Sleep & CPAP Report — 2026-08-28

🚩 Flags

• 2-night streak of elevated obstructive events. OA count was 12 tonight and 14 last night, back-to-back — both well above the 30-day average of 8.2. Prior to Aug 27, only two nights in the last month (Aug 20 and Aug 21) hit ≥12 OAs. AHI has now printed above 2.4 for two consecutive nights (2.43 tonight, 3.17 last night) vs a 30-day baseline of 1.79. This meets the "3-night streak" trigger if tomorrow also lands above ~10 OAs — watch it.
• Central apneas: 3 tonight, 3 last night. Two consecutive nights tied for the highest CA count of the last 30 days. 30-day CA average is ~1. Small absolute numbers, but the pairing with elevated OA is worth noting.
• Pressure delivery ran unusually low. Pressure_Avg 6.08 cmH₂O (vs 30-day 7.45, -18%) and Pressure_90th 8.90 (vs 9.94, -10%) — the lowest 90th-percentile pressure of the last two weeks. The machine did not ramp up despite the elevated OA count. Worth checking whether pressure settings were changed, or whether events clustered outside the algorithm's response window.
• Leak was essentially perfect. Leak_Avg 0.28 L/min and Leak_90th 0.0 L/min are the best in the recent window (30-day avg Leak_Avg 0.81, 90th 2.48). Whatever is driving the OA increase, it is not mask fit.
• Sleep architecture stayed strong. Deep 13.5% (+0.5 pp vs 30d), REM 24.7% (+0.9 pp vs 30d), awake time only 6.0 min (vs 30-day 16.2), efficiency 98.4%. The respiratory-event spike is not translating into fragmentation — consistent with events clustering rather than distributing across the night.

Summary Table — Night of 2026-08-28

Metric	Value
Sleep start	23:19:12 CDT
Time in bed	387.9 min (6h 28m)
Total sleep	381.9 min (6h 22m)
Sleep efficiency	98.4%
Awake time	6.0 min
Core (N1+N2) sleep	235.8 min / 61.8%
Deep (N3) sleep	51.7 min / 13.5%
REM	94.3 min / 24.7%
CPAP usage hours	6.58
Session count	1
AHI	2.43
RDI	2.43
Obstructive apneas (OA)	12
Hypopneas (H)	1
Central apneas (CA)	3
Unclassified apneas (UA)	0
RERA	0
Pressure avg	6.08 cmH₂O
Pressure min	4.00 cmH₂O
Pressure 90th	8.90 cmH₂O
Pressure max	12.88 cmH₂O
Leak avg	0.28 L/min
Leak 90th	0.0 L/min
Leak max	10.8 L/min
Flow-limit p95	0.04
Flow-limit min >0.10	8.1
Flow-limit min >0.20	1.8

Trailing Averages Comparison

Metric	Tonight	7-day avg	Δ vs 7d	30-day avg	Δ vs 30d
AHI	2.43	1.86	+30%	1.79	+36%
OA count	12	9.0	+33%	8.2	+46%
CA count	3	1.3	+131%	1.0	+200%
Leak_Avg (L/min)	0.28	0.65	−57%	0.81	−66%
Leak_90th (L/min)	0.0	1.37	−100%	2.48	−100%
Pressure_Avg	6.08	7.37	−17%	7.45	−18%
Pressure_90th	8.90	9.80	−9%	9.94	−10%
Usage hours	6.58	6.81	−3%	6.72	−2%
Deep sleep %	13.5%	12.9%	+0.6 pp	13.1%	+0.5 pp
REM %	24.7%	25.9%	−1.2 pp	23.8%	+0.9 pp
Awake min	6.0	8.7	−31%	16.2	−63%
Total sleep (min)	381.9	393.8	−3%	384.9	−1%

Week-over-Week Trend

Metric	Recent 7d (Aug 22–28)	Prior 7d (Aug 15–21)	Direction
AHI	1.81	1.68	↑ +7.5%
OA count	8.57	8.29	↑ +3.4%
Leak_Avg	0.66	0.91	↓ −27.7%
Leak_90th	1.37	2.40	↓ −42.9%
Pressure_Avg	7.14	7.90	↓ −9.6%
Pressure_90th	9.62	10.62	↓ −9.4%
Deep sleep %	12.83	11.95	↑ +7.3%
REM %	25.80	23.01	↑ +12.1%
Awake min	9.18	13.46	↓ −31.9%

Underlying weekly trend is favorable — leak is materially down, deep and REM are up, awake time is down. The last two nights' OA spike is riding on top of an otherwise-improving week; that's what makes it stand out.

Recent 5-Night Detail

Date	AHI	OA	CA	Leak_Avg	Leak_90th	P_Avg	Hours
2026-08-28	2.43	12	3	0.28	0.0	6.08	6.58
2026-08-27	3.17	14	3	0.83	3.6	7.59	6.63
2026-08-26	1.33	7	1	0.72	0.0	—	6.77
2026-08-25	1.61	7	2	0.73	2.4	—	6.85
2026-08-24	1.81	9	0	0.37	0.0	—	6.63

Correlation Analysis (54 matched nights)

Pair	the user r
AHI ↔ Deep sleep %	+0.14 (weak positive)
AHI ↔ Awake minutes	+0.04 (none)
Leak_Avg ↔ Deep sleep %	−0.18 (weak negative)
Leak_Avg ↔ Awake minutes	−0.06 (none)
Leak_90th ↔ Deep sleep %	−0.21 (weak negative)
Leak_90th ↔ Awake minutes	−0.04 (none)

High- vs low-leak bucket comparison (split at Leak_90th median = 2.4 L/min): - High-leak nights (n=18): mean deep sleep 13.1%, awake 14.4 min - Low-leak nights (n=36): mean deep sleep 15.8%, awake 17.1 min

Directional pattern is consistent with prior reports: higher leak weakly associates with lower deep sleep, but the effect is small and noisy. Tonight — extremely low leak paired with elevated OA — is exactly the pattern the correlations would not predict.

Interpretation

For the second night in a row, obstructive events ran well above baseline while everything about mask fit and airflow looked pristine. Two features of tonight complicate the story:

1. Pressure delivery was noticeably lower than typical (avg 6.08 vs a 30-day 7.45). The AirSense algorithm chose not to ramp; either the events were too short to trigger a response, or something changed in the pressure ceiling/settings. It's worth verifying that no accidental setting change occurred (min pressure did hit the 4.0 floor, which is the machine minimum).
2. Sleep architecture is holding up. Deep and REM are both fine, awake time is minimal — so subjectively you'd likely feel rested. The events are being tolerated, not fragmenting the night.

The most actionable read: watch tomorrow. If night 3 also runs ≥10 OAs, this becomes a genuine pattern worth investigating (positional shift, congestion, alcohol, recent weight change, or a device-side setting drift). If tomorrow returns to the 7–9 OA baseline, treat Aug 27–28 as a two-night blip while leak was coincidentally excellent.

Caveats

• SpO2/pulse columns remain empty (device isn't recording those channels).
• Sleep-stage percentages are Apple Watch estimates (accelerometer + HR), not lab EEG.
• Single CPAP session, no mid-night mask-off artifacts to filter.
• Pressure_Min of 4.00 cmH₂O suggests the machine sat at its minimum for at least part of the night — normal at ramp-in but the 90th also being 8.90 (below usual) indicates the algorithm never demanded much pressure tonight.