Evidence-based physiotherapy and rehabilitation resources
Editorial & review policy
Evidence-based physiotherapy and rehabilitation resources
Editorial & review policyLearn physiotherapy · BPT-306 and BPT-404
Heart rate, blood pressure, oxygen saturation, perceived exertion. Four numbers you record without thinking. Each one has a well-documented way of being wrong that is not random error — it is bias in a known direction, of a known size, and it is usually the clinician's technique or the equation that causes it.
In one line. The errors in routine monitoring are systematic, not random — which means averaging more readings does not remove them, and which means they are fixable by changing what you do.
Three of the four findings on this page come from randomised or pooled evidence large enough to be treated as settled. The fourth — perceived exertion — comes out of it better than most people expect.
The single largest measurement error on this page is a cuff
In a randomised crossover trial of 195 adults, using a regular cuff on a patient who needed an extra-large cuff raised systolic blood pressure by 19.5 mmHg (95% CI 16.1 to 22.9). For those needing a large cuff it added 4.8 mmHg (3.0 to 6.6); for those needing a small cuff it subtracted 3.6 mmHg (−5.6 to −1.7). [4] Nothing else a physiotherapist does routinely produces an error of that size.
Two randomised crossover trials, both in adults, quantify what technique costs you.
| What was varied | Effect on systolic | Effect on diastolic |
|---|---|---|
| Arm resting on the lap versus supported on a desk at heart level |
+3.9 mmHg (95% CI 2.5 to 5.2) | +4.0 mmHg (95% CI 3.1 to 5.0) |
| Arm unsupported at the side versus supported on a desk |
+6.5 mmHg (95% CI 5.1 to 7.9) | +4.4 mmHg (95% CI 3.4 to 5.4) |
| Regular cuff, patient needs extra-large | +19.5 mmHg (95% CI 16.1 to 22.9) | — |
| Regular cuff, patient needs large | +4.8 mmHg (95% CI 3.0 to 6.6) | — |
| Regular cuff, patient needs small | −3.6 mmHg (95% CI −5.6 to −1.7) | — |
The arm-position trial was designed carefully enough to be worth trusting: 133 participants, each measured four times including a repeat in the reference position, so that ordinary within-person variability was subtracted out rather than assumed away. The authors' conclusion is that the positions clinicians actually use produce substantial overestimation and may lead to misdiagnosis of hypertension. [3]
Note the direction in the cuff trial. Undercuffing a large arm inflates the reading; overcuffing a small arm deflates it. The patients most likely to be told they have hypertension they do not have are those with the largest arms — and the physiotherapy caseload in cardiac and metabolic rehabilitation is not a randomly sized sample. [4]
The equation every student learns has no derivation worth the name. The meta-analysis that examined it pooled 351 studies, 492 groups and 18,712 subjects and derived instead:
Maximal heart rate
HRmax = 208 − 0.7 × age — with age alone explaining most of the variance (r = −0.90). The equation was then cross-validated in a separate laboratory study of 514 healthy subjects, which gave 209 − 0.7 × age: virtually identical. It did not differ between men and women, and was not influenced by wide variation in habitual physical activity. [1]
The practical consequence is stated plainly by the authors: 220 − age underestimates maximal heart rate in older adults, and therefore underestimates the physical stress imposed during exercise testing and the intensity that should be prescribed. [1] For a 70-year-old the two equations differ by enough to change a prescription.
In children the error runs the other way, which is the part most people miss. A meta-analysis of 20 effects from seven articles in participants under 18 found age-based equations inaccurate by 6.3 bpm overall, with the 220 − age form overestimating maximal heart rate by 12.4 bpm (effect size 0.95) while the 208 − 0.7 × age form underestimated it by 2.7 bpm (effect size −0.34). The reviewers' conclusion is that adult-derived equations are not applicable to children at all, and that if one must be used it should be the latter. [2]
A systematic review assessed with QUADAS-2 and GRADE screened the literature to March 2023 and found 44 studies covering at least 222,644 participants and 733,722 paired SpO2–SaO2 measurements. Its conclusion is that pulse oximetry can overestimate true arterial oxygen saturation in people with darker skin tones, and that the magnitude is likely to be greater when saturation is lower — that is, worst when it matters most. [5]
Three caveats the review states itself, and which must travel with the finding: a high risk of bias was detected in 64% of studies; only 11 (25%) studies actually measured skin tone, covering 2,353 (1.1%) participants, with the rest reporting ethnicity as a proxy; and several studies reported no inaccuracy related to skin tone. Meta-analysis of the data was not possible. [5]
A separate meta-analysis put numbers on the consequence. Across 15 studies with 732,505 paired oximetry measurements from 207,464 hospitalised patients, undetected (occult) hypoxaemia was more common than in White patients: pooled prevalence ratio 1.67 (95% CI 1.47 to 1.90) in Black patients, and 1.39 (95% CI 1.19 to 1.64) in patients identifying as Asian, Latinx, Indigenous, multiracial or another group. [6]
Read that second figure carefully
These studies rely on self-identified race or ethnicity, which the review itself flags as obscuring the underlying variation — the physical mechanism concerns skin pigmentation and light absorption, not identity. And the review found no evidence from outpatient settings. [6] Most physiotherapy happens in outpatient settings, so the honest position is that the inpatient finding is strong, and its transfer to your clinic is untested rather than established.
Of the four numbers, the subjective one has the least disappointing evidence. A review found 950 studies citing the original session-RPE proposal, of which 36 examined its validity and reliability using the modified category-ratio scale. Those studies confirmed validity, good reliability and internal consistency across several sports and physical activities, in men and women, in children, adolescents and adults, and across levels of expertise. [7]
The review notes the method can stand alone for monitoring training load, though some authors recommend combining it with a physiological parameter such as heart rate. [7] Given how much of this page is about heart rate being mismeasured or miscalculated, a patient-reported number that has been validated 36 times deserves more respect than it usually gets in an Indian outpatient department.
Because it is the one hanging on the machine. On a patient needing an extra-large cuff that adds 19.5 mmHg to systolic. [4] Measure the mid-arm circumference and match the cuff; it is a ten-second habit that removes the largest error in this article.
Both inflate the reading — the lap by 3.9 mmHg systolic and the unsupported side position by 6.5 mmHg. [3] Support the arm on a surface with the mid-cuff at heart level. The trial was built specifically to separate this from ordinary variability, so it is not explained away by "blood pressure moves around".
It underestimates maximal heart rate in older adults, [1] and in children the same form overestimates it by 12.4 bpm. [2] Both errors push a prescription in a clinically meaningful direction, and both are avoided by using 208 − 0.7 × age — while remembering it is still a population estimate with real error around it.
Even the better equation is a regression line through group means from 351 studies. [1] An individual sits somewhere around that line, not on it. If a decision genuinely depends on maximal heart rate, it needs measuring, not predicting — and if measuring is not possible, perceived exertion is a defensible alternative rather than a lesser one. [7]
Oximeters can overestimate true saturation in darker skin, and more so as saturation falls. [5] The clinically useful response is not to distrust the device, but to weight the rest of the picture — work of breathing, ability to speak, symptoms, trend within the same session on the same finger — rather than treating a single displayed number as the endpoint of the assessment.
Every effect on this page has a direction. Taking three readings with the wrong cuff, or three readings with the arm hanging, gives you a precise estimate of the wrong value. This is the difference between random error, which repetition reduces, and bias, which it does not.
Evidence grade: High for blood pressure technique, moderate elsewhere.
The two blood pressure findings are randomised crossover trials with within-person controls, reporting confidence intervals that exclude no effect by a wide margin. [3][4] The heart rate equation rests on a meta-analysis of 351 studies plus an independent laboratory cross-validation of 514 subjects that reproduced it almost exactly — about as good as a prediction equation gets. [1]
The oximetry evidence is the weakest presented here despite the enormous sample, and the review says so: high risk of bias in most studies, skin tone actually measured in a small minority of participants, some studies finding no effect, and meta-analysis not possible. [5] The prevalence figures are firmer [6] but are hospital data. The perceived-exertion review is a narrative review of validity studies rather than a meta-analysis. [7]
Sourcing note. This is a curriculum resource written against sources that can be opened and checked. It reproduces no figures or tables from any textbook.
208 − 0.7 × age, in adults and in children, on the evidence here. [1][2] But treat the result as the centre of a range rather than a target. The children's review found an average error of 6.3 bpm even for the equations it recommends. [2]
It is when it sits at a threshold. The arm-position trial enrolled participants of whom a substantial proportion already had systolic pressure at or above 130 mmHg, and the authors concluded the positions may lead to misdiagnosis and overestimation of hypertension. [3] A patient who is reclassified because their arm was hanging has been harmed by technique.
No, and the review does not say that. It says the device can overestimate true saturation in darker skin tones, that the effect is likely larger at lower saturations, and that the evidence base has substantial limitations. [5] Use it as one input, follow the trend within a session on the same finger, and let clinical signs override a reassuring number.
Honestly, it is untested. The meta-analysis found no evidence from outpatient settings, and its data are from hospitalised patients categorised by self-identified race or ethnicity rather than measured skin tone. [6] The device physics does not change when a patient walks in rather than being admitted, so the concern is reasonable — but calling it established for outpatients would be going beyond what was studied.
The session-RPE method has validity and good reliability confirmed across 36 studies in men and women of different age categories and expertise levels, and the review notes it can be used as a standalone method for monitoring training load, though some recommend pairing it with heart rate. [7] For how much load to prescribe once you can measure it, see exercise dose.
Often for reasons on this page rather than physiology: a different cuff, a different arm position, a different chair height. Both trials here were designed to isolate exactly those variables, and both found effects large enough to explain a disagreement. [3][4] Standardise the setup before you attribute the difference to the patient.
How to use this
This page reports how large the routine measurement errors actually are, in millimetres of mercury and beats per minute, from randomised and pooled evidence. Faculty may use this page in teaching with attribution. It carries its review date and its next review date, so you can see at a glance whether it is current before you put it in front of a cohort.