Sage Journals: Discover world-class research

Abstract

Background:

Adsorption of insulin to infusion sets impacts patient therapeutic outcomes and, unaccounted for, may exacerbate persistent hyperglycemia or result in therapy-induced hypoglycemia. This article aims to provide recommendations for clinicians involved in intensive care and/or outpatient pump therapy contexts.

Methods:

A dynamic adsorption model is used to evaluate time-varying insulin concentration in the infusion set outflow. Hourly and daily percentage insulin loss to adsorption is examined for neonatal, pediatric, and adult intensive care patients, as well as outpatient children and adults weighing 30, 50, and 80 kg. A short review of preconditioning methods is included.

Results:

Insulin adsorption in outpatient pump therapy is most pronounced in the first hour, where as much as 80% of the intended insulin dose may be lost to adsorption. Subsequently, insulin adsorptive loss is typically negligible. Overall, extra care should be taken in the first 1-6 h of a new infusion set, particularly in children or teenagers. Typically, insulin adsorption in the adult intensive care unit is negligible unless infused at low flow rates (<2 mL/h). Insulin adsorption is significant in pediatric and neonatal intensive care, resulting in delivery concentrations as low as 5%-50% of that intended. Thus, it is recommended that preconditioning of insulin delivery lines be carried out prior to infusion initiation in this context. However, no preconditioning method completely removes adsorption, and care should still be taken in the first 1-6 h of insulin dosing.

Conclusions:

Recommendations made in this article are dependent on the insulin concentration and flow rate used in each clinical context.

Keywords

insulin adsorption insulin binding insulin delivery pump therapy intensive care

Introduction

Adsorption is the binding of a substance to a material surface, a phenomenon extremely common in the delivery of protein-based therapies.^1,2 In particular, the adsorption of insulin to materials involved in its delivery has been well noted over the last half century,^3-16 but is not well understood in a clinical context. The aim of this article is to highlight the problem, and present recommendations based on a model-based approach, with the intention of unifying the diverse literature in this field.

This article first presents a description of the problem and recommendations for a clinical response, followed by the model- and literature-based analysis from which the recommendations are drawn. A final section summarizes key literature around methods to minimize the impact of adsorption, and their efficacy and/or limitations. The overall problem is addessed in the context of insulin infusions in the adult intensive care unit (ICU), neonatal intensive care unit (NICU) and/or pediatric intensive care unit (PICU), and outpatient diabetes.

Straight to the Point: Problem and Recommendations

The Problem: “Dude, Where’s My Insulin?”

Insulin adsorption reduces delivery of insulin to the patient, a phenomenon that is almost always unaccounted for in insulin dosing regimens. Insulin adsorption, or binding, to delivery surfaces, such as glassware,^17,18 syringes,^4,19 IV bags,^5,13,18,20 delivery tubing, and catheters,^{3,6,8-10,16,18,21} removes insulin that would otherwise be delivered to the patients. This phenomena is different from, but similar to, aggregation, which is the binding of insulin to itself. The result is under-delivery of insulin, which has had clinically significant effects on insulin dosing.^7,10

The Actual Problem: “Ignorance is Not Bliss”

Clinically, unknown under-delivery of insulin appears as persistent hyperglycemia and/or lowered insulin sensitivity. A worst-case scenario could see over-response to apparent insulin-resistance, where higher insulin doses combined with saturation of adsorptive losses may result in over-dosing and hypoglycemia. Thus, unaccounted for insulin adsorption represents unmanaged risk.

Pre-flushing or pre-priming of lines is used to mitigate the effects of insulin adsorption. However, this practice is not universally applied.^9,22 Nor is there any consistency in approach, using insulin, flushes or soaks, albumin or other proteins, all of which have different effects and degrees of efficacy.

Clinical Recommendations

The recommendations in Table 1 show the main clinical recommendations based on clinical context, which is a function of infusion set material and volume, insulin concentration, and flow rate. Recommendations are based on an experimental data-based model and a range of predicted insulin losses for each context, which in turn determine whether adsorption is an issue that needs to be addressed. Details on the model and methods are provided in subsequent sections.

Table 1.

Clinical Recommendations to Minimize the Impact of Insulin Adsorption on Delivered Insulin Therapy.

Context	Practice	Adsorption impact	Recommendation
Adult ICU	1 U/mL insulin in saline, ~1.6 mL infusion lines (typically PE).	• Adsorption can result in 2%-12% under-delivery of insulin per hour in the first 1-5 h of an infusion at a low flow rate (<2 mL/h). • At flow rates >2 mL/h, adsorptive losses are mostly <1%, and thus negligible.	No particular action required.
Neonatal/pediatric ICU	0.2-0.5 U/mL, 0.3-4 mL infusion lines (PE or PVC).	• Insulin adsorption is greater in PVC lines. • In PE lines, insulin adsorption can mean under-delivery of insulin by up to 90%-95% at low flow rates (0.1 mL/h), meaning only 5%-10% of intended insulin delivery is actually achieved. • Higher flow rates are less impacted by adsorptive loss, thus at a flow rate of 1 mL/h, the greatest insulin loss from an hour’s delivery is ~20%, and 80% of intended insulin is actually delivered. • It should be noted that this peak loss can occur two or more hours into the infusion.	Recommendation: Implementation of some method to mitigate insulin adsorption. No existing method will completely prevent adsorption, so care should be taken to avoid over-dosing insulin at the 3-6 h mark. • Flushing lines with insulin solution has some effect, but the fast flows (~1-10 mL/min) associated with timely flushing of insulin in a clinical setting may not allow adequate time for insulin adsorption to occur. Additional soaking is therefore recommended, for as long as can be warranted by clinical workflow. • Delivery of insulin in protein-rich solutions can significantly reduce adsorption. Insufficient data exist in literature as to the optimal procedure. • In our experience, infusion sets are often changed every 24 h. Consideration of longer intervals between changes of infusion sets may be warranted.
Outpatient pump therapy	100 U/mL, lines from 60-110 cm.	• Much, if not most, of the insulin may be lost in the first hour of insulin delivery. • After the first 1-2 h of infusion, delivery is likely ~100%.	Recommendation: Given that preconditioning lines in this context can be expensive and/or difficult, it is recommended that caution be exercised while dosing insulin (infusion and/or bolus) in the first 3-6 h of a new line. • Adsorptive effects are most pronounced where daily insulin doses are smaller. Thus, greater consideration of adsorption should be taken in children and/or teenagers.

Abbreviations: ICU, intensive care unit; PE, polyethylene; PVC, polyvinyl chloride.

Model-Based Analysis

Methods

Two contexts are considered: intensive care unit (ICU) infusion of insulin and outpatient pump therapy. Three lines are considered for each context and are denoted “Low,” “Medium,” and “High” volume. ICU line dimensions are pulled from Refs^8-10 and are considered reflective of the range of lines across neonatal, pediatric, and adult intensive care. Pump infusion sets are categorized by typical lengths available commercially, with an assumed internal radius based on Ref.²³ as tubing inner diameter is not commonly stated. Derived characteristics, such as volume, surface area, or radius, as appropriate, are calculated assuming a circular cylinder, and connection geometries and/or cannula are ignored. These values are given in Table 2.

Table 2.

Typical Internal Geometric Characteristics of Infusion Lines.

ICU lines	Dead space
ICU lines	Low	Medium	High
Volume (mL)	0.4	1.6	4
Length (cm)	150	206	203
Derived qualities (approximate)
Radius (mm)	0.3	0.5	0.8
Surface area (m²)	0.0027	0.0064	0.0101
SA:Vol ratio (m²:1 mL)	0.0069:1	0.0040:1	0.0024:1
Pump infusion sets^a	Low	Medium	High
Assumed radius^b (mm)	0.1	0.1	0.1
Length (cm)	60	80	110
Derived qualities (approximate)
Volume (mL)	0.07	0.09	0.12
Surface area (m²)	0.0007	0.0010	0.0013
SA:Vol ratio (m²:1 mL)	0.0105:1	0.0105:1	0.0105:1

Abbreviations: ICU, intensive care unit; SA, surface area.

Tubing only, not including cannula.

Based on an inner diameter of 0.19 mm as per Ref.²³

The most common materials used are polyethylene (PE) and polyvinyl chloride (PVC). For this analysis, all lines are assumed to be PE or PE-lined, as the most conservative case, since PE adsorbs less insulin than PVC.²⁴ A model-based analysis (see Appendix 1 for model) is carried out to assess the impact of adsorption at three different concentrations: 0.2, 1.0, and 100 U/mL. The model in Appendix 1 is used to calculate time-dependent insulin concentration at the infusion line outlet, representing the time profile of insulin delivered to the patient. Appendix 2 translates concentrations and flow rates in mL/h to U/kg/h, for comparison across different ICU contexts.

The material’s total capacity for insulin adsorption (U/m²) is assumed hyperbolic, based on relationships observed in previous work,²⁴ and the resulting equation in Figure 1 is used. In the case of very low flow pump infusions (of the order of μL/h), reflecting significant extrapolation of the relationship in Figure 1, adsorptive capacity is constrained to 500 U/m², which is approximately twice the maximum observed across both materials in Ref.²⁴

Figure 1.

Flow dependence of total adsorptive capacity of polyethylene.

Results

Intensive care

Model-based results

Figure 2 and Table 3 show that insulin adsorptive loss is a function of concentration, flow rate, and line size:

Figure 2.

Infusion line outlet concentration (delivered insulin) for three different intensive care unit lines (defined in Table 2), with inlet concentration: (a) 1.0 IU/mL and (b) 0.2 U/mL. Bar charts show total insulin delivery per hour as a percentage of expected/intended delivery, and bars are overlaid on the same axes so are not representative of sums to 100%. Dashed lines represent results with adsorptive capacity (B_eq) constrained to a hypothetical lower maximum of 100 U/m².

Table 3.

Case Study for a Medium Volume Infusion Line.

Flow rate			Adsorptive capacity		Expected 24 h delivery (U)	% Hourly delivery			% of Cumulative delivery
(mL/h)	70 kg BW (U/kg/h)	30 kg BW (U/kg/h)	Capacity: B_eq (U/m²)	Total bound = B_eq × SA (U)	Expected 24 h delivery (U)	Hour 1: % delivery	Hour 3: % delivery	Hour 6: % delivery	% of 1-3 h delivery	% of 1-6 h delivery	% of 1-12 h delivery	% of 1-24 h delivery
0.5	0.007	0.017	40.1	0.26	12	96	88	96.2	91.6	92.2	95.7	97.8
1	0.014	0.033	21.3	0.14	24	97.9	97.7	99.5	96.7	97.5	98.9	99.4
2	0.029	0.043	11.9	0.08	48	98.9	99.3	99.9	99	99.4	99.7	99.8
3	0.043	0.1	8.8	0.07	72	99.3	99.7	100	99.5	99.7	99.8	99.9

Abbreviations: BW, body weight; B_eq, surface adsorptive capacity in U/m²; SA, surface area; U, units of insulin.

Lower input concentration exacerbates loss impact: Adsorptive loss had lower overall impact at concentrations of 1 U/mL, where Figure 2a shows the minimum delivery is of the order of 90% of expected, matching literature.⁸ By comparison, infusions at the lower concentration of 0.2 U/mL had slightly lower minimum delivery (78% vs 88%) under the same flow rate and infusion set conditions (Vol:Med in Figure 2a).

More insulin is lost overall at lower flow rates: The total insulin lost to the line (surface adsorptive capacity) is a function of flow rate, based on results in Ref.²⁴ The greatest total loss occurs at lower flow rates, decreasing hyperbolically with increasing flow.

Lower flow rates take longer to reach the minima: Insulin concentration at the outlet do not necessarily occur immediately after infusion initialization, in part due to time taken for priming-volume turnover. Figure 2b shows this delay can be much greater than 5 h in low flow low concentration scenarios.

Larger surface areas affect total loss: Higher surface areas result in greater adsorptive loss of insulin due to a larger number of surface binding sites. In Figure 2, the highest volume line had the highest surface area, and the effect of insulin loss was prolonged.

Clinical takeaway

ICU: Panel a in Figure 2, and the medium volume line in particular, is representative of likely insulin adsorptive effects on adult ICU delivery of insulin (Table 3). The effects are most significant during the first 1-3 h, and likely have negligible or very little clinical impact, since typical ICU flow rates (Appendix 2) are greater than 1 mL/h at 1 U/mL concentration. Total insulin adsorption to the medium volume line is ≤0.25 U, representing a very small proportion of daily insulin requirements (~2% at 0.5 mL/h, 0.1%-0.5% at higher flow rates). Overall, insulin adsorptive effects have only small to negligible clinical significance for adult ICU delivery of insulin.

NICU/PICU: Panel b in Figure 2 is representative of neonatal or pediatric intensive care insulin infusions. Results in Figure 1b with the low volume infusion line approximate previous experimental results by Ref.¹⁰ Adsorptive effects are much more significant in this cohort, as losses and their impact are exacerbated by low flow and low concentration settings. Overall insulin loss may represent a significant proportion of daily insulin delivery. Thus, pre-priming/preconditioning of infusion sets is important in this clinical context, as the lower flow and lower concentration combination exacerbate adsorption.

Pump therapy

Model-based results

Input insulin concentration is 100 U/mL, and thus very low flow rates are required to delivery 0.01-0.02 U/kg/h basal insulin. As these infusion rates are of the order of μL/h rather than mL/h, the adsorptive capacity of the material is constrained to a maximum of 500 U/m², as in Figure 1. The upper limit on adsorptive capacity of 500 U/m² was chosen as being similar or greater than the maximum adsorptive capacities seen in PE and PVC,²⁴ presenting the clinically conservative case of the maximal possible adsorptive loss.

Figure 3 shows rapid insulin adsorptive loss, with insulin adsorption typically saturating within an hour. Given the extremely low flow rates, the minimum insulin recovery is 0 U/mL at all flow rates examined, and remains <5% for typically 5-55 min. In Table 4, such a loss might represent around 1%-5% of total insulin for the first day of delivery, but the 0.25 U bound can represent 31% (80 kg person at 0.01 U/kg/h)—80% (30 kg person at 0.01 U/kg/h) of the first hour’s intended insulin delivery. If the limit on adsorptive capacity was half that modeled here (250 U/m²) then it might be expected that 15%-40% of the first hour’s intended insulin delivery. If higher, as posed in Table 5, the impact could be higher.

Figure 3.

Model-based estimation of delivered insulin concentration at the outlet of different length infusion sets for wearable insulin pumps. Dimensions for each size (low, medium, and high) are in Table 2. Note that bars are overlaid on the same axes so are not representative of sums to 100%. Adsorptive capacity is limited to a maximum 500 U/m², based on previous observations.²⁴

Table 4.

Case Studies—Impact of Infusion Line Adsorptive Losses for a Weight-Based Basal Delivery.

Line volume	Total insulin bound (U)^a	Percent of daily insulin lost to adsorption
		0.01 U/kg/h basal			0.02 U/kg/h basal
		Weight (kg)			Weight (kg)
		30	50	80	30	50	80
High	0.31	4.4	2.6	1.6	2.2	1.3	0.8
Med	0.25	3.5	2.1	1.3	1.7	1.0	0.7
Low	0.19	2.6	1.6	1.0	1.3	0.8	0.5

Assumes a constrained maximum adsorptive loss of 500 U/m².

Abbreviation: U, units of insulin.

Table 5.

Impact of Model Extrapolation to Very Low Flow Rates—Upper Extremes on Potential Insulin Loss if Limit on Adsorptive Capacity is Removed.

Basal rate: 0.01 U/kg/h		Surface adsorptive capacity: B_eq (U/m²)	Total U adsorbed	% of daily insulin lost to adsorption
				Weight (kg)
				30	50	80
Flow rate (mL/h)	0.0025	7522.5	3.8	52.5	31.5	19.7
	0.005	3762.5	1.9	26.3	15.8	9.9
	0.0075	2509.167	1.3	17.5	10.5	6.6
	0.01	1882.5	0.9	13.1	7.9	4.9
	0.0125	1506.5	0.8	10.5	6.3	3.9
Constrained model-derived adsorption capacity		1000	0.5	7.0	4.2	2.6
		700	0.4	4.9	2.9	1.8
		500	0.25	3.5	2.1	1.3

Adsorption capacities were derived using the relationship for polyethylene in Figure 1 on a “medium line.” For context, no adsorptive capacities greater than 1000 U/m² have been observed at flow rates of 0.05-4 mL/h.²⁴

Abbreviation: U, units of insulin.

Clinical takeaway

Total adsorption to insulin lines is currently difficult to estimate due to uncertainties in dynamics at low flow rates and no reported test data. It is likely of the order of 1%-5% of total insulin requirements for the first day, but may be much higher in children. The greatest impact is seen in the first hour, where potentially as much as 30%-80% of the intended hour’s dose of insulin may not be delivered. Thus, caution should be exercised in the first 1-2 h of insulin delivery when a new infusion line is inserted.

How Might I Mitigate This Issue?

Having identified the impact of adsorption, the logical follow on is “how do I minimize its effect?” Various attempts to mitigate the impact of adsorption have been explored, including pre-priming the line with flushes or soaks with insulin (at delivery or higher concentration), or with other protein-rich solutions such as albumin.^6,9-11,25,26 While these methods are able to reduce the impact of adsorption, they are not able to completely remove it. Equally, their impact has not been assessed completely, and may be method specific. This section aims to review and summarize the experimental evidence available for different priming methods and their efficacy.

Table 6 summarizes a few key studies in this area. The majority of them are carried out in the context of the NICU, as the low concentrations and flow rates used in the NICU mean adsorption has greatest impact on insulin delivery.

Table 6.

Summary of Literature—Methods to Reduce the Impact of Insulin Adsorption.

	Flow rate (mL/h)	Conc. (U/mL)	Vol. (mL)	Priming methods	Results: insulin concentration at the outlet (primed vs unprimed)
	Flow rate (mL/h)	Conc. (U/mL)	Vol. (mL)	Priming methods	Time t = 0	Time t = 1 h	Hours 2-24
Fuloria et al¹⁰	0.05	0.2	0.3	Tubing is first primed with 5 U/mL insulin, then allowed to dwell for 20 min before being flushed with saline.	PVC: ~90% vs ~40% PE: ~100% vs 40%	PVC: ~70% vs 20% PE: N/A	All flow rates: ~80%-100% vs a steady increase from ~10% to 90% in unprimed tubing
Fuloria et al¹⁰	2	0.2	0.3		PVC: ~90% vs ~40% PE: ~100% vs 40%	PVC: ~40% vs 20% PE: ~80% vs 20%
Hewson et al⁹	1	0.2	4	20 mL flush	60% vs ~25%	~50% vs ~20%	All methods: 50%-80% vs a steady increase from ~10% to 70% in unprimed tubing
				1 h soak	~40% vs ~25%	~50% vs ~20%
				1 h soak followed by a 5 mL flush	~50% vs ~25%	~55% vs ~20%
				Mixed with albumin, no flush or soak	~50% vs ~25%	~75% vs ~20%	~constant 75%
Goldberg et al²⁶	N/A “flush” only	1	100 inch	10 mL flush	~85%	Not recorded—a limitation of this study: minimum insulin concentration often occurs more than 15 min after the initialization of an insulin infusion ⁸
				20 mL flush	~97%
				50 mL flush	~100%
Newby and Holmes²⁵	5	0.1	23	Tubing was filled to its priming volume, then preconditioned in two ways: (1) a 25 mL flush, and (2) a 30 min soak followed by an additional 25 mL flush.	53.4% vs 73.8% (method 1 vs method 2)	68.5% vs 21.1% (method 1 vs method 2)
	Flow rate (mL/h)	Conc. (U/mL)	Vol. (mL)	Priming methods	Results: Insulin concentration in 2 mL samples representing 2 h of insulin output (primed vs unprimed)
	Flow rate (mL/h)	Conc. (U/mL)	Vol. (mL)	Priming methods	0-2 h	2-4 h	4-6 h
Simeon et al¹¹	0.5	0.5	5 inch microbore	3 mL flush	67.2% vs 55.9%	65.8% vs 62.3%	74.1% vs 72.1%
Simeon et al¹¹	0.5	0.25	5 inch microbore	3 mL flush	70.9% vs 55.9%	63.0% vs 62.3%	71.4% vs 72.1%

Abbreviation: U, units of insulin.

A publication by Goldberg et al²⁶ is a perhaps more widely used standard for insulin preconditioning in critical care, recommending a 20 mL flush with insulin to mitigate the effects of insulin adsorption. However, their results do not seem to translate to the NICU setting with lower concentrations and flow rates, as seen in Hewson et al.⁹ A further limitation of the Goldberg et al²⁶ study is that they only measure the concentration in the first samples taken from the end of the line post-flush (t = 0,), and other studies have shown that the minimum insulin concentration delivered often occurs more than 15 min after the initialization of an insulin infusion,^8-10 due to fast flush/priming vs much slower infusion flow rates.

Summary of Mitigation Methods and Efficacy

Preconditioning of lines using flushes, soaks, or delivery alongside other proteins are effective ways to reduce the impact of insulin adsorption, but they do not completely mitigate the problem. In addition, they may present barriers to implementation in a clinical setting:

Flushes: Flushing the line with insulin solution prior to infusion can reduce insulin adsorption effects,^11,25,26 as it offers the opportunity to presaturate binding sites on the material surface. A limitation of this method is faster flow rates in flushing (1-10 mL/min vs 0.1-5 mL/h delivery rate) tend to result in lower overall adsorption (Figure 1), so flushing the line at flow rates convenient to clinical practice will have lower effect on adsorption and pre-saturation compared to additional soaking²⁵ or slower flushes.²⁷ In addition, the insulin solution flushed is “wasted,” which may impose a cost consideration in some contexts.

Soaks: Soaking the line with an insulin solution prior to infusion can reduce adsorption,^9,10,25 and is more effective than a flush alone,²⁵ but requires delays in the initiation of insulin therapy by 20-60 min. Such delays may be detrimental to patients or impose clinical burden on staff.

Sacrificial proteins: co-delivery of insulin alongside other proteins, such as albumin, is a very effective way to reduce insulin adsorption. However, delivery in albumin or plasma may present health concerns in some contexts.

Limitations and Disclaimers

This analysis is model based, rather than directly experimental. It is thus able to describe adsorption dynamics at a broad range of flow rates and conditions not feasible experimentally. The models presented have had preliminary validation using experimental data reported in literature,^24,28,29 and so are considered indicative of adsorption dynamics across the range of intensive care flow rates analyzed. Their extrapolation to the very slow infusion rates used in pump therapy may be less reflective of actual dynamics, as no study has examined adsorption at very low flow rates. The lack of clinically focused experiments in this low flow region force extrapolation, but also point out the need for focused clinical experiments in this region to add precision to the recommendations made.

However, the studies used are perhaps most comparable to “bath” studies from a material science perspective, where the adsorptive capacities reported broadly match those at intensive care type flow rates.²⁸ Most studies in this area use a short acting insulin (summarized in Refs^24,28), though one study also compared a short and a rapid acting insulin, with extremely comparable results.³ Overall, results presented here provide strong indications of likely outcomes in intensive care contexts, and a reasonable estimation of the range of possible outcomes when the models are extrapolated to pump therapy.

Footnotes

Appendix 1: Model

A two-state compartment model is used to describe insulin “free” in solution (F), and bound to the material surface (B) within the infusion tubing. The infusion tubing is discretized into N subcylinders. By conservation of mass:

(1)

{\dot{F}}_{i} V_{i} + {\dot{V}}_{i} F_{i} = F_{i - 1} q_{i - 1} - F_{i} q_{i} - A_{i}

(2)

\dot{B} S_{A i} + {\dot{S_{A}}}_{i} B = A_{i}

where $F_{i - 1}$ is the input insulin concentration (U/mL) to the subvolume, $q_{i}$ is the flow rate (mL/min) out of the subvolume, and $V_{i}$ and $S_{A i}$ are the subvolume’s volume (mL) and material surface area (m²), respectively. The adsorptive loss A_i is modeled as a predator prey term, which describes protein receptor type dynamics:

(3)

A_{i} = S_{A i} K_{1} F_{i} (B_{eq} - B_{i})

The rate constant for adsorption to the material surface (mL/U.min) is $K_{1},$ and $B_{eq}$ is the maximum binding capacity of the surface (U/m²). Adsorption is assumed irreversible, as the reverse mechanism is still debated and may be negligible.

The first element has input concentration $F_{o}$ , and the outlet concentration, N elements later, is $F_{N} .$ $K_{1}$ (0.01 mL/U.min) and $B_{eq}$ are assumed constant across the entire length. The initial conditions for each element are assumed to be $F_{IC} = F_{i - 1}$ U/mL and $B_{IC} = 0$ U/m² when there is no preconditioning. During tube fill, $\dot{V} = q_{i - 1}$ was used to capture the subvolume filling rate for each element, and $q_{i} = 0$ constrained unfilled volumes to no outflow. Time t = 0 is defined as the time that the first few drops of insulin are leaving the outlet of the infusion set tubing after filling. N = 150 subvolumes are used based on a pervious convergence analysis (Knopp et al²⁴).

Total adsorptive capacity of a material surface is modeled as a hyperbolic function of flow rate, based on previous results.²⁴ For PE materials:

(4)

B_{eq} = 2.5 + 18.8 / q

The model-based approach is presented in more depth in Refs.^24,27,28

Appendix 2: Translation of Flow Rates to U/kg/h for Intensive Care Patients

Conc. (U/mL)	Flow rate (mL/h)	Flow rate (U/kg/h)
		Body weight (kg)
		1	3	10	30	50	70
0.2	0.1	0.020	0.007	0.002	0.001	0.000	0.000
	0.2	0.040	0.013	0.004	0.001	0.001	0.001
	0.5	0.100	0.033	0.010	0.003	0.002	0.001
	1	0.200	0.067	0.020	0.007	0.004	0.003
	2	0.400	0.133	0.040	0.013	0.008	0.006
	3	0.600	0.200	0.060	0.020	0.012	0.009
1.0	0.1	0.100	0.033	0.010	0.003	0.002	0.001
	0.2	0.200	0.067	0.020	0.007	0.004	0.003
	0.5	0.500	0.167	0.050	0.017	0.010	0.007
	1	1.000	0.333	0.100	0.033	0.020	0.014
	2	2.000	0.667	0.200	0.067	0.040	0.029
	3	3.000	1.000	0.300	0.100	0.060	0.043
	4	4.000	1.333	0.400	0.133	0.080	0.057

Abbreviation: U, units of insulin.

Note.Flow rates in bold are more common to clinical practice.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: NZ National Science 7 Science for Technological Innovation (SfTI) and Callaghan Innovation NZ, grant #CRS-S3-2016, Royal Society of New Zealand Cook Fellowship #JCF-UOC-1501, Tertiary Education Commission of NZ, MedTech Centre of Research Excellence, #3705718.

ORCID iD

Jennifer L. Knopp

References

Darcy

PF.

Drug-interactions with medical plastics. Drug Intell Clin Pharm. 1983;17(10):726-731.

Tzannis

Hrushesky

WJM

Wood

Przybycien

TM.

Adsorption of a formulated protein on a drug delivery device surface. J Colloid Interface Sci. 1997;189(2):216-228.

Masse

Maton

Genay

, et al. In vitro assessment of the influence of intravenous extension set materials on insulin aspart drug delivery. PLoS One. 2018;13(8):e0201623.

Nayef

Khan

Brook

MA.

The stability of insulin solutions in syringes is improved by ensuring lower molecular weight silicone lubricants are absent. Heliyon. 2017;3(3):e00264.

Tsao

Lin

Chen

CY.

Quantitative analysis of insulin in total parenteral nutrition bag in Taiwan. J Food Drug Anal. 2016;24(1):214-219.

Thompson

Vital-Carona

Faustino

EV.

The effect of tubing dwell time on insulin adsorption during intravenous insulin infusions. Diabetes Technol Ther. 2012;14(10):912-916.

Jakobsson

Shulman

Gill

, et al. The impact of insulin adsorption onto the infusion sets in the adult intensive care unit. J Diabetes Sci Technol. 2009;3(1):213-214.

Zahid

Taylor

Gill

Maguire

Shulman

Adsorption of insulin onto infusion sets used in adult intensive care unit and neonatal care settings. Diabetes Res Clin Pract. 2008;80(3):e11-e13.

Hewson

Nawadra

Oliver

Odgers

Plummer

Simmer

Insulin infusions in the neonatal unit: delivery variation due to adsorption. J Paediatr Child Health. 2000;36(3):216-220.

10.

Fuloria

Friedberg

DuRant

Aschner

JL.

Effect of flow rate and insulin priming on the recovery of insulin from microbore infusion tubing. Pediatrics. 1998;102(6):1401-1406.

11.

Simeon

Geffner

Levin

Lindsey

AM.

Continuous insulin infusions in neonates: pharmacologic availability of insulin in intravenous solutions. J Pediatr. 1994;124(5 Pt 1):818-820.

12.

Sato

Ebert

Kim

SW.

Prevention of insulin self-association and surface adsorption. J Pharm Sci. 1983;72(3):228-232.

13.

Twardowski

Nolph

McGary

Moore

HL.

Influence of temperature and time on insulin adsorption to plastic bags. Am J Hosp Pharm. 1983;40(4):583-586.

14.

Schildt

Ahlgren

Berghem

Wendt

Adsorption of insulin by infusion materials. Acta Anaesthesiol Scand. 1978;22(5):556-562.

15.

Peterson

Caldwell

Hoffman

Insulin adsorbance to polyvinylchloride surfaces with implications for constant-infusion therapy. Diabetes. 1976;25(1):72-74.

16.

Weisenfeld

Podolsky

Goldsmith

Ziff

Adsorption of insulin to infusion bottles and tubing. Diabetes. 1968;17(12):766-771.

17.

Hill

JB.

The adsorption of I131-insulin to glass. Endocrinology. 1959;65:515-517.

18.

Petty

Cunningham

NL.

Insulin adsorption by glass infusion bottles, polyvinylchloride infusion containers, and intravenous tubing. Anesthesiology. 1974;40(4):400-404.

19.

Jones

Kaufmann

Middaugh

CR.

Silicone oil induced aggregation of proteins. J Pharm Sci. 2005;94(4):918-927.

20.

Illum

Bundgaard

Sorption of drugs by plastic infusion bags. Int J Pharm. 1982;10(4):339-351.

21.

Ley

Ammann

Herder

Dickhaus

Hartmann

Kindgen-Milles

Insulin adsorption to catheter materials used for intensive insulin therapy in critically ill patients: polyethylene versus polyurethane– possible cause of variation in glucose control?

Open Crit Care Med J. 2014;7(1):1-6.

22.

Alsweiler

Kuschel

Bloomfield

FH.

Survey of the management of neonatal hyperglycaemia in Australasia. J Paediatr Child Health. 2007;43(9):632-635.

23.

Lindpointner

Korsatko

Kohler

, et al. Glucose levels at the site of subcutaneous insulin administration and their relationship to plasma levels. Diabetes Care. 2010;33(4):833-838.

24.

Knopp

Bishop

Lerios

Chase

. Capacity of infusion lines for insulin adsorption: effect of flow rate on total adsorption [published online ahead of print September 27, 2019]. J Diabetes Sci Technol. doi:10.1177/1932296819876924.

25.

Newby

Holmes

DT.

Effect of tubing flush or preconditioning on available insulin concentration for IV infusion: a pilot project. Can J Hosp Pharm. 2017;70(4):320-321.

26.

Goldberg

Kedves

Walter

Groszmann

Belous

Inzucch

SE.

“Waste not, want not”: determining the optimal priming volume for intravenous insulin infusions. Diabetes Technol Ther. 2006;8(5):598-601.

27.

Knopp

Bishop

Chase

J.G

. A finite element model for insulin adsorption in ICU infusion sets. 2019 41st Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). Berlin, Germany, IEEE; 2019:1682-1685.

28.

Knopp

Hardy

Vergeer

Chase

JG.

Modelling insulin adsorption in intravenous infusion sets in the intensive care unit. IFAC J Syst Control. 2019;8:100042.

29.

Hardy

Chase

Knopp

JL.

Model based insulin absorption into intravenous infusion sets in adult and neonatal intensive care unit’s. IFAC-PapersOnLine. 2018;51(27):24-29.

Clinical Recommendations for Managing the Impact of Insulin Adsorptive Loss in Hospital and Diabetes Care

Abstract

Background:

Methods:

Results:

Conclusions:

Keywords

Introduction

Straight to the Point: Problem and Recommendations

The Problem: “Dude, Where’s My Insulin?”

The Actual Problem: “Ignorance is Not Bliss”

Clinical Recommendations

Model-Based Analysis

Methods

Results

Intensive care

Model-based results

Clinical takeaway

Pump therapy

Model-based results

Clinical takeaway

How Might I Mitigate This Issue?

Summary of Mitigation Methods and Efficacy

Limitations and Disclaimers

Footnotes

Appendix 1: Model

Appendix 2: Translation of Flow Rates to U/kg/h for Intensive Care Patients

Declaration of Conflicting Interests

Funding

ORCID iD

References