If you absolutely need to trigger a handler every time then here are two options:
1) run a noop shell command which will always report as changed
- name: trigger nginx-restart
2) use debug along with changed_when: to trigger a handler
- debug: msg="trigger nginx-restart"
Also of note for Option 1 and Check Mode: You may want to use
check_mode: no if using Ansible version 2.2 or higher or
always_run: yes if using earlier versions than that so that the task does not get skipped over in check mode. From my manual testing it looks like the handlers remain in check mode, but please be careful as your case may differ.