 part of the sessions on the Data Quality Online Academy. So, some questions about this app. Have you seen it? Have you used it? Do all the analytic users have access to this app in your country? Are you sure? Go and check. So some of the learning objectives, we will look at what is the most common use of the Data Quality app, the value of validation rules, how to translate validation rules into actual use, and some issues associated with the Data Quality app. There will be no DHR is online training of how to move your cursor here and what to select there. You can use the manual. So the DQ app contains the following. It allows you to do a validation rule analysis, a standard deviation analysis, a min-max outlier analysis, and a follow-up analysis. So, let's talk about validation rules to start. Who's responsible? But talk about validation rules sort of get everybody looking away from me. And the feeling is, well, you know, they're not important so it doesn't matter. Nobody runs them anyway. And if they do and you find something, nobody does anything. You can't change anything. So why bother to do this? If we start with validations, we start asking the questions about who's responsible for thinking about them and then defining them. Is the person or the people or the group, is it the HMISDHH2 configuration expert? Is it a program manager? Is it a district information officer? Who has to do the thinking and the defining and the configuration? Was it done by one person who takes the responsibility? Then we ask, well, who's responsible for running these rules? The data entry clock, the program manager, facility or district information officer or higher level who's responsible for running them. If we want to understand the logic of validation rules, we need a method. It is a method to assess the quality of data entered in DHHISD. And these rules are based on a set of predefined rules set for the data. Unfortunately, the validation rules do not allow you to fully figure out whether or not the values reported are completely accurate. It is only based on what is in DHHISD2. And if a value is 25, you don't know what the 25 clients they saw and they transferred from the client record onto the tally sheet onto the input form and they captured 25. You are just given the 25. So how do we define a validation rule? We use the term an expression. An expression is a relationship between a number of data elements that talk to each other. This validation rule expression has a left side, a right side, and in the middle we have an operator. And so some of the common operator use is less than or equal to, equal to or greater than and equal to. So some of the operators that we see most commonly. Validation rules are created on base what you know is true. Malaria oddity tested cannot be more than malaria oddity positive. Because in order to be oddity positive, you had to be tested. Another example is live birth weighing less than 2.5 kg cannot cannot be more than live births. Validation rules are created based on what you know is true. For instructions on how to set up a validation rule, read the manual. They actually very good and we don't read them often enough. Some simple guides to validation rules. Validation rules are assigned into groups. This is for analysis. Validation rules need to be easy to use and easy to understand. We use these common operators less than equal to or greater than one. Are we sure that everybody understands what these signs mean and which way the error points? There's also the importance field of low, medium and high. Have we decided how to assign our validation rules into the importance field or do we just select the default? Validation rules also need to be consistent. In other words, the way that they set up in written always have the smaller number first. It is easier to understand. It's intuitively easier and it makes it easier to grasp the concept. Validation rules should be run after data entry before the complete button is ticked. Validation rules should also be run after most or all of the monthly data has been captured. So then there are some questions about who's responsible for this. The facility information officer, the district information officer, the program officer, the program manager. Do you have a standard operating procedure to guide the process of running these rules and taking action? Validation rules, how to define them. A validation rules is an expression that defines a relationship between a number of data elements. An expression has a left side, a right side and then in the middle an operator. So some of the common operators used are less than or equal to, equal to or greater than and equal to. Validation rules are created based on what you know to be true. Malaria RDT positive cannot be more than malaria RDT tested. For instructions on how to set up a validation rule, look in the DHA's two manuals. It's actually very good. So how to run a validation rules. You select an org unit. You select a time period. You select your group and you click validate. Depending on the size of your database and the number of rules to be run, be careful in what you select. Try and avoid crashing the system. Once you have run the validation rules, you get a report which can be downloaded into different formats. Excel is the easiest. So this is the output once you have run the report. So on the left you start off with the name of the facility, the time period, the importance. Remember we discussed who decides low, medium or high. Then the validation rule and in this output is blank. Then what is the left value? What was the operator and the value on the right side? And then there is a detail icon. And if you double click on the detail icon, you will get the rule explained to you. So this rule says that child, children five to nine years old with acute respiratory illness should be greater than or equal to child five to nine years with pneumonia. So the left hand side, child with symptoms of acute respiratory illness five to nine years, there were no children seen. And the right hand side of the no children seen, 11 were diagnosed with pneumonia. When you see something like this where the is a missing value and then a value, you need to determine should the 11 have been the acute respiratory and the pneumonia. So you actually need to go and find out why is there missing value and what is the correct response to how do you fix this validation rule. And there are many ways of going about sorting this out. If you get a list which you download into Excel, this is what the report will look like. Facility, the time period, the rule, the left side, the value, the operator, the value on the right side and the right side description. And if you've downloaded it into Excel, you can sort it to according to organization units, and you can give to supervisors and program managers for action. So this is how you sort out your validation rules. Then the other services available in the data quality app, the standard deviation analysis. Now we mainly use the WHODQ tool for extreme outliers and missing data. The min-max outlier analysis, this you can only use if you have set min-max values and you need to set them first and the next presentation will go through that process. And the last aspect in the data quality app is the follow-up analysis. It's used to mark data that is correct but does not fit the pattern or triggers a validation rule, and we will have a slide about that. This is marked for follow-up. When you double click a data value, you will get a data information window. And top left-hand corner, you have a space for a comment, and if you click on the star, you can save it. You highlight the star and you save. And this comment in the star is useful to explain when a validation rule has been triggered. It explains why the data doesn't look as you wanted to look. And then you can run the report on the follow-up stars. So, some discussions. What are the barriers to running validation rules? Have you looked at the data and set the most possible validation rules? Have you done everything that could be done? Do validation rules pick up all the data mistakes? We've spoken about that earlier. When obvious data entry errors are identified, why are they not corrected? Who is responsible for chasing up the data entry errors? What happens when these errors are not corrected? What is the result? So, here are some exercises associated. So, confirm access to the data quality app in the country DHA's two instances. Obtain a list of all validation rules and the groups that they fit into. You may have to ask the IT people to help you. Review all your validation rules. Review them for consistency in terms of the smaller value first. We've spoken about that. Then I want you to run at least two validation rule groups for the last six months. Write a short report on the rules that are violated. As part of your short report, consider an action plan. Do some rules need to be rewritten to make them easier to understand? That could be part of your action plan. Are some rules being triggered consistently? And is that a problem of data quality? Or is that a problem with the rule? And then I want you to think of one rule that you can add to the current list. Thank you.