Wrist HR vs Chest Strap: What Community PPCS Prescription Assumes

Community sub-symptom prescriptions often treat wrist optical heart rate as interchangeable with the chest-strap channel used in evidence. Validation studies report device- and intensity-dependent limits of agreement tens of bpm wide—wider than a typical prescription band.
Published

July 3, 2026

Caution

Not medical advice. Educational use only. Data from published literature.

The prescription is already on the fridge: twenty minutes, most days, stay near a heart-rate number the clinic wrote down. The adolescent has a consumer watch. The parent has a job and a headache of their own. Nobody in the kitchen is checking whether the watch uses the same measurement channel the trial used.

They are thinking: the number on the watch is the number from the trial.

Kitchen math and trial math are not the same problem.

Community notes often copy trial heart-rate targets without copying trial measurement class. Validation studies report wrist–reference disagreement tens of bpm wide—often wider than a typical written band (Figures 1–7 below).

The narrower question is whether a wrist optical readout is statistically compatible with the chest-strap channel the evidence actually used, once the prescription is written as a heart-rate band. A school athletic trainer whose athletes all wear watches faces that every week. So does a community pediatrician who wrote “keep HR under 140” and assumes the family’s Apple Watch will enforce it, or a parent who advances the session because the watch “looked fine” while symptoms creep up the next morning.

Leddy’s adolescent RCT monitored home sessions with a Polar H7 chest strap, not a consumer wrist sensor PMID: 30715132. Published limits of agreement for wrist devices are often tens of bpm wide—sometimes >100 bpm—while a typical written HR band is much narrower (Muggeridge graded-intensity PDF; Sartor et al. PMID: 29881626; see source panel below). If you prescribe a %HR or target-HR band, use a chest strap; if only a wrist device is available, gate intensity with a symptom cap and/or RPE (defined later), and treat the watch digit as ancillary. Figures 1–7 plot the GLP claim panel that supports those moves: LOA intervals, a Bland–Altman schematic with published bounds, LOA widths, bandwidth ratio (R), bias versus width, lab→field paired widths, and same-study device contrast. Figure 4 and the SOP are the shortest path to the clinic decision.

1. Why the watch digit feels like evidence

Buffalo-style and related programs often prescribe at a fraction of a measured or estimated heart-rate threshold—commonly on the order of 80% of the HR that provoked symptoms, 20 minutes, most days of the week PMID: 30742254 PMID: 32079897. That wording sets an intensity target. It says nothing about whether a consumer wearable can enforce the target.

The Leddy 2019 RCT is explicit about the home channel: participants exercised “at the prescribed target heart rate (HR) while wearing a provided Polar H7 Bluetooth Heart Rate Sensor and Fitness Tracker to monitor HR” PMID: 30715132. In that trial, chest-strap monitoring is how the written target became an observable session—not a methods footnote.

Wrist devices do not measure cardiac electrical activity; they rely on photoplethysmography, and validation work reports that wrist-worn monitors fail to provide accurate readings during treadmill exercise PMID: 28709155. The kitchen assumption—watch number = trial number—quietly swaps measurement classes while keeping the same prescription language.

That swap is easy to miss because the interface looks scientific. A glowing digit updates every second. Parents can screenshot it. Coaches can ask athletes to “stay in zone.” None of those social facts changes the agreement statistics in Figures 1–4.

1.1 Clinic, access, and mixed messages

Sub-symptom aerobic exercise is no longer fringe in adolescent concussion care. In community settings the open question is usually how hard to go, and how anyone knows. When the answer is written as a heart-rate band, the monitoring channel becomes part of the prescription—whether or not the note says so.

When there is no treadmill, clinicians fall back to bike tests, age-predicted bands, RPE, and symptom caps—and should prefer a chest strap on any %HRmax path. What those approaches could not yet quantify—because the wearable validation panel had not been assembled—is how wide wrist–reference disagreement can be, or how often “acceptable” average error coexists with a corridor too wide for a narrow band—that measurement layer follows in Figures 1–7.

Two clinicians can both “use heart rate” and still disagree in practice. One issues a Polar-class strap and treats the number as a ceiling. Another assumes the family’s watch is close enough and advances dose because the log “looked green.” The adolescent hears the same diagnosis and the same written percentage, then lives two different real intensities. Families get blamed for noncompliance when the real problem is often unmeasured channel error.

1.2 Trial prescription vs validation studies
Literature Question Example
Prescription evidence What device did the clinical evidence use? Leddy home Polar H7 chest strap
Validation studies How wide is wrist–reference disagreement? Bias, LOA, (W_{}), (R) in Figures 1–7

The prescription literature answers what the evidence assumed you could measure. Validation studies answer how wrong a consumer wrist reading can be relative to a reference. Mixing them produced a published error: a validation-style ±10 bpm margin was attributed to Worts et al. 2019, but full-text search does not recover that claim (refuted in the GLP handoff verification). Worts still matters for prescription framing—BCTT-anchored programming at 80% of provocation HR PMID: 30742254—and supplies none of the validation literature’s numeric corridor.

The clinical concern survives the errata: community wrist readouts may not match trial chest-strap conditions, so intensity can be mis-ranked. Only the evidentiary anchor changes—from an unsupported Worts margin to the validation panel plotted below.

1.3 What “close enough” would have to mean

Suppose a clinician writes “stay between 125 and 140 bpm.” That band is 15 bpm wide. For a wrist channel to decide whether a session stayed inside that band, its individual-level disagreement with reference cannot routinely span a wider corridor than the band itself—when the error corridor is wider than the prescription band, the watch cannot adjudicate the band, regardless of how neat the ratio looks on paper.

Figures 1–4 make that requirement visible with published limits of agreement (Figure 4 plots an illustrative ratio (R = W_{}/W_{}) for readers who want a single scalar). Figures 5–7 show why metrics that sound reassuring—high correlation, “acceptable” MAPE, near-zero bias—can still leave the bandwidth test unmet.

Band-width sensitivity (15 vs 30 bpm example)

(R) scales inversely with (W_{}). The table uses the same published LOA widths as Figures 1–4:

Device arm (W_{}) (R) at 15 bpm (R) at 30 bpm
Lab Fizzo ~12 0.8 0.4
Polar OH1 39 2.6 1.3
Muggeridge TM 22 1.5 0.7
PE field 72.6 4.8 2.4
Fitbit Charge 3 79 5.3 2.6
Apple Watch (Sartor) 135.5 9.0 4.5

Widen the example band and Muggeridge treadmill flips from fail to pass at (R=1)—which is why the rule is a teaching scalar, not a universal clearance test. Buffalo-style 80% of provocation HR bands are often wider than 15 bpm once threshold is individualized; recompute before treating Path 3 as safe. The default for home HR-band prescriptions stays the same: strap for the band; wrist ancillary.

2. Agreement statistics (what the figures plot)

Bias, limits of agreement, derived widths, and the bandwidth ratio have to sit on the same page before the clinical story in Section 1 becomes operational. The definitions below feed Figures 1–7.

Statistic Definition Role in figures
Bias Mean (device − reference), bpm Marker on LOA interval; x-axis in Fig 5
LOA Published limits of agreement, bpm Horizontal interval in Fig 1–2
(W_{}) (-) Figs 3, 5, 6, 7
(R) (W_{}/W_{}) with (W_{}=15) bpm (illustrative band); (R>1) means error corridor wider than example band Fig 4

All plotted values are GLP verified published summaries. No patient-level points are available in the handoff corpus, so Figure 2 is a Bland–Altman schematic annotated with published bounds rather than a re-analysis of raw differences. Published corridors still matter without the scatter: when a paper reports LOA of −46 to 33 bpm, the clinical reader already knows the channel is blurry relative to a 15 bpm band.

The source panel below is the single numeric ledger for every figure. If a number does not appear here, it should not appear in the plots.

Identifier correction (Muggeridge / Sun–Fizzo rows)

GLP identifier_warnings flagged that database PMID tags for these two rows might not match PubMed titles. Independent PubMed DOI lookup confirmed the tags had been swapped with each other and with an unrelated third paper: the Muggeridge Polar OH1/Fitbit Charge 3 study’s correct identifier is PMID 33764310 (JMIR Mhealth Uhealth 2021;9(3):e25313), and the Sun/Liu Fizzo PE-lessons study’s correct identifier is PMID 32663136 (JMIR Mhealth Uhealth 2020;8(8):e17699). The previously-referenced PMID 34957939 belongs to an unrelated Apple Watch 6/Polar Vantage V/Fitbit Sense validation study (Hajj-Boutros et al.) and does not appear in this piece. Quotes and numbers were PDF-verified independent of this labeling error and are unaffected; only the citation identifiers changed.

Device / setting Bias LOA (bpm) (W) Source (PDF-verified)
Polar OH1 vs H10 −1 −20 to 19 39 Muggeridge et al. PMID: 33764310
Fitbit Charge 3 vs H10 −7 −46 to 33 79 Muggeridge et al. PMID: 33764310
Fizzo vs Polar — PE field −2.60 −38.89 to 33.69 72.6 Sun & Liu PMID: 32663136
Fizzo vs Polar — lab running +0.64 −5.18 to 6.45 11.6 Sun & Liu PMID: 32663136
Apple Watch vs chest-strap ref. — −58 to 77.5 135.5 Sartor et al. PMID: 29881626
Polar M600 — treadmill — −10.7 to 11.3 22 Horton et al. PMID: 29135785
Polar M600 — wider arm — −39.6 to 27.2 66.8 Horton et al. PMID: 29135785
Figure 1 — LOA interval (forest) plot
Figure 1. Limits of agreement (bpm) for device − reference. Horizontal segment = published LOA; orange diamond = bias when reported; vertical dashed line = 0 (perfect mean agreement). Study-specific; not a pooled meta-analysis.
Device − reference (bpm) 0 −60 −30 +30 +60 +80 Polar OH1 Fitbit Charge 3 Fizzo PE (field) Fizzo lab M600 treadmill M600 wide arm Apple Watch ◆ bias

Reading Figure 1. Longer segments mean wider individual-level disagreement. Lab Fizzo is short and hugs zero—the look of a channel that could, in principle, police a narrow band. Fitbit, PE field, and Apple Watch segments span tens of bpm on both sides of zero. Polar OH1 sits in between: better than the consumer watch arms, still wider than a 15 bpm example prescription band. The forest layout is intentional: it is the standard way agreement studies display LOA, and it prevents the false comfort of reading only a single “average error” number.

Figure 2 — Bland–Altman schematic (published bounds)
Figure 2. Bland–Altman schematic. Vertical axis = device − reference (bpm). Horizontal dashed lines = published LOA for three arms; solid orange = bias. No patient-level points (not available); bounds are verified published summaries.
Device − reference (bpm) Mean of methods (schematic) 0 OH1 +19 −20 Fitbit +33 −46 PE field bias LOA

Reading Figure 2. A prescription band lives near the zero line. When LOA lines sit far above and below zero, a single watch reading can land well outside the band even if bias is small (PE field).

Figure 3 — LOA widths (W_{})
Figure 3. Derived LOA width W = LOA_high − LOA_low (bpm). Gold = example prescription band width (15 bpm).
0 50 100 140 bpm Rx 15 Lab 11.6 M600 TM 22 OH1 39 M600 wide 66.8 PE field 72.6 Fitbit 79 Apple Watch 135.5 Rx band Lab Fizzo M600 TM OH1 M600 wide PE field Fitbit AW Sartor
Figure 4 — Bandwidth ratio (R=W_{}/15)
Figure 4. Illustrative bandwidth ratio R = W_LOA / W_Rx with example W_Rx = 15 bpm (not a universal clinical cut-point). Red dashed line at R = 1. Illustrative rule used in this piece: if R > 1, the channel cannot default as adjudicator of that example band.
R = 1 0 4 8 Lab 0.8 M600 TM 1.5 OH1 2.6 M600 wide 4.5 PE 4.8 Fitbit 5.3 AW 9.0

Reading Figure 4. Only the laboratory arm is green ((R<1)). Every community-relevant optical row fails the rule.

Figure 5 — Bias versus LOA width
Figure 5. Bias (x) versus LOA width (y) for arms with both statistics. PE field sits near bias = 0 but high width—small mean error does not imply a usable narrow band.
Bias (bpm) W_LOA (bpm) −8 0 +2 0 40 80 OH1 Fitbit PE field Lab
Figure 6 — Lab versus field (paired widths)
Figure 6. Same device family (Fizzo vs Polar): laboratory running W = 11.6 bpm versus PE-lesson field W = 72.6 bpm (bias −2.60; ICC 0.742; MAPE 8.89%). Paired contrast within one PDF.
Lab PE field 11.6 72.6 W_LOA (bpm) ≈ 6× wider in field
Figure 7 — Same-study device contrast (Muggeridge)
Figure 7. Polar OH1 vs Fitbit Charge 3 vs Polar H10 in one graded-intensity protocol: bias, LOA width, and correlation. Holding design fixed, device choice roughly doubles W_LOA.
|Bias| (bpm) W_LOA (bpm) r OH1 1 39 0.95 Fitbit 7 79 0.80 MAPE for both devices reported “acceptable (<5%)” in the same paper—coexisting with these widths. Fitbit vigorous cycling: r=0.183, MAPE=−16.4% (intensity collapse).

Gillinov et al. report concordance coefficients that vary by device (Apple Watch (r_c = 0.92), TomTom Spark (r_c = 0.83)) and modality (“on the elliptical trainer without arm levers, only the Apple Watch was accurate ((r_c = 0.94))”) PMID: 28709155. Horton’s treadmill-versus-wider-arm contrast in the source panel and Figure 3 (widths 22 vs 66.8 bpm) makes the same point inside one optical device: modality changes the corridor PMID: 29135785.

“Any watch” is too vague to be a measurement class. Use Figures 1–4 for bandwidth; Figures 5–7 when a summary statistic looks reassuring but the corridor is still wide.

3. Three operational disagreements (device channel)

The figures settle a measurement argument. In clinic, the same numbers still get argued about—for reasons that have more to do with device class than with taste.

Disagreement 1: “The watch is good enough for home.”

Parents and some clinicians treat consumer wearables as close substitutes for trial monitors because the devices are ubiquitous and the digit is continuous. Figures 1–4 say otherwise for narrow %HR bands. Even Polar OH1—among the better optical arms in the panel—has (W_{}=39) bpm and (R=2.6) against a 15 bpm example band. Fitbit Charge 3 reaches (W=79) and (R=5.3). Apple Watch progressive-exercise LOA width is 135.5 bpm.

“Good enough” would require stating a tolerance: good enough for what decision? Motivation and rough pacing are one decision class. Adjudicating membership in a 15 bpm prescription band is another. The panel supports caution for the second class. For a school program without strap budget, I would write symptoms + RPE as the primary stop rule and treat the watch as a log—not keep Path 3 because the athlete already owns an Apple Watch.

Disagreement 2: “MAPE under 5% means we can trust the zone.”

Muggeridge et al. PMID: 33764310 deemed overall MAPE “acceptable (<5%)” for both OH1 and Fitbit while publishing LOA widths of 39 and 79 bpm. MAPE summarizes relative average error. A narrow band needs control of individual-level scatter. Figure 2’s Bland–Altman schematic is the right mental model: the band lives near zero; the LOA lines may not.

The same paper’s intensity collapse for Fitbit (vigorous cycling (r=0.183), MAPE=−16.4%) shows that a device can look acceptable in an overall summary and still fail in the exact intensity neighborhood where a PPCS prescription often lives—submaximal aerobic work that is not quiet rest.

Disagreement 3: “If bias is small, the reading is safe.”

Figure 5 places PE field near bias = −2.60 bpm with (W=72.6) bpm. Figure 6 shows the same device family moving from lab width 11.6 to field width 72.6. Small bias means the center of the error distribution is near zero; each session’s error can still be large. For a stop/go rule keyed to a narrow HR ceiling, scatter dominates.

Direction of bias (secondary to width)

Reported mean biases in the community-facing arms are small: Polar OH1 −1 bpm, Fitbit −7 bpm, PE field −2.60 bpm. Against LOA widths of 39–135 bpm, those centroids sit inside the scatter that already breaks narrow-band adjudication (Figures 1–4). Negative bias could occasionally produce false reassurance—a watch reading 135 when true rate is 142—but that is a footnote to width, not a second pillar of the argument. The operational move does not change: if the prescription is a tight HR band, do not let a wrist digit be the stop rule; use strap, symptoms, or RPE.

4. School, clinic, and home

Match the control variable to what the setting can actually measure. Do not start from whichever device the family already owns and reverse-engineer the rule.

Below, Path 2 is a home heart-rate band read by chest strap; Path 3 is the same band read from a wrist watch; Paths 4–5 are symptom cap or RPE with no HR adjudication. The full five-path table is in §5.

School athletic training room

No treadmill; referral wait measured in weeks; athletes already wear watches; staff can supervise short after-school sessions.

Copying a clinic HR band onto the watch face and treating green digits as adherence invents precision the channel does not have. If you need a numeric HR ceiling, issue or require a chest strap for those sessions (Path 2). If you cannot, write symptom cap + RPE as the primary controls and demote the watch to optional logging (Paths 4–5). Document wrist_optical so the next clinician does not invent precision that was never measured.

Outpatient PT / community clinic

Bike available; periodic visits; can teach device use; may see PPCS at 4–16 weeks.

Do not assume the home watch continues the in-clinic strap session without saying so. Keep the same channel at home as in the evidence you are imitating. If the plan is %HR or target HR, send the athlete home with a strap protocol. Use clinic time to practice donning, Bluetooth pairing, and the stop rule. If the family will only use a watch, rewrite the plan around symptom/RPE controls instead of pretending the channel matches Leddy’s Polar H7 arm PMID: 30715132.

Home / primary care only

No graded test; parent supervision; variable monitor quality; need plain-language rules.

“Keep HR under 140” with no device class and no symptom rule leaves intensity ungoverned. Lead with symptom cap and talk-test/RPE language. If a strap can be obtained cheaply and the family will use it, a conservative HR ceiling can be added as a secondary guardrail. Otherwise keep an unvalidated wrist digit from becoming the only governor of intensity.

Vignette

A 15-year-old has headache and cognitive fog six weeks after sport-related concussion. The clinic writes: walk 20 minutes, five days a week, keep heart rate near 80% of the provocation HR measured on a prior exertional test—translated for the family as “stay under about 140.” The family owns an Apple Watch. No chest strap is provided.

Under Figures 1–4, that plan has a silent measurement gap. Sartor’s progressive-exercise LOA width for Apple Watch HR versus a chest-strap reference is 135.5 bpm ((R=9.0) against a 15 bpm example band) PMID: 29881626. Even if another optical device were substituted, Muggeridge’s Fitbit arm ((W=79)) and PE field arm ((W=72.6)) still fail the bandwidth rule for narrow adjudication.

Work through the plan in order. The control variable is a HR band, as written. If a chest-strap class channel is available this week, issue it and keep the band (Path 2). If not, rewrite around symptoms and RPE (Paths 4–5); the watch can stay for motivation, but the digit is not the stop rule. If dose advances because “sessions were completed,” record which channel defined completion—a wrist-only green log is weak evidence of zone adherence when (W_{} W_{}).

What to tell the family:
“The clinic wrote a heart-rate ceiling because the test used heart rate. At home that means the chest strap we showed you—not the watch—if we are going to use the number. If the strap is not on, ignore the watch for go/stop and use the symptom and talk-test lines on the plan (full sentences, RPE about 11–14). The watch can stay on for motivation; it does not decide the session.”

The symptom/talk-test stop rule is an operational convention (about +2 points on a 0–10 scale is common in community prescribing); it is outside this piece’s verified HR-agreement panel.

5. Five monitoring paths

Path Device Deployability Compatible with narrow %HR band?
1. BCTT + clinical monitor Facility chest / clinical Low Yes — for setting the target
2. %HR + chest strap Polar H7–class Medium Yes — matches RCT home channel
3. %HR + wrist PPG Consumer watch High No as default for narrow bands ((R>1) at 15 bpm; see sensitivity table)
4. Symptom cap None Highest N/A — different control variable
5. RPE 11–14 None High N/A — different control variable

Paths 1–2 preserve the chest-strap measurement class the prescription evidence used. Path 3 maximizes deployability but imports the validation literature’s LOA problem. Paths 4–5 switch the control system when Path 3 fails the bandwidth rule—the same move as when treadmill access is missing, now applied to monitoring rather than to threshold testing.

Symptom cap (full definition)
  1. Rate concussion-related symptoms 0–10 at session start (or the clinic’s usual scale).
  2. Recheck during aerobic exercise.
  3. Stop when symptoms rise beyond the clinic’s written symptom-limited allow-band (community prescribing often operationalizes that as about +2 points on a 0–10 scale—an operational convention, outside this piece’s GLP verified HR-agreement panel).
  4. No heart-rate device is required.

Symptoms make a rough intensity meter at best. When HR agreement is too wide, a symptom-gated stop rule is still an explicit, teachable control that does not pretend to deliver trial-grade %HR accuracy. Prescription-facing reviews discuss symptom-limited and home-exercise frameworks alongside HR-based programming PMID: 30742254, without supplying a wrist-PPG accuracy margin.

RPE band (full definition)
  1. Use a Borg-type rating of perceived exertion (classic 6–20 scale).
  2. Keep steady aerobic work in an 11–14 band: roughly “light to somewhat hard,” able to speak in short sentences.
  3. Stop or downshift if RPE climbs past the band or the symptom cap trips.
  4. A watch may stay on the wrist for motivation; it does not adjudicate the session when Path 3 is rejected.

RPE inherits subjective-scale noise, not Table B’s LOA—a different error model, and one families can be coached on without claiming ECG-grade precision.

After Figures 1–4, an HR-band prescription maps to a chest strap (Path 2). Wrist PPG alone maps to Paths 4–5, with watch HR kept ancillary.

6. SOP checklist

One-page SOP (monitoring channel)
Clinician Parent Adolescent
□ Write control variable: HR band / symptom cap / RPE / mix □ Know which rule actually stops the session □ Rate symptoms 0–10 at start and mid-session
□ If HR band: issue/require chest strap (H7-class) □ Strap on before the walk; phone paired □ Number is a ceiling, not a goal
□ If only wrist: rewrite plan to symptom/RPE primacy □ Ignore watch HR for go/stop (LOA too wide for band) □ Stop per written symptom allow-band
□ Document device class in the note □ Log device used each day □ Talk-test / RPE 11–14
□ Progression notes name the channel □ Do not advance duration and intensity same day □ Report next-morning symptom carryover
□ Revisit channel if plan plateaus or flares □ Photo the written plan for coach/school □ No contact sport without clearance

Short version for parents:
“If we gave a heart-rate number, use the chest strap. Wrist watches can disagree with reference heart rate by many tens of beats during exercise—and they often read low, so a ‘safe’ digit can hide a higher true rate. If you only have a watch, ignore the HR digit for go/stop: use the symptom allow-band and talk-test / RPE 11–14 rules on your written plan.”

Tie-in to progression logic

Some programs advance dose when the next scheduled session is completed (a session-ramp rule—call it Logic C here: each completed session authorizes the next dose step). That logic assumes the readout you trust. If the readout is a wrist digit with (R>1), “completed in zone” is a soft claim. Either attach a strap-class channel or define completion with symptoms/RPE.

7. Errata — Worts ±10 bpm attribution

Originally published elsewhere in this series: wrist PPG “may carry ±10 bpm error versus chest ECG used in BCTT,” attributed to Worts et al. PMID: 30742254.

Correction: that numeric attribution is refuted on full text. Worts supports BCTT-anchored 80% provocation-HR programming (prescription evidence). Device-error magnitudes come from validation LOA/bias—Figures 1–7 in this piece.

Layer After correction
Clinical concern (community wrist ≠ trial chest conditions) Retained
±10 bpm from Worts Removed
Replacement anchors Figures 1–7 (Gillinov mechanism; Muggeridge / Sun–Fizzo / Sartor / Horton agreement stats)

This page states the correction in full.

8. Limitations

  1. No patient-level re-analysis. Widths are computed from published LOA bounds. Figure 2 is explicitly schematic.

  2. No pooled meta-LOA. Rows are study-specific; devices, protocols, and populations differ. Descriptive min/median/max language is not a random-effects summary.

  3. Population mismatch. Validation samples are largely healthy adults or school PE cohorts, not adolescent PPCS trials. Small mean biases may not transfer; LOA widths do.

  4. Identifier correction. GLP’s database had swapped PMID tags across three papers (Muggeridge; Sun/Liu; and an unrelated Hajj-Boutros et al. Apple Watch 6 study). Verified by PubMed DOI lookup: Muggeridge = PMID 33764310; Sun/Liu = PMID 32663136. Quotes and derived numbers were PDF-verified independent of this labeling error and are unaffected.

  5. Symptom cap / RPE. Different error models (subjective scales). The “about +2 points” allow-band is an operational convention in community prescribing; it does not appear in this piece’s verified HR-agreement panel.

  6. Example band width. (W_{}=15) bpm and the (R>1) threshold are illustrative decision aids for this piece, not validated clinical cut-points. A wider written band changes (R) (see sensitivity table in §1.3); it does not make wrist and chest channels interchangeable.

  7. Figure panel. Figures 1–7 derive from one verified source table (~7 device arms). Multiple views restate the same LOA widths—interval, schematic, bar, ratio, bias scatter, lab→field, cross-device—useful for different reader entry points, not seven independent datasets.

9. Closing

Community PPCS care often copies trial heart-rate targets without copying trial heart-rate measurement class. The flagship adolescent RCT monitored home sessions with a chest strap PMID: 30715132. Once you plot the validation panel, optical agreement under exercise and field conditions turns out to be wide, device-specific, and scene-dependent—LOA widths from lab-like ~12 bpm to progressive-exercise ~136 bpm; same-study Fitbit width 79 bpm with −7 bpm mean bias; PE field width ~73 bpm despite near-zero bias; overall MAPE “<5%” sitting beside those widths.

Use a chest strap when the written plan is a heart-rate band. When only a wrist device is available, rewrite the plan around symptoms and effort, and treat the watch digit as ancillary—not as the number that decides the session.

Series: Foundation Stack · methods · wearable HR error margins