Across more than a decade of published studies, smartphone scoliometer apps have agreed with the physical scoliometer used in clinics to within roughly one degree, with reliability scores consistently above 0.90. That is accurate enough for screening and for tracking change over time — but no scoliometer, digital or physical, measures the Cobb angle or diagnoses scoliosis.
What “accurate” actually means here
People asking whether a scoliometer app is accurate are usually asking one of four different questions, and they have different answers. Separating them is the whole of this page.
| The question | What it really asks | Short answer |
|---|---|---|
| Validity | Does the app give the same number as a physical scoliometer on the same back? | Yes, typically within about 1°. |
| Repeatability | If the same person measures twice, do they get the same number? | Yes, when technique is consistent. |
| Reproducibility | Do two different people get the same number? | Largely yes, but this is where most real-world error creeps in. |
| Diagnostic accuracy | Does the reading tell me whether scoliosis is present, and how severe? | No. ATR is a screening signal, not a diagnosis. |
The first three are properties of the instrument and the technique. The fourth is a property of the method — and it is the one that a better app cannot fix, because ATR and the radiographic Cobb angle are simply not the same measurement.
What the research shows
Agreement with the physical scoliometer
The category was first formally validated in 2012, when Franko and colleagues tested a smartphone scoliometer app against the hand-held device and established smartphone scoliometry as a valid, low-cost alternative. Two years later, Balg and colleagues measured 34 patients with both instruments and found the app agreed with the physical scoliometer to within 0.4°, with an intraclass correlation coefficient (ICC) of 0.947 — concluding the app was valid for clinical evaluation.
For context, an ICC above 0.90 is generally read as excellent agreement, and 0.4° is smaller than the difference you get from re-positioning the same patient.
Agreement between different observers
Qiao and colleagues ran a head-to-head comparison of manual and smartphone-aided trunk rotation measurement and found the smartphone matched the manual scoliometer on both intra-observer and inter-observer reliability. In other words, switching to a phone did not introduce a new source of disagreement between examiners — the variability that exists is the variability the forward-bend test has always had.
Accuracy when a parent takes the measurement
This is the finding that matters most for home screening. In a 2022 study of 50 adolescents with idiopathic scoliosis, a smartphone app was used during the forward-bend test by both a spine surgeon and a parent. Correlations against the reference scoliometer were 0.97 for the surgeon and 0.92 for the parent, with all ICCs above 0.92. Driscoll and colleagues reported a similar result across 39 adolescents, where a parent achieved an ICC of 0.91.
A parent is measurably less consistent than a surgeon. But the gap is small, and a 0.92 correlation is more than adequate for the job a home reading is being asked to do: noticing asymmetry, and noticing change.
| Study | What was compared | Result |
|---|---|---|
| Franko et al., 2012 | Smartphone app vs physical scoliometer | First formal validation of the app category |
| Balg et al., 2014 | iPhone vs scoliometer, 34 patients | Agreement within 0.4°, ICC 0.947 |
| Qiao et al., 2014 | Manual vs smartphone-aided ATR | Matched intra- and inter-observer reliability |
| Driscoll et al., 2014 | Smartphone ATR, 39 adolescents | Close to surgeon; parent ICC 0.91 |
| van West et al., 2022 | Surgeon vs parent using a phone, 50 adolescents | Correlations 0.97 and 0.92; all ICCs above 0.92 |
Where the error actually comes from
Once agreement between instruments is inside a degree, the instrument stops being the interesting variable. Almost all of the spread in real-world readings comes from how the measurement is taken.
| Source of error | Typical effect | How to reduce it |
|---|---|---|
| Bending too far forward | Rounds the spine and can flatten the very rotation you are trying to see | Bend only until the back is roughly horizontal at the level being measured |
| Uneven weight or a twisted stance | Introduces rotation that is postural, not structural | Feet together, knees straight, arms hanging freely, weight even |
| Measuring at a different spinal level | ATR varies along the spine; a few centimetres changes the number | Sweep the whole back each time and record the single highest reading |
| Phone not flat on the back | Reads the phone’s tilt rather than the trunk’s | Rest the long edge across the back, perpendicular to the spine, with light even contact |
| A thick or uneven phone case | Can rock the device and add a degree or two | Remove the case, or use the same case every single time |
| Different people measuring | The largest practical source of variation | Have the same person take every reading where possible |
| Time of day and activity | Small variation is normal | Measure at a similar time, under similar conditions |
The practical implication is that a series of readings taken the same way by the same person is far more informative than a single reading taken carefully. Consistency beats precision when you are watching for change.
What accuracy does not mean
An app can reproduce the scoliometer almost perfectly and still not tell you whether someone has scoliosis. Three limits are worth stating plainly.
ATR is not the Cobb angle. Trunk rotation is a surface measurement of asymmetry; the Cobb angle is measured on a radiograph. They correlate — one study of patients with idiopathic scoliosis reported a correlation of about 0.7 — but the relationship is loose enough that you cannot convert one into the other for an individual. Our ATR versus Cobb angle guide covers this in detail.
Screening thresholds trade false positives against missed cases. The same study reported roughly 87% sensitivity at a 5° trunk-rotation threshold. That means a 5° cut-off catches most, but not all, significant curves — and referring at 5° rather than 7° means referring more children who turn out not to need treatment. Both thresholds are defensible; neither is perfect.
A normal reading is not a clearance. Some curves, particularly single lumbar curves and curves in children who are still growing quickly, produce little surface rotation. A low ATR alongside visible asymmetry, uneven shoulders, a prominent shoulder blade or a waist that looks different on the two sides still warrants a professional look.
Getting the most accurate reading you can
- Set the scene the same way each time. Good light, a firm floor, the person barefoot with their back exposed or in a fitted top.
- Calibrate the app on a level surface before you start, so the phone’s zero is true.
- Position carefully. Feet together, knees straight, arms hanging with palms together, bending forward slowly from the hips.
- Sweep, do not spot-check. Move from the upper back down to the lower back, pausing at each level, and note the highest number.
- Repeat the whole measurement two or three times and record the highest consistent value rather than a single outlier.
- Log it. Date, level of the spine, and the reading. A number without a trend line is hard to interpret.
Our step-by-step measurement guide and the Adam’s Forward Bend Test guide walk through the positioning in more detail.
When a reading needs professional review
Seek an assessment from a doctor, physiotherapist or chiropractor experienced in scoliosis if any of the following apply:
- A reading of 5° or more, particularly in a child or adolescent who is still growing.
- A reading of 7° or more at any age.
- An increase of 2° or more between measurements taken the same way.
- Visible asymmetry, back pain, or a change in posture, regardless of the number.
Our guide to what counts as a normal scoliometer reading explains how these thresholds are used.
Download the Scoliometer App
Available on iPhone and Android.
Trusted by clinicians and families in 75+ countries.
Frequently asked questions
- Is a smartphone scoliometer as accurate as a real scoliometer?
- In published comparisons, yes for practical purposes. Studies have reported agreement within roughly 0.4° and ICC values above 0.90 between smartphone apps and the hand-held scoliometer. The remaining variation comes mainly from positioning and technique rather than from the device.
- How accurate is a phone reading taken by a parent at home?
- A 2022 study of 50 adolescents found a parent using a smartphone during the forward-bend test correlated 0.92 with the reference scoliometer, against 0.97 for a spine surgeon. That is enough to screen and to monitor change, though a clinician will still be more consistent.
- Does the model of phone affect the result?
- The inclinometer sensors in modern smartphones are more than precise enough for a measurement reported in whole degrees. What matters more is that the phone sits flat on the back, so a thick or uneven case can matter more than the handset itself. Use the same phone and the same case each time.
- Can the app tell me my Cobb angle?
- No. The app measures the Angle of Trunk Rotation from the surface of the back. The Cobb angle is measured from a radiograph. The two correlate but cannot be converted into one another for an individual.
- Can a scoliometer app diagnose scoliosis?
- No. It is a screening and monitoring aid. A raised reading indicates trunk asymmetry that may warrant a professional assessment; only a clinician, usually with imaging, can diagnose scoliosis.
- Why did I get two different readings a minute apart?
- Almost always positioning. A change in how far the person bends, where along the spine you measure, or how evenly they are standing will move the number. Repeat the whole sweep two or three times and use the highest consistent value.
- Can a normal reading miss scoliosis?
- It can. Screening at a 5° threshold has been reported at roughly 87% sensitivity, and some curves produce little surface rotation. If there is visible asymmetry or symptoms, seek an assessment regardless of the reading.
- Has the Scoliometer App itself been clinically validated?
- The research supports smartphone-based ATR measurement as a method, evaluated across a range of apps and approaches. The Scoliometer App implements that method, but it has not been the subject of its own product-specific clinical trial, and we do not claim otherwise.