Thanks for the writeup! I work in the industry and have a few groups of questions that I'm very curious about. 

1. Usage over time
   1. How does usage trend over time? Do people continue to use the program?
      1. Every health tech product I've been exposed to has trouble getting off the ground in the first 2-6 weeks before seeing wider adoption.
   2. For the 3.3->1.8 hours a day time savings (is it per day or per shift?), does this pass your gut check? Since you presumably have access to how much time someone spent using the monitors (is it password protected by user or open for anyone?) it seems like an easy stat to pull. If for instance, people are spending ~1 hour a day on IMPALA we can assume that this survey result is false as they'd presumably spend >.8 hours a day responding to alerts from GOAL3 and following up with actual patients.
   3. Did any hospitals you implemented with have existing bedside monitors?
2. Did hospitals respond unexpectedly to the program?
   1. Have any sites reduced nurse staffing in the months to years following implementation? Or changed how and where they staffed nurses?
   2. For the main study, could the type of patients sent to the HDU have changed post live? That would explain the "observed crude mortality r[ising] slightly" and some of Martijn's post.
3. GOAL3 Tech questions
   1. Does IMPALA integrate with anything? HL7 FHIR is increasingly seeing adoption in Africa including the countries you listed. (this source is bad but makes the point) Without integration, you risk hitting an implementation wall, where IMPALA is a good fit for hospitals without any monitoring system but is difficult to manage for any hospital with a partial monitoring system or EHR. Have you run into this scenario?
   2. Can IMPALA track patients across admissions? This may help measure post-discharge mortality, but then you'd have to wade into unique patient identifiers and correctly linking them. Plus, twins.

Thank you again! I think your program is amazing and has a lot of potential.