Viewing entries tagged

How does case-mix drive staffing ratings?

Changes in patient acuity change the case-mix or “expected” hours used to determine staffing stars. That’s not news for anyone who reads this blog. The problem is the relationship is difficult to tease out. This is because CMS does not publish the exact criteria they use to select assessments that are included in calculating acuity and don’t give us any documentation showing the ones they used.

This leads to questions like “If my case-mix increased by X, how much pressure would that put on my staffing rating?” This is a common question for anyone who is considering implementing a program to improve documentation. ADL coding is a classic example of this. Improved ADL documentation nearly always increases case-mix, which increases the number of hours required to maintain or improve staffing ratings. (For the record, never let anything get in the way of improving your documentation. Better, more complete documentation should always be the goal, regardless of star ratings or quality metrics.)

Let’s see if we can create a surrogate measure.

Estimating Overall Case-Mix Hours

For this analysis we combined the detailed Medicaid assessment data from the state of North Carolina with the latest data extract from Nursing Home Compare. Once those were linked up we just use a simple linear regression analysis like so:

What that chart is telling you is that there appears to be a relationship Medicaid case-mix index and overall nursing case-mix (or required hours). This makes intuitive sense because at least some of the Medicaid assessments are used to calculate case-mix hours. (Or even all, who really knows?)

Specifically, we are explaining slightly less than 60% of the variation in the data using just Medicaid case-mix. This isn’t too bad if you consider in North Carolina we use a “point in time date”, often called a “picture date” in other states. CMS uses a much more complex criteria to select assessments to include.

The slope of our regression line is 1.307. This means that for every 1 point change in Medicaid case-mix, the overall nursing hours expected by CMS for staffing ratings increases by 1.3 hours per patient per day. If you happen to be close to the lower cut-point for staffing ratings and your acuity increases, you could drop a star. (As luck would have it, we have a great tool for checking how close you are to the next staffing cut point. Check out Staffing Navigator™. It’s free!)

How do I use this?

Using this estimate, we can figure out the break even point between case-mix “slope” and hourly nursing pay. Bear with me a moment:

First we can use the CMI “slope” to estimate the change in Medicaid pay like this: (I am using North Carolina data as well as the 34 grouper for this experiment. I am ignoring changes in Medicare Part A.)

Here is the equation to calculate daily change in revenue for Medicaid for a given change in case-mix:

CodeCogsEqn (7).gif

Here’s the equation for the change in nursing expense for a given change in case-mix index using our slope from the regression analysis:

CodeCogsEqn (8).gif

To find the break-even point we just set those two equations equal and solve. We end up with this beauty. Fee free to admire it’s elegant simplicity. I’ll wait. (This will set the mood as you take it in.)

CodeCogsEqn (9).gif

So if you believe your case-mix “slope” is $77 per point, divide that by 1.307 and you get $58.91. That means that as long as your fully loaded expense to hire and pay nursing staff (ALL nursing staff) is less than $58.91 per hour per nurse/assistant, an increase in case-mix index will cover the cost of increased nursing hours. (This is an oversimplification since nursing hours can’t typically be adjusted in fine increments, very large facilities excepted.)

If the average fully loaded pay of all nursing staff in your facility is greater than $58.91 per hour, you have my sympathy but with all due respect, check your math. Additionally, $77 is probably low. The take-away: Increased acuity (either real or due to improved documentation) may indeed increase the need for nursing staff to maintain a star rating, but increased pay from Medicaid should always cover it.

The sub-take-away: Always be closing improving documentation!

P.S. You can get an exact number for slope by examining your reimbursement report. It may be as high as 100. If you aren’t comfortable with that, let me know. I will help you.

Want to talk? Let’s.

Staffing Navigator™ - A Case Study

In today’s posting I am going to explain how to use Staffing Navigator™ to diagnose and troubleshoot a staffing issue that is causing problems in a 5 Star Rating. This post will also explain staffing cut points which can be confusing.

If you haven’t already, download Staffing Navigator™. It’s free in the app stores.

Hypothetical Problem Statement

Let’s say we work with a facility that bounces between 4 and 5 stars on Nursing Home Compare. It’s become a problem for one reason: This is a facility that markets itself on being 5 stars. They can’t have the rating occasionally slipping to 4 stars.

This is what most of us would consider to be a nice problem to have but it’s still a problem to solve. Let’s use Staffing Navigator™ to troubleshoot.

First, we select the facility from the list and check out the current situation. Right away we can see a few things:

Figure 1

Figure 1

  • The survey is quite good at 4 stars. Getting that survey to 5 stars would be one way to solve the problem.

  • Neither Quality nor Staffing are high enough to get an extra star, but staffing might be close. We’ll need to investigate further.

  • This building has average size and acuity, so changing the staffing rating should be reasonable to do. (I’m basing this on the SEI. See the app for details.)

(If you aren’t familiar with the rules, touch the word “Staffing” for documentation that includes the rules and a handy table showing how RN and overall ratings translate into staffing stars.)

For the sake of our example, let’s say the facility recently had a survey that is going to be with us for a while. So while improving the survey rating is the most important thing to do, it isn’t the most expedient.

Figure 2

Figure 2

Next, scroll down to the Adjusted Hours section. I find this part of the app is the most confusing for people so let’s spend a little time here.

The hours you report on your PBJ are not used directly to determine your staffing ratings. The hours are first adjusted based on the acuity of your residents and the national average number of nursing hours. After those adjustments are made, then the cut points are applied and you see a rating. That’s why your adjusted hours don’t match reported hours. Adjusted hours, not reported hours, drive ratings.

Second, we need to understand CPP. CPP is telling us our position between the cut points. If CPP is 99%, that means you are right on the cusp of the next higher staffing rating. Likewise, a CPP of 1% means you are on the edge of moving down a rating. A CPP means of 50% you are exactly between cut points.

This is important because even when your PBJ doesn’t change at all, the number or residents in your building and their acuity does. Remember we use adjusted hours not reported hours. If your CPP is either very low or very high you will find staffing ratings bouncing around from quarter to quarter.

Figure 3

Figure 3

Let’s take a look at our adjusted hours again. I’ve highlighted the RN CPP of 92.0%. This means we are very close to the next higher rating for RNs.

If we take a look at the star rating table, we can see that we’re currently at an RN staffing rating of 2. Given our CPP of 92%, we would occasionally expect to drift into an RN rating of 3, which would give us 4 overall staffing stars and trigger an additional overall star. (Again, see the rules below the table in Figure 4 in the app.) We’ve identified a possible cause of the problem.

Figure 4

Figure 4

Now that we understand what’s happening, we can start to make some recommendations. The first thing we want to do is select an RN target rating of 3. After we do that, we check the results.

Figure 5

Figure 5

You can see in figure 5 that the optimizer is trying to reduce LPN hours to get to the least expensive way to get an RN rating of 3 and a total rating of 5. Since we’re just trying to work on RN hours, this isn’t really what we want. We want the optimizer to leave the LPNs and CNAs alone and let us focus on the RNs.

To do this, touch the “Minimum Hours” button at the bottom of the app and then touch the “Use Current” button and then touch “Back to Detail”. This tells the optimizer “Don’t make any changes unless I change a target.”

Back at the detail, we can see that’s exactly what’s happened. (See Figure 6.) The optimizer is telling us we need to add 10% of a full time RN to all shifts to get to 3 RN stars. But we aren’t quite done.

Figure 6

Figure 6

Looking at the adjusted hours, notice that the CPP is 0.0%. If we’re trying to protect against fluctuations in acuity and caseload, this isn’t going to help. We need to push away from the cut points.

Figure 7

Figure 7

Fortunately, this is easy to do. We need to tell the optimizer we want more margin. Touch the “Minimum Hours” button and check the “CPP Target”. It’s set to 0% which means the optimizer will stop exactly when it hits the cut point. This isn’t what we want. In this case we want plenty of margin so we’ll choose 50%. That would place us exactly between the cut points.

Figure 8

Figure 8

Notice in Figure 8 that we now have an optimized CPP of 49.8%. That’s plenty of margin. (Also notice that the CPP target of 50% caused the overall hours to be increased as well. If we want 50% margin for RN, it doesn’t make sense to keep the CPP for total nursing hours at 7.6% so both are increased.)

Figure 9

Figure 9

Here are the final optimized hours. (Figure 9) We now safely have a 3 RN rating and a 5 total rating which gives us 4 staffing stars. (Is this the cheapest way to get a 4 star staffing rating? Let’s call that homework. Staffing Navigator™ can easily do that too.)

Figure 10

Figure 10

If you’d like additional help with Staffing Navigator™, or quality measures, or rehab obviously, then contact us!

Introducing: Staffing Navigator™!

Today we’re introducing a new software tool: Staffing Navigator™! This tool is intended to help operators understand how acuity, number of residents, nursing rates and mix (RN, LPN and CNA) work together to determine your staffing stars. Additionally you can use Staffing Navigator™ to:

  • Determine the ideal mix of RN, LPN and CNA to achieve any star level

  • See how close you currently are to the next highest (and lowest) star

  • Understand how the acuity and caseload of ANY facility compares to any other

  • See how many additional (or fewer) nursing hours it would take to hit any staffing level

  • Run complex nursing staffing scenarios

  • See acuity, number of residents and staffing levels for any facility in the US!

In the remainder of this posting I am going to explain in detail how to use this tool. But before I do that, let me just say: This is an advanced tool intended for users who understand staffing planning, star ratings, hourly and fully loaded nursing rates. It is a tool we use to help people understand what it would take to improve star ratings. You may need additional training to effectively use this tool. (Let me know if you’d like to do some private training using this new tool.)

If you’re still interested, great! Let’s get started. First things first, download Staffing Navigator™ from your app store. (Yes, it is free!, Spread the word!) Click one of the links to the right to install it.

Get it on Google Play

A tour of Staffing Navigator™


Now that you’ve installed and before I explain how to use it, let’s take a tour of the interface.

Once you have selected a facility, you will see the screen to the right. Note the letters by each section.

Current Ratings (A Left)

The current ratings are taken directly from the Nursing Home Compare website. This information is for reference and does not change as you work on different scenarios.

SEI (A Right)

SEI is a metric developed by Broad River Rehab. It tells you how difficult it will be for this facility to change star ratings compared to all other facilities in the country. An SEI of 50.0 would mean that half the SNFs in the country are more costly to change staffing star ratings and half are less.

SEI takes into account both acuity and number of patients. The higher either of these, the more expensive it will be to increase your staffing star rating. You can touch the box for more information on SEI.

Targets (B)

We’ll cover the targets much more in depth later. For now, just understand that when you first select a facility, these will both be set to you current staffing ratings. Note that the small star(s) under the dropdowns tell you what your staffing star rating would be if your staffing ratings matched the dropdowns.

Labor Hours (C)

The labor hours section shows you the current actual labor hours per patient per day for the three nurse types. The next two columns show you the optimized hours and the difference between optimized and current hours. (Much more on optimization later.) The last column shows you the increase or decrease in staff to get to the optimized level of staffing.

Labor Expense (D)

Labor Expense (highlighted) shows your current labor expense based on actual hours reported and the rates you have configured. (More on rates later.) Like the labor hours section, this area shows you the difference between your actual hours and optimized hours.

Adjusted Hours (E)

The adjusted hours (not highlighted) show you current and optimized adjusted hours as well as how close you are to the cut points. If CPP equals 50%, that means you are exactly centered between the cut points for the current staffing level. Likewise, if CPP is 99%, you are very close to the next star rating.

Minimum Hours & Rates (F)

These will both make more sense to discuss as we talk about optimization, but the navigation buttons are at the bottom of the screen. For now we’ll just say that you can configure rates and minimum hours to do many different types of analysis.

Let’s move on to actually using Staffing Navigator™!


How to use Staffing Navigator™

Okay, let’s get to some usage examples. Select a facility on your app and follow along. I would only caution you that selecting a one-star overall facility can be misleading: if the survey score is one then the overall is going to be one regardless of other factors. Likewise, facilities with an RN rating of one can be confusing as well. Make sure you have a good understanding of the star rating rules before you study edge cases.


I’m using this facility for my example. It’s got 2 overall stars with a two rating across the board for staffing. The SEI is 89.7 which means this building has high acuity, a relatively high number of residents and it’s going to be expensive to increase the star rating.


Scrolling down, we can see the optimizer is suggesting changes to all three nursing types. Although we haven’t changed the RN and overall staffing targets, the hours can are still optimized for the current star rating, based on the minimum nursing hours and the rates, both of which can be configured to your needs.

The rates are straightforward and easy to understand. By default the rates are set to national average rates from the Bureau of Labor Statistics. More on these later.

The minimum nursing hours are a little more complicated. CMS publishes something called case-mix hours which are essentially the hours CMS expects to see given the reported acuity of the residents in the building. (These used to be called expected hours but have recently been changed to case-mix hours.)

By default, Staffing Navigator™ uses these case-mix hours as the minimum hours for each nursing type. You can see in my example facility both RN and CNA hours are less then than the case-mix hours so both were increased. LPN hours are higher than case-mix targets so LPN hours have been reduced. (I’m oversimplifying here. There are actually a few more things going on, but you get the idea.)

You can easily change the minimum hours to the actual hours, which will cause the optimization not to reduce any hours. Use this if you don’t want to reduce hours or staffing. Just keep in mind that you aren’t really optimizing at that point, but the tool will do whatever you ask. This technique is useful if you want to simply reach the next star level and you aren’t as concerned about the cost.

You can also reduce the minimum hours to zero, but beware: star ratings favor RNs, and CNA rates are typically lower than LPNs. If you set the minimums to zero you will not get any LPN hours. This isn’t a bug, it’s just the natural outcome of star rating rules and the economics of nursing.


Based on this optimization, you could save $329.94 per day while maintaining a two star rating.

Back to the rates: the default rates are national averages and are not fully loaded. I highly recommend you update these rates for your market and use fully loaded rates if you plan on using the labor expenses for more than just differential comparison.

Below that, you can see that initially the RN hours were 5.6% away from the lower cut point for 2 stars. A small upward change in overall acuity or an increase in residents will cause RN hours to fall below that cut point and result in an RN staffing rating of one which would drop this facility to one staffing star. This is a serious concern.

After optimization, you can see we have much greater room for fluctuations in acuity and occupancy. You will find this isn’t always the case when doing optimizations however, as we’ll see next.

Simulating an Increase in Staffing Star Rating

So for my example facility you can see we have two overall stars. For a lot of markets, having less than 3 overall stars will have a negative effect on Part A admissions so getting one extra overall star could be very important. We can get a bonus star by getting our staffing stars up to four and we want to know how much that would cost. This would help answer the question: “Should we focus on increasing staffing or improving quality measures?” (Note that for this building, an additional quality star would result in a bonus star as well.)


Our first hint should be the SEI of 89.7. That tells us right away that this is going to be expensive. The next thing to consider is what actions do we want to take. If you touch the blue circle or the section title “Adjusted Hours” you’ll see the cut point table. I’ve highlighted both where we are currently as well as three potential place we could go to get 4 staffing stars.

All we have to do now is change the dropdowns and compare the results. This type of analysis is called differential analysis: we don’t care about the exact dollar amounts, just which on is lowest. If we decide we’re actually going through with this, we would need fully loaded rates and we’d want to check facility acuity and caseload over time to see if this data accurately represents normal running conditions or is this quarter an outlier. In other words, this tool is pointing you in a direction only. Due diligence is important.

Scenario RN Total Change in Expense
1 ★★★ ★★★★★ $2,362.05
2 ★★★★ ★★★ $1,330.37
3 ★★★★★ ★★ $2,729.77

You can see that scenario 2 is the least expensive. Let’s take a closer look at the results.

We chose an RN rating of 4 and an overall rating of 3. You can see the optimized staffing changes to the right. We’d need to hire about 60% of a full time CNA for all shifts and a challenging 2.4 full time RNs for every shift. (You don’t need to spread those RN hours over all shifts, you just need the hours at some point during the day.)

You can see the optimization reduced the LPN headcount by nearly 1.5, which offsets the cost somewhat.

Speaking of cost, this is by no means inexpensive. Using national average rates, (again, not fully loaded) we’re looking at an increase of $485K annually.

Remember that SEI of 89.7?

Lastly, look at the adjusted hours. Notice that the optimizer hit the nursing target with minimal expense, which means you are very close to the lower cut point. If you were planning to implement this change, be aware that these are the minimum required hours to get the star rating. There is no margin for error built in.

Other Analysis

With a little practice and thought, there are hundreds of different scenarios you can do with this tool. You can also request private training sessions or custom analysis of your facilities.

PDPM, Staffing & STRIVE

There is a problem on the horizon. Under PDPM, the staffing study used to determine how many nursing hours to expect for a given RUG level will not longer work. That study, called STRIVE, was originally done between 2005 and 2009. The data from that study is used to estimate how many nursing hours (RN, LPN and aide) to expect for a given RUG level.

The staffing portion of the 5 star rating system depends on expected nursing hours, now called case-mix hours, to scale actual nursing hours by patient acuity prior to assigning stars. (See this post for more detail.)

We’ve contacted CMS to ask how this is going to be handled. As of 4/1/2019, the answer was “We haven’t decided.” CMS has a lot on its plate with the changes to the 5 star system rolling out this month and PDPM happening on October 1st.

A Proposal

One approach CMS could take is to simply mirror the idea used in the nursing portion of PDPM. They could simply combine the nursing RUGs and average the nursing times and use those.


  • It’s easy to understand and implement.

  • It doesn’t required major overhaul of the star rating system.

  • It’s cheap and fast


  • It isn’t really rigorous. It could be argued that this is a leap of faith.

I will focus on the con for a moment. While this idea isn’t statistically rigorous, I would argue that the original STRIVE study has some results that raise eyebrows anyway. I am humbly suggesting that the STRIVE study isn’t perfect. That isn’t an excuse to make decisions that aren’t supported by data, but we need to be pragmatic here. Another STRIVE study would be hugely expensive and there isn’t time prior to PDPM anyway.

Also, PDPM already combines these nursing categories. I haven’t seen any justification for this other than to reduce the overwhelming number of patient classifications. Since we’ve already made that decision, this feels like the natural way to go.

What would it look like?

I simply took the required minutes for the RUGs that have been combined and averaged them. All other RUGs are left unchanged. (Note: I am using unweighted averages here. I don’t have the data to do weighting. I would strongly suggest weighting these.)

This image shows the expected overall nursing hours by RUG for all of PDPM. I have overlaid the distribution for total nursing hours for every home in the country.

Averaging the nursing hours (click to enlarge)

How would it change my star rating?

That depends on the acuity at your facility, but the answer is probably not much. (Warning: I am now speculating.) If you watch the animation a few times, you’ll see that overall most RUGs get pulled towards the center of the existing distribution. Unless you have extremely high or low acuity, you’re probably likely to get pulled towards the center, slightly. (Remember: higher acuity will either put negative pressure on your staffing star rating or cause you to spend more in staffing for the same rating. Read the last two sections of this posting for an explanation.) More speculation: Your normal variation in acuity is probably greater than the change you’d see from this proposal.


Perfect is the enemy of good. Although this is clearly a compromise, it’s probably good enough. It definitely isn’t conceptually any more difficult than what we’re doing pre April 2019 change, and only requires a comparatively small leap of faith.

There are much bigger issues to work on, even in 5 star. (Transparency around which assessments are included in the 5 star rating would be an example.) Let’s just fix this and move on.

If you want Broad River Rehab to analyze your staffing and help you figure out how to get that next star, contact me. I’d love to hear from you.

5 Star Staffing Ratings

We’re getting a lot of inquiries about 5 star ratings. It’s understandable because of the changes coming in April. One question we’re getting asked is “What would it take to improve my staffing rating?” or “How can I get another star?”

In this post we’re going to take a look at that, using the new staffing cut points and the latest data from the nursing home compare data. If you are unfamiliar with how those work or the April update, check out this post.

As our example, let’s assume you own a nursing home and your star ratings look like this:

Category Rating
RN Staffing

Note: This is an actual facility I have chosen as an illustration. I have no affiliation with this building which I will keep anonymous. The data is real.

As you can see, overall this facility is 3 stars. It received one bonus star due to excellent quality metrics but did not get an extra star from staffing. (4 overall staffing stars would give another bonus star in this case.) So a natural question would be: How many more nurses would it take to generate that bonus star and get this building to 4 overall stars? (We’re ignoring the elephant in the room regarding survey scores.)

The Details

To answer our question we need to know a few things. First off we need to know that stars are assigned based on adjusted hours using the following table:

Cut Points as of April 2019 (click to enlarge)

Next we need to know is where we stand. Our example facility has a adjusted RN hours of 0.574 and overall adjusted nursing hours of 3.467. Looking that up in the table, you can see we have 3 RN stars and 2 total nursing which nets 3 stars for staffing.

Now we need to understand how adjusted hours are calculated.

CodeCogsEqn (1).png

Adjusted hours are simply scaled by the national average hours and the expected hours. (More on both of those later.)

Looking back at the table, there are several ways we can get 4 overall stars. We can keep RN stars at 3 and get overall nursing hours way up. We could run RN hours way up and keep overall nursing hours the same, or even lower. Or we could shoot for 4 RN stars and 4 total nursing stars. That seems like the logical choice: we’ll have more margin when any type of nursing hours fluctuate. (Keep in mind that case-mix hours are moving as well.)

So we now have an objective: raise the adjusted RN hours to ≥ 0.724 AND raise the overall adjusted nursing hours to ≥ 4.038.

Let’s add some additional requirements:

  • Aide, LPN and RN hours should at least meet or exceed the case-mix hours. (These used to be called expected hours so you can see why we’d like to get as close to those as possible.)

  • Minimize the cost. Cost is the reason there aren’t more nurses in facilities anyway so we might as well optimize to get the maximum 5 star rating we can get for our dollar.

For our example facility I am assuming that the hourly costs per nurse type are as follows. Keep in mind that the exact numbers don’t matter much, just the relationship. (Aide hours cost less than LPN hours which cost less than RN hours.)

NurseHourly Rate

The Disclosure

To solve this problem I am going to use a technique called linear programming. I’m not going through all the details here because that’s not the point of this post.

The Results

NurseOriginal HoursOptimized HoursFTE ChangeDaily

You can see in an attempt to minimize the expense, the optimization reduced the number of LPNs and increased CNAs, because under the star rating system, LPNs and CNAs are the same. (Obviously that’s not true in the facility, just the star ratings.)

Okay, at first glance you’re probably thinking: “$2,408.74 per day?!” The short answer is “yes”. It costs a lot to increase staffing. It’s also the future with PDPM. Check out this article.

Why so much?

One thing I said I would come back to is the “case-mix hours”. Remember this equation?

CodeCogsEqn (1).png

The case-mix hours in the denominator represent the acuity of the patients in your facility. In fact, those hours are useful to compare one facility to any other in the country. You can’t know exactly what RUGs were billed but you can be sure that higher case-mix hours mean more nursing hours are expected. (Acuity, in this case, is based on the STRIVE Study which maps nursing hours to the RUGS IV 66 grouper. It’s actually pretty interesting and a little controversial. Maybe that’s a topic for another time.)

Case-mix hours or “expected” number of total nursing hours per day for all skilled nursing facilities

Case-mix hours or “expected” number of total nursing hours per day for all skilled nursing facilities

This is a histogram of the expected overall nursing hours for every facility in the US. Notice the tight distribution.

One thing I didn’t share earlier about our example facility is the overall case-mix hours are 4.12158. Take a moment to find 4.12 on that histogram. Our example facility is in the 99th percentile for case-mix, or expected hours. (That means our example facility has higher acuity that 99.1% of the facilities in the US.)

If you refer to the equation from earlier, you can see that higher case-mix hours lower your adjusted hours, meaning you need more hours for a given star rating if you have higher average acuity. This is important. Read the part in bold again. Acuity is a moving target and changes each time new star ratings are released, but if your building has a tendency to trend higher, you’ll pay for it in either a lower star rating or higher labor costs.

How much higher? In this case, if our example facility were in the 50th percentile for acuity instead of the 99th, it would become a 5 star staffing facility immediately. In fact they could actually have fewer RNs and LPNs on average (with slightly more CNAs) and still be five stars for staffing. The bottom line is that if this facility were in the 50th percentile, they could potentially save $3,016.93 per day in labor! That’s a $5,425.67 daily difference from our optimized plan to get 4 stars.

Taken further, if this facility were in the 1st percentile instead of the 99th, the potential labor savings would be $5,371.08 per day or $1.96M annually.

While you can’t just change acuity, do keep in mind the effect it is having on your star rating.

Final Thoughts

One more thing I didn’t mention is your adjusted hours are also affected by the national average numbers. That means that if the national averages were to rise for some reason, like say, PDPM, then your adjusted hours would go up without you making any changes. Don’t say CMS never gave you anything.

Contact us today if you would like to analyze your staffing star ratings.