 I'm going to present how you can quickly train staff at district and even facility level to make use every month of the WHO data quality tool. First I want to share with you a single slide that explains why data quality assurance should be decentralized and it should be routine. Let me define some of these terms. By data quality assurance I mean a continuous process of checking the quality of the data investigating where there are suspicious values and if necessary editing the data to remove any errors. By decentralized we mean that data quality assurance is best done at the level of the district and the health facility and this process should be routine that it should be done every month as data are entered into the system. This is a bottom-up approach to data quality assurance. It is more efficient for data quality problems to be fixed closer to where the data are first recorded and reported rather than wait for staff at higher levels regional or national to review an attempt to fix these data quality problems. This is for several reasons. First of all at the district level the volume of data to review is smaller especially if the data are reviewed every month rather than waiting months later to identify and investigate suspicious values. Secondly investigation of suspicious values is more practical for staff at a decentralized level. The district staff have access to the paper records. They have established relationships with the facility staff and they can more easily communicate with them to undertake investigations. And finally DHIS2 data belong to the district and district staff have the permission to edit them and remove any errors. So let's get started with a demonstration of how district and perhaps facility staff can be trained quickly to use the DHIS2 data quality review tools. First of all this user needs to understand how to launch the WHO data quality tool. Users at district level do not need to understand how to use all the functions of the WHO data quality tool. In fact they don't necessarily need to be instructed in how to use all of the four tabs of the dashboard. They can simply understand the power of the outliers page of the dashboard. They will first need a bit of orientation to the settings window of the dashboard. They should understand that they can make the settings window appear and disappear by clicking on the settings icon. For review of the quality of data at district level it's best to configure a special group of indicators that are most important for review at district level. So the data quality reviewer should set data to district group and they should set period to last month if the review is being done at the end of the month when you have data entered for last month. If the review is done at the beginning of the month then it's probably best to set the period to the month before last. Now how should the organization unit of the settings window be configured? If the user is at the level of the district then the organization unit is going to automatically be set to show the data just for their district. However if you are practicing the instruction of district staff at national level you'll have to configure the boundary of the organization unit to the specific district that you are supporting for data quality review. In this case we've selected district A1. Again at district level they won't have to go through this process because DHIS2 will automatically set the window to show data only for their district. The disaggregation should ideally be set to facility level. In the case of this DHIS2 instance facility is the only level below district so it's the disaggregation is automatically set to facility as soon as we set the district as soon as we set the boundary to a single district. But you may have two levels below the level of the district in which case the menu will show those two levels and you should select facility. By disaggregating to facility we'll see how it simplifies the process for the person doing the data quality review at district level. To free up space on the screen it's best to hide the settings window. Users at district level will need a bit of explanation to understand the table the result table that appears. They'll need to understand that all of the data are for their district and that each row represents data for one health facility in their district and one of the indicators in the district group. The data in the row are for the 12 months leading up to the period that has been selected. Users need to understand that values that are highlighted in red are suspicious and cells that are pale yellow have missing values. The results table is lengthy so users need to understand how to filter it in order to identify only the most suspicious values. And the filtering process involves as we have learned to click on options and then they can include the Z score in the table. Users need to understand that the Z score is simply a measure of how suspicious the value in red is with a higher Z score in this case the modified Z score showing that the value is more suspicious. We can now filter the table to show only values that have a modified Z score that is quite high by clicking on outliers and modified Z score and extreme. We now have a relatively small table of the most important suspicious values for the reviewer at district level to investigate. They should follow up on at least the first page of the filtered results table. For some of these values let's take for example the value that is highlighted in red in the first row and the user can be taught how to use the menu icon and click on visualize in order to see this chart. For for this value they may have an explanation. They may understand that this facility indeed had an increase in outpatient attendance. So not all values that are highlighted in red will require extensive investigation. But the point is that the data quality reviewer needs to understand why it is that that value is so different from the other values in the row. Again they can be taught how to click on the menu icon at the end of a row and visualize. And they will almost certainly see that some of the values that are highlighted in red are so suspicious that they are likely erroneous, that they are errors and deserve to be followed up with a phone call or some other communication to people at facility level. In addition to investigating the suspicious values the reviewer at district level needs to look into the most important missing values. And to use the outlier tab to identify those missing values they should be taught to turn off the filter, click on missing data and now the results table shows at the top. Those indicators and those health facilities which have the largest volume of missing data and it may be worthwhile for them to make attempts to mobilize these additional data. Contact in this case the district hospital and see if they can get these missing reports. After all the values from the district hospital are so high that if the reports are missing it will affect the total performance as measured at the district level. Finally the user at district level should understand how to use the download button so that they can download to their computer and visualize a spreadsheet that they can use even when they are not connected to the internet and they could print this out and make use of it in supervision. The user at district level may already understand how to use the sorting functions of Excel but you might need to refresh them that if they click on data and then sort this table can be sorted according to the values of one row. In this case let's select the maximum modified z score and largest to smallest and now we see at the top the values that are most suspicious or they might filter this to find the most important missing values sorting in this case by the modified by the gap weight largest to smallest and here again we see the facilities that are responsible for the most important missing data. In summary to instruct staff at district or facility level to use the tool in a short amount of time perhaps even during the supervision visit the instruction should focus on how they can quickly learn how to make use of the outlier page of the dashboard and how they can by screening the data routinely every month identify and investigate the most important missing and suspicious values and in this way maintain the quality of the data in a bottom up approach.