Tuesday, September 23, 2014

Getting Performance Management Right--Let the robots do the work!

The amount of public health a taxpayer dollar can buy ultimately depends on how efficiently and effectively services are delivered at the local level. But it is really hard to get 3,000 local health departments to behave themselves. For this reason, public health is engaged in a major effort to improve its ability to manage it own performance. The Public Health Foundation, the Public Health Accreditation Board and the CDC--among many other public health partners--have performance management initiatives. A common goal of all these initiatives is to encourage local health departments to engage in performance management.

I worry that if this effort isn't done right, it could be counterproductive. What if a national performance management effort makes 3,000 local health departments waste their time on busy-work projects dreamed up at the federal level? Because becoming more effective is so important, performance management is part of the strategic plan of the Mid-Michigan District Health Department. But in order to make sure we are not wasting our time, and are really becoming more efficient, we take what is, perhaps, a unique approach.

First, let me say that we have already made a lot of progress building our performance management system and I am very proud of the high level of staff engagement in the process. One reason for our success is that we have a firm foundation. Many of the programs we operate already contain elements of performance management. The MiWIC and MCIR databases, for example, enable WIC and immunization program managers to easily run revealing reports. Michigan's minimum program requirements for Family Planning, Children's Special Healthcare Services, and Hearing and Vision Screening, among others, all contain elements of performance management. Many Environmental Health programs are moving in this direction. Another reason we are having success is that we have a dedicated a full time position (Ross Pope) to performance management.

But, as we work to extend and deepen performance management, we have to make many difficult choices that will have long-term consequences. The two most important choices are "What to measure?" and "How to measure?" MMDHD's work is guided by a clear theory of how performance management works that tells us what and how to measure. Its two principles are:
  1. Only measure something if the information will change your behavior. Another way of saying this is "Measure just enough to get what you need and then stop".
  2. Automate everything. If it takes a lot of staff time to collect information it probably isn't worth it. Our motto is "Let the robots do the work."
What you really want to measure are key indicators of the efficiency of your own workflows. How quickly are clients being served? What proportion of clients are getting services that meet your gold standard? What proportion of billable services are actually being collected? The answers to such questions have immediate relevance to what your staff are doing now.

A performance management system delivers information to people. You want the acquisition of the information to be effortless. And the information should be so relevant to what they are doing that it changes their behavior almost instantly. Consider two really basic and funny examples:

When you are driving in your car, you have on the dashboard in front of you information about the performance of your vehicle--the speedometer and fuel gauge, for example. You glance at the dashboard frequently and change what you are doing in response to the information. If the speed limit is 65 and your speedometer says 75, you take your foot off the gas just a little (why some people don't is a topic for another day). What if, instead of displaying your speed or the amount of gas in your tank, the dashboard displayed the national traffic fatality rate for the current year compared to previous years, or information about automobile manufacturers' recalls of defective vehicles? The fatality rate and recalls are very important information. But they change so slowly you would soon stop looking at the dashboard. Information on fatalities and recalls wouldn't tell you anything about the performance of your own vehicle. That information is irrelevant.

What if you didn't have a speedometer in your car but you wanted to know your speed? One thing you could do is use a stopwatch to get the time between mile markers and calculate your speed that way. If the stopwatch told you it took you a minute to go from one mile marker to the next, you would probably realize you were going 60 miles per hour. What if the stop watch said 55 seconds? You would probably realize you were going faster than 60 miles an hour. But are you going faster than the speed limit of 65? Could you do the math in your head? Now you are flying down the road a mile a minute, fiddling with devices and not concentrating on what you are doing. The effort of calculating your speed has made you guilty of distracted driving. The attempt to measure your performance has actually degraded your performance.

I have seen plenty of examples of public health agencies pushing in the wrong direction on performance management. The National Association of County and City Health Officials (NACCHO) recently gave a webinar on performance management that featured a prominent local health department that explained that it took 7 FTEs for them to maintain their performance management system. Many of the things they measured were nothing other than things they should have been measuring as part of routine surveillance anyway, like mortality and morbidity data. They had not identified things that would change the way they did their day-to-day operations. In my mind, this system seemed to have huge costs and questionable benefits.

Consider this: the Public Health Accreditation Board requires local health departments to engage in performance management. But it doesn't define what that means! If a health department has a highly distracting and uninformative operation it calls performance management, would they be accreditable?  Shouldn't we insist that a performance management system actually improve performance?

Let me end by bragging about the MMDHD team's recent achievement. This is the report you can see in the graphic above. The report consists of a variety of productivity data on a couple of dozen different programs, which is available for any period of time at the push of a button. You can see caseloads blooming or tanking, failures to bill, etc. instantly. No one creates these reports. The work of setting up the programming has been done, and now the report generates itself from the data on our server on demand. Clinic staff regularly consult this report to plan their work. Next we need to extend this level of automation to our environmental health and administrative functions. Let the robots do the work!