Equipment, Gas Supply and Patient Data

In depth

The one-page summary above is what the PDF prints. These sections go further.

Medical gas identity and connections

Federal labeling rules (21 CFR 201.328) require high-pressure medical gas cylinders to be colored on the shoulder so the color shows from above, and portable cryogenic containers to carry a 360-degree wraparound label naming the gas. Color helps recognition but does not state a mixture, and colors differ by country (one Indian review describes oxygen cylinders with a black body and white shoulder). The label is what identifies the contents, so a cylinder with a torn or unreadable label is set aside and reported, even if the shoulder colors and the regulator fit look right.

Indexed connections are a second barrier. Small cylinders use a yoke with the Pin Index Safety System (oxygen pins 2 and 5); low-pressure connections such as wall outlets and flowmeter inlets use the Diameter Index Safety System, in which thread diameters differ by gas. A fitting that will not connect suggests the wrong gas: the response is to stop and read the label, never to change, adapt or force the fitting.

FDA and NIOSH (2006) traced fires at the regulator-valve junction to oxygen leaking across reused nylon crush gaskets, which can ignite from flow friction. Their advice: crack the cylinder valve briefly before attaching the regulator to blow debris out of the port; fit one clean seal, either a new single-use crush gasket or the metal-bound elastomeric washer made for repeated use; keep the valve, regulator and gasket free of oil and grease; tighten the T-handle by hand, never with a wrench; and open the post valve slowly, closing it at once if gas escapes at the junction.

Cylinders are kept upright and secured against falling, moved on a cart rather than rolled, inspected for damage, and never modified at the pressure relief device (CGA). Oxygen supports combustion: see PFTs and Oxygen Delivery for fire precautions and the cylinder factors used in duration math. Worked example: an E cylinder at 1,000 psi running at 4 L/min lasts about 1,000 × 0.28 ÷ 4 = 70 minutes.

Flowmeters and blenders in detail

A Thorpe tube is a float in a vertical tapered tube; it depends on gravity and reads correctly only upright. FDA classifies two kinds. The uncompensated tube is calibrated with its outlet open to the atmosphere, so back pressure from a kinked line or a high-resistance device changes the reading, classically making it read lower than the flow actually delivered. The compensated tube is calibrated at a reference outlet pressure (the needle valve sits downstream of the float) and stays accurate with downstream resistance; the float jumps briefly when it is plugged into a live outlet with the valve closed.

A Bourdon gauge flowmeter is a pressure gauge behind a fixed orifice, calibrated as flow with the outlet open (21 CFR 868.2300). It works in any position, which suits transport cylinders, but it cannot sense a downstream obstruction: a kink raises the pressure, so the needle reads the same or higher while the patient receives less. A cylinder needs a regulator ahead of a flowmeter.

Air-oxygen blenders proportion two 50-psi supplies. In one cleared design (a single 510(k) summary) the alarm sounds when the supply pressures differ by about 20 psi, and the higher-pressure gas then flows straight to the outlet until the difference is back to about 6 psi; other designs and settings differ. At a 21% setting an oxygen loss may not alarm, and a blender may stay silent if both supplies fall together. The response to a blender alarm is to check wall outlets, hoses and cylinders and to analyze the delivered oxygen, not to recalibrate the blender.

Liquid oxygen, concentrators, piped-gas failure and MRI

Bulk liquid oxygen is stored at about −160 to −180 °C (it boils at −183 °C); one liter of liquid yields about 861 L of gas, and it slowly evaporates even when unused, which matters for home liquid reservoirs. Supply is regulated to about 50 to 55 psi for flowmeters and ventilators. Oxygen-failure alarms sense pressure, not flow, so they do not detect a wrong gas in the line.

Concentrators pass room air through zeolite sieve beds that adsorb nitrogen and deliver about 92 to 96% oxygen when working well. Open Critical Care states that a device should have a low-oxygen alarm below 82%; WHO specifications require more than 82% at maximum rated flow, so individual devices differ. Common units deliver 3, 5, 8 or 10 L/min, and sieve beds have a finite life (about 25,000 hours is quoted). A concentrator needs electricity, so a home patient keeps a backup cylinder and tells the fire department, electric company and telephone company that oxygen is in use (MedlinePlus). A low-purity alarm means switching to the backup cylinder and calling the equipment provider.

If piped oxygen fails, each ventilated patient is moved to a cylinder supply or a manual resuscitator in order of need, the delivered FiO₂ is verified, and the cylinder inventory is managed until the system is restored. Before MRI, the ventilator, cylinder, pumps and monitors must be labeled for that MRI environment: the field may always be on, a steel oxygen tank at the door can be pulled into the bore, and freeing it may require quenching the magnet.

Oxygen analyzers

Electrochemical analyzers measure oxygen partial pressure, displayed as a percentage. A galvanic (fuel) cell makes its own current from a lead anode that is consumed, responds in about 20 seconds and is unaffected by water vapor. A polarographic (Clark) cell uses a membrane-covered electrode driven by the analyzer battery; its membrane deteriorates over about 3 years. Both are calibrated at two points, room air (21%) and 100% oxygen, before use; a cell that drifts or cannot reach 100% is replaced. Paramagnetic analyzers, used in anesthesia machines, are fast enough for breath-by-breath readings but need a water trap.

Reprocessing, sharps and leak tests

The Spaulding scheme sets the minimum: items entering sterile tissue or the vascular system are sterilized (biopsy forceps break the mucosa and count as critical); items touching mucous membranes or nonintact skin, including respiratory therapy and anesthesia equipment, bronchoscopes and laryngoscope blades, get at least high-level disinfection; items touching intact skin get low-level disinfection. Cleaning comes first, because disinfection fails on soiled surfaces.

For flexible scopes the sequence is a leak test before immersion, cleaning, high-level disinfection or sterilization, rinse, alcohol rinse and forced-air drying, and storage that protects the scope from recontamination. Tap water can recontaminate equipment with nontuberculous mycobacteria, Legionella and Pseudomonas, which is why CDC prefers a sterile-water rinse and accepts tap or filtered water only when followed by alcohol and forced air.

OSHA forbids bending, recapping or removing contaminated needles unless no alternative exists, and then only with a one-handed technique or a device; shearing or breaking needles is forbidden. Sharps containers are closable, puncture resistant and leakproof, upright, not overfilled, and as close as feasible to where sharps are used.

Blood gas quality control, worked through

Control results are plotted on a Levey-Jennings chart against the mean and standard deviations. In the classic multirule, a value beyond 2 SD is only a warning that prompts a look at the other rules. The run is rejected only if a rejection rule fails. Random error (scatter) is usually shown by 1-3s or R-4s; systematic error (bias, as from an aging electrode, protein on a membrane or calibration drift) by 2-2s, 4-1s or 10x. R-4s is read within one run; 4-1s and 10x are often read across runs and control levels, and computer QC can apply them without waiting for a 2 SD warning. A steady climb of one control over several runs, still inside the limits, is building systematic error and is fixed before it fails.

CLIA (42 CFR 493.1267) requires calibration or calibration verification as the manufacturer specifies, one control sample each 8 hours of testing with low and high values on each day of testing, a control each time patients are tested unless the instrument verifies its own calibration at least every 30 minutes, and documentation of everything. Electronic QC checks the sensors and electronics; control material still has to run on that schedule. After a failure, the laboratory must evaluate all patient results from the failed run and since the last acceptable run, act and document (493.1282). Proficiency testing samples are run in the routine workload by the staff who normally test, with no discussion with other laboratories, and records are kept 2 years (493.801).

Reading the chest radiograph

Check the film before the lungs. Rotation: the medial clavicle heads should be equidistant from the spinous processes; rotation distorts the mediastinum and the lung density. Inspiration: about ten posterior ribs at the midclavicular line. Projection: portable films are AP, magnify the heart by about 20% and are almost always suboptimal.

Lines and tubes: ETT tip about 3 to 5 cm above the carina (the carina lies at T4 to T5; one source cited gives 2 to 5 cm, and Kelly says only above the carina), knowing that neck flexion moves it toward the carina and extension away, by about 2 cm; central venous catheter tip at the cavoatrial junction (too high risks thrombosis and malfunction, too low arrhythmia); pulmonary artery catheter in a main pulmonary artery and never beyond the hilum; IABP tip about 2 cm below the left subclavian origin; enteric tube side port below the gastroesophageal junction, and a tube ending above the diaphragm may be in the airway.

Pneumothorax shows as a lung edge with no markings beyond it; end-expiratory views help detect small ones. In a supine patient air collects anteriorly and at the base, and an abnormally deep, lucent costophrenic angle (deep sulcus sign) may be the only clue. Tension is diagnosed clinically, not by waiting for a film; ultrasound is more sensitive than a plain film.

About 75 mL of fluid blunts the posterior costophrenic angle on an upright film, and a meniscus rises along the lateral wall; a supine effusion only makes one side look hazier. The opaque hemithorax: shift toward the opacity means collapse (a mucus plug, or a tube in the right mainstem bronchus occluding the left); shift away means a massive effusion (more than about 4 L) or mass; no shift suggests consolidation, or an effusion over a collapsed lung.

Chest examination and breathing patterns

Airless lung conducts sound, so consolidation gives increased fremitus, bronchial breath sounds, egophony (E heard as A) and whispered pectoriloquy over a dull area. Pleural fluid separates lung from wall: dull, with decreased or absent fremitus and breath sounds, and egophony at the upper border where compressed lung lies. Pneumothorax also lowers fremitus and breath sounds but is hyperresonant. Crackles suggest atelectasis, edema or interstitial disease; wheezes small-airway narrowing; a rub pleural inflammation.

Stridor is high-pitched and mainly inspiratory when the obstruction is in the extrathoracic upper airway; it can be expiratory or biphasic, and it signals a possible airway emergency. Breathing patterns: Cheyne-Stokes waxes and wanes in cycles with apnea (heart failure, neurologic disease, drugs); Biot breathing is irregular apnea alternating with 4 or 5 equal deep breaths that start and stop abruptly (central nervous system injury such as meningitis); Kussmaul breathing is deep and regular (metabolic acidosis); paradoxical breathing (the abdomen moves inward on inspiration) signals respiratory muscle fatigue or weakness.

Clubbing is diagnosed by a nail profile angle over 180° or a phalangeal depth ratio over 1; it is most common with lung cancer and is also seen in cystic fibrosis and idiopathic pulmonary fibrosis; a benign hereditary form starts early in life.

Central cyanosis (whole body, including tongue and mucous membranes) needs at least about 5 g/dL of unsaturated hemoglobin, roughly an arterial saturation below 85%; its absence does not exclude hypoxemia, and in severe anemia it may never appear. Peripheral cyanosis of the lips, fingers and toes reflects slow flow with high extraction, as in low cardiac output, while the arterial saturation stays normal; if vigorous massage improves nail-bed color, the cyanosis is likely peripheral.

Arterial sampling and specimen quality

Modified Allen test: with the hand raised and the fist clenched, both arteries are compressed until the palm blanches, the hand is opened and the ulnar artery released. Color back within 5 to 15 seconds is normal; slower means poor ulnar collateral flow, and another site (usually the other radial) is chosen. The test has marked interobserver variability and predicts hand ischemia poorly (pooled sensitivity 77%, specificity 93%), and Doppler is the better check where available, but exams and many policies still expect it before a radial puncture.

Capillary samples do not assess oxygenation, and venous values do not replace arterial PaO₂, PaCO₂ or pH (AARC 2013). Sampling and pre-analytic errors are covered on ABG Interpretation.

A sputum sample is worth culturing when the Gram stain shows more than 25 neutrophils and fewer than 10 squamous epithelial cells per low-power (×100) field. Many squamous cells with few neutrophils means saliva: the result would grow mouth flora, so a new deep-cough or induced specimen is collected.

Pulse oximetry in depth

In critically ill patients with an SaO₂ of 90% or more, oximeter bias is under 2% with precision under 3%; agreement worsens below 90%. FDA lists poor circulation, skin pigmentation, skin thickness, skin temperature, current tobacco use and nail polish as factors that affect readings. FDA issued its safety communication in 2021, held advisory panels in November 2022 and February 2024, and issued draft guidance on January 7, 2025 that is not yet final.

Sjoding (2020) defined occult hypoxemia as an SaO₂ below 88% despite an SpO₂ of 92 to 96%. It occurred in 11.7% of such paired readings in Black patients vs 3.6% in White patients in one hospital, and in 17.0% vs 6.2% across 178 hospitals. The practical rule: when a decision hinges on a few points, or the patient looks worse than the number, get an arterial sample on a CO-oximeter.

Low perfusion and motion are the commonest causes of error, and the plethysmographic waveform is the quality check. In critically ill patients with low perfusion, forehead reflectance sensors agreed better with arterial values than finger sensors. Black, purple and dark-blue nail polish raised readings slightly with wide error, and turning the probe sideways did not remove it.

Two-wavelength oximeters count carboxyhemoglobin as oxyhemoglobin, so SpO₂ stays falsely reassuring in carbon monoxide poisoning. Multiwavelength pulse CO-oximeters estimate COHb, but in suspected poisoning their limits of agreement were about −12% to +14%, they could not measure COHb below an SaO₂ of 85%, and pooled sensitivity at a 10% cutoff was only 0.65. Laboratory CO-oximetry confirms or excludes poisoning while 100% oxygen continues.

Respiratory muscle testing and the stress index

MIP and MEP are volitional tests. MIP is measured after a full exhalation to residual volume (values about 30% higher than from FRC) and MEP from total lung capacity, each a maximal effort held 1 to 2 seconds, sitting, through a mouthpiece with a small (about 2 mm) leak that stops the cheeks and a closed glottis from adding pressure. Do 3 to 8 maneuvers and record the highest; repeat efforts should agree within about 10%. Lower limits of normal: MIP 60 cm H₂O in women and 80 in men (reported as negative pressures), MEP 120 and 150. Air leaks around the mouthpiece in facial weakness give falsely low values.

In intubated or uncooperative patients a one-way valve lets the patient exhale but not inhale, so each effort starts from a lower volume with rising drive; the highest value usually comes after 15 to 20 seconds, and the measurement takes about 25 seconds. A brief occlusion underestimates strength and should not drive a weaning decision on its own.

The stress index describes the shape of the airway pressure-time curve during volume control with constant (square-wave) flow in a passive patient. Volume enters at a steady rate, so the slope reflects compliance. A straight rise (index 1) suggests neither tidal recruitment nor overdistension; a curve bending upward (over 1) means compliance falls as the lung fills, suggesting overdistension, and PEEP, tidal volume or both are lowered; a curve bending downward (under 1) suggests tidal recruitment and room to raise PEEP. Patient effort invalidates it. Plateau pressure and auto-PEEP are on Ventilator Modes.

Exam pearls

• Label over color, color over fit: a matching regulator never proves the contents.
• Hissing at the yoke: close the valve before touching the connection.
• Bourdon reads high when kinked; uncompensated Thorpe reads low.
• Blender alarm means a supply problem, not an FiO₂ problem; still analyze FiO₂.
• Galvanic analyzer: two-point calibration at 21% and 100%.
• Failed QC: fix, rerun, then look back to the last good run.
• Opaque hemithorax: toward = collapse, away = fluid.
• Supine trauma film with a deep lucent angle: pneumothorax.
• SpO₂ 93% in a darkly pigmented, distressed patient: check SaO₂.
• Stress index over 1: too much distension.