Patient identification is still proving to be quite an issue with the records that Health Workers are submitting, making it difficult to be sure that a record for a follow up visit is attached to the right patient.
As not everyone has an id number we can use, originally we asked health workers to identify patients by the id number they enter in their log book. This, we thought, had the advantage that we could easily then match up the database records to the paper records. The combination of this and the health post name (selected from a text list in the form, but stored as code number in the database) should have given us a unique identification for each patient. Only the first registration form contains the full name, on the other visit forms, we just ask for the health post name and the id number, plus the year of birth and age to use as checks for the data.
Using the year of birth and age checks we can identify where patient ids may have been entered incorrectly, but we are seeing quite a lot of errors. In theory rectifying these errors shouldn’t be very time consuming or difficult, assuming that follow up calls to the HEWs are made soon after the error is made. Unfortunately, delays to following up these errors, mean that now it will be quite difficult to fix all the errors.
On each patient visit form, we recently also added the patient first name, as an aid to matching errors back to their correct registration forms.
Some of the problems we have come across include:
- Two (or more) patients being registered with the same ID number
- Patient visit forms being entered with the wrong ID – and so getting matched to the wrong patient registration record
- Patients being re-registered with a new id number, especially when they may attend a visit at a different health post, or in a health centre. HEWs issue a registration card to each patient when they are first registered. If the patient later visits a different facility, the health post name and id from the card should be used, but seems this is not always happening and patients are getting re-registered. This makes it very difficult to track whether patients are following up on referral advice.
- Some health post have restarted the numbering in their log books (the new year in the Ethiopian calendar started in September), so we are starting to see the same id number being re-used for new patients (although this wasn’t meant to happen)
Given the lack of reliable identification numbers, it was probably inevitable that we would have experienced some errors with correctly matching records up. I would have hoped that with quick follow up to rectify errors, the health workers would have soon got used to taking extra care when entering patient id information.
There are other options we could have taken for patient identification, but these may have also had their own drawbacks. For example:
- pre-registering all patients in a given area – though this seems like substantial work; or
- providing a set of pre-generated bar codes or numbers (with check digits), which the HEWs can issue when the see a new patient. A check digit mechanism, would have really helped ensure mistakes in entering id numbers were minimised – though it may not have avoided the same numbers being reused for different patients. In retrospect I think this is the approach we should have taken.
Another factor which may have contributed to this problem is that we’re forcing ODK to do something that it probably wasn’t really designed for. ODK is a general data collection tool, each form is an independent entity, not necessarily designed to link up records entered from different forms. Some other recent mHealth tools, have a front-end so the user needs to click on a particular patient to enter the a visit record. But this requires some form of synchronization of the data between the phone and the main database, to ensure that all the patients a health worker may visit have their records already stored on the phone, otherwise it may lead (again) to patients being re-registered.
In Ethiopia, there seem to be some efforts to resolve this identification issue, for example the national Health Management Information System (HMIS) or Family Folder system, but these aren’t fully rolled out to all the health posts we’re working in, so we wouldn’t be able to take advantage of these. It seems to me that for these types of mHealth tools to work well and generate good quality reliable data, then a reliable and consistent system for patient identification is required, but hopefully this will be coming soon in Ethiopia.