Step 4
Tests & what they mean
Three checks worth asking about
Each one starts from results most people already have. Each ends the same way: what was noticed, what it might mean, and the test that would settle it. None of them is a diagnosis. Each is a better question.
Check one: is the kidney number telling the whole story?
- Noticed: a kidney result, creatinine, that looks reassuring in someone who has lost muscle, is very slight, or has been unwell for a long time.
- Might mean: creatinine comes from muscle. Less muscle makes less creatinine, so the number can look good even when the kidneys are filtering less. It is like judging a factory's output by its exhaust when the factory has cut a shift.
- Test: ask whether a cystatin C test fits. It measures kidney filtering without depending on muscle. A big gap between the two estimates is itself worth raising. Strong evidence Source: the KDIGO 2024 kidney guideline. Grade A.
Check two: was it really just dehydration?
- Noticed: an odd result put down to "probably dehydration."
- Might mean: dehydration leaves fingerprints in the same blood test. Think of soup simmered down. The things that can't leave the pot, like albumin and hemoglobin, rise together. If they didn't rise together, dehydration may not explain it. If they did, it may.
- Test: ask to repeat the test under standard conditions: same lab, same time of day, well hydrated, same fasting. A clean repeat either confirms the excuse or retires it. Strong evidence Reading two estimates together and flagging when they disagree is our own method, not yet tested at scale. Not yet tested Source for the method: Dill and Costill, J Appl Physiol 1974. Grade A.
Check three: could breathing at night be part of it?
- Noticed: two ordinary numbers creeping up together over the years. One is bicarbonate, the "CO₂" line on a chemistry panel. The other is hematocrit, the share of blood made of red cells. Or: waking to urinate at night, morning headaches, unrefreshing sleep, or snoring that someone else has noticed.
- Might mean: shallow breathing during sleep can leave the body holding a little more carbon dioxide, and making a few more red cells. It is like a city short of buses running more of them. Kidney disease and some medicines move the same numbers: water pills, testosterone, and some diabetes medicines. So the pattern is a question, not an answer.
- Test: ask whether a sleep study fits. Reading these two numbers together as an early sign is our own method, not yet tested. Not yet tested
Published cut-offs exist for narrow groups. For example, a bicarbonate below 27 helps rule out one kind of night-time under-breathing. That applies to people with obesity who already have disordered breathing in sleep, and the guideline itself calls the evidence very low quality. Good evidence
The drift sizes our checks watch for have not yet been tested against large clinical databases, so we don't print them. Until they are, the checks point to a question, never to a number.
Sources: Mokhlesi et al., American Thoracic Society guideline, Am J Respir Crit Care Med 2019. Grade B. The cut-off is in millimoles per litre.
A lab result measures how crowded something is in your blood. So hydration, and the draw itself, can move it without your health changing at all. The full chapter on what a lab number really means →
Tests that must be shipped away. For rare conditions such as C3 glomerulopathy, key tests are run by a few specialty labs. Your sample must be drawn, frozen and shipped exactly right. You deserve the very best, and you can help your local lab give it. Getting specialty tests done right →
The early-warning idea
How can a warning hide in tests you already had? Each result is usually read alone, once, against a normal range. Read together and over time, the same numbers can say more.
- Moving, but still "normal." A result can stay inside the normal range while drifting the same way, visit after visit. The trend is the signal, and no single result shows it.
- Normal alone, telling together. Two results can each look fine while the relationship between them does not. Some well-known checks are simply one result divided by another.
- Explained away. An odd result gets put down to the day of the draw, such as "probably dehydration," when the same test often holds enough to check that.
We are building checks for each of these, one at a time, so you and your doctor can read what is already there. Check one is just below.
The full chapter, layered from plain words to the formulas →
Early warning · check one
"It's probably just dehydration." You may hear this when a result comes back odd. It might be true. But the same blood test that raised the question usually holds enough to check it, so it doesn't have to stay a guess.
When you are short on water, your blood gets more concentrated, like soup simmered down. Everything that can't leave the pot rises together: albumin, the main blood protein; hemoglobin and hematocrit, your red cells; and usually sodium and blood urea nitrogen, often printed as BUN. Most standard panels already measure these.
So the question to ask is simple: did they all rise together, compared with an earlier result from the same lab?
An example with invented numbers:
| Marker | Earlier, felt well | Case A | Case B |
|---|---|---|---|
| Albumin | 4.0 | 4.4 ↑ | 4.0 |
| Hemoglobin | 12.0 | 13.2 ↑ | 12.1 |
| Sodium | 139 | 144 ↑ | 136 ↓ |
Case A: everything rose about 10 percent together. The blood looks concentrated, and other results from that draw may read about that much high. Case B: nothing moved together, and sodium fell. Dehydration is unlikely to explain the odd result, so it deserves a real look.
It works both ways. Too much water dilutes the blood and can make a result look better than it is.
This is a way to ask a sharper question, not a way to read your own results. Your doctor makes the comparison.
Talk it through before your visit. A short conversation about the day your blood was drawn. You can speak or tap. You can say "I'm not sure" to anything, wander off-topic, because details matter, or stop at any moment. At the end you get lines for your one-page sheet.
Nothing you say or type is saved or sent anywhere by this page. One caution: some browsers send spoken words to their maker's servers to turn them into text. If that matters to you, type instead.
Get a baseline before treatment starts
A treatment can only be judged against where you started. A disease can only be called "progressing" against a number from before. One result is an event; two are a direction. If treatment begins with no baseline, everyone is guessing for the next year.
Before the first dose of anything, ask for a complete set of numbers, drawn under standard conditions, written down and kept. Without it, a later change cannot be read as better, worse, or noise.
Ask for the previous numbers too, from a year ago and two years ago. A value can sit inside the normal range for years while travelling steadily across it. Only the older results show that.
"Before we start, can we lock in a complete baseline drawn under standard conditions, and pull my results from the last two years, so we can tell later whether this is working?"
A testing regimen, followed religiously
A trend only means something if the measurement is the same each time: same tests, fixed schedule, same lab, same conditions. That means fasting or not, well hydrated, drawn the same way. A test done differently each visit produces noise that looks like change.
Two hospitals can measure the "same" test in different units, with no agreed reference. So results from different places may not be comparable, even under one name. Pick one lab and stay with it.
Draw under the same conditions every time. This is where the hydration chapter matters: sitting time, morning water, the tourniquet. Write the schedule down. The one reliable record of your trend is often the copy you keep, not the one scattered across portals.
"Can we set a fixed schedule and use the same lab each time, so the trend is real and not an artefact of how it was drawn?"
The tests that aren't routine — and why to insist
The standard panel is built for the common case, and misses the uncommon one. The tests that catch what routine misses are the ones you often have to ask for by name. They cost more, and insurers resist them. It is a fight worth having, because the cause decides the treatment.
- Serum free light chains — the high-sensitivity M-protein screen; can be abnormal when the routine electrophoresis looks normal.
- C3 and C4 measured together, plus a full complement panel — the pattern, not one number.
- The right biopsy staining — kappa/lambda immunofluorescence; a pronase-digested paraffin technique catches deposits the standard read misses.
- Your own prior results — the trend across years, requested, not assumed.
Not every "nothing" is the same.
Sometimes, even after careful testing, no cause is found. That can be a good day. An earned nothing comes from a test that was able to find the problem, done in a way that could have found it: the right test, the sensitive version, the right sample, the right conditions. It closes a door, so the search can move on. And its numbers become your new starting point.
An asserted nothing is different. It is an explanation offered without a test that could have proved it wrong. One example is "probably dehydration" with no repeat. Another is a normal routine test standing in for a sensitive one that was never run. It may well be right. It just hasn't been checked yet.
The question that tells them apart is simple, and any clinician will recognise it: "Was this a test that could have found the problem?"
The insurer, approached as a partner
These tests and treatments cost money, and the reflex is to expect a fight. It does not have to be one. Approach the insurer with respect and a clear reason, including how it serves their own interest. Then the answer is often yes, and the cooperation is genuine.
- Open with respect. The person reading it can say yes — make it easy to.
- Give a thorough, specific rationale: which test, why, what it rules in or out, and what decision it changes.
- Show their interest plainly. The sensitive test now is cheaper than the missed diagnosis and the crisis it becomes: the hospital stay, the dialysis, the escalation. Aligned incentives, stated.
- Make it easy to grant: name the code, the guideline, and the specialist backing it.
Done this way the cooperation can be phenomenal — not because you demanded, but because you handed them a good reason and an easy way to say yes.
Nothing here is a diagnosis. If anything is urgent, call your doctor or emergency services now. That means chest pain, trouble breathing, a stopped or sharply reduced flow of urine, or fever with back pain.