Patch pump purgatory: the mechanics of a patch pump on skin

The patch pumps that manufacturers are all barrelling towards, off the back of the Omnipod’s success, may all come with their own peculiar purgatory. In this article we look at the mechanical issues at the heart of patch pump design, and at what manufacturers could do about them. If you don’t want to read the whole article, then the following points highlight what the evidence suggests within this article.

TL;DR

• Patch pumps remove the tubing, and with it a familiar set of infusion problems. They bring a mechanical problem of their own: a rigid shell several centimetres long sits on soft skin, with the cannula fixed into it at one point.
• The cannula is held rigidly, so whenever the shell tilts, the cannula tilts with it. Most everyday loads tilt the shell. In a simple model, lying on the far end of a pump with its cannula at the other end lifts the cannula hub 0.75 to 1.5 mm and turns the cannula 3 to 4 degrees, and a garment catching the cannula end lifts the hub 1.75 mm.
• Repeated through a day and a night, that levers the cannula back and forth in its tract and could widen it. That offers a mechanism for insulin leaking back to the skin (tunnelling) and may add to the decline many Omnipod users report on day three, which tubed-set studies also show.
• Design decides how much the cannula moves. A cannula near the middle of the shell lifts less than one at the end, as we theorise with the Medtrum Nano against the Omnipod and the pumps that copy its layout, and a shell bonded to its adhesive right to the edges moves less than one with a loose margin. None of this has been measured on a patch pump, and the article sets out the bench tests that would settle it.

Patch Pumps

A patch pump puts the pump and its reservoir in one rigid shell with the cannula, and sticks the whole thing to the skin. Losing the tubing is what makes it attractive. It also creates a mechanical arrangement that tubed sets never had: a stiff body several centimetres long, resting on soft tissue, with a cannula a few millimetres long fixed to one spot on it. Earlier this year I wrote a hypothesis paper on the mechanics of Omnipod-style pods ^[1]. This article extends that reasoning to newer patch pumps, of which Equil and the Hippo P1 follow the Omnipod’s layout and the Medtrum Nano does not. It then turns to two things Omnipod users already know about: pods that seem to work less well on their third day, and the Tegaderm some people put under the pod to stop that.

The pump as a beam on a soft bed

Engineers have a standard model for a stiff object resting on something soft: a beam on an elastic foundation. Think of the skin and fat under the pump as a mattress of small springs, each pushing back in proportion to how far it is pressed. The pump is far stiffer than the tissue, so it does not bend. It can only sink and tilt.

Start with the simplest load. You lie on your front with the pump under you, and the mattress spreads your weight evenly across its top. That is a uniformly distributed load. On an even bed every spring under the pump is pressed by the same amount, so the pump sinks straight down without tilting. The skin goes down with it and the cannula goes down with the skin, so nothing moves relative to anything else. This is the load a patch pump handles best.

A force concentrated on one spot behaves differently, and what matters is how far that spot is from the middle of the pump. The pump sinks and tilts, and as long as the force lands within the middle third of the pump’s length, every spring under it is still being pressed: the whole adhesive pad stays in compression. Civil engineers use the same middle-third rule for foundations. For a 59 mm Equil the middle third runs 9.8 mm either side of centre. The shorter pumps have less room: 8.7 mm on the 52 mm Omnipod and 6.7 mm on the 40 mm Medtrum Nano ^[2].

Once the force lands outside the middle third, the far end of the pump tries to lift off the skin. The only thing holding the skin to it there is the adhesive, now pulling where it was pressing. Most of the loads a patch pump meets in daily life land outside the middle third: an elbow, the edge of a pillow, a seat belt, a door frame. An end-mounted cannula sits at that far end.

Uneven support under the pump

The textbook case assumes every spring is the same. Under a patch pump they are not. Fat is thicker in some places than others, and a thin layer over muscle or bone is much stiffer than a thick one. Sites such as the back of the arm are curved. Skin stretches more easily in some directions than in others, and the adhesive is stiffest where it is bonded to the shell and floppy at its fringe.

An uneven bed makes even a uniform load tilt the pump. Suppose the fat under an Equil runs from 10 mm thick at one end to 20 mm at the other, so the bed at the thick end is half as stiff. A load spread evenly over the top, enough to sink the middle of the pump 1 mm, sinks the thick end 1.33 mm and the thin end 0.67 mm ^[2]. That is a tilt of 0.65 degrees. On a bed that varies smoothly the effect stays small.

Curvature matters more. On a curved site a flat pump rests on a crown, in firm contact through the middle and held to the skin at its ends mostly by the adhesive. The middle-third rule then applies only to the part in firm contact. If an Equil is firmly supported over its central 30 mm, an illustrative figure, its middle third shrinks from 9.8 mm either side of centre to 5 mm, and almost any load that is not dead centre puts one end into peel.

The photograph below fits that picture. It was taken after removing an Omnipod, and shows a broad red footprint crossed by curved ridges where the shell pressed into the skin, with a separate, more focal mark at the insertion site. The load under the pod was uneven, and the insertion region had its own response. A photograph of the surface cannot show what the cannula did underneath, so I treat this as the observation that prompted the question and nothing more.

When the pump moves

Take a concrete case. You lie on your side and your weight goes through the end of an Equil furthest from its cannula. Assume that end sinks 2 mm. That figure is a modelling choice; abdominal fat will give more than that under a firm press, and a lean arm less.

On an even bed, with the skin resisting being pulled up as firmly as it resists being pressed in, the pump tilts about 2.9 degrees. It pivots about a point 9.8 mm beyond its middle on the cannula side. The far end lifts 1.0 mm, half the depth of the press. Equil’s cannula sits about 5 mm in from that end (my estimate from the photographs below), and its hub lifts 0.75 mm.

That is the best case. Skin held up only by adhesive tents easily, and it resists being pulled far less than it resists being pressed. If it pulls back half as firmly, the hub lifts 1.07 mm. A quarter as firmly, and it lifts 1.50 mm, a quarter of the length of a 6 mm cannula. Nobody has measured that ratio for skin under a patch pump adhesive, so the range is 0.75 to 1.5 mm ^[2].

Those numbers describe the skin surface under the hub. How much of the movement reaches the tissue around the cannula has not been measured, since the tissue moves with the skin to some extent. A fraction of that, repeated each time the wearer rolls over, still moves the cannula in and out of its tract.

The direction of the cannula changes what that lift does. With Equil’s 90 degree cannula, all of it is axial: the cannula slides along its own tract, dragging on the tract wall, which is where local inflammation would start. The Omnipod’s 45 degree cannula, and the Hippo P1’s similar one, gets 0.7 to 1.5 mm of lift from the same press, split evenly between movement along the tract and movement across it. The across component bends a soft cannula and presses it into one side of the entry point.

The cannula turns with the pump

None of these pumps has a joint between the cannula and the shell. Equil’s leaflet has the cannula inserted into the plastic base plate ^[3], and the Medtrum Nano’s steel cannula is driven in from the pump by a button on its top ^[4]. In each design the hub is part of the housing and moves with it ^[1]. So when the shell tilts, the cannula tilts with it, through the same angle.

In the worked example the Equil tilts 2.9 to 3.7 degrees, and the tip of a 6 mm cannula swings 0.31 to 0.39 mm sideways while the hub lifts ^[2]. Pressing the cannula end instead pushes the hub 1.75 mm into the skin and turns the cannula the other way. Rolling from one side to the other during the night levers the cannula back and forth about its hub, and its tip sweeps a band of tissue about twice that width, 0.6 to 0.8 mm across.

That is how the tract widens. Tissue around a freshly inserted cannula fits it closely. Each tilt levers the cannula against one wall of the tract and opens a gap on the other, and each lift drags it partway out along its length and back. Repeated, those movements cut a channel larger than the cannula, shaped like a cone with its point at the hub and its widest part at the tip. A soft cannula can bend a little along its length, but its root is held at the shell’s angle, so it is levered all the same. Insulin can then track back up the channel to the skin. That is the leakage Walsh and Heinemann describe for tubed sets ^[5], and the one Omnipod users call tunnelling.

A sideways knock adds shear on top. The force on the top of the pump reaches the skin through the adhesive as a sideways pull, and the skin slides over the fat beneath it. The cannula is fixed at the skin and free at the tip, so it is pushed across its length whatever its angle ^[1].

The script that produces these numbers is in the repository alongside this article ^[2]. Every figure scales with the assumed indentation, so a 4 mm press doubles them.

Where manufacturers put the cannula

The Omnipod 5 pod measures about 52 x 39 x 14.5 mm without adhesive and weighs about 26 g empty ^[6]. Its soft cannula goes in to a depth of 6.5 mm, at 45 degrees according to the platform’s technical description or at less than 50 degrees in Insulet’s own FAQ ^{[7, 8]}. The insertion site sits towards the narrow end of the pod.

Equil, from MicroTech Medical in Hangzhou, is semi-disposable. A reusable pump body with a rechargeable battery clips onto a disposable base plate that carries the adhesive, with the reservoir changed alongside it. The South African distributor gives 5.9 x 4.0 x 1.1 cm and 23 g with battery and filled reservoir ^[9]. The infusion-set leaflet covers soft cannulas of 6 mm and 9 mm (HRN-S-60 and HRN-S-90), inserted with a separate spring inserter that clips onto the base plate ^[3]. The leaflet’s drawings show the cannula port at one end of the base, with the inserter pressed straight down onto it, and the cannula goes in at 90 degrees to the skin. I handled one on a conference stand recently, and the cannula was where I expected it to be.

The Hippo P1 patch pump is a lot like an Omnipod from underneath, with a similarly angled cannula. I have not found published dimensions or an insertion angle for it, so where numbers are needed the Omnipod’s figures stand in for it.

The Medtrum Nano is laid out differently. It is smaller, at 40.5 x 31.5 x 11.5 mm and 13.8 g, with a stainless steel cannula inserted manually at 90 degrees ^[10] that is 5 to 6 mm long ^[4]. I measured where it sits. The cannula is about 10 mm from the head end of the pump and 30 mm from the reservoir end, and centred across the width. That puts it a quarter of the way along, halfway between the end and the middle. On the end-mounted pumps the cannula is about 5 mm from the end.

Whether the Equil and Hippo P1 layouts are imitation or the same packaging logic arrived at independently, I can’t say. The drive and reservoir take up most of a patch pump, and putting the cannula at the end keeps it clear of them. The mechanical consequence is the same either way. The Medtrum Nano shows that a patch pump does not have to put it there.

Position matters because a press on one end puts the pivot on the other side of the middle, and an end-mounted cannula sits about as far beyond it as the pump allows. A cannula in the middle is pushed in by a press on either end, across the range of skin stiffness used here, but it still turns with the shell, so its tip sweeps as far as one at the end. Moving the cannula to the middle removes the lift and leaves the sweep, which depends on the tilt and the cannula’s length.

The Medtrum Nano sits between the two. With its cannula 10 mm from the middle, the same 2 mm press on the reservoir end lifts the hub by 0.25 mm on the even bed and 0.87 mm with skin that pulls back a quarter as firmly, against 0.75 to 1.50 mm on Equil ^[2]. All of it is axial, since the cannula is vertical. Pressing the head end instead pushes the hub into the skin. The sweep works against the Nano. A shorter pump tilts further for the same indentation, 4.3 to 5.5 degrees against 2.9 to 3.7 on Equil, so the tip of its 6 mm steel cannula swings 0.45 to 0.57 mm with each press ^[2]. A 30-gauge cannula is about 0.3 mm across ^[11], so pressing each end in turn sweeps a band roughly three times its width. Steel does not bend at all, so the whole length is levered. The one direct observation of cannula movement damaging tissue is of steel tips in pigs ^[12]. The Nano trades lift at the hub for more sweep at the tip, and nobody has measured which matters more. And of course, with the steel tip, that sweep is likely to cause a different kind of sharp object damage.

There is a catch. A patch pump can rotate about several axes depending on where the force arrives, and a sideways knock tips it about an edge. No single cannula position is best for every direction of load. For presses along the length of the pump, a central position is never at the far end of the longest lever.

Height matters too. A sideways knock on the top of a patch pump acts at its full height above the skin, and the tipping moment scales with that height. Equil is listed at 11 mm ^[9] against the Omnipod’s 14.5 mm, which helps against knocks and does nothing for the end-press case, where what counts is how far the load lands from the middle.

Equil adds one feature the Omnipod does not have: the pump body is a separate part clipped to the base plate. Any play in that joint is one more place for movement, though it might equally absorb some of it. Whether Microtech have considered this is unknown.

Pulling up on a pump

Everything so far has been about pressing. A woman who wears patch pumps read a draft of this article and pointed out the load it had left out: pulling. A garment catches a corner of the pump as it comes off and lifts it: underwear and trousers at the stomach, lower back and thighs, sleeves on the arms. In her experience it happens most when going to the toilet.

A pull is a press turned upside down, with one difference. Under a press, the tissue under the pressed end pushes back firmly and only the far end depends on the adhesive. Under a pull, the end being pulled is held only by the adhesive and by the skin’s resistance to being lifted, which is the weaker of the two.

The same model handles it ^[2]. Pull one end of an Equil up with enough force to lift that end 2 mm, on skin that resists pulling as firmly as pushing. Pulled at the far end, the pump pivots and the cannula end is pushed in 0.74 mm. Pulled at the cannula end, the hub comes up 1.75 mm, nearly as far as the edge being pulled, because it sits only 5 mm from it. A corner pull at the cannula end lifts the hub 1.0 mm and tilts the pump about two axes at once. In each case the cannula turns through about 3 degrees, as it does under a press, so a pull adds to the sweep at the tip.

On skin that resists pulling a quarter as firmly, the same force lifts the pulled end 7 mm and the hub of an end-mounted cannula more than 6 mm, which is longer than the cannula. At that force either the adhesive lets go or the cannula comes out; the model cannot say which. The Medtrum Nano’s cannula, 10 mm in from its end, comes up 1.25 mm when that end is pulled 2 mm, against 1.75 mm on Equil.

How the shell is fixed to the adhesive

The Omnipod’s shell is not bonded to its adhesive pad over the whole pad. Along the straight edges there is a margin of approximately 5 mm where the shell can lift away from the adhesive, and the shell is fully fixed only at the cannula end. The Medtrum Nano’s base is bonded to the adhesive right to its edges.

A margin like that changes how far the shell can move. Where the shell pulls away from the skin over an unbonded strip, nothing holds it down, so it lifts off the pad. With the Omnipod’s footprint and a 5 mm margin, the same pull on a cannula-end corner lifts that corner 3.0 mm instead of 2.0, and the hub 1.32 mm instead of 0.98 ^[2]. A pull on the cannula end lifts the hub 2.26 mm instead of 1.71. Even a press on the far end lifts the hub further, 0.86 mm against 0.71, because the side margins no longer help hold the cannula end down.

There may be a benefit. A shell that lifts off its pad is not peeling the pad off the skin, so the adhesive may survive repeated pulls better. The cost falls on the cannula. The movement the margin allows is taken up at the one place the shell is fully fixed, which is the cannula end.

Day three

Many Omnipod users say the third day of a pod is when things go wrong. “On day 3 my blood sugar shoots up like I didn’t even take insulin,” one wrote last year, correcting with “3-4 times my usual amount” ^[13]. A parent of a six-year-old on Omnipod 5 described glucose “uniformly running higher than the first two days. Same ratios, same routines, same everything” ^[14]. Not everyone has it; a reply to that post said they almost always use the extra eight hours without trouble. Some have built workarounds. Before Omnipod 5, one person ran a separate day-3 basal programme 50% higher than usual ^[15]. Others have had prescriptions changed to two-day pods ^{[13, 14, 15]}, and a UK poster’s diabetes nurse suggested 48-hour changes ^[16]. These are self-reports, unchecked against any device data.

I could find no published analysis of glucose by day of wear, for Omnipod or any other patch pump. The data to do it exist, since Omnipod 5 records insulin delivery and sensor glucose against each pod.

For tubed infusion sets the analysis has been done several times, and it shows the same pattern. In Thethi’s study, 20 adults kept one set in for up to 100 hours while wearing a retrospective CGM ^[17]. Average glucose was 6.8 mmol/L on day 2, 8.0 on day 3, 8.6 on day 4 and 9.1 on day 5. Time above 10 mmol/L went from 14.5% of the day to 38.3%. Total daily insulin rose from 48.5 U on day 3 to 55.3 U on day 5, so glucose climbed while people were taking more insulin, and 7 of the 20 had at least one test period stopped early because glucose went above 16.7 mmol/L. A pilot the same year was titled “trouble starts on day 3” ^[18].

Real-world use looks similar. In CareLink records from 45 adults, time in range fell from 74.2% on day 2 to 66.5% on day 4 ^[19]. Even Medtronic’s seven-day set, in a trial of 259 people, had its best day on day 2 at 77.7% time in range and drifted to 71.6% by day 7, with insulin dose essentially unchanged ^[20]. That table covers only sets that lasted the full seven days. The studies that found no change either analysed only the sets that lasted the course ^[21] or averaged the whole week against the first three days ^[22], which hides a late decline. In the seven-day trial, day 1 was also worse than day 2 ^[20], so day three is where a decline from a day-2 best becomes noticeable.

What changes at the site

The puzzle is that insulin is absorbed faster as a set ages. In the most detailed pharmacokinetic study, time to peak insulin after a bolus fell from 67.5 minutes on the day of insertion to 32.5 minutes on day 3 with a standard Teflon set ^[23]. Swan found the same direction with glucose clamps between day 1 and day 4 ^[24].

Faster sounds like an improvement. The catch is that the dose also finishes sooner. In Swan’s study the insulin effect fell back to half its peak 42 minutes sooner on day 4 (95% CI 18 to 66), although total effect was similar. In Kastner’s study, total insulin reaching the blood over five hours was about 16% lower on day 3 than on day 0 for the standard set (not tested on its own), and about 24% lower by day 7 with both set types pooled. That study analysed six people, so the size is uncertain; the direction matches the glucose data.

The tissue is the likely reason. In pigs, an area of inflammation forms around the cannula and grows over the first four days of wear, and around Teflon keeps growing to day 7, even with no insulin infused: the sets were capped after a small saline test bolus ^[12]. As it thickens, the insulin depot gets smaller and the pressure needed to push insulin in rises ^[25]. Biopsies of people’s pump sites show fibrosis and fat necrosis alongside inflammation with signs of allergic sensitisation. The authors name “insulin preservatives, plastic materials, and adhesive glue” as the leading candidates ^[26].

Where the patch pump comes in

Movement is one of the proposed drivers of that inflammation and of leakage. Walsh and Heinemann, writing about tubed sets, describe tugs on the tubing and dropped pumps producing micro-movements of the cannula under the skin that “may delay wound healing or cause inflammation”. With a Teflon cannula, they suggest, the movement “may create a space along the cannula for partial leakage of insulin to the surface, resulting in hyperglycemia” ^[5]. Omnipod users call that leakage tunnelling. Walsh and Heinemann also say they could find no systematic measurement of it. The only direct observation of movement damaging tissue is of steel cannulas in pigs ^[12].

A patch pump turns ordinary loads into that kind of movement. Each time it is lain on and released, several times a night, the hub lifts by 0.75 to 1.5 mm and the cannula is levered through 3 to 4 degrees, widening the tract as described above. By the morning of day three the tract has been through two nights of it, on top of the inflammation that forms around any cannula. Motion would add to the day-3 decline; it is not needed to explain it. If it contributes, the decline should be steeper on patch pumps than on tubed sets and steeper at sites people lie on. Neither comparison has been published.

Tegaderm under the pod

Many Omnipod users stick a piece of Tegaderm, a thin transparent polyurethane film with an acrylate adhesive ^{[27, 28]}, to the skin before applying the pod. Insulet’s placement guide lists Tegaderm HP, along with Cavilon, Skin Tac and hydrocolloid bandages, as a barrier “to help limit skin irritation”. It tells users to cut a small hole where the cannula goes, and it says Insulet has not tested any of the products ^[29]. The official reason is skin. Omnipod has caused allergic contact dermatitis from isobornyl acrylate, found at the highest concentrations in the pod body and not the adhesive patch ^[30], and a barrier between device and skin is a standard workaround ^[31].

Users report something else. Many skip the hole and fire the cannula straight through the film. “It has pretty much completely solved the day 3 sensitivity issues for me,” wrote one ^[32]. “Tegaderm stopped all the tunneling problems I had with my Omnipod,” wrote another ^[33]. A person prescribed two-day pods said it had let them reach three days ^[15]. One user offered a mechanism: “If you cut a hole, the cannula can wiggle around just like it would if you didn’t use the film” ^[34]. These are anecdotes from people who tried something and chose to post about it.

Why it might work

The most direct explanation is a seal. When the introducer needle punches through intact film, the film closes around the cannula. Because the film is itself stuck to the skin around that point, it forms a collar at the skin surface. Insulin tracking back up the cannula meets a sealed surface where there used to be an open hole under the pod’s adhesive. Users who say cutting a hole “defeats the purpose” are describing this ^[35]. A seal would act on leakage. It would do nothing for absorption deeper in the tissue.

The second explanation is mechanical and follows from the lever argument. With the cannula through the film, the film grips the shaft where it enters the skin. The shell still levers the hub, but the entry point is now tied to a sheet anchored over several square centimetres, often well beyond the pod’s own pad, so the mouth of the tract is harder to prise open and the skin around it tents less. That fits the wiggle the user described. Whether it does depends on how stiff the film is compared with skin, and I could not find manufacturer figures for thickness or stretch to check it. A film also adds another bonded layer that could slide. Nobody has measured which effect wins.

Third, the film keeps the pod’s adhesive, and whatever has migrated into it, off the skin. That should reduce dermatitis under the pod. Whether surface inflammation changes absorption from a cannula tip several millimetres down in the fat has not been measured; the biopsy study is the only link, and it is indirect ^[26]. Barriers work for some people with device allergies and not for others, and the dressings can sensitise in their own right ^[31].

Fourth, selection: the people who post about a fix are mostly the ones it worked for.

If the first two explanations carry the effect, Tegaderm works by limiting the motion an end-mounted cannula is exposed to, and it does so at the skin. A patch pump with its cannula nearer the middle, like the Medtrum Nano, should need it less.

None of this makes it a recommendation. Firing the cannula through the film goes against Insulet’s instruction to cut a hole. One user reported a sharper, more painful insertion through film ^[36], and a review of device dermatitis notes that perforating a dressing for the cannula “sometimes prevents correct placing of the device” ^[31]. Some users have been told that firing through film could carry fragments of it into the tissue; I found no evidence either way.

What manufacturers could do

None of the following has been tested on a patch pump, and each is a hypothesis that the mechanics above suggests.

Move the cannula towards the middle. For a press on either end, the middle of the pump sinks and an end lifts, so a central cannula is pushed in where an end-mounted one is pulled out. It still turns with the shell. The Medtrum Nano shows the layout can be packaged; its quarter-way position already cuts the hub lift to between a third and three fifths of Equil’s in the example above.

Decouple the cannula from the shell. Every pump here holds the cannula rigidly, so it turns whenever the shell does. A hub that lets the cannula pivot, or a short flexible section between the shell and the cannula, would stop the shell’s tilt reaching the tract, as long as the pump stays stuck down ^[1]. Of the changes here, it is the one that addresses the sweep.

Hold the skin down where the pump lifts. The lift at the far end, and the whole of a pull, is resisted only by adhesive, and adhesive in peel is at its weakest at an edge. An adhesive pad that extends well beyond the shell, with a firmly bonded zone around the insertion site, would spread that pull over more skin and keep the skin at the entry point still ^[1]. That is roughly what Tegaderm users are improvising. Bonding the shell to the adhesive right to its edges stops the shell lifting off the pad; an unbonded margin may spare the adhesive, but it lets more movement reach the cannula.

Keep the shell low, and choose the cannula with its position in mind. Height sets the tipping moment from a sideways knock. A steel cannula is levered through the tissue along its whole length when the shell tilts; a soft one gives a little along its length, but its root is levered all the same.

All of these can be measured on a bench before anyone wears them, which is the subject of the last section.

Testing the idea

If the mechanism is real, end-mounted cannulas should show more site problems on the side people sleep on. The failures should look like slow, partial under-delivery, with rising glucose and no occlusion alarm, and the day-3 decline should be steeper on patch pumps than on tubed sets. Pods worn over Tegaderm with the cannula fired through it should show a shallower decline than pods worn without it. The Medtrum Nano offers a comparison that needs no new device: a patch pump with its cannula a quarter of the way along, which should show fewer of these problems if the lever matters, although its steel cannula and smaller size differ too. None of that is established. The infusion-set literature documents kinking, dislodgement, leakage along the tract and occlusions that go unnoticed ^{[37, 38, 39, 40]}, but it was mostly written about tubed sets, and none of it shows the same pathway in tubeless patch pumps.

The measurement that would settle it is straightforward to describe. Pumps mounted on a skin analogue over a soft fat layer can be pressed and pulled at known points, ends and corners included, while pumping at a basal rate, with shell and hub motion tracked optically. Sideways knocks are a second set of tests. The numbers that matter are how far the hub moves and the cannula turns for each millimetre of shell movement, and how wide the tract becomes after repeated cycles. Both should be compared across cannula positions. Ultrasound of the cannula in a person rolling over would be the in-vivo version. The day-3 question needs no new hardware at all: Omnipod 5 system records, split by pod day and body site, would show whether the decline exists in pods and how large it is.

Omnipod has a large body of clinical evidence behind it, and nothing here shows that it or any of the newer pods is unsafe. The model does suggest that where the cannula sits on a patch pump decides how much it moves. Equil and the Hippo P1 appear to have inherited the Omnipod’s choice without published testing of whether it is the right one, and the Medtrum Nano shows that a different choice can be packaged.

Acknowledgements

Thanks to Deniz Cengiz for the photographs of the Omnipod adhesive pad. Thanks to Mary Murphy for the inspiration to write the article, and for her thoughts on clothes catching on patch pumps.

References

1. Street T. Mechanical considerations in tubeless patch-pump design: lever effects, pod motion and cannula micro-displacement in generic Omnipod-style systems. Preprint, version 1.0, 9 September 2026.
2. scripts/worked_example.py and scripts/pull_model.py in this repository.
3. MicroTech Medical (Hangzhou). Infusion Set HRN-S-60 / HRN-S-90 instructions for use, 1001-PMTL-066 V02. https://insulinpatchpump.co.za/wp-content/uploads/2023/01/Equil-Insulin-Patch-Pump_Infusion-Set_English-1.pdf. Accessed 5 October 2026.
4. Street T. Medtrum’s Nano: an alternative patch pump. Diabettech. 16 May 2024. https://www.diabettech.com/medtrums-nano-an-alternative-patch-pump/
5. Walsh J, Heinemann L. Optimizing duration of usage of insulin infusion sets: impact of mechanical stress on infusion sites and identifying individuals with IIS issues. J Diabetes Sci Technol. 2025;19(1):3-4. doi:10.1177/19322968241233607.
6. Forlenza GP, Buckingham BA, Brown SA, et al. First outpatient evaluation of a tubeless automated insulin delivery system with customizable glucose targets in children and adults with type 1 diabetes. Diabetes Technol Ther. 2021;23(6):410-424. doi:10.1089/dia.2020.0546.
7. Zisser HC. The OmniPod Insulin Management System: the latest innovation in insulin pump therapy. Diabetes Ther. 2010;1:10-24. doi:10.1007/s13300-010-0004-6.
8. Insulet Corporation. Omnipod System FAQ: cannula insertion depth and angle. Omnipod UK. https://www.omnipod.com/en-gb/current-customers/resources/omnipod-system-faq. Accessed 9 September 2026.
9. Equil Insulin Patch Pump South Africa. Patient support. https://insulinpatchpump.co.za/patient-support/. Accessed 9 October 2026.
10. Medtrum. TouchCare Nano insulin patch pump and Nano CGM fact sheet, v1.00. 2021. https://abcd.care/sites/default/files/Meetings/EXTOD/Sponsors/1632392287touchcare-nano-insulin-patch-pump-nano-cgm-fact-sheet-v1-00-2-pdf1632392287.pdf. Accessed 9 October 2026.
11. International Organization for Standardization. ISO 9626:2016 Stainless steel needle tubing for the manufacture of medical devices: requirements and test methods. Geneva: ISO; 2016.
12. Hauzenberger JR, Münzker J, Kotzbeck P, et al. Systematic in vivo evaluation of the time-dependent inflammatory response to steel and Teflon insulin infusion catheters. Sci Rep. 2018;8:1132. doi:10.1038/s41598-017-18790-0.
13. r/Omnipod. How do I make my Omnipod keep working on day 3? 3 May 2025. https://www.reddit.com/r/Omnipod/comments/1ke3ads/
14. r/Omnipod. Day 3 effectiveness. 19 February 2026. https://www.reddit.com/r/Omnipod/comments/1r94hj4/
15. r/Omnipod. Highs on day 3. 25-27 October 2025. https://www.reddit.com/r/Omnipod/comments/1ofh7j3/
16. Diabetes.co.uk forum. Omnipod high sugars. 15-20 January 2023. https://www.diabetes.co.uk/forum/threads/omnipod-high-sugars.193544/
17. Thethi TK, Rao A, Kawji H, et al. Consequences of delayed pump infusion line change in patients with type 1 diabetes mellitus treated with continuous subcutaneous insulin infusion. J Diabetes Complications. 2010;24(2):73-78. doi:10.1016/j.jdiacomp.2009.03.002.
18. Schmid V, Hohberg C, Borchert M, Forst T, Pfützner A. Pilot study for assessment of optimal frequency for changing catheters in insulin pump therapy: trouble starts on day 3. J Diabetes Sci Technol. 2010;4(4):976-982. doi:10.1177/193229681000400429.
19. Yoshimura K, Urai S, Hozumi K, et al. Observational study of the impact of infusion set replacement in insulin pump users on glycemic management. J Diabetes Sci Technol. 2025. doi:10.1177/19322968251345837.
20. Brazg R, Garg SK, Bhargava A, et al. Evaluation of extended infusion set performance in adults with type 1 diabetes: infusion set survival rate and glycemic outcomes from a pivotal trial. Diabetes Technol Ther. 2022;24(8):535-543. doi:10.1089/dia.2021.0540.
21. Patel PJ, Benasi K, Ferrari G, et al. Randomized trial of infusion set function: steel versus teflon. Diabetes Technol Ther. 2014;16(1):15-19. doi:10.1089/dia.2013.0119.
22. Waldenmaier D, Zschornack E, Buhr A, Pleus S, Haug C, Freckmann G. A prospective study of insulin infusion set use for up to 7 days: early replacement reasons and impact on glycemic control. Diabetes Technol Ther. 2020;22(10):734-741. doi:10.1089/dia.2019.0445.
23. Kastner JR, Bailey TS, Strange P, et al. Progressive acceleration of insulin exposure over 7 days of infusion set wear. Diabetes Technol Ther. 2023;25(2):143-147. doi:10.1089/dia.2022.0323.
24. Swan KL, Dziura JD, Steil GM, et al. Effect of age of infusion site and type of rapid-acting analog on pharmacodynamic parameters of insulin boluses in youth with type 1 diabetes receiving insulin pump therapy. Diabetes Care. 2009;32(2):240-244. doi:10.2337/dc08-0595.
25. Eisler G, Kastner JR, Torjman MC, et al. In vivo investigation of the tissue response to commercial Teflon insulin infusion sets in large swine for 14 days: the effect of angle of insertion on tissue histology and insulin spread within the subcutaneous tissue. BMJ Open Diabetes Res Care. 2019;7(1):e000881. doi:10.1136/bmjdrc-2019-000881.
26. Kalus A, Shinohara MM, Wang R, et al. Evaluation of insulin pump infusion sites in type 1 diabetes: the DERMIS study. Diabetes Care. 2023;46(9):1626-1632. doi:10.2337/dc23-0426.
27. 3M. Tegaderm Transparent Film Dressing instructions for use, 34-8721-7382-7. 2017. https://multimedia.3m.com/mws/media/498994O/3m-tegaderm-transparent-film-dressing-frame-style.pdf. Accessed 5 October 2026.
28. 3M Colombia. Ficha técnica: 3M Tegaderm Estándar 1622W, 1624W, 1626W, 1627, 1628 y 1629. https://assets.solventum.com/is/content/mmmspinco/tegaderm-standard-1622w-1624w-1626w-1627-1628-1629-data-sheet-ms-ipss-es-copdf. Accessed 5 October 2026.
29. Insulet Corporation. Pod site preparation and placement, INS-ODS-06-2019-00035 V5.0. https://www.omnipod.com/sites/default/files/2021-04/Omnipod_Pod-Placement-Guide_English.pdf. Accessed 5 October 2026.
30. Raison-Peyron N, Mowitz M, Bonardel N, Aerts O, Bruze M. Allergic contact dermatitis caused by isobornyl acrylate in OmniPod, an innovative tubeless insulin pump. Contact Dermatitis. 2018;79(2):76-80. doi:10.1111/cod.12995.
31. de Groot A, van Oers EM, Ipenburg NA, Rustemeyer T. Allergic contact dermatitis caused by glucose sensors and insulin pumps: a full review. Part 2. Contact Dermatitis. 2025;92(3):164-175. doi:10.1111/cod.14697.
32. r/Omnipod. Tegaderm question. 8 January 2026. https://www.reddit.com/r/Omnipod/comments/1q6ycnn/
33. r/Omnipod. How does Tegaderm help with Omnipod? 24-25 June 2026. https://www.reddit.com/r/Omnipod/comments/1uenptz/
34. r/Omnipod. Tegaderm question. 20-21 July 2026. https://www.reddit.com/r/Omnipod/comments/1v1a8lf/
35. r/Omnipod. Can someone please explain to me how they use Tegaderm? 29-31 March 2026. https://www.reddit.com/r/Omnipod/comments/1s6wi88/
36. r/Omnipod. Tegaderm and Omnipod, painful insertion. 23 June 2024. https://www.reddit.com/r/Omnipod/comments/1dmdkcs/
37. Heinemann L, Krinelke L. Insulin infusion set: the Achilles heel of continuous subcutaneous insulin infusion. J Diabetes Sci Technol. 2012;6(4):954-964. doi:10.1177/193229681200600429.
38. Deiss D, Adolfsson P, Alkemade-van Zomeren M, et al. Insulin infusion set use: European perspectives and recommendations. Diabetes Technol Ther. 2016;18(9):517-524. doi:10.1089/dia.2016.07281.sf.
39. Klonoff DC, Freckmann G, Heinemann L. Insulin pump occlusions: for patients who have been around the (infusion) block. J Diabetes Sci Technol. 2017;11(3):451-454. doi:10.1177/1932296817700545.
40. Zhang G, Romo-Anselmo E, Kwa T, Cohen O, Vigersky R, Chattaraj S. Advances in insulin infusion set in the new era of automated insulin delivery: a systematic review. J Diabetes Sci Technol. 2023;17(2):302-313. doi:10.1177/19322968221145731.

Omnipod is a trademark of Insulet Corporation and Equil a trademark of MicroTech Medical (Hangzhou) Co., Ltd. Both names are used descriptively. The animations and diagrams are conceptual and are not manufacturer drawings.