DHIS2 Organisation Units

Every value in DHIS2 belongs to an organisation unit, and every organisation unit sits in one hierarchy. Data is entered at the bottom and added up towards the top, so the hierarchy has to be right before anything else is built.

1. 1Level 1, national. One unit, the root. Everything hangs from it.
2. 2Level 2, district. Data from the chiefdoms under it is added up here.
3. 3Level 3, chiefdom. The same again, one level down.
4. 4Level 4, facility. Where data is entered. Nothing is entered above this level.

1. 1The tab. The orange line shows where you are.
2. 2Side menu. The same items as the cards.
3. 3Organisation unit. The units and the hierarchy itself.
4. 4Organisation unit group. Groups of units, such as every CHC.
5. 5Group set. Holds the groups, for example Facility type.
6. 6Organisation unit level. Where the levels are named, in Step 6.

The first unit has no parent. Everything else will sit under it, so give it the name of the country or the programme area.

1. 1Parent. Empty for the root. For every other unit, click its parent in this tree first.
2. 2Name (*). Sierra Leone.
3. 3Short name (*). Used in reports and charts.
4. 4Code. Optional, but useful when data comes from another system.
5. 5Opening date (*). Must be earlier than any data you will enter.
6. 6Coordinates. Leave empty for the root.
7. 7SAVE. Creates the unit.

1. 1Name. Sierra Leone.
2. 2Short name. The same, since it is already short.
3. 3Opening date. 1970-01-01 is the usual choice, so no data is ever rejected for being too early.

Every unit except the root needs a parent. Click the parent in the tree, then fill in the form.

1. 1Parent. Sierra Leone is selected, so Bo will sit under it.
2. 2Name. Bo.
3. 3Short name. Bo.
4. 4Code. SL_BO, so the district is easy to match later.
5. 5Opening date. The same date as the root, unless the district opened later.

Repeat the same two clicks one level down. The facility is where coordinates matter, because a facility without them never appears on a map.

1. 1Parent. Badjia, the chiefdom.
2. 2Name. Ngelehun CHC.
3. 3Code. A stable code, often the national facility code.
4. 4Opening date. When the facility opened. Data before this date is rejected.
5. 5Coordinates. Longitude first, then latitude, in square brackets.

The levels exist as soon as units are created, but they have no names until you give them some. Those names appear in every analysis app, so do it now.

1. 1Organisation unit level. The side menu item.
2. 2The list. One row per level that exists.
3. 3Level number. 1 at the top, counting down the hierarchy.
4. 4Name. National, District, Chiefdom, Facility.
5. 5Offline levels. How many levels the tree loads at once. 1 is fine.

A group holds units of the same kind, for example every CHC. Groups are what let a chart show only health centres, or only hospitals.

1. 1Name. CHC.
2. 2Short name. CHC.
3. 3Code. OUG_CHC.
4. 4Available units. Everything not yet in the group.
5. 5Selected units. The facilities of this type.
6. 6SAVE. Creates the group.

A group set is the question the groups answer. Facility type is one question, ownership is another. A unit belongs to one group per group set.

1. 1Name. Facility type.
2. 2Short name. The same.
3. 3Compulsory. Ticked, so every facility must belong to one of the groups.
4. 4Data dimension. Ticked, so charts can be split by facility type.
5. 5The groups. CHC, CHP, MCHP and Hospital.
6. 6SAVE. Creates the group set.

Creating units by hand is fine for a handful. For a district with two hundred facilities it is not, so the rest of this guide imports them from a file.

The importer reads the columns by position, so the order matters and the first row is ignored. Only name is required, but parent is what builds the hierarchy.

1. 1name. The full name of the unit.
2. 2parent. The UID of the unit above it.
3. 3coordinates. Longitude first, in square brackets, for facilities.
4. 4The header row. Ignored on import, but keep it so the columns stay in order.
5. 5One row per unit. And one file per level, which is the next step.

1. 1Format. CSV.
2. 2Import type. Metadata.
3. 3CSV object type. Organisation unit. Getting this wrong is the most common mistake.
4. 4Identifier. UID, because the parent column holds UIDs.
5. 5Dry run. Ticked. Nothing is written, but everything is checked.
6. 6The file. One level at a time: districts, then chiefdoms, then facilities.
7. 7IMPORT. Runs the check.

A dry run either passes or tells you exactly what is wrong.

1. 1Status. OK, dry run. Nothing was written yet.
2. 2Created. How many units would be created.
3. 3Total. It should match the number of rows in your file.

1. 1Status. ERROR. Nothing was written.
2. 2Ignored. The rows that could not be used.
3. 3The messages. A missing parent UID, or a missing opening date. Fix the file and run it again.

A parent must exist before its children, so the files go in from the top down. After each import, copy the new UIDs into the parent column of the next file.

1. 1Dry run. Untick it only after the dry run passed.
2. 2The file. The next level down.
3. 3IMPORT. Writes the units.

The tree is the proof. Every facility should sit under a chiefdom, and every chiefdom under a district.

1. 1The root. One only.
2. 2Districts. Directly under the root.
3. 3Chiefdoms. Under their district.
4. 4Facilities. Under their chiefdom, where data is entered.
5. 5What next. Assign the data set to these facilities, as in the data set topic.