There are 6 current MU EP CQM’s that have an exclusion for palliative care:
CMS 61, 64, 69, 149, 159, and 160.
They handle the exclusion in this manner:

CMS 61: "Procedure, Order: Palliative Care" during "Measurement Period"
CMS 64: "Procedure, Order: Palliative Care" during "Measurement Period"
CMS 69: "Procedure, Order: Palliative Care" starts before end of "Occurrence A of Encounter, Performed: BMI Encounter Code Set"
CMS 149: "Intervention, Performed: Palliative Care" overlaps Occurrence A of $DEMEncounters149
CMS 159: Intervention, Order: Palliative Care" <1 year(s) starts before end of "Measurement Period"
CMS 160: "Intervention, Order: Palliative Care" satisfies all >= 4 month(s) starts after start of "Measurement Period" < 8 month(s) starts after start of "Measurement Period"

To put these variants in non-technical language, from least restrictive to most restrictive:
CMS 61, 64, and 159 all require an order for palliative care to be placed at any time during the measurement period.
CMS 160 requires the order to be placed between the 5th and 8th month of the measurement period.
CMS 69 requires the order to be placed before or during the index encounter.
CMS 149 requires the provision of palliative care, not just an order, and requires that the provision span the index encounter.

In an EHR, it is unlikely that a patient will have an order in the record of the form “Commence palliative care”. There may be an order for “Referral to hospice”, but a referral to hospice and acceptance to hospice care or the actual provision of care is not the same thing. The patient may have a change in the resuscitation status to “allow natural death”, but a change in the status or an order for a change in status is not the same as an order for palliative care and will not be counted. You might then wonder “OK, what is a good way to determine that a patient is in palliative care” and in fact there is no consistent and accepted way to do so. One good option is to look for the encounter code “Z51.5” from ICD-10, but this does not appear to be covered by the “Intervention, Order” or “Procedure, Order” constructs.
Identifying patients receiving palliative care is going to be an inexact science, and we encourage measure developers to be as liberal as possible to accept false positives and minimize false negatives, maximizing the number of exclusions for patients who might be in palliative care. We also encourage measure developers to harmonize existing measures to define a single exclusion definition.