When someone collapses in a grocery store aisle, nobody standing over them debates medical philosophy. They call 911. A system takes over that was designed for exactly that moment, and it works.
That system gets called conventional medicine, mainstream medicine, Western medicine or allopathic medicine. Most of those labels make it sound like a belief system. It isn’t one. Conventional medicine is a set of rules: who may practice, what training they have to finish, what proof a treatment must produce before anyone can sell it, and who keeps watching after it goes on sale.
The rules are open to anyone who can meet them. That’s the part people miss. The National Institutes of Health runs an entire center devoted to studying yoga, acupuncture, meditation and herbal products. Conventional medicine is a filter they can pass through, not the rival of natural approaches.
This article covers what the filter is made of, what it catches well, and what slips by it.
The five parts you use
Think of the system as five doors, each built for a different kind of problem.
Primary care is the front door. Americans made about 1 billion visits to physician offices in 2019, according to the Centers for Disease Control and Prevention’s National Ambulatory Medical Care Survey. That’s roughly 321 visits for every 100 people, and 50.3% of them went to primary care physicians. A primary care clinician’s real job isn’t any single disease. It’s holding the whole picture over years and deciding what needs more firepower.
Specialists trade breadth for depth. A rheumatologist sees more lupus in a month than most family doctors see in a decade. The cost of that depth is that no specialist is responsible for the rest of you. How referrals work is its own subject, covered elsewhere in this library.
Hospitals exist for problems that need continuous monitoring or an intervention you can’t do in a clinic room. Admission is a judgment about how often someone needs to lay hands on you, not a judgment about how sick you feel.
Emergency departments answer one question quickly: is this about to kill you? The CDC’s National Hospital Ambulatory Medical Care Survey counted 155.4 million emergency department visits in 2022, or 47.3 visits per 100 people. Only 11.5% ended in a hospital admission. So roughly nine out of ten people who go to an ER go home the same day. That’s the department working correctly, not wastefully.
Pharmacy is the part people use most and ask least. Federal survey data from 2017 through March 2020 found 49.9% of people had used at least one prescription drug in the past 30 days, 24.7% had used three or more, and 13.5% had used five or more. Drug therapy was involved in 71.9% of office visits in 2019. Your pharmacist is often the most reachable clinician you have, and usually the cheapest.
Four questions that tell you which door to knock on
- Is this getting worse by the hour or by the month? Hours means emergency. Months means an appointment, and the appointment will go better than the ER would.
- Does anyone already know my history? If yes, start there. Starting over with a stranger costs you time and repeat testing.
- Am I asking for a diagnosis or for a treatment I already know I need? Those go to different places.
- Is this about a medication? A pharmacist can usually answer faster than a scheduled visit, including whether two things you take interact.
How a drug gets approved, and what the test proves
The Food and Drug Administration describes five stages: lab discovery, preclinical testing in animals for basic safety questions, clinical research in people, FDA review, and post-market safety monitoring.
Before a single human takes the drug, the company files an Investigational New Drug application containing animal and toxicity data, manufacturing details, trial protocols and information about the investigators. FDA gets 30 days to review it and can halt the study.
Then come the phases, with the agency’s own figures:
- Phase 1: 20 to 100 people, several months. Purpose is safety and dosing. About 70% of drugs move on.
- Phase 2: up to several hundred people who have the condition, several months to two years. Purpose is whether it works and what it does to people. About 33% move on.
- Phase 3: 300 to 3,000 people, one to four years. Purpose is effectiveness and tracking adverse reactions. About 25% to 30% move on.
Multiply those three survival rates and you get a number worth sitting with. Fewer than one in ten drugs that begin human testing reach approval. Most candidates die in the building.
What survives goes into a New Drug Application. FDA first checks whether the submission is even complete and can refuse to file it. If it’s complete, the review team has six to ten months to decide. Inspectors visit the trial sites looking for data that was fabricated, manipulated or withheld. When the paperwork doesn’t settle a question, FDA convenes an advisory committee of outside experts that includes a patient representative.
Now the honest part. A successful trial proves that this product beat its specific comparison on a specific measured outcome, in the specific people enrolled, over the months the study ran. It does not prove the drug is the best of your available options. It does not prove the benefit is worth the price. And it does not prove much about people the trial excluded, which often means older adults, pregnant patients and anyone carrying three conditions at once.
FDA says so itself. The agency’s own page on post-market monitoring states that a product’s safety picture “evolves over the months and even years” after it reaches the market. That’s why MedWatch exists, and why FDA built the Sentinel system to scan large electronic health databases for safety signals in close to real time.
The training pipeline, in plain terms
The Federation of State Medical Boards, which represents the state boards that license physicians, lays out the path.
It starts with an MD or DO degree, typically four years after a bachelor’s degree, from one of the 155 allopathic or 37 osteopathic medical schools in the country. The degree alone permits nothing. No state hands over a full, unrestricted license until a graduate has put in at least a year of postgraduate training, and some states set that bar at two or three. Depending on the specialty a resident picks, the training stretches anywhere from three years to seven.
Along the way come the licensing exams: the USMLE sequence for MD graduates, COMLEX-USA for most DO graduates. Boards cap the number of attempts and set deadlines for finishing the sequence.
Then the parts people don’t picture. Boards assess physical, mental and moral fitness. The application asks about criminal convictions, any malpractice judgment or settlement, where you have worked, and health problems that could get in the way of practicing safely. Many boards run criminal background checks and query the Federation’s disciplinary data bank. A license also expires. Most run on a one or two year cycle, and continuing medical education is the price of keeping one in nearly every state.
Add it up and the shortest route from a bachelor’s degree to independent practice is about seven years. The long routes run past eleven.
One distinction is worth carrying out of this section, because it causes real confusion. A medical license covers the general practice of medicine. It is not specialty-specific. Certification by a specialty board is a private credential layered on top, and no state makes it a condition of licensure. Most physicians get certified anyway, largely because hospitals usually require it before granting privileges. Decoding credentials in depth is a separate article here.
What conventional medicine does better than anything else
A pattern runs through its genuine wins. Clear target, short clock, measurable endpoint.
Trauma and acute emergencies. A ruptured appendix, a collapsed lung, a hemorrhage. These are mechanical problems with mechanical answers, and the answers work within hours.
Infection. Identify the organism, pick the drug that kills it, confirm the fever breaks. The feedback loop closes in days.
Surgery. Few interventions in any tradition can match what a hip replacement does for someone who can’t walk a block.
Cancer treatment. The National Cancer Institute’s SEER program puts five-year relative survival for cancer of any site at 70.5%, based on 2016 through 2022 data. In its longest-running registries, observed five-year survival rose from about 49.9% for people diagnosed in 1975 to about 73.6% for those diagnosed in 2018. That gain came from decades of trials, most of which failed.
Diagnostics. Imaging and laboratory testing can see inside a living person. Nothing else can. This is arguably the single largest contribution, because it tells every other approach what it’s dealing with.
Time-critical cardiovascular care. Cardiovascular disease still caused 919,032 deaths in 2023, about one in every three US deaths, and the CDC estimates about 805,000 Americans have a heart attack each year, roughly 605,000 of them first attacks. The system’s answer to an individual heart attack is now a stopwatch and a protocol rather than a bedside vigil.
Where it runs slow, weak, or out of time
Now the other column, stated as plainly.
Chronic conditions. The CDC reports that about three in four American adults have at least one chronic condition, more than half have two or more, and over 90% of adults 65 and older have at least one. A system optimized for discrete problems with endpoints is working against its own design here. Chronic disease has no endpoint. It has a trajectory.
Time. This is not a soft complaint, and there’s arithmetic behind it. A 2020 study in the American Journal of Public Health by Nia Privett and Sarah Guerrier calculated what it would take for one physician to deliver all the preventive services the US Preventive Services Task Force recommends to a typical panel of 2,500 adults. It would take 8.6 hours of every working day, which works out to 131% of the time a physician has. That’s just prevention, before anyone walks in sick.
The same journal published the same calculation in 2003, when Kathryn Yarnall and colleagues found it would take 7.4 hours a day. Two decades later, with fewer recommendations on the list, it takes more time, not less. So when your doctor doesn’t spend fifteen minutes on your diet, you’re looking at a math problem, not indifference.
Conditions without a clear marker. Persistent fatigue, long-standing dizziness, widespread pain, gut symptoms that don’t match any scan. The system is built around tests, and it is noticeably weaker when the test comes back normal and the symptom is still real. Patients in this situation often get bounced between specialists, which is the system telling you it doesn’t have a box for you.
What never gets studied at all. The FDA machine runs on applications. A company files an IND, then an NDA. Nobody files a New Drug Application for a walking program, a sleep schedule or a change in diet. Those get researched, but by academic investigators on public money, usually in smaller and shorter studies with less funding behind replication. For scale, the National Center for Complementary and Integrative Health, established in October 1998 as the NIH’s lead agency for research in this area, operated on $170.3 million and about 102 full-time staff in fiscal 2024. It is one of 27 institutes, centers and offices at NIH.
“Evidence-based” is a method, not a brand
The phrase gets used as a marketing sticker. Underneath it is an actual procedure, and the best way to see it is to watch the system rule against itself.
The US Preventive Services Task Force grades every recommendation. An A means high certainty of substantial net benefit. A B means high certainty of moderate benefit, or moderate certainty of moderate-to-substantial benefit. A C means offer it selectively, because the net benefit is small. A D means recommend against it, because there’s no net benefit or the harms outweigh it. An I statement means the evidence is insufficient to judge, and if you get the service anyway, you should understand that uncertainty. The Task Force also labels its own level of certainty as high, moderate or low. Cost is not part of any of it.
A system that can only say yes isn’t a method. The D grades and the I statements are the proof that this one is.
The term itself was formalized in a 1996 editorial in BMJ by David Sackett and four colleagues, titled “Evidence based medicine: what it is and what it isn’t.” People were already misusing the phrase thirty years ago. The title says so.
And the method is genuinely open at the input end. NCCIH’s own summary of the research reports that acupuncture may help with chronic pain and with tension and migraine headaches, that meditation may help with blood pressure, anxiety, depression, some bowel symptoms and insomnia, and that tai chi may improve balance. Those conclusions came from randomized trials and systematic reviews, the same tools used to evaluate drugs. Nothing about the origin of a treatment keeps it out. Failing the test does.
The trade-off nobody states out loud
Rigor buys one thing and costs another, and both halves are real.
What it buys: a phase 3 trial of 300 to 3,000 people running one to four years can reliably detect a moderate benefit and rule out a large harm. That’s a genuine protection, and it is the reason the modern pharmacy is not a catalog of hopeful guesses.
What it costs: those same trials are too small and too brief to catch rare harms, which is exactly why FDA keeps monitoring after approval. They usually pit one product against one comparison rather than sorting six real-world options. They routinely exclude the complicated patients who will take the drug. And they only ever examine what somebody paid to submit.
So the approval process is a filter with a particular mesh size. It catches moderate benefits in middle-aged trial participants very well. It catches rare harms late. It never sees what nobody submitted at all. Those three sentences explain most of what frustrates people about mainstream medicine and most of what protects them.
The signs that mean go now, not later
Everything above is about how a system behaves on average. Here is where averages stop mattering.
Get emergency care immediately for chest pressure, tightness or pain lasting more than a few minutes, especially with sweating, nausea or shortness of breath. Same for sudden weakness or numbness on one side, a drooping face, trouble speaking or understanding speech, or a sudden severe headache. Stroke treatment is measured in minutes, and the window closes.
Also go now for trouble breathing at rest, a high fever with a stiff neck or new confusion, or severe abdominal pain that keeps escalating.
Make an appointment soon, not eventually, for a new lump, a mole that is changing, bleeding you can’t account for, weight loss you didn’t intend, or any symptom you’ve been managing on your own for several weeks with no change. These are the problems where conventional diagnostics earn their keep, and where waiting costs the most.
One more. If you already have a diagnosis with a time-sensitive treatment, such as an early-stage cancer, a serious infection or an unstable fracture, delay is itself a choice with consequences. Adding something to that treatment is a different decision from postponing it.
The most useful thing to take from all of this: conventional medicine’s strength and its weakness come from the same place. It is a system that demands proof before it acts. That makes it very good at problems where proof can be gathered in a few years, and slow at nearly everything else. Figuring out which kind of problem you have is what tells you which door to use first, and when to stop waiting for one to open.
Sources
- U.S. Food and Drug Administration. “The Drug Development Process.” fda.gov
- U.S. Food and Drug Administration. “Step 3: Clinical Research.” fda.gov
- U.S. Food and Drug Administration. “Step 4: FDA Drug Review.” fda.gov
- U.S. Food and Drug Administration. “Step 5: FDA Post-Market Drug Safety Monitoring.” fda.gov
- Federation of State Medical Boards. “About Physician Licensure.” fsmb.org
- Centers for Disease Control and Prevention, National Center for Health Statistics. “FastStats: Physician Office Visits.” cdc.gov
- Centers for Disease Control and Prevention, National Center for Health Statistics. “FastStats: Emergency Department Visits.” cdc.gov
- Centers for Disease Control and Prevention, National Center for Health Statistics. “FastStats: Therapeutic Drug Use.” cdc.gov
- National Cancer Institute, Surveillance, Epidemiology, and End Results Program. “Cancer Stat Facts: Cancer of Any Site.” seer.cancer.gov
- Centers for Disease Control and Prevention. “About Chronic Diseases.” cdc.gov
- Centers for Disease Control and Prevention. “Heart Disease Facts.” cdc.gov
- U.S. Preventive Services Task Force. “Grade Definitions.” uspreventiveservicestaskforce.org
- National Center for Complementary and Integrative Health. “About NCCIH.” nccih.nih.gov
- National Center for Complementary and Integrative Health. “Complementary, Alternative, or Integrative Health: What's In a Name?.” nccih.nih.gov
- PubMed. “Privett N, Guerrier S. "Estimation of the Time Needed to Deliver the 2020 USPSTF Preventive Care Recommendations in Primary Care." American Journal of Public Health, 2020. DOI: 10.2105/AJPH.2020.305967.” doi.org
- PubMed. “Yarnall KS et al. "Primary care: is there enough time for prevention?" American Journal of Public Health, 2003. DOI: 10.2105/ajph.93.4.635.” doi.org
- PubMed. “Sackett DL et al. "Evidence based medicine: what it is and what it isn't." BMJ, 1996. DOI: 10.1136/bmj.312.7023.71.” doi.org