
Return to Breastfeeding Transformations (Ch. 1)
Creating a descriptive profile of child feeding practices will allow you to produce a child feeding profile for a district, region, or country and compare it to an "ideal practices" profile based on WHO child feeding recommendations. The profile will enable you to examine age specific child feeding practices in comparison to recommended practices that are known to be nutritionally beneficial. A child feeding profile will also allow you to identify negative practices and prioritize childfeeding interventions. The two primary recommendations that can be examined using this profile method are:
Exclusive Breastfeeding up to 4  6 months (influenced by both the percentage of infants who are never breastfed and the duration of exclusive breastfeeding in the population)  
Timely introduction of appropriate complementary foods at 6 months (by this age infants should be receiving solid foods in addition to breast milk) 
We will use the keast4j.sav data to develop a child feeding practices profile for Eastern Kenya. The first step is the creation of age groups.
Computing age groups and output for profile development
1. For analysis of child feeding practices we are most interested in the first two years of life, so that the first step in data analysis and preparation of the profile will be to select only those cases 24 months or younger for analysis.
Under data, go to select cases. Select if age in months is less than or equal to 24 (age<=24).
2. Now you will want to create a new variable that will allow you to plot child feeding practices by age groups of 3 months (0 through 2 months, 3 thorough 5 months, and so on).
Under Transform, select Recode into different variable. Select the age in months variable (age) as your input variable and type cfage(for child feeding age group) as the name of the output variable and click on change.
Next select Old and New Values. For old values select range, putting 0 in the first field and 2 in the next field (this is establishing a range of the months 0 through 2) and give this a new value of 1 (your first child feeding age group) by typing 1 in the new value field and clicking on Add. Then continue to create age groups of 3 months: The old value range 3 to 5 months has a new value of 2, old value range 6 to 8 has a new value of 3 and so on. The last new value will be 9 (you have created 9 age groups!) and will only contain the old value of 24 because we have selected only children less than 24 months. Now click Continue and then OK.
Be sure to repeat your newly created values under the data menu, define variable routine so that you will have a reference for future use of your newly created variable cfage. To do this highlight the variable you have created and select define variable from the data menu. Now click the Labels button and type in the label you want to give that variable in the variable label box. Tab down to value and type the number corresponding to the first value (e.g. 1 is the value for the first age group). Next tab down to value label and enter the corresponding value (e.g. 0 to 2 months is the value label for the first age group). Click Add. When all values have been entered and added, then click continue and then OK. Note that this routine should be repeated for each new variable created below.
Feeding Categories
First, a decision must be made on how many feeding categories will be included in the profile. This decision should be based on the availability of data on specific foods given to children under 24 months and on the level of specificity you will need to analyze child feeding practices and develop appropriate programmatic responses. It is advised that as few categories as possible be used in order to simplify the profile and allow for ease in interpretation by nontechnical staff. You can combine similar food/ liquid variables to create a core set of new child feeding variables. The following 5 categories are recommended:
Not breastfed Breastfed with staple cereal (maize, rice, wheat, sorghum, etc.) Breastfed with milk (all sources other than breastmilk) Breastfed with water/juice Exclusively breastfed
Obviously, some children will fall into multiple categories (ex. breastfed with both staple and other milk source). For the purpose of developing our profile, however, will want each child to fall into only 1 category. Therefore a hierarchy must be established. The following hierarchy is recommended.
Those not breastfed > category 1 Those breastfed with staple cereal and any combination of other foods/ liquids > category 2 Of the remaining children under 24 months (breastfeed, but not receiving staple cereal and therefore do not fit into category 1 or 2), those breastfed plus given milk other than breast milk or milk other than breast milk plus water/juice > category 3
Of the remaining children under 24 months after this step (not fitting into categories 1, 2, or 3), those who are breastfed with water/juice only > category 4 The final category should thus consist of those children under 24 months who do not fit into categories 14. This should be made up of those children who are given breast milk and nothing else (exclusively breastfed) >
category 5
COMPUTING STEPS we will want to use current data only so will use the variable bfnow (currently breastfeeding) as opposed to data based on recall.
EXERCISE for CATEGORY 1 NOT Breastfeeding: Establish those who are breastfed and those that are not breastfed.
1.1 Under the Statistics menu, choose Summarize, then choose Cross tabs
1.2 Now put the bfnow breastfeeding variable (remember, 1=yes and 0=no) under the columns heading by selecting it from the list of variables and using the arrows to move it under the columns heading.
1.3 Similarly, select the cfage variable you have just created and put under the rows heading.
1.4 Now click on OK.
1.5 Your results should be:
(note: if your column headings are numbers rather than the words yes/no, you can go back to the Data Editor screen, select view and check off value labels. This will allow your output files to display the value labels for numeric values (yes=1, no=0). Also note that blanks and/or omission of an age group means that no child in the data set falls into that category.
So overall, 23 of the 313 children who meet the criteria (24 months and younger) do not breastfeed. The table provides disaggregated counts of those not breastfeeding in each age category (19). From this information, the percentage not breastfeeding in each age category and overall can be calculated (e.g. overall = 23/331 = 7.3%
Now, according to your hierarchy (see above), you will want to create an output than will give you the number of children in each feeding category from among those that are breastfed (290). This will allow you to compute percentages at each age category. Remember each case can only go under 1 feeding category so that the sum of the feeding categories in each age group will = 100%.
EXERCISE for CATEGORY 2 Breastfeeding plus Uji: The second feeding category will be those who are breastfeeding, plus uji (staple cereal) and any combination of other foods/liquids.
2.1 You will need to create a new variable like this: highlight the ammennow variable on the SPSS data screen.
2.2 In the Data menu click on Insert variable (this will place the new variable next to your breastfeeding variable)
2.3 Highlight the new variable column by clicking on the top of the column
2.4 Select Define variable from the Data menu a type bfuji in the cell for variable name
2.5 Click on Labels and type bfuji under variable label.
2.6 Tab down to Value and type 1, then tab to Value Label and type ‘yes’ and click Add
2.7 Now type 0 under Value and ‘no’ under Value Label and Add
2.8 Now click Continue and then OK
2.9 Highlight the variable you have just created (bfuji) and click on Transform, Compute
2.10 Under Target variable type bfuji.
2.11 Select porridch (gave child porridge/ uji) from the list of variables and move it to the right side of the = sign
2.12 Click on Select If and If case satisfies condition:
2.13 Select bfnow from the variable list and move it using the arrow.
2.14 After the bfnow type =1 (bfnow=1).
2.15 Click Continue and then OK.
To create a crosstabulation follow the instructions under the first feeding category above and substitute bfuji for bfnow under columns. Your results should be:
This provides you with the total number in this category (290) and the total percentage in this category. It also provides the same information disaggregated by age group.
EXERCISE for CATEGORY 3 Breastfed, plus milk: Of the remaining children under 24 months (breastfeed, but not receiving staple cereal (uji) and therefore do not fit into category 1 or 2, those who are breastfed and given milk (any type) other than breast milk or milk other than breast milk and water/juice are in category 3.
3.1 Again will need to create a new variable by highlighting the ammennow variable and selecting under the Data menu, Insert variable (this will place the new variable next to your breastfeeding variable bfnow and your bfuji variable)
3.2 Highlight the new variable column by clicking on the top of the column
3.2 Click on Define variable and enter bfmilk in the cell for Variable name
3.3 Click on Labels and type bfmilk under Variable label
3.4 Now type 0 under Value and No under Value Label and Add
3.5 Tab down to Value and type 1, then tab to Value Label and type Yes and click Add
3.6 Click Continue and then OK
3.7 Highlight the variable you have just created (bfmilk) and click on Transform and Compute
3.8 Under Target variable type bfmilk
3.9 Select powdmlkch from the list of variables and move it to the right side of the = sign, type + and move formlch over, and type + again and move the variable milkch over (the formula should read: powdmlkch+formlch+milkch) Note that you may also type this formula into the equation box and avoid the ‘select’ and 'move' routine just described.
3.10 Now click on Select if and If the condition is satisfied:
3.11 Select bfnow from the variable list and move it using the arrow
3.12 After the bfnow type =1 (bfnow=1)
3.13 Add the & symbol and select the bfuji variable from the variable list moving it with the arrow. Then add an = 0 (The box should now read ‘bfnow=1 & bfuji=0’)
3.14 Click Continue and then OK
3.15 Now to dichotomize the bfmilk variable (yes/no) by clicking on Transform, Recode,
Into Same Variable.3.16 Move bfmilk from the variable list to the right and click on Old and New Values
3.17 Under old value type 0 and tab to New value and type 0, then click on Add
3.18 Go back to Old values, select Range and type 1 in the first range box and 3 in the second (1  3) and type 1 in the new value box, then click on Add
3.19 Click on Continue and then OK (Note dichotomizing the variable will be important for future analysis and to maintain consistency with the variable labels that you created earlier)
To create a crosstabulation follow the instructions under number one and substitute bfmilk under columns. Your results should be:
Note that the total number presented here is 64, the same number under the column NO in the BFUJI table. This is a good way to check that all remaining cases are being included in the output table (since only those cases that did not fall into category 1 or 2 are considered here). Therefore the overall total for out next table (category 4 should only be 41, those who didn’t fall into category 1, 2, or 3)
EXERCISE for CATEGORY 4 Breastfed with water/juice: Of the remaining 78 children under 24 months after the previous computing step (not fitting into categories 1, 2, or 3), those who are breastfed plus water/juice only be coded as breastfed with water/juice, category 4
4.1 Again will need to create a new variable by highlighting the ammennow variable and clicking on Data, Insert variable (this will place the new variable next to your bfmilk variable)
4.2 Highlight the new variable column by clicking on the top of the column
4.3 Select Define variable from the Data menu a type bfwater in the cell for Variable name
4.4 Click on Labels and type bfwater under Variable label
4.5 Type 0 under Value and No under Value Label and Add. Tab down to Value and type 1, then tab to Value Label and type Yes and click Add
4.6 Click Continue and then OK.
4.7 Highlight the variable you have just created (bfwater) and click on Transform, Compute
4.8 Under Target variable type bfwater
4.9 Select waterch from the list of variables and move it to the right side of the = sign, type + and move glcwtch over, type + and move juicech over, and finally type + and move the variable othliqch over so that the formula reads ‘ waterch+glcwtch+juicech+othliqch’. Note that you may also type this formula into the equation box and avoid the ‘select’ and move routine just described.
4.10 Now click on Data, Select If, and If case satisfies condition
4.11 Select bfnow from the variable list and move it using the arrow. After the bfnow type =1 (bfnow=1). Now add the & symbol and then select the bfuji variable from the variable list moving it with the arrow. Then type an =0 with no spaces. Add another & and select and move the bfmilk variable over. Then type =0. Note that you can type this information instead of using the select and move routine. The box should now read ‘bfnow=1&bfuji=0&bfmilk=0’
4.11 Click Continue and then OK
4.12 To dichotomize the bfwater variable (yes/no), click on Transform, Recode, Into Same Variable.
4.13 Move bfwater from the variable list to the right and click on Old and New Values
4.14 Under Old value type 0 and tab to New value and type 0, then click on Add
4.15 Go back to Old values, select range and type 1 in the first range box and 4 in the second (1  4) and type 1 in the new value box, then click on Add
4.16 Click on Continue and then OK. (Note dichotomizing the variable will be important for future analysis and to maintain consistency with the variable labels that you created earlier).
To create a crosstabulation follow the instructions under number one and substitute bfwater under columns. Your results should be:
EXERCISE for CATEGORY 5 Exclusive Breastfeeding: The final category should thus consist of those 15 children less than 24 months who do not fit into categories 14. This should be made up of those children who are given breast milk and nothing else (exclusively breastfed) and will be category 5. Although by default all remaining children must fall into this category, follow the computing steps to create a variable for exclusive breastfeeding and to double check your work.
5.1 Again will need to create a new variable by highlighting the ammennow variable and clicking on Data, Insert variable (this will place the new variable next to your bfwater variable)
5.2 Highlight the new variable column by clicking on the top of the column
5.3 Select Data,Define variable and type bfxcl in the cell for Variable name
5.4 Click on Labels and type bfxcl under Variable label
5.5 Type 0 under Value and No under Value Label and Add. Tab down to Value and type 1, then tab to Value Label and type Yes and click Add
5.6 Click Continue and then OK.
5.7 Highlight the variable you have just created (bfxcl) and click on Transform, Compute
5.8 Under Target variable type bfxcl, type 1as Numerical Expression.
5.9 Now click on Select if, and If condition is satisfied:
5.10 Find bfnow from the variable list and move it using the arrow. After the bfnow type =1 with no spaces in between (bfnow=1). Now add & and then select the bfuji variable from the variable list moving it with the arrow. Then type an =0 with no spaces. Add another & and select and move the bfmilk variable over and type =0. Finally type & again and then move the bfwater over and type =0. Note again that you can type this information instead of using the select and move routine. The box should now read ‘bfnow=1&bfuji=0&bfmilk=0&bfwater=0’.
5.11 Click Continue and then OK.
5.12 To dichotomize the bfxcl variable (yes/no), click on Transform, Recode, Into DIFFERENT Variables
5.13 Move bfxcl from the variable list to the right and type bfxcl2 and click on Old and New Values
5.14 Under Old value click system missing and tab to New value and type 0, then click on Add
5.15 Go back to Old values, select value 1 and in the New value also select the value 1. So that now, all of those that were not coded as exclusively breastfed are now coded as NOT exclusively breastfed (or 0).
Note that this variable is already dichotomized with the values 1=yes and 0=no. To create a crosstabulation follow the instructions under number one and substitute bfxcl2 under columns. Your results should be:
EXERCISE for CREATING NEW VARIABLE 'CATEGORY' Using the Output created in the previous steps, create a feeding category using percentage in each feeding category within each age group. For each age group simply divide the number in each breastfeeding category by the total number in that age group. Remember that the denominator for this calculation is listed as the total for each age group in the first output, currently breastfeeding (yes/no). By creating a new variable 'Category' and placing and running crosstabs, we can determine the breakdown of each age group by feeding practice with ease.6. 1 Click on the Data menu, Define Variable. In the Variable Name box type: category.
6.2 Click on Transform, Compute.
6.3 Under Target Variable, type category. In the Numerical Expression box type 1.
6.4 Now click on Select If, and If condition is satisfied:
6.5 Find bfnow form the variable list and move it using the arrow. After bfnow type =0. Click Continue, then o.k.
6.6 When you are asked to Change existing variable? Click o.k.
6.7 For the rest of the feeding variables you will follow the same Transform, Compute steps with the target variable being category. For Numerical Expressions you will type: 2: For Select If bfuji =1
3: For Select If bfmilk = 1
4: For Select If bfwater = 1
5: For Select If bfxcl2 = 16.8 Note For each new category, click o.k. when asked Change existing variable?
6.9 Click Statistics, Summaries, Crosstabs. In the row type cfage, in the column type category.
6.10 Click the Cells box. Under Counts, make sure the Observed box is checked. Click on the Row box under the Percentage box. This will compute percentages of children in each feeding category.
The output is shown here:
If you choose to create your summary in Microsoft Excel or another spreadsheet program, remember that you may enter the equation into the appropriate cell and the software will do the calculation for you. To doublecheck you calculations you may add each row across (the sum for each age group) to ensure it equals 100%.
There are several ways to present this information graphically:
Open Microsoft Excel. Enter the information from your analysis exactly as it appears here.
exclusively bf
bf plus uji (porridge)
bf plus milk
bf plus water
not bf
0 to 2
31.00%
10.30%
3.40%
51.70%
3.40%
3 to 5
4.50%
70.50%
11.40%
11.40%
2.30%
6 to 8
4.30%
78.30%
13.00%
4.30%
0.00%
9 to 11
0.00%
87.50%
10.00%
2.50%
0.00%
12 to 14
2.10%
79.20%
6.30%
4.20%
8.30%
15 to 17
0.00%
85.70%
5.70%
2.90%
5.70%
18 to 20
0.00%
69.70%
3.00%
0.00%
27.30%
21 to 23
3.40%
82.80%
3.40%
0.00%
10.30%
24
0.00%
66.70%
0.00%
0.00%
33.33%
Highlight all cells containing information and select chart from the insert menu. Select the Area chart and next. Follow instructions to preview. Continue to click next until you reach the chart title and axis labels screen. Enter appropriate headings (e.g. Age Group in Months for Xaxis and percentage for Yaxis). Click finish.
Your child feeding profile should be similar to the one above. Another option is to hand draw the graph using graph paper.
ANALYSIS
Compare your results to the recommended child feeding practices profile Which is a higher priority for a child feeding intervention, exclusive breastfeeding to 46 months or timely introduction of complementary foods at 6 months (separate, but critical child feeding practices)? Are a large enough percentage of children in the last age category continuing to breastfeed? What interventions might be proposed to address the highest priorities in terms of child feeding practices?