Changing the shape and color of a patient’s prescription medication can mean the difference between a patient who sticks with a prescribed medication and one who does not. Researchers at the Brigham and Women’s Hospital published the findings from their year-long study of heart-attack patients in the Annals of Internal Medicine Monday.

Lead researcher Aaron Kesselheim and his team focused on generic drugs for their study because of their prevalence and the price and payer incentives patients have to take them, and used data from insurer UnitedHealth to follow patient behaviors for the study. They also tracked heart-disease patients, a group in which more than 50% of patients with cardiovascular disease or who are at high risk of heart disease “adhere poorly to their prescribed medications, even in the year after myocardial infarction.”

In short: altering a drug’s appearance increased the likelihood that a patient would stop taking the medication by 34%.

A new shape triggered greater noncompliance than a color change, and researchers suspect that it could be because “shape changes may exert a stronger effect because the visual acuity needed to distinguish among pill colors may be diminished among patients with chronic diseases—particularly elderly ones” since identifying a shape taps into two senses (sight and touch), which ups the odds of noticing something has changed, whereas only sight is used to identify color.

Adherence to any drug—brand or generic—is a complex behavior that is fed or diminished by factors including cost, access, and the physician-patient relationship. Non-adherence is costly—the IMS Institute for Health Informatics says lack of fidelity to a treatment regimen is tied to around $105 billion in avoidable costs, and the National Community Pharmacists Association estimates that non-adherence triggers around $290 billion in downstream health-related costs.

The key FDA requirement for generics is that the medications be the bioequivalent of the branded drugs they mimic. Researchers suggest that if the generics do not look just like the branded drugs, it could be prudent to require generics at least look like each other.

Being sensitive to what drugs look like is a habit drugmakers have instructed patients to look out for, so they know they are receiving the right medication. This plays somewhat into the tension between branded drugmakers who benefit from branded prescriptions and payers, who want the freedom to substitute generics unless a prescription is written as “dispense-as-written.” Yet the lack of visual continuity takes on greater importance in the overall generics space because patients often vary their prescription channels, which means the lack of consistency surfaces as much for patients who receive their medications from one source as it does those who rotate suppliers.

Recent manufacturing mix-ups, such as when epilepsy medication showed up in a patient’s bottle of the generic diabetes medication Glumetza, also indicate that patients are right to be iffy about medications that do not look like the previous batch.

Of the drugs researchers tracked—beta blockers, angiotensin II receptor blockers and statins—they found generic statins changed shape and color the most, even though there were only three surveyed generics, as opposed to beta blockers, which changed the least.

Researchers acknowledge that while doctors can warn patients, medications may look different from fill to fill, or that pharmacies can slap a label on a pill bottle to explain the drug’s appearance, such interventions are not enough. With bottles, for example, they note “warning stickers on pill bottles are often numerous and are often ignored.”

They also note that it would not require an act of Congress for the FDA to require generics makers to agree on a drug’s look and stick with it. They assert the FDA’s power to require one look is built into the Food, Drug and Cosmetic Act’s phrasing that allows the regulator to bounce a generic if the “composition of the drug is unsafe under such conditions because of… the manner in which the inactive ingredients are included.”