Diabetic Retinopathy Screening - Look Into My Eyes

It ain’t hard to tell I got money
You can see it if you’re blind
You can see it when it shine
You see it then it’s mine
Always on my hustle, always on my grind (screening) 24/7, 365
— Wiz Khalifa in "Look Into My Eyes" or an Opthamologist

We are truly blessed to be living at this exact moment in time. No, not this generation or century but this exact week. Well at least until Sunday. That is because humanity will be able to witness the conclusion of Breaking Bad - the best TV show that ever graced our multimedia-saturated minds. I can't help but think about Walt and his journey this week as it all ends in what will hopefully be a fitting conclusion for such an amazing character. If you haven't watched the show let me introduce you briefly to Walt.

He's a middle aged man who was formerly a brilliant scientist but one who was eventually relegated to teaching high school chemistry. He finds out one day he has lung cancer and immediately contemplates his inevitable departure from this earth. In doing this, he looks to find a way to take care of his family's financial future. This leads to cooking Methamphetamine and eventually creating an empire with his scientific prowess while ruining some other lives in the process. But it was all for the good of his family.

That's kind of how screening goes in medicine. 

The general public, and probably the majority of physicians, "shotgun" screen patients because they genuinely believe that it is doing some good. Grabbing a TSH, CBC, and BMP never hurt anyone, right? But for a healthy 31 year old female, there wouldn't really be a need for any of that. Those few tests add up too, kind of like the small lies that Walt told that eventually forced him into bigger, more dangerous lies. If we screen people for no good reason we're just wasting money that could be put to actual good use. It's no secret that there are specialist physician groups that advocate for more aggressive screening and it just so happens the screening would involve them seeing more patients than they normally would. Again, I'm not saying specialist organizations are a bunch of Dr. Evils, but when you're living in arena called the American Health Care System it can lead to a dog eat dog mentality. 

That long winded intro leads me to my question this week: 

How often do I need to screen my type II diabetic patients for retinopathy?

Current guidelines for diabetic retinopathy (DR) screening are a bit nebulous. The American Diabetic Association (ADA) recommends yearly screening initially then essentially gives the power to the Opthalmologist to determine the frequency. It would theoretically be based on the presence or absence of retinopathy or signs that it may be developing. Unfortunately, the USPSTF doesn't have any official stance on this topic so the mainstay recommendation that people typically follow is the ADA's. The question is, how many of us (family physicians) actually adhere to this? I for one will admit that retinopathy checks are often the last on my to-do lists for my diabetic patients. Right after med management, diet, exercise, more med management, microalbumin/creatinine, pneumovax, foot checks, and more med management.

I clearly haven't learned my lesson yet as I still thought "simple question must have simple answer, right?" The answer to this question may be the most complex yet and it starts with me having to define two things that were and still are very confusing to me.  The first is quality-adjusted life years (QALY). QALY is essentially a number that is assigned to the quality of life one leads over a year. For instance, 1.0 equals perfect health and 0 equals death. Everything else falls in between. The lovely example Wikipedia uses is 0.5 equaling being bedridden. 

The second thing that needs to be clarified is utility value. Utility value is very similar to QALY in that it tries to objectively measure the impact of a disease process on a patient's ability to function in their everyday life. Again, the 1 and 0 values for QALY also apply to utility value as well, it's just not used in terms of years.

So now that we are armed with that information, let's jump into the data. The data on topic essentially boils down to a debate between two doctors and a couple of incidence studies. Let's start with the back and forth between Drs. Javitt and Vijan. Dr. Javitt was the first to publish back in 1994 in Diabetes Care [1]. His study was essentially a cost analysis of diabetic retinopathy screening for America. Using his own model and theoretical population based on epidemiological studies and clinical trials he came to the conclusion that America would save a lot of money if they did the "recommended" opthalmologic care (yearly screening) for diabetics. Close to $500 million to be exact. In addition, he argued this would save 94,000+ person-years of sight. 


Great! Looks like we'll save a bunch of money. But wait, Dr. Vijan comes by in 2000 and really starts beefin' with Javitt and his work. He publishes a paper in JAMA that showed that there really wasn't a big difference between yearly screening versus every 3 years. [2] He sighted some examples saying that a 45 year old diabetic patient with an A1c of 11% gained on average 21 days of sight when screened annually versus every 3 years. Also, a 65 year old patient with an A1c of 7% would gain only three days of sight. Finally, he showed that the low risk group (75yo, A1c 7%) would cost $211,570 per QALY gained while the high risk group (45yo, A1c 11%) would cost $40,530 per QALY gained. Extrapolating to the presumed US population he found it cost $107,510 per QALY gained for annual versus every other year screening and $49,760 per QALY for every other year versus every three years.

That's a lot of numbers.

To put things a bit more into perspective the "accepted" definition of a cost-effective intervention is at $50,000 per QALY gained. So based on that we could see that annual screening is only cost effective in the high-risk group as well as  if you were to stretch out screening to every 3 years. In the end, Dr. Vijan discussed how he thought having tailored screening recommendations such as annual for high risk and every 2-3 years for low risk might be beneficial. He also pointed out that some of the aggressive screening recommendations might be overzealous since controlling for risk factors like hypertension has been shown to substantially reduce retinopathy risk. And hey, we're Family Medicine physicians, if there's one thing we know it's hypertension.

Pretty convincing eh? At least I thought so.


But back comes Dr. Javitt as he is not one to be down for the count. In that same year, Dr. Javitt posted a response in JAMA, that kind of like his initial recommendations, were a bit aggressive. [3] Some of the main arguments were that Dr. Vijan's study used the NHANES III population which only had a single photograph of each eye (sensitivity for retinopathy, 60%) versus the gold standard 7-field photography. He also chided him for not using the "internationally adopted simulation mode for testing public health interventions" which was surprise... a model Dr. Javitt created in 1997 (Diabetes Care). The best argument in my opinion is that Dr. Vijan used a utility value of 0.69 for vision loss instead of the 0.48 that Dr. Javitt used. What's funny is that most of the data COMPLETELY REVOLVES AROUND THESE TWO NUMBERS.

If you used 0.48 as your value for vision loss (close to complete blindness), then annual screening becomes much more effective. However, if you use 0.69 (typically a value used for vision < 20/200) then the opposite is true. So it basically comes down to what you consider to be vision loss: the legal definition of blindness or actual blindness. 

After I saw that, I essentially gave up as I quickly realized there were way too many subjective moving pieces (utility value, cost of an eye exam/medicare reimbursement, what is high risk or low risk, the best methods for eye exam, etc) to adequately analyze this argument anymore.  

I'm tired and the three of you still reading this are likely as well. So let me wrap this up real quick. 

I looked at two incidence studies, one from the UK and one from Wisconsin, that give a bit more clinical insight into this dilemma. The UK study [4] in 2003 looked at 20,570 screening events in type II diabetic patients who were seen at general practitioner offices. They followed the incidence of sight threatening (moderate preproliferative) diabetic retinopathy out to 5 years. In patients with no history of retinopathy, the incidence was 0.3% (95% CI 0.1-0.5) in year one and a cumulative incidence of 3.9% (CI 2.8-5.0) at year five. To have a 95% probability to remain free of any sight threatening retinopathy they found the mean screening intervals to be 5.4 years (95% CI 4.7-6.3) for no history of retinopathy, 1 year (95% CI 0.7-1.3) for a history of any retinopathy findings and 0.3 years (95% CI 0.2-0.5) for those already with sight threatening findings.

The Wisconsin study look at 3 different populations stratified by age (< or > 30 yo) and whether they were already using insulin. They then looked at incidence over 10 years and the progression of retinopathy. Unfortunately, UW's journal database becomes stingier by the day and I didn't have access to the full article. Reportedly (via UpToDate author), there was no progression to proliferative retinopathy over four years in patients with no baseline retinopathy. Take it for what you will as I can't speak much more to that.


So what did I learn? There's no such thing as an easy answer anymore with this blog. This was definitely one of the more excruciating dives into literature so far. I was also reminded (which I am daily) that there are people much more intelligent than myself who can talk about theoretical models and calculations all day. I know it seemed like I was taking shots at Dr. Javitt but really it was all in good fun. Any of these authors could talk circles around me all day and I can't help if I have a non-treatment bias. Plus, I have to make up for the likely unnecessary MRI and Echo I ordered this week (see Maureen! :D).

What I am taking away from my own practice though is that I will probably adopt a tiered system. My higher risk (<45 yo, insulin dependent) patients will likely be referred for yearly screening while my lower risk patients will be referred for one every 2-3 years. The incidence data I discussed is good to make you feel better when you forget to screen a patient after 4-5 years, but not something that will sway me enough to push out the intervals even further. Finally, if I was forced to choose a utility value (again, this is highly subjective) for myself, it would be 0.69 as I consider 20/200 to be vision loss in my book. 

TL;DR: No right or wrong answer - personally will screen high risk patients yearly and low risk patients every 2-3 years. 

Now get hyped with the Breaking Bad finale trailer below as all bad things come to an end this Sunday (Warning: spoilers contained if you aren't up to date).