Home » Data » Dataset use in Stata » analyzing blood pressure and blood glucose
| analyzing blood pressure and blood glucose [message #9994] |
Tue, 14 June 2016 12:33  |
DHS user
Messages: 111
Registered: February 2013
|
Senior Member |
|
|
I am trying to reconcile how to identify the appropriate sample to analyze for the blood pressure and blood glucose sample (which is half of the households selected for the male survey, see section 1.6.1 and 1.6.4 in final DHS report) in the 2013 Namibia DHS.
Table 17.1 (pg 237 in the DHS report) shows that 2,584 women and 2,163 men age 35-64 were eligible for these tests. Among these individuals, 80.7% of women and 70.7% of men had their blood pressure measured, and 75% of women and 63.8% of men had their blood glucose measured.
This would equal the following sample sizes which I can more or less match exactly in the datasets I downloaded based on identifying the eligible participants who consented and had no reported issues with the blood sample.
Women (n) Men (n)
Blood Pressure 2093 1536
Blood Glucose 1938 1384
Our question is why are the following numbers reported in measured blood pressure tables "17.4.1 Blood pressure status: Women" "17.4.2 Blood pressure status: Men" and "17.7.1 Prevalence of diabetes by background characteristics: Women" and "17.7.2 Prevalence of diabetes by background characteristics: Men".
Women (n) Men (n)
Blood Pressure 2048 1406
Blood Glucose 1873 1,221
It is not clear how using the dataset to identify eligible males and females in households selected for the male interview and consented and had biologically plausible values changes from the sample listed in Table 17.1 to the sample size in the final prevalence tables: 17.4.1 and 17.4.2 for blood pressure and 17.7.1 and 17.7.2 for blood glucose.
Would someone be able to help me reconcile this?
|
|
|
|
| Re: analyzing blood pressure and blood glucose [message #9998 is a reply to message #9994] |
Tue, 14 June 2016 18:56  |
Trevor-DHS
Messages: 805
Registered: January 2013
|
Senior Member |
|
|
I reviewed the information you provided and the data in the datasets, and I'm giving some summaries based on the Stata data below:
1) The sample is all men and women age 35-64 in the half of households that were selected for the men's questionnaire.
. tab hv104 if hv027==1 & hv105>=35 & hv105<= 64,m
sex of |
household |
member | Freq. Percent Cum.
------------+-----------------------------------
male | 2,163 45.57 45.57
female | 2,584 54.43 100.00
------------+-----------------------------------
Total | 4,747 100.00
From this you can see that 2,584 women and 2,163 men were eligible for these tests. These numbers are unweighted as they are the actual numbers tested in the survey.
2) From these cases, a total of 3,634 (2,085 women and 1,530 men) granted consent for blood pressure testing, and 3,316 (1,937 women and 1,379 men) granted consent for blood glucose testing (unweighted again). These numbers translated into the percentages as shown below.
. tab sh333 hv104 if hv027==1 & hv105>=35 & hv105<= 64, col m
blood |
pressure | sex of household
measuremen | member
t granted | male female | Total
-----------+----------------------+----------
no | 19 19 | 38
| 0.88 0.74 | 0.80
-----------+----------------------+----------
yes | 1,530 2,085 | 3,615
| 70.74 80.69 | 76.15
-----------+----------------------+----------
. | 614 480 | 1,094
| 28.39 18.58 | 23.05
-----------+----------------------+----------
Total | 2,163 2,584 | 4,747
| 100.00 100.00 | 100.00
. tab sh336e hv104 if hv027==1 & hv105>=35 & hv105<= 64, col m
read consent | sex of household
statement ( blood | member
glucose) | male female | Total
----------------------+----------------------+----------
granted | 1,379 1,937 | 3,316
| 63.75 74.96 | 69.85
----------------------+----------------------+----------
refused | 9 10 | 19
| 0.42 0.39 | 0.40
----------------------+----------------------+----------
respondent not presen | 46 21 | 67
| 2.13 0.81 | 1.41
----------------------+----------------------+----------
. | 729 616 | 1,345
| 33.70 23.84 | 28.33
----------------------+----------------------+----------
Total | 2,163 2,584 | 4,747
| 100.00 100.00 | 100.00
3) The numbers presented in the coverage of testing are unweighted numbers as they related to the testing in the survey, and not to estimates for the population. The numbers in table 17.4.1 and 17.4.2, and in 17.7.1 and 17.7.2 are weighted as the percentage estimates are for the population. These tables also include slightly less cases than above as some cases had out of range values and are excluded from the tables see the extra conditions used below.
. tab hv104 if sh333 == 1 & sh335aa >= 30 & sh335aa <= 300 & sh335ab >= 0 & sh335ab <= 160 [iw=hv005/1000000], m
sex of |
household |
member | Freq. Percent Cum.
------------+-----------------------------------
male | 1,405.9812 40.71 40.71
female | 2,047.6235 59.29 100.00
------------+-----------------------------------
Total | 3,453.6047 100.00
. tab hv104 if sh336e == 1 & sh336k >= 0 & sh336k <= 222 [iw=hv005/1000000], m
sex of |
household |
member | Freq. Percent Cum.
------------+-----------------------------------
male | 1,221.3445 39.47 39.47
female | 1,873.0363 60.53 100.00
------------+-----------------------------------
Total | 3,094.3808 100.00
I hope this helps explain the differences that you see.
|
|
|
|
Goto Forum:
Current Time: Thu Jan 30 17:15:05 Coordinated Universal Time 2025
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
