The Sobering Truth of Perfect Health Metrics in an Imperfect World

In today’s data-driven healthcare landscape, algorithms and metrics are hailed as tools to optimize patient outcomes. They’re designed to also reduce costs and streamline the delivery of care. Predictive models assist in disease risk assessment, AI-driven diagnostics enhance accuracy, and demographic data informs public health strategies. 

It should be simple… Use data to improve healthcare outcomes and streamline systems. 

But there’s a catch. This approach assumes the data we’re working with is complete and fair by default. This isn’t always the case, of course, taking into account systemic issues that plague the framework. 

Blind Spots and Biases in the Code

Healthcare algorithms are only as good as the data they’re built upon. Predictive models are a case in point. They’re used to assess everything from hospital readmission risk to patient pain levels. However, they’re frequently developed using datasets that underrepresent marginalized populations. 

The result? Well-meaning systems that reinforce existing disparities. 

A striking example emerged in 2019, with a widely used risk-predicting algorithm. It was supposed to flag patients for extra care to avoid critical conditions becoming a bigger problem. However, it was found to show bias and discrimination against people of color. 

Since Black people tend to spend less on healthcare, the system scheduled them for fewer follow-up visits by default. This happened even though they presented with more serious health issues than their white counterparts. 

These kinds of oversights don’t just misclassify risk, but can translate to fewer resources, less support, and delayed interventions. In an industry where decisions can literally be life-or-death, gaps in the code become gaps in care. 

When Efficiency is the Goal, Not the Guide

Metrics like “cost per patient” or “treatment adherence rates” are often used to measure healthcare success. But numbers like these don’t always tell the full story. Take long-acting contraceptives, for example. 

These methods are often promoted in public health initiatives as affordable, low-maintenance solutions. Sometimes, they’re the preferred suggestion for people with limited access to care. While they do cut down on the need for regular doctor visits, they may not address people’s needs or long-term health risks. 

One case that’s drawn attention is Depo-Provera, a hormonal birth control injection. Lawsuits have recently raised concerns about serious side effects linked to prolonged use.  Specifically, the Depo-Provera lawsuit outlines health conditions, like brain tumors, bone density loss, hormonal imbalance, and infertility. 

According to TruLaw, Depo-Provera contains a synthetic form of progesterone, which makes the risk of developing brain tumors five times higher. In fact, women who use Depo-Provera for at least a year, have a 555% increased risk of meningioma. 

This is only one of many cases that highlight the stark disconnect between efficiency-driven healthcare strategies and patient-centered care. 

Quantified Doesn’t Mean Qualified

Metrics can be useful tools, but they only tell part of the story. They don’t always give us a holistic view of the human side of healthcare. Things like whether patients understand their treatment options or what they personally value in their care. Harder to measure, but just as important. 

A 2023 report from Time, using exclusive results from the Harris Poll, revealed that 7 out of 10 American adults felt the healthcare system failed them. The main concerns included high costs, long wait times, and not enough focus on preventative care. 

Additionally, the Commonwealth Fund emphasized the importance of diversity among healthcare workers. The lack of representation in care is what underscores the marked lack of equity in the delivery of that care. 

It’s therefore clear how important it is to look beyond quantitative metrics to fully understand and improve patient care. Although data can steer decisions, numbers alone aren’t enough. To create systems that truly serve everyone, we also need to consider real perspectives and lived experiences that numbers don’t capture. 

Building a more equitable healthcare system means acknowledging where today’s data-driven models fall short. That includes bringing in more diverse data, being transparent about how algorithms are developed, and centralizing what actually matters to patients. 

By acknowledging and addressing the imperfections in our data and metrics, we get closer to a reality where healthcare truly does serve everyone equally. 

Where Tech Goes From Here

As data continues to shape the future of healthcare, it’s increasingly pressing that we remember there’s a real person behind every metric. When the algorithms get it wrong or inputs don’t reflect real-world experiences, entire communities go unseen. As it stands, AI is falling into a very human pattern of bias and discrimination. 

The only way tech will truly be “better” or “advanced” is if we program it to ask better questions for adequate representation. A smarter healthcare system will become a reality once it recognizes who’s missing from the equation, and makes provision for that. Because equity isn’t a side feature. That’s the whole point.