 Okay, so the session objectives are as I just mentioned to understand the key principles associated with data quality and define what a validation rule in the HIS 2 is. Understand how validation rules can be used to measure consistency in the HIS 2. Execute validation rule review in data entry. Execute validation rule review in bulk via validation analysis. This will be the first part of the session today, then we'll discuss on how validation rules can be compared against calculated thresholds. And then execute these analysis using the validation rules with defined thresholds, followed by sending out notifications for these validation rules. So based on the WHO data quality guidelines, these are the four components of data quality. Completeness, timeliness, consistency. So completeness and timeliness of data is once we submit how complete is our data which is submitted and how timely did we submit the data. For this session we'll discuss more on the consistency where in validation rules will help. So consistency can be either internal consistency of reported data, wherein we compare the data reported within the data entry form or within the HIS 2. So say two fields in the same form we are comparing them. External consistency could be when we compare our data with external sources of data. Example, we import the survey data in the HIS 2 and we enter data and then we are comparing both the figures that will be external consistency. Or we could also compare with population estimates. We would review the denominators mostly for the immunization coverage indicators that we have we would review what are our denominators what is our population data and then validate the values that we are entering. So we'll go through each of these type of validations in our session. This also allows for these various aspects of data quality to be checked. In some cases, additional work may be required to ensure available data is necessary for comparison. So maybe for some cases for external consistency we might have to prepare external data and import it in the HIS 2 so that we could compare for external consistency. There might be an additional work to be done if we have to compare it through survey data or some other application data. The HIS 2 is not a substitute for every possible data quality check that can be performed. However, it can enhance the quality of data that is being entered substantially by using the built-in features to review the data prior to analysis interpretation dissemination and feedback. So all the time types of data quality check that could be done by big statistical software, but it does ensure that in a way where does ensure that the good quality data is being available in the system. These are the four components of data quality, completeness, correctness, timeliness and consistency. So completeness reports to understand how complete the data is in the system, correctness to understand if the data is precise or not, timeliness reports to understand if data is being received late or on time. And consistency which we focus on today, validation rules which compare actual values against other data elements or thresholds or survey data or other source data. So validation rules consist of a left side value, a right side value and an operator. So as you see this example, we are comparing malaria cases treated to suspected malaria cases. So malaria cases treated is our left side and suspected malaria cases is our right side. Now the cases treated could either be same as the cases suspected or less than that because you might not read all of those cases. So our validation rule would be defined in a way which says malaria cases treated, our operator here would be less than or equal to suspected malaria cases. So the validation rules tell us what should be true. They'll say that in a normal scenario, in an ideal situation, malaria cases treated would be less than or equal to suspected malaria cases. We will not create the rule otherwise just to say that rotavirus doses administered are greater than rotavirus outline. No. Our rule should be what should be the exactly true situation in the field. If you think about validation rules, we could actually consider different examples. So there could be logical rules. These are basically when we enter the data and the data and reform to measure internal consistency. These are logical rules like cases treated from malaria should be less than or equal to suspected malaria cases. Or we could compare rules using a threshold value. So here in we take the component of thresholds or population data or outliers or estimated figures. So then the example here is rotavirus to administer should be less than or equal to rotavirus threshold. If we are comparing a threshold for rotavirus doses, so a vaccine which is given to children, if we have a threshold value available based on our previous data, we can set in those thresholds in the system and then compare our values against those. So when we know that this is a threshold, this is the maximum value for a vaccine to be given. And we can compare and say that the number of vaccines given to children should either be less than or equal to the threshold value. This will be further clarified while I take few examples and run through the application, but this is just an introduction to the concept that we will be using. Now these validation rules that we create, we can use the data from aggregate data entry, which I'll just demonstrate. From the event capture or from the tracker capture. So we could say that from the tracker capture we are aggregating doses administered to individual child, and then we've set in a threshold value in aggregate. So we could compare those two figures aggregated data from tracker could be compared to the threshold values, and we can run our validation rules. And there are these ways of evaluating these rules, we can either view them in the data entry, which I'll show you after this, or we could schedule them and have them run on an automatic schedule, which we can define, or we can go to validation analysis and trigger these rules ourselves. So for today's demo, I'll show you how we can run these validation rules in data entry and how we can trigger them using the validation rule analysis. So I'll stop here to show these features for common logical rules. The example one that you see here, I'll just show you the logical rules few examples for logical validation rules. And then we move to the threshold rules after that. I'm sure you are using slack for raising your queries but then if there is some issue that we need to our question that we need to discuss maybe you could ask that before I start the demo. If not, then I think I'll start right so once I log into the the tries to I can go to data entry to see how this validation rules could be reviewed. So here in the data entry, I will select one facility, maybe the Cardinal Hospital HIV data set. So this is my data entry form to run a validation I have two options. I could either you see on the top right corner there is an option called run validation. You could either click on validation here to run a validation check. Or in the bottom of the screen you see an option here and validation. You would either use this run validation to run your validations. Or you see that we have a complete and incomplete button here. So these also trigger validations so complete button actually triggers two things one is it also checks for validation alerts and displays them as well as it also captures the date when you are completing the data set. So this will help you in both timeliness and the validation of the data which is entered so then the timeliness and consistency both can be measured through the complete button. Okay, let's come back to run validation. Now for this data entry form if I run validation. So here on this pop up you see a validation error result. Explains to us there's a description which tells us what is the validation error or validation rule which was there. And what are the left side operator and right side. So let's look at this rule. It says HIV test positive male plus HIV test positive female should either be less than or equal to HIV test performed male plus HIV test performed female. So to simplify it says that HIV test positive cases could either be equal to or less than the total test that have been performed for HIV. So what we see is our left side which is the positive cases is 467 while the test performed is only 273 so which not ideally be possible. So we can close this go back to our data entry and check where the issue is. So if you look carefully in the data entry form we have HIV test performed and HIV test positive. So as we see male and female test are 99174 while positive cases are 123 and 344. So let's try to correct this data. Maybe I'll change 50 and maybe this was 34 and by mistake I went to 344. So after correcting this data if I run validation again. So then it says that the data entry screen successfully passed validation. So this is the first step of checking the validations wherein we write when we enter the data we could run validation and check for any validation issues in the data, which is basically checking for internal consistency in my data. This check could be done at data entry or unit by facility by facility, but maybe someone at a higher level at a district level wants to see the validation checks for entire facilities under him. Or maybe I at Cardinal Hospital I want to see validation checks for like last six months or last year. The option is I go on each period and then run validation and an easier option could be to run validations in bulk, which can be done through data quality. So there's an app called data quality. I'll go to this app. So during our practice sessions we have each of these steps clearly mentioned so don't worry if you think that it's not very clear we'll take one more example and then we'll practice together during the hands on. session. So here in the validation rule analysis here we need to go to run validation. And maybe I could just select my district and select last one month. And I only run the validations for HIV. So it says it passed successfully maybe let me take a. And now I see in the month of March, I have this validation rule which was broken which was showing an error to see the details further I can click on this icon on the end. And here you see the details. So this is my validation rule HIV test positive should be less than equal to performed. And on my left side positive HIV female was 300 male was 899 while test performed were only these these details will help me understand in detail what was the issue with the validation. Any questions or do you want me to repeat something maybe my co facilitators could help me know in case a section is to be repeated. If not, then let's look at one more validation rule. We could take an example of a rule which compares a lot of data fields to each other. So I'll go back to my data entry form. And this time I take an example of immunization for the last month. Try running the validation now. Let's take this first example, DPT hepatitis B and HIV so DPT doses given should be less than total used and wasted. So it says that the total vaccines which are used or either wasted is 188 while the doses which are given is 191. So this basically compares the doses administered to the stock which was available at the facility. And of course if you had this talk of just 188 vaccines, you cannot administer 191 vaccines to the children. So let's look at the data entry in detail. So this is our DPT so this rule then what it does is it adds up all these values. So DPT doses given to under one over one, DPT two, DPT three all these doses given. So they all sum up to I think what our rule said was 191. So this total for DPT 80 plus 59 plus 52 is 191. While we go to the stock section at the bottom. This one. It says I had an opening balance of 1124 vaccines and closing balance of 936 vaccines which means I used only 188 vaccines. So I still have 936 vaccines in my stock and only 188 I used. So if I just use 188 vaccines I cannot administer 191 vaccines. So maybe we can say we had an opening balance of 1130 and there was an error here. Now try to run the validation. So that validation no longer exists. I can further correct this one. I think this is some other comparison which is being done some other validation with someone has created. And then we could just correct our values and this gets removed. Let me check which is the other rule which is causing the issue. I think there's some other validation which is an error. Yeah, so this way we could just compare so this validation will actually helps you to look at multiple values and compare them to the another value which is also a combination of multiple values. Tomorrow we look at how these complicated validation rules could be set up in the application. Now just to revise we go back to our data quality and try to run the same validation analysis there. So in the data quality validation rule analysis I go to run validation and then I select and I just select one desert district over last three months maybe because we have some data there only run for immunization and validate it. We have so many of these errors and if we look at the details this will show us that these are the validations that were these were the doses for hepatitis B which were given. And this is the calculation that we've used for the ending balance. Yeah, and then we could look at all of these and accordingly take the actions in the data entry by correcting this data. I think I'll just stop here if there are any questions or else we could do a little hands on for running this validation ourselves. If you go to the page and then you go to the exercises download you should be able to download the. Not an exercise but how you can run through the same steps that I've just done and see if you are able to run these validations review them and correct them and rerun the data.