The Lone Dog: Quality control in Laser Vision Correction

The patient undergoing laser vision correction (LVC) has faith in his surgeon. The surgeon has faith in the excimer laser. For half a million dollars, shouldn't it work properly every time?

Let's say you are Lou “Big Pupil” Schmo, and your ophthalmologist, Joe Surgeon, is a capable, thoughtful and honest guy who collaborated with you on the key decisions in LVC; educated you about the general risks of the procedure and those additional risks specific to you; and programmed the laser's computer properly using your data, not that of the patient next in line. He also performed the surgery properly from his perspective. What is his perspective? If you have LASIK, he needs to make the corneal flap properly. For either LASIK or PRK, he must monitor the status of the stromal bed during the treatment and ensure that the laser is implementing its program while properly aligned on your eye. Joe then needs to reposition your flap (if made), and care for you properly after surgery.

Despite Honest Joe's excellent care, Lou's vision after LASIK is terrible. He now has a best-corrected visual acuity (i.e., the best visual acuity he can possibly attain) of 20/40, whereas before surgery it was 20/20. The starbursts at night prevent him from driving and the ghosting during the day prevents him from reading. His life is catastrophically disrupted and — like other LVC casualties — he's heading fast for acute situational depression (the kind that needs pills, not pep talks).

“Nothing,” reply some clearly unfazed surgeons (and/or their attorneys). “Those side effects are in the consent form. Lots of people get them. Don't be so fussy.”

The Lone Dog would like to bite the heinie of one of these charmers, or better yet arrange a cornea-swap between them and Lou (were such a thing was possible).

What sets us humans apart as a species is our ability to cognate, think, intellectualize, comprehend, analyze, and figure it out. So if you aren't thinking, you aren't human. Therefore anyone who tells Lou in the face of his manifest symptoms that nothing went wrong, and that he should have someone re-read the consent form to him, is a sub-human. To avoid sub-human rating or being likened to lawyers, surgeons should try to figure this out.

Lou was an excellent candidate. Also, he did a good job fixating on the blinking red light during the laser treatments. He was absolutely obedient to all of Joe's post-op instructions and nothing untoward happened during that time. As insiders say, “The patient was cooperative and compliant, and healing was uneventful.”

Joe did it right. The pre-op evaluation was thorough and accurate and the laser was properly programmed, with an optical zone diameter chosen to minimize Lou's risk of night vision complaints. The flap was clean, the stromal bed was dry, and the post-op care was top notch. As insiders say, “The procedure was uneventful and the post-operative care was appropriate.”

So now comes the time to “figure it out”. Hmm . . . Clearly it wasn't Joe, who has done thousands of essentially identical procedures without a Lou-like outcome. Probably it has something to do with Ingredient #1, the patient. Every human body is unique, and patients like Lou simple responded uniquely badly to LVC. It's a mystery. We're working on it.

But what about the Other Mysterious Ingredient, that Great Black Box, the excimer laser . . .

Facts about excimer lasers: they cost a bunch, or used to anyway; they are protected by one zillion patents apiece and only certified engineers can get anywhere near their guts; they have soothing calibration rituals that are conducted before each treatment and at various other times as well; and they run a bunch of self-diagnostics and turn themselves off if they're sick. They're always perfect. Really.

As industry reps and laser engineers say, “The laser ablation routine was performed according to specifications.”

Note however that no currently marketed laser can actually measure the effect of the laser pulses on the cornea during the ablation itself.

Laser vision correction is a multi-step process, one step of which is the actual laser treatment. The laser is supposed to ablate (vaporize or remove) a certain amount of the corneal stroma (the thick central tissue layer of the cornea) so that the curvature of the corneal surface is reshaped in a tightly specified fashion.

Ideally we would like to consistently and objectively measure the laser's performance during the actual treatment to be sure that the ablation is occurring as programmed. At the present time research is underway to develop systems for “real time” monitoring of the effect of the laser pulse energy on the shape of the cornea, but these are a long way from clinical use. Excimer laser energy is invisible and the changes in the shape of the cornea are too subtle to be visualized by the surgeon, even with high magnification. This places the patient (and, in a sense, the surgeon) at the mercy of the laser.

To understand how a patient can be most unhappy after a seemingly successful laser refractive surgery, consider the following analogy:

You own a white pickup truck. You want it painted blue, so you take it to a body shop and select your color (analogous to choosing your desired post-operative prescription). When the vehicle is delivered, it's blue. However, it has runs, drips, fingerprints, dog hair stuck in the paint, and poorly covered areas. You notice that the hood and the trunk are not the same tone of blue. To your bafflement, the technician states that the job has been performed according to specification, because, after all, your truck is now blue.

Similarly, before LVC Lou could barely read the 20/400 letter (the “big E”) without his glasses. Now he can read the 20/40 line, which is 10 times better. Despite technically receiving the desired surgical outcome, Lou's vision quality is severely reduced. He sees the equivalent of drips, fingerprints and dog hair overlying everything he looks at.

In addition to the visual misery, an LVC casualty can be further distressed because he seems to be the only person who thinks that anything is “wrong” with his result. The surgeon and his office staff point to that small line of letters and explain that he is one of the fortunate (~80%) who have achieved glasses-free driving-standard vision without requiring an enhancement. His job is to be delighted that he can now read the alarm clock on waking in the morning. He is letting everyone down. “We're sorry, but we find nothing on examination that can explain your complaints.” [Sub-text, particularly if the patient is a woman: you're making them up. Alternative sub-text, particularly if patient is an engineer: you're awfully picky.]

The source of disagreement between the patient and the doctor (or paint technician) regarding the quality of the surgery (or paint job) generally rests upon the inadequacies of the types of tests that are performed post-operatively. When a surgeon tests visual acuity (the “letter chart” with black letters against a white background), only one aspect of the cornea's optical performance — which is primarily a function of the central 1 to 2 millimeters of its surface — is being measured. Many types of vision quality loss are caused by optical irregularities of the cornea outside this small central region but within the area treated by the laser. So far, clinical tests which might measure the effect of these irregularities (such as low-light contrast sensitivity) are infrequently used. In addition, Lou's newly sculpted corneas are attached to Lou's eyeballs, and thence to Lou's brain. Believe it or not, the brains of some LVC patients are blamed for “inadequate cortical adaptation”, meaning that they are not getting used to starbursts, ghosting, desaturation, and other visual garbage that a less neurotic or less demanding patient would readily tolerate and learn to ignore.

Surgeons also perform an examination with a piece of equipment called a slit lamp, which is a microscope sitting on its side on wheels; the slit lamp allows the corneas to be viewed at varying levels of magnification, up to 40 times magnification in some cases. Slit lamps are used to detect opacities in the cornea or structural irregularities such as striae (a type of flap wrinkling problem that occurs after LASIK). Using the slit lamp, the corneal reshaping that the laser is supposed to accomplish is visible only in a “global” way as a generalized flattening (for myopic surgery) or steepening (for hyperopic surgery) of the central part of the cornea. Many patients hear “Your corneas look great!” when there are no scars, wrinkles, or hazing visible on slit lamp examination. This is unhelpful and/or irrelevant because the “drips, smudges and dog hair” that the patient sees can be due to corneal surface irregularities far finer than the slit lamp (which has been around for about 100 years nearly unchanged) is designed to detect.

In the case of Big Pupil Lou and other LVC unfortunates, can we examine the corneas in a fashion that directly measures the critical effect that the laser was supposed to have in reshaping the corneal surface, and skip Lou's brain? The answer is yes, with a test that has been around since long before the idea of excimer laser refractive surgery was scribbled on a napkin, but considerably less long than slit lamps: corneal topography.

The cornea is a curved surface. Curved surfaces can be described in terms of elevation. Suppose you are in a building with a dome, and you are standing directly under the center of the dome. With some help, you can measure how high above you the ceiling is. If you walk away from the center and repeat the measurement it will decrease, since the ceiling is closer to you because the roof is curving downwards. By walking around under the dome and taking many measurements, recording your location and the ceiling height (elevation) that you found at each location, you could develop a comprehensive numerical map of the dome's shape. A corneal topographer is capable of this precise function: describing the shape of the corneal surface in terms of a huge set of elevation data (typically thousands of data points).

Prior to refractive surgery, corneal topographies are performed to look for irregularities of the corneal surface – bulges, sinkholes, and weird slopes — which may cause minimal perceptible symptoms but which could seriously interfere with the laser's ability to accurately reshape the cornea. The laser software for standard ablations is expecting a cornea whose curvature follows some basic rules, the most important being radial symmetry. Radial symmetry means that the piece of dome ceiling 5 feet north of center and the piece of dome ceiling 5 feet south of center are the same height.

Using topography after a laser ablation, we have a readily available, standardized, reproducible, objective test of the laser's performance. The laser is supposed to convert the cornea from one curved shape to a different, specified curved shape. So if you compare a pre-operative topography to a post-operative topography taken fairly soon after surgery (before any unexpected biomechanical responses can have modified the laser-defined curvature) you will have a decent objective assessment of the laser's ability to deliver what the surgeon expects and the patient paid for.

So everyone does this, right? Actually, it is quite rare for a surgeon to routinely perform post-operative topographies soon after surgery. This makes even less sense than it sounds like to the layperson, because it prevents the surgeon from identifying increasingly bad laser behavior that is not causing severe symptoms until it really trashes someone out, in which case the habitual (and even vaguely defensible, in the absence of convincing data to the contrary since no one bothered to take topographies) explanation is that the patient healed poorly.

In the early days of LVC, potential refractive surgeons were almost entirely dependant upon Laser Company credentialing seminars and lectures from “refractive surgery experts” for information about how to perform these procedures, including appropriate pre- and post-operative testing. Since some of the early Food and Drug Administration clinical trials did not require analysis and reporting of post-operative topographies and correlation with vision quality [the LD could digress for another 5 pages on the lunacy of this omission] the manufacturers eluded the necessity of discussing the matter and the experts (all paid consultants to those manufacturers) did not promote it. Over time, a professional perception developed that post-operative topographies were of no clinical value. This was partly due to the blissful ignorance of the typical refractive surgeon, who never got post-op topographies on happy patients, and thus learned too little about a “good” topographical result to be in a position to identify a “bad” one. It was also aided and abetted by a few loud-mouths who frequented the columns of the throwaway journals and the podiums of refractive surgery symposia, who argued that two patients could have almost the same topography and one would be ecstatic and the other unhappy, so really it must be that mysterious issue of “cortical adaptation” or “dry eye” or some other factors that distinguished these patients. It was definitely not related to 1] the difference between the size of the optical zone (correctly curved central region of cornea) visible on topography and the size of the patient's pupils, or 2] central curvature irregularities in the unhappy patient's corneas that were evident when the topographer's color printout was set to a finer scale to show more detail.

And so topography, never having made it into the early post-operative examination routine, was not understood well enough to help a physician analyse an unexplained poor outcome in a case where he was convinced that his technical performance during surgery was unimpeachable. And he was left holding the bag. At that point he had one good option and one bad option, from his perspective. Good option: sadly tell the patient that he “healed badly” or “regressed” or “needed an enhancement”. Bad option: critically review the patient's pre-operative evaluation for overlooked risk factors; get a post-op topography; conclude that the laser didn't actually create the intended ablation profile (i.e., didn't properly reshape the curvature of the cornea over the entire treated area); and call the Quality Control guys at (insert name of Laser Co. here).

Blaming the patient's healing response is a useful option because Lou “Big Pupil” Schmo is frightened, exhausted, confused, and almost certainly not capable of a multi-million dollar “persecute Joe Surgeon MD” pay-back.

Blaming the laser is not a useful idea because the (insert name of Laser Co. here) has an elaborate and well-funded mechanism in place to make it difficult, scary, and ultimately fruitless for Doctor Joe to make the call. The LD, tapping the anecdotal data stream in the refractive surgery industry (RSI), has heard of one determined surgeon who took it to the top and beyond, facing off against a particular Laser Co. and spending gobs of his own money to do it. The Laser Co. ignored Joe, and Joe eventually went away but he didn't quit. Convinced that this laser was unpredictably but not infrequently seriously harming patients, he called the guys at the Food and Drug Administration. They never called him back.

This doc reportedly couldn't get any other surgeons to stand up with him, because they were all afraid of hurting the image of the RSI. Is the LD to understand from this that hurting patients is not considered by surgeons to be a way of hurting the image of the RSI?

The laser ablation is an open-loop engineering process because there is no real-time feedback about whether the amount of corneal tissue that needs to be removed, and is programmed to be removed, is actually removed. There are a variety of ways that the laser can fail to create the proper ablation in the corneal stroma which are undetectable at the time of surgery and not visible on slit-lamp examination thereafter.

All brands of laser come with elaborate calibration systems, with rules and regulations about how often certain types of calibration need to be performed. Calibration is more than a series of rote tasks: the tests must be conducted correctly, the results interpreted accurately, and calibration errors corrected properly. It is impossible that calibration is performed “perfectly” and that the laser performs “perfectly” every time. Since “perfect” is impossible, “impossible” is also impossible. Example of a preposterous statement from a Laser Co. rep: “It is impossible that our laser ever creates a decentered ablation.”

Also, the types of calibration performed may test laser actions that are a long way up-stream from the actual ablation. The best calibration would be a process that closely simulates an actual laser refractive surgery and then measures the desired end result, which is a change in the curvature of a curved surface. For some lasers, the ablation profile that is planned for the patient is first lasered into a piece of poly-methylmethacrylate (PMMA) plastic. The surgeon can check the accuracy of the ablation in a general way and look for roughness or pocking of the plastic to indicate that the ablation isn't homogeneous. Unfortunately the PMMA is flat, perfectly stationary, and not affected by moisture (NB). This is still a better effort than some lasers, which satisfy themselves by vaporizing a circle out of paper or film, which merely proves that the laser aperture is spitting out energy in the general direction of the future location of the cornea.

Common ablation errors visible on the topographies of patients with poor outcomes include decentration, under-ablation of one part of the cornea, under-ablation of everything but the most central cornea (leading to an undersized optical zone), over-ablation creating pock-marks, and ablation so randomly bad as to defy a descriptive term. There are even known cases where most of the laser energy just flat out missed the eye, although everything under observation during surgery seemed to go fine. No one knew the ablation was missing in action until post-operative topographies and corneal thickness measurements were performed.

These problems occur even in the presence of impeccable surgical technique and meticulous calibration. Since the surgeon isn't getting post-op topographies routinely, any repeated patterns of ablation error occurring in his patients aren't being detected until a patient like Lou comes along with his life-disrupting symptoms.

If you were Lou, would you agree to surgery if the consent form contained a statement that the laser could invisibly malfunction in an unknown way, for no good reason that anyone can find or head off, and then you're stuck until better technology comes along?

If you were Joe, would you agree to buy an excimer laser if the contract contained a statement that the laser could invisibly malfunction in an unknown way, for no good reason that anyone can find or head off, and then your patient is stuck until better technology comes along?

All the late-generation lasers have a tracking system that is supposed to keep the laser “lined up” with your eye. Think of it as Nintendo where the computer keeps the blaster aimed at the alien for you. With non-tracking lasers, it was the patient's responsibility to look at that little red light, and the surgeon's responsibility to take his foot off the pedal if the eye was rolling around too much since this would create an essentially random ablation pattern. Many patients balked at the idea of being partly responsible for centration, reasoning accurately that when you are scared, flat on your back, can't see anyway because you have no superficial cornea, and your eyelids are jammed open by wire prongs the task may exceed the patient's ability.

Trackers have accordingly been heavily marketed to the public, often with graphics showing a red line squiggling all over the cornea (non-tracking) vs a little red spot maintaining perfect location in the center of the cornea.

And trackers are a good thing, particularly in the hands of surgeons who understand their limitations. However . . .

The laser isn't an artificial intelligence, and doesn't look down through its aperture and say, "Aha! An eye!" A photograph of your eye is taken before surgery (usually by a technician although occasionally by your surgeon), and this is fed into the computer, which macerates it and comes up with some positioning instructions for the moving elements of the laser system. Enter human error. A tracker photo misalignment of 1 mm (possibly less) has a reasonable chance of leading to an irregular ablation. Since the cornea is 11-12 mm across for most adult humans, a 1 mm error is less than 10% of the total corneal diameter. If your technician is inexperienced or sleepy this may be unnoticed, in which case you're out of luck.

Tracker drift or “registration error” can occur. Without getting heavy on the engineering, believe me that a machine which contains swivelling mirrors so small that they are moved electromagnetically might accumulate errors due to wear and tear. Or maybe somebody bumped into it. Lou the Patient and Joe the Surgeon are believing that the laser is tracking Lou's moving corneas properly and delivering a correctly centered ablation. If the tracker has a drift bias in a particular direction (e.g., downward towards the patient's toes), the ablation ends up off center in the same direction. Small degrees of decentration are often asymptomatic, depending on the patient's pupil size and attempted degree of prescription correction. So tracker drift can go unnoticed by docs using that laser who do not perform routine post-operative topographies, until Big Pupil Lou comes along.

Trackers can also be too much of a good thing. As the cornea rolls away from a straight-up position, its surface starts to slope in a way that the laser doesn't recognize (because it is programmed to treat a radially symmetric cornea). If you deliver a certain amount of energy to a corneal surface that has a greater slope than the programmers used in their calculations, you will get less total re-shaping in that area than you need. So you would actually prefer that the laser's tracker not follow the eye too far in any direction away from straight up; instead it should interrupt the ablation routine until the surgeon gets things under control. This can lead to a lot of interruptions. One of the Nidek lasers has a pseudo-tracker which causes so many interruptions that typically surgeons turn it off. More commonly, however, surgeons let the tracker do all of the decision making about the position of the eye during the treatment, not understanding that directing the laser pulse to the proper area of cornea and having that pulse remove the correct amount of tissue are distinct and conflicting technical issues.

As an amusing example of the two-facedness of RSI advertising, consider the Alcon Summit Autonomous LADARvision laser. This laser is marketed as having a super-tracker (LADAR is an acronym for the full title of this tracker) which can follow the cornea as it rolls from the eyebrows to the belly button. At an Alcon training seminar, the audience is repeatedly (and I do mean repeatedly) told by the trainer that “Centration is the responsibility of the surgeon.”

As alluded to previously, idiosyncratic patient healing factors are blamed for poor results with a frequency that can only be described as ignorant. If the RSI would suspend its comfortable assumption that it really was the patient's fault, it might learn something about laser QC and improve the equipment, making the procedure safer and more reliable for future patients.

The human corneal stroma is a “highly conserved” tissue. This means that, from human to human, the chemical and physical structure of this tissue is nearly identical. Some individuals have diseases of certain layers of the cornea which can affect the outcome of LVC, but for now let's consider only normal patients.

If each patient's corneal stroma was ablated at its own personal rate by the excimer energy, LVC would never work. The central premise of this procedure is that a certain amount of laser energy delivered to the stroma will remove a certain depth of tissue (known as the etch rate), and that the etch rate can be presumed to be the same for all humans as long as environmental conditions such as room humidity are identical.

When an ablation ends up sloped, pocked, warped, or mangled these changes are usually visible to the patient immediately, and they are also detectable by topography. When the topography is performed soon after otherwise uncomplicated surgery it is improbable that the problem is due to “bad healing”. Sometimes the first topography isn't obtained for weeks or months after surgery. If the healing appeared to be uneventful (see Ingredient #1 above), there are no sound clinical grounds for deciding that some undetectable “patient healing factor” led to the poor result. This is particularly true if the patient has been complaining vigorously since immediately after surgery about the “runs, drips, fingerprints, and dog hair” disturbing his vision.

This article has primarily been in the context of a surgeon who is technically excellent at the actual LASIK procedure, and it has not discussed other physician factors like proper patient selection, surgical planning, and post-operative care. However, when considering a bad ablation, possibly the most frequent surgeon error is failure to maintain an even state of moisture on the area of corneal stroma that is going to be treated. Excimer laser energy is rapidly dissipated when it travels through water (think of it as the water molecules soaking up the energy), so if the patient's stromal bed is unevenly wet the ablation will end up with shallow and deep spots, depending on the location of the puddles.

Laser Companies try hard to avoid mentioning the possibility of poor surgical technique because this would aggravate their customers (surgeons), and because each company requires specific training and certification — which it provides — to use its laser. Surgeons have to perform a certain number of cases while being observed by representatives of the Laser Co. as part of the certification process. However, if backed deeply into a corner, the Laser Companies will blame the surgeon rather than fessing up that their lasers aren't perfect all the time. If your surgery was performed by Dr. Scary Harry Flap-n-Zap (whom you're suing), this can create a useful rift between the surgeon and the laser manufacturer. Unfortunately the time, money, and unswerving intent required to push a multi-national corporation into a tight squeeze is beyond the resources of most private individuals and most plaintiffs' attorneys.

In the LD's considered opinion, many patients — even “satisfied” patients — get laser ablations not as good as they were entitled and even believed to get despite surgery executed with superior technique, because 1a] high-contrast visual acuity and residual prescription are the only “tests” the typical refractive surgeon routinely uses to assess the result of the operation, 1b] these tests are markedly inadequate quality control measurements for the laser's performance, 2a] due to the mechanical complexity of the lasers, most surgeons do not understand how various types of laser errors can occur and what the effects would be on the ablation, 2b] they certainly do not know what topography changes and patient symptoms would result, and 3a] it is usually futile to report suspected laser errors to manufacturers (What are they going to do, admit it?), plus 3b] it is risky to try, especially if you are leasing the laser or took advantage of convenient manufacturer financing to buy it. Remember that laser ablation errors will almost always be attributed to the patient's healing process, since supposedly laser ablation errors don't exist thanks to calibration and tracking and the general genius of the engineers at (insert Laser Co. name here).

If you are considering LVC, the message you should take from this article is that a laser is capable of invisible malfunctions, which create the type of problem that is currently not remedied with glasses nor readily treated surgically: irregular ablations leading to irregular curvature of the cornea after surgery. So when you sit down to talk to Joe Surgeon, it behooves you to quiz him carefully about how well he knows the laser, how involved he is with its maintenance and operation, whether he has ever suspected or confirmed that the laser malfunctioned in a fashion that wasn't detected by either the laser or the surgeon at the time of treatment (and what remedial steps were taken, and how he can be sure that the same problem won't happen again), and whether he routinely or semi-routinely gets post-operative topographies on his LVC patients. “It always works perfectly” is not an accurate answer. “I've never had a laser problem” means that he's never suspected and investigated the laser. “Topographies really don't tell you much” means you should cut and run.

Did you find this editorial useful? Write to the editor and let us know your opinion.

Back to Keratoscoop

Column archives