro codes very helpful, sorry. Not all the variables represented in the Tables are represented in the programming codes. I would really appreciate it if I can get more clarity on:
1. how the total number of respondents on Table 15.4 was arrived at if from Table C.2.1, for example, the number of eligible men aged 15-54 is over 3,900, when I removed those aged below 35 years, I end up with 2,510 men.
2. I am interested in knowing whether the denominator was obtained by extracting those 35 years and above from datasets BDIR61FL and BDMR61FL for women and men, respectively? if not, how?
3. What variables identify men and women aged 35 years or older who were eligible for blood glucose measurement. I have seen a few publications use 8835 residents (women: 4311; men: 4524) aged 35 years or older were eligible for blood glucose measurement. I would like to validate this information.
Hope this is helpful?
Best wishes
JB
|
|
|
|
|
|
| Re: Replicating BDHS Table on History of Diabetes [message #13794 is a reply to message #13704] |
Wed, 27 December 2017 12:22 |
Bridgette-DHS
Messages: 3208
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS Stata Specialist, Tom Pullum and Lindsay Mallick, a researcher on the DHS analysis team.
We apologize for the delay in this reply to your forum posting of December 11. Because of the specialized topic, and staff travel, we are not able to give a complete answer.
1) Only one in every third household was selected to participate in the biomarkers questionnaire. All men and women age 35 and older in the households were selected to participate in the biomarkers portion of the survey. This included older women and men who were not interviewed in the women's or men's individual questionnaire.
2) Therefore, you should not use the IR or the MR file. Instead, you should use the PR file to answer your questions related to blood glucose measurement.
3) The variables sh230 through sh284 contain the information you will need to match these tables. These include, age, sex, consent to testing, and results of the tests. If you tabulate variable sh231 (unweighted), you will be able to match the denominators in table 15.1.
Please try to work from this advice. If you still have difficulties, please feel free to post again, preferably with as specific a question as possible.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
