single bar chart to include bars of multiple indicator variables

[kmstephenson]

I have a wide dataset that includes many chronic condition indicators, and I want to see all of these on the same bar chart.

For example:

Patient_ID        Start_Dt          End_Dt            chf_ind    copd_ind     diabetes_ind        ckd_ind     ..... etc

        1                 1/1/2012      12/31/2012            1             0                        0                         0

        2                 1/1/2012      12/31/2012            0             0                        1                         1

        3                 1/1/2012      12/31/2012            1             1                        1                         1

What I did is keep this dataset with all indicator variables as the fact table and then transposed the data into 1 row per patient per time frame per condition (indicator variable=1).

For, example:

Patient_ID      Start_Dt          End_Dt           condition

        1             1/1/2012        12/31/2012          CHF

        2             1/1/2012         12/31/2012         DIABETES

        2             1/1/2012         12/31/2012         CKD

        3             1/1/2012         12/31/2012         CHF

        3             1/1/2012         12/31/2012         COPD

        3             1/1/2012         12/31/2012         DIABETES

I used the transposed variable condition for the bar chart to be able to see in descending order the frequency of conditions. However, if I select any condition within the bar chart, it does not show me what other conditions these patients have. For example, I want to be able to select the CHF bar and see that of patients with CHF, the other coexisting conditions they have are COPD and DIABETES. I had to create listboxes and buttons to select the indicator variable from the fact table to update the bar chart with this information.

Do you have any recommendations for another way of doing this? Where you have a number of indicator variables that you want to see on the same chart like this?

Thank you!

[mark_casselman]

Hi,

We did this a while ago. The trick is to make a data island with conditions, and select all customers that have that condition with the P()-function.

Script:

data:

LOAD * INLINE [

Patient_ID, Start_Dt , End_Dt , condition

1 , 1/1/2012 , 12/31/2012 , CHF

2 , 1/1/2012 , 12/31/2012 , DIABETES

2 , 1/1/2012 , 12/31/2012 , CKD

3 , 1/1/2012 , 12/31/2012 , CHF

3 , 1/1/2012 , 12/31/2012 , COPD

3 , 1/1/2012 , 12/31/2012 , DIABETES

];

condition_data_island:

LOAD distinct condition as selected_condition

resident data;

Chart (the expression is in the title):

Mark

[kmstephenson]

Hi Mark,

Could you provide the qvw of what you pasted in the response above? It isn't working for me so I wanted to double check a few things.

Thank you!

[mark_casselman · Accepted solution]

Hi,

Here it is.

Mark

[kmstephenson]

Okay, thanks! Do you know of any way to use this approach with a cycle group? For example, I happen to have 5 different ways that chronic conditions are measured (the condition indicators above are grouped into frailty conditions, conditions that result in high mortality rates, currently managed, hccs etc), so I was using a cycle group that had these different condition type fields.

[mark_casselman]

KM,

I'm not sure I understand your additional question. Do you still want te select a specific condition (like DIABETES) and only see the 'related' conditions from one group (like FRAILTY), or do you want to select a group of conditions and see all related conditions ? Perhaps, you could give me a more detailed example of data and expected result ?

Greets,

Mark

[kmstephenson]

Hi Mark,

From the example above, let's just say CHF and COPD belonged to one condition type, and DIABETES and CKD had other characteristics belonging to another condition type. I would have two separate data tables, one for each condition type. For example:

data_conditiontype1:

LOAD * INLINE

[ Patient_ID, Start_Dt , End_Dt , conditiontype1

1 , 1/1/2012 , 12/31/2012 , CHF

3 , 1/1/2012 , 12/31/2012 , CHF

3 , 1/1/2012 , 12/31/2012 , COPD

];

data_conditiontype2

LOAD * INLINE

[ Patient_ID, Start_Dt , End_Dt , conditiontype2

2 , 1/1/2012 , 12/31/2012 , DIABETES

2 , 1/1/2012 , 12/31/2012 , CKD

3 , 1/1/2012 , 12/31/2012 , DIABETES

];

Condition types 1 and two would be in the same cycle group. Does that make sense?

[mark_casselman]

Hi KM,

If these groups are relevant, i would add them to your data. You can add them in de data itself:

LOAD * INLINE

[ Patient_ID, Start_Dt , End_Dt , conditiontype, conditiongroup

1 , 1/1/2012 , 12/31/2012 , CHF, Group1

3 , 1/1/2012 , 12/31/2012 , CHF, Group1

3 , 1/1/2012 , 12/31/2012 , COPD, Group1

];

Then you can use the conditiongroup var directly in your charts or as a cyclic group. (extra remark : as you can see you should change the var names conditiontype1 and conditiontype2 in conditiontype so that conditiontypes of both tables end up in the same var in one table).

If you have a lot of data, you can also load a seperate table like this (QV will link the tables automatically):

LOAD * Inline [

conditiontype, conditiongroup

CHF, Group1

COPD,Group1

DIABETES, Group2

CKD,Group2

];

Mark

[kmstephenson]

Hi Mark,

What would the cycle group be? I want to be able to cycle through conditiongroup to see condition types all in one table like the original example you posted above. From my understanding, if conditiongroup was the cycle group, then the bar chart would contain only 1 bar for Group 1, and cycle through to 1 bar for Group 2, etc. I want all conditions for each condition group to be shown when that condition group is selected

[mark_casselman]

KM,

You were correct to put the condition_types in 2 different variables to enable cycle groups.

I added them to the example.

Ok ?

Mark