ICU and You · Procedures · Number 5
Choosing the patient, the vein and the size — and several things you were taught that will not survive contact with the evidence
A PICC is the easiest central line to put in and the hardest one to justify. Almost nothing goes wrong at insertion. Most of what goes wrong happens over the following fortnight, to a patient somebody else is looking after, for reasons that were determined before the needle went anywhere near the arm.
Which makes this a procedure where the thinking beforehand matters more than the hands. So this piece spends its first half on decisions and its second half on technique, and that proportion is deliberate.
| Device | Tip sits | Realistic dwell |
|---|---|---|
| Peripheral cannula | Wherever it is | 3–5 days |
| Long peripheral | Mid upper arm, short of the axilla | Up to about 4 weeks |
| Midline | Axilla or infraclavicular — not central | 2–4 weeks |
| PICC | Cavoatrial junction | Weeks to months |
| Non-tunnelled CVC | Cavoatrial junction | Days to weeks |
| Tunnelled CVC / port | Cavoatrial junction | Months to years |
A midline is not a central line. It goes in the same way, it looks much the same coming out of the arm, and it confers none of a PICC's tolerance for difficult infusates. Vasopressors, concentrated potassium and parenteral nutrition do not belong in one.
The naming around these devices is genuinely inconsistent between services — “extended dwell”, “long peripheral” and “midline” are used interchangeably and wrongly. Do not trust the label on the handover sheet. Ask the only question that matters: where is the tip?
There is a structured answer to this, and it is worth knowing by name because it will be quoted at you. MAGIC — the Michigan Appropriateness Guide for Intravenous Catheters — used a formal consensus method with a fifteen-member multispecialty panel to rate device choice against indication and duration.
| Expected duration | Appropriate device |
|---|---|
| 5 days or fewer | Peripheral cannula. PICC is rated inappropriate. |
| 6–14 days | Midline preferred over PICC |
| 15 days or more | PICC appropriate |
| 31 days or more | Tunnelled catheter or port |
That table assumes the infusate could go peripherally. If it could not, the calculus changes immediately — that is the one thing that justifies a PICC at short notice.
Multi-lumen PICCs placed “so we have a spare” are rated inappropriate. So is a second lumen to separate sampling from infusion. Single lumen is the default, and lumen count is an independent risk factor for thrombosis. Intensive care asks for triple lumens by reflex, and most of the time the reflex is wrong.
And be honest about the weak link. MAGIC's own panel conceded that clinicians predict expected duration of access badly — which is the single input the whole algorithm runs on. In intensive care that prediction is worse than anywhere.
There is a paediatric version, miniMAGIC, partly Australian in authorship, and it is not simply the adult table with smaller numbers. In neonates a PICC becomes appropriate at eight days rather than fifteen, and implanted ports are rated inappropriate at any duration. Midlines in infants are rated uncertain — the adult midline evidence does not transfer. And PICCs for paediatric malignancy and for parenteral nutrition are also rated uncertain rather than appropriate, which is a real divergence from what adult practice would predict.
You have been taught that osmolarity above 600 mOsm/L, or a pH below 5 or above 9, mandates central access. Those numbers are quoted with great confidence. Here is where they come from.
A 1998 experiment in the ear veins of male Japanese white rabbits. The tolerated osmolality was duration-dependent: about 820 mOsm/kg at eight hours, 690 at twelve, and 550 at twenty-four.
“600” is a rounded midpoint of that curve — and the finding that actually mattered, that the threshold moves with how long you run the infusion, is dropped every time the number is repeated.
The American nutrition societies say 900 for peripheral parenteral nutrition. The European guideline says 800–850 and adds, in terms, that there is not enough evidence to indicate a clear cut-off. Thresholds in use across organisations run from 500 to 850. The international infusion standards grade the whole question at their lowest evidence tier.
As for pH under 5 or over 9 — I could not find a primary source for those numbers at all. They appear to be inherited convention.
What is solid is vesicant status. Anthracyclines, vinca alkaloids, concentrated vasopressors and concentrated potassium cause tissue necrosis if they extravasate, and that is mechanism rather than arithmetic. MAGIC's approach of reducing the whole question to a binary — peripherally compatible or not — is more honest than a numeric threshold pretending to a precision nobody has.
This is the section most likely to save somebody something they cannot get back.
Every arm vein you use is a fistula somebody may need in three years. Australian guidance advises vein preservation from CKD stage 3b — an eGFR under 45 — in anyone likely to require dialysis, and states that PICCs should be avoided in chronic kidney disease and considered only when there is no alternative.
MAGIC independently arrived at the same eGFR threshold, which is the strongest argument for the rule. Preserve the non-dominant forearm; use the dorsal veins of the dominant hand for peripheral access; and where prolonged access is genuinely needed, a small-bore internal jugular line is preferred to a PICC.
The evidence is observational and confounded — patients who get PICCs are sicker, and sicker patients have worse fistulas. But prior PICC is associated with roughly a threefold odds of not having a functioning fistula, and the alternative costs almost nothing. That asymmetry is what makes the recommendation reasonable despite the evidence being imperfect.
Infection guidance prefers the subclavian site for non-tunnelled central lines. Renal vein preservation prefers the internal jugular and specifically warns against subclavian, because subclavian catheterisation is associated with central venous stenosis in the region of 48 per cent against 22 per cent without.
These two guidelines genuinely disagree. In a patient heading for dialysis, the renal consideration should win. Trainees should know the documents conflict rather than assume they have misread one.
For scale: PICCs themselves are associated with central vein stenosis in about 4.8 per cent and occlusion in 2.7 — real, but an order of magnitude below subclavian.
“Never use the arm on the side of a mastectomy or axillary clearance” is taught everywhere. The best study — nearly 7,900 cannulas — found two complications in 2,743 contralateral placements against two in 5,153 ipsilateral ones, and concluded that avoiding the ipsilateral arm was an unnecessary practice. ANZCA guidance says there is no contraindication to ipsilateral skin puncture.
But do not over-extrapolate. That evidence is about cannulas and venepuncture. A PICC is a large-bore, long-dwell device with a measurable thrombosis rate, and nobody has shown it is safe in a cleared axilla. The honest position: the blanket ban on ipsilateral needles is not supported; caution about ipsilateral PICCs remains sensible and is also unevidenced. Established lymphoedema is a reasonable relative contraindication on mechanical grounds.
The remaining contraindications are more straightforward. Absolute: infection or burn at the site; known thrombosis or occlusion of the target or draining vein; a vein too small for the smallest acceptable catheter; superior vena cava obstruction; refusal. Relative: an ipsilateral pacemaker or defibrillator (lead-related venous narrowing is common, so use the other arm — sensible anatomy rather than evidence), prior ipsilateral central stenosis, and active bacteraemia, which is a question of timing rather than of site.
Coagulopathy, incidentally, is an argument for a PICC rather than against one — the insertion site is compressible, which is more than can be said for a subclavian.
Largest calibre, most direct route to the axillary vein, straight course with fewer obstructing valves, and it sits medially and superficially — away from the brachial artery and median nerve.
Paired, smaller, and sitting in the neurovascular bundle immediately beside the brachial artery and the median nerve. Identify both on short-axis ultrasound before you needle anything.
Smaller and more variable, and it joins the axillary vein at a sharp angle at the deltopectoral groove. If your catheter stops at the shoulder, this is usually why.
Aim for the middle third of the medial upper arm, roughly 12 to 15 cm above the medial epicondyle. Too low and the site sits in the antecubital region where arm circumference changes with elbow flexion and vessels converge; too high and you are near the axilla, where hair, moisture and movement defeat the dressing. There is a trademarked scheme that divides the arm into zones and names them by colour — the underlying anatomy is sound, the branded protocol rests on a forty-catheter personal series, so learn the reasoning rather than the trademark.
The rule is that the catheter should occupy no more than 45 per cent of the vein's diameter. The number comes from an Australian prospective cohort of 136 patients that recorded four symptomatic thromboses. Above 45 per cent against at or below, the relative risk was 13 — with a confidence interval running from 1.45 to 122.8.
That is a signal, not an estimate. The authors themselves described the previously used 33 to 50 per cent range as arbitrary. Use 45, because it is the best number available and the international standards have adopted it — but do not quote it as though it were established.
The better-evidenced version of the same idea is simply absolute catheter diameter. A meta-analysis of forty studies found symptomatic DVT rising steadily with gauge: 0.89 per cent at 3 French, 3.26 at 4, 5.46 at 5, and 10.66 at 6. Heterogeneity was high, so hold the absolute values loosely — but the gradient is unambiguous and it is entirely within your control.
Smallest diameter that will do the job. Fewest lumens. Basilic vein. Ultrasound for every step. Everything else is downstream of those.
Lower third of the superior vena cava, at the cavoatrial junction, or the upper right atrium. The principle underneath, which explains both failure modes at once, is that the tip should lie parallel to the vessel wall in high-flow blood.
NSW guidance states that if real-time tip confirmation can be verified during insertion, post-procedure chest radiographs are not required, and describes the intracavitary ECG method as accurate, precise and cost-effective.
Four qualifications. It is state-level guidance, not national — check your own service. The ECG method needs a discernible P wave, so atrial fibrillation and paced rhythms defeat it, which is not a rare situation in intensive care. A film is still needed if there is any concern about a mechanical complication. And paediatric practice is more conservative — the major Australian children's hospitals require radiographic confirmation before use, full stop.
One paediatric detail worth carrying even if you mostly do adults: for a lower-limb PICC, the tip sits at or just above the diaphragm, and L1 is specifically avoided because that is where the renal veins enter.
Two pieces of standard teaching need correcting here, and they point in opposite directions.
The landmark meta-analysis — 64 studies, 29,503 patients — found PICCs carried about 2.5 times the odds of DVT compared with centrally inserted catheters, with a pooled incidence in the critically ill of nearly 14 per cent. A later analysis of over 109,000 patients reproduced the odds ratio almost exactly.
Two things qualify it. Restricted to randomised trials, the PICC-versus-CVC difference was not statistically significant — the association is driven by observational data, in which sicker and less mobile patients preferentially receive PICCs. And when analysis is restricted to studies using modern technique — ultrasound, documented tip verification, deliberate size selection — the pooled rate falls to 2.4 per cent.
So the useful message is not a frightening odds ratio. It is that the risk is substantially determined by how you place the line.
The belief that PICCs are less infective than central lines is an artefact of where they are used. The meta-analysis that is usually cited in their favour found an overall relative risk of 0.62 — but that advantage was almost entirely in outpatients (RR 0.22). In hospitalised patients it was marginal. And per catheter-day, the incidence rate ratio was 0.91 with a confidence interval from 0.46 to 1.79 — no difference at all.
Choose a PICC for infusate, duration or access. Do not choose one because you think it is cleaner.
The rest, with the honest caveat that several of these have no trustworthy denominator: occlusion is common and variably reported; dislodgement affects around 6 per cent of PICCs in Australian data and is the dominant mode of failure; catheter fracture and embolism exist only as case reports, so anyone quoting you a rate is guessing; and insertion arrhythmia is well understood mechanistically but has never been reliably counted.
A meta-analysis of 3,689 children (neonates excluded) found overall complications 6 per cent, infection 3 per cent, thromboembolism 1 per cent — roughly an order of magnitude below adult intensive care — but mechanical failure 8 per cent, which is higher.
That inversion is the paediatric signature. In an adult you are mostly worrying about the vein. In a child you are mostly worrying about the catheter.
Almost all of what follows is practice-based consensus rather than trial evidence. Treat it as technique, and treat anyone who tells you otherwise with suspicion.
Nothing here is novel and all of it is audited.
Fewer lumens than you think you need. The smallest catheter that will do the job. Basilic, middle third of the arm, under ultrasound. Tip at the cavoatrial junction and parallel to the wall. Ask about kidneys before you touch the arm. And take it out the day it stops being necessary.
What do you do when the catheter stops at the shoulder?
Everybody who places enough of these has a sequence — something about the arm, something about the head, a flush at the right moment, a point at which they stop and resite rather than persisting. Very little of it is written down anywhere, which is exactly why it is worth collecting.
I would also like to hear about the PICC you wish had never been inserted — the one that went in easily, for a good enough reason at the time, and cost the patient something later. Those are the ones this piece exists for.
And the one I would most like to publish: what are the three things you would say to somebody starting out on PICC insertion? Tips, tricks, cautions, dos and don'ts — whatever three you would actually say if a registrar turned to you before their first one. The habit you have never had to think about. The thing you only learned by getting it wrong. The line you draw for yourself about when to stop and ask for help.
Three is deliberate. It is short enough that you will send it, and long enough that you have to choose — and what you leave out will be as interesting as what you keep.
The button opens a reply straight to me. If your device blocks it, write to icuandyou@icloud.com instead.