Physical compatibility of insulin aspart, lidocaine, alprostadil and vancomycin with individualised two-in-one parenteral nutrition used in the neonatal intensive care unit Authors: Thomas G Van Gelder, Daniël Vergoossen, Inge A Zonnenberg, Arief Lalmohamed, Marloes I Koole-Oostveen, Elisabeth M L van Reij, Arjen van Loon, Toine C G Egberts Published: 2026 ABSTRACT AbstractObjectivesCritically ill newborn infants often require simultaneous administration of multiple intravenous (IV) solutions through the same catheter lumen, making compatibility of these solutions crucial in neonatal intensive care units (NICUs). This study aimed to investigate the physical compatibility of insulin aspart, lidocaine, alprostadil and vancomycin with individualised two-in-one parenteral nutrition (PN).MethodsThe study was conducted at the hospital pharmacy’s drug compounding facility of the University Medical Centre Utrecht. Two PN formulations were prepared with different electrolyte concentrations (PN1 with high electrolytes and PN2 with low electrolytes), each with either 0% or 30% w/v glucose, resulting in four solutions for testing. Each solution was then mixed with the selected IV drugs in a 1:1 ratio. Compatibility was assessed through visible particle testing, pH measurements and subvisible particle testing at multiple time points (T=0, T=1, T=4 hours).ResultsNo visible particles were detected in any combinations. However, insulin and lidocaine combinations exceeded the subvisible particle threshold of 6000 particles ≥10 µm per container volume at T=0 hours, with insulin also exceeding the threshold in a specific PN combination at T=4 hours. pH measurements indicated minimal shifts in the PN solutions, suggesting significant buffering capacity.ConclusionAlprostadil and vancomycin IV solutions are physically compatible with two individualised neonatal PN solutions, with high as well as low glucose concentrations. Combinations of PN with lidocaine or insulin form subvisible particles, which could have clinical implications if administered in large volumes over extended periods of time. In clinical scenarios where there is no other option but to administer lidocaine or insulin through the same catheter lumen as PN using a Y-site connector, the use of an in-line filter is advised. Our study adds important compatibility data that can guide clinical practice in NICU settings. However, the broader application of these results requires careful consideration of the unique characteristics of each neonatal PN solution and drug combination. WHAT IS ALREADY KNOWN ON THIS TOPIC Co-administration of parenteral nutrition (PN) and intravenous (IV) drugs through the same catheter lumen is common in neonatal intensive care units (NICUs) due to limited venous access. However, incompatibilities between PN solutions and IV drugs can lead to particle formation, posing potential risks to neonates. Studies on the compatibility of PN solutions with specific IV drugs are lacking. Co-administration of parenteral nutrition (PN) and intravenous (IV) drugs through the same catheter lumen is common in neonatal intensive care units (NICUs) due to limited venous access. However, incompatibilities between PN solutions and IV drugs can lead to particle formation, posing potential risks to neonates. Studies on the compatibility of PN solutions with specific IV drugs are lacking. WHAT THIS STUDY ADDS This study shows that vancomycin and alprostadil are physically compatible with individualised neonatal PN solutions, while insulin aspart and lidocaine form subvisible particles that exceed pharmacopoeial thresholds. These findings highlight the importance of careful consideration of compatibility when co-administering IV drugs with PN solutions. This study shows that vancomycin and alprostadil are physically compatible with individualised neonatal PN solutions, while insulin aspart and lidocaine form subvisible particles that exceed pharmacopoeial thresholds. These findings highlight the importance of careful consideration of compatibility when co-administering IV drugs with PN solutions. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY The results provide critical data for NICU clinicians and pharmacists, emphasising the need for in-line filters when co-administering insulin or lidocaine with PN. This study also underscores the importance of standardised compatibility testing methods and provides a foundation for future research into clinically relevant compatibility protocols. The results provide critical data for NICU clinicians and pharmacists, emphasising the need for in-line filters when co-administering insulin or lidocaine with PN. This study also underscores the importance of standardised compatibility testing methods and provides a foundation for future research into clinically relevant compatibility protocols. Introduction Critically ill newborn infants often require simultaneous administration of multiple intravenous (IV) drug solutions. Due to limited venous access possibilities, the desire to administer drugs through the same catheter lumen (facilitated by, for example, Y-site connectors) is therefore not uncommon. This makes incompatibility of different IV drugs a critical consideration in neonatal intensive care units (NICUs). Incompatibility can lead to particle formation due to precipitation of compounds following chemical reactions or due to pH shifts affecting solubility and conformational changes.1 These physicochemical issues are particularly relevant in the NICU. Neonates often require continuous administration of parenteral nutrition (PN), which contains ingredients such as calcium and phosphate that are known to precipitate under certain conditions.24 Also, due to fluid restrictions in neonates, IV drugs are given as concentrated as possible, pushing the limits of their solubility. Incompatibilities can have serious clinical implications for neonates, primarily due to the risk of particle formation.5 One key concern is the obstruction of pulmonary vessels, as seen in the notorious example of ceftriaxone-calcium precipitation, which has led to fatal outcomes in neonates.6 Furthermore, particles in infusion solutions like PN can cause pulmonary artery granuloma after prolonged infusion.7 Randomised controlled trials have shown that in-line IV filters might reduce serious complications caused by particle contamination of infusion solutions in neonates.8 9 Another risk involves the occlusion of catheter lines, which prevents the administration of critical medications, necessitates line replacement and can potentially delay treatment.10 Determining whether two IV solutions such as PN and IV drugs can be combined often relies on evidence from scientific publications and handbooks. When reviewing this evidence, it is important to note the significant differences between neonatal and adult PN. Neonatal PN contains more glucose per litre, more mmol calcium per litre and has a lower pH than adult PN, all of which might influence compatibility with IV drugs.11 12 Consequently, extrapolating data from adult PN to the NICU setting should be approached with caution. Also, PN solutions for neonates are often individualised, leading to variations that complicate data extrapolation even further.13 Although information on the compatibility of neonatal PN solutions with IV drugs is available, there is still a lack of documented compatibility data for many other drug and specific neonatal PN combinations. Given this background, the aim of this study was to investigate the physical compatibility of four commonly used IV drugs in the NICU (insulin aspart, lidocaine, alprostadil and vancomycin) with four locally compounded individualised neonatal PN solutions. Methods Setting The study was conducted at the hospital pharmacy’s drug compounding facility of the University Medical Centre Utrecht (UMCU), The Netherlands, and at the NICU of the Wilhelmina Children’s Hospital, which is integrated within the UMCU and has 24 intensive care beds. Subvisible particle testing was performed at the Laboratory of the Royal Dutch Pharmacists Association (KNMP), known as “Laboratorium der Nederlandse Apothekers” (LNA). Individualised parenteral nutrition In our NICU we use one of two locally designed PN standard solutions containing amino acids, electrolytes, trace elements and water-soluble vitamins, tailored to the neonate’s age and electrolyte balance. The composition of the two PN solutions is based on the ESPGHAN/ESPEN/ESPR guidelines on paediatric parenteral nutrition.14 Directly following birth, ‘PN 2’ is preferred for 2–3 days due to its lower electrolyte concentration, addressing insensible fluid loss in premature newborns. Once the fluid balance stabilises, ‘PN 1’, with higher electrolytes, is used. After selecting the appropriate PN standard solution, neonatologists specify the volume and glucose concentration (typically 10–20% w/v) to add, using software integrated within the computerised physician order entry system. The order is processed in the hospital pharmacy’s compounding facility, where a pharmacy technician uses the BAXA ExactaMix Pro 2400 Compounder to mix all the components. A clinical pharmacist verifies the final product before delivery to the NICU. Lipids are administered as a separate infusion and join the PN and other infusions at a Y-site connection, downstream of a 0.2 µm filter, which the lipids bypass. The infusion setup in our NICU, providing context for the clinical conditions under which PN and IV drugs are administered, is shown in online supplemental figure S1. Given that neonatologists can add different amounts of glucose to either standard solution PN 1 or PN 2, we decided to examine two extremes: standard solutions without any added glucose (0% w/v) and those with a final glucose concentration of 30% w/v. This approach resulted in four PN solutions for the current study, referred to as PN 1–0% w/v glucose, PN 1–30% w/v glucose, PN 2–0% w/v glucose and PN 2–30% w/v glucose. Table 1 shows the amounts of glucose, amino acids, electrolytes, trace elements, water-soluble vitamins and water per litre, along with the pH of the four PN solutions tested in this study, while table 2 shows the manufacturing information for all the components. Table 1Composition of the four tested parenteral nutrition solutions per litreComponentPN 1 (high in electrolytes)PN 2 (low in electrolytes)0% w/v glucose30% w/v glucose0% w/v glucose30% w/v glucoseGlucose anhydrous (g)03000300Amino acids* total (g)71.6 (709 mL)28.6 (284 mL)75.0 (743 mL)30.0 (297 mL)Alanine (g)5.72.35.92.4Arginine (g)6.02.46.22.5Aspartic acid (g)4.31.74.51.8Cysteine (g)1.30.51.40.6Glutamic acid (g)7.12.87.43.0Glycine (g)2.81.13.01.2Histidine (g)2.71.12.81.1Isoleucine (g)4.81.95.02.0Leucine (g)7.12.87.43.0Lysine (g)7.83.18.23.3Methionine (g)1.70.71.80.7Ornithine hydrochloride (g)2.30.92.40.9Phenylalanine (g)3.01.23.11.2Proline (g)2.10.92.20.9Serine (g)2.81.13.01.2Taurine (g)0.40.20.40.2Threonine (g)2.61.02.71.1Tryptophan (g)1.40.61.50.6Tyrosine (g)0.30.10.30.1Valine (g)5.42.25.62.3Electrolytes (mL)1987913956Sodium (mmol)53.721.57.02.8Potassium (mmol)71.728.717.57.0Magnesium (mmol)4.31.75.72.3Calcium (mmol)24.59.825.610.3Phosphate (mmol)34.513.821.08.4Acetate (mmol)42.917.20.00.0Chloride (mmol)99.839.962.725.1Trace elements (mL)†2911.62911.6Zinc (mg)7.252.907.252.90Copper (mg)0.580.230.580.23Manganese (mg)0.030.010.030.01Selenium (mg)0.060.020.060.02Fluorine (mg)1.650.661.650.66Iodine (mg)0.030.010.030.01Water-soluble vitamins (mL)‡2911.62911.6Aqua ad iniectabilia (mL)37614.860624pH5.405.055.094.75*Primene 10%.†Peditrace.‡Soluvit N. Composition of the four tested parenteral nutrition solutions per litre Primene 10%. Peditrace. Soluvit N. Table 2Manufacturer details of components of locally compounded parenteral nutrition solutionsComponentManufacturerLot numberGlucose 50% 1000 mLA+Apotheek (compounding pharmacy)30-11-2024, 31-07-2025Aqua Destillata 3 LBraun Medical BV28-02-2026, 31-03-2026Primene 10% 250 mLBaxter BV31-10-2024, 31-03-2025Soluvit NFresenius Kabi Nederland BV31-12-2023, 31-07-2024Peditrace 10 mLFresenius Kabi Nederland BV31-03-2025, 28-02-2026Calcium chloride 0.225 mmol/mL 500 mLApotheek A15 (compounding pharmacy)31-10-2025Magnesium chloride 0.5 mmol/mL 250 mLApotheek A15 (compounding pharmacy)30-04-2025Potassium sodium phosphate (1.25/0.5/1.5 mmol/mL) 100 mLApotheek A15 (compounding pharmacy)29-2-2024Potassium acetate 1 mmol/mL 100 mLApotheek A15 (compounding pharmacy)31-01-2026Sodium chloride 29.2% (5 mmol/mL) 50 mLApotheek A15 (compounding pharmacy)31-10-2025 Manufacturer details of components of locally compounded parenteral nutrition solutions Intravenous drugs After reviewing current clinical practice we chose four frequently used drugs to test compatibility with the PN solutions, considering both their high prescription rates and their potential sensitivity to incompatibility and pH variations: insulin aspart (NovoRapid), lidocaine, alprostadil and vancomycin. The concentration, diluent, pH and manufacturer information of the tested drug solutions in this study are shown in table 3. Also, the table contains the infusion time and the maximum volume per syringe that would be administered in a clinical setting. The insulin aspart solution also contains human albumin (Albuman) in a concentration of 2 mg/mL to prevent adsorption of insulin onto catheter material.15 Vancomycin is provided as powder for concentrate for infusion solution and was reconstituted following the instructions in the Summary of Product Characteristics, which do not include a filtration step. Table 3Intravenous drugs investigated in this study, including information on concentration before mixing with PN, the diluent used, the pH of the solution, infusion time and the maximum volume administered per syringeDrugManufacturerConcentration before combinationDiluentpHInfusion timeMaximum volume per syringeInsulin aspart(+ human albumin)Excipients: glycerol, fenol, M-cresol, zinc chloride, disodium hydrogen phosphate dihydrate, sodium chloride, hydrochloric acid, sodium hydroxide, water for injectionsExcipients human albumin: sodium caprylate, sodium chloride, sodium hydroxide, hydrochloric acid, water for injectionsNovo Nordisk BV (NovoRapid)Lot no. NZF4W23Human albumin: Prothya Biosolutions Netherlands B.V. (Albuman)Lot no. 23A18H163A, 23E01H163A1 IE/mL (human albumin: 2 mg/mL)Sodium chloride 0.9% w/v6.49Continuous (24 hours)12 mL (of which 0.12 mL is human albumin)LidocaineExcipients: sodium chloride, hydrochloric acid, sodium hydroxide, water for injectionsFresenius Kabi Nederland BVLot no. 18X46807.5 mg/mLGlucose 5% w/v5.84Continuous (24 hours)26 mLAlprostadilExcipients: ethanolPfizer BV (Prostin)Lot no. GN139820 μg/mLGlucose 5% w/v4.46Continuous (24 hours)12 mLVancomycinExcipients: noneHikma Farmaceutica S.A.Lot no. 2241149.1, 2301098.110 mg/mLGlucose 5% w/v3.291 hour5 mL Intravenous drugs investigated in this study, including information on concentration before mixing with PN, the diluent used, the pH of the solution, infusion time and the maximum volume administered per syringe Compatibility assessment For all compatibility testing procedures we mixed drug and PN solutions in a 1:1 ratio. The solutions were mixed within sterile type 1 glass vials, which were then gently swirled multiple times. We tested compatibility immediately following mixing (T=0), at 1 hour (T=1) to simulate slow administration situations and at 4 hours (T=4) to ensure we observed late occurring effects. We also conducted all testing procedures for the stock solutions of the IV drugs and PN solutions before mixing them together. Compatibility was evaluated with three tests for all solutions. Visible particle test: Following the European Pharmacopoeia method for detecting visible particles (Ph. Eur. 11th edition, 2.9.20 ‘Particulate Contamination: Visible Particles’), we used the Red Range LV28 Liquid Inspection System in a GMP class C clean-room with a controlled temperature between 15°C and 25°C to inspect the mixtures for visible particles, incorporating both false negative (known incompatible solutions: furosemide and ceftriaxone) and false positive (glucose 5% w/v and aqua ad iniectabilia) controls to validate our findings.16 Visible particle test: Following the European Pharmacopoeia method for detecting visible particles (Ph. Eur. 11th edition, 2.9.20 ‘Particulate Contamination: Visible Particles’), we used the Red Range LV28 Liquid Inspection System in a GMP class C clean-room with a controlled temperature between 15°C and 25°C to inspect the mixtures for visible particles, incorporating both false negative (known incompatible solutions: furosemide and ceftriaxone) and false positive (glucose 5% w/v and aqua ad iniectabilia) controls to validate our findings.16 pH test: We measured the pH of solutions using a Metrohm 913 pH metre equipped with a combined electrode and temperature sensor to evaluate pH shifts on mixing drugs with PN solutions. pH test: We measured the pH of solutions using a Metrohm 913 pH metre equipped with a combined electrode and temperature sensor to evaluate pH shifts on mixing drugs with PN solutions. Subvisible particle test: Subvisible particle testing was conducted at the LNA, with sample preparation in a laminar air flow cabinet to prevent particle contamination. The laboratory conditions were controlled at a consistent ambient temperature of 21±3°C. We used the AccuSizer A2000 Particle Sizing System manufactured by Entegris/PSS. To ensure proper mixing, a magnetic stirrer was used with a magnet placed in the solution. The magnet was cleaned prior to each use to prevent contamination. We followed the ‘Light Obscuration Particle Count Test’ (Ph. Eur. method 2.9.19) to assess particle contamination (aligning with USP <788> Particulate Matter in Injections).17 18 The test criteria differ based on the nominal volume of the drug container: Test 1.A for volumes over 100 mL and Test 1.B for volumes under 100 mL. For our compatibility testing, we defined the ‘container’ as the maximum volume of the mixed PN and IV drug solution that a neonate would receive: 24 mL for mixed solutions with insulin and alprostadil, 10 mL for vancomycin, and 52 mL for lidocaine. These volumes are twice the values shown in table 3, as the drugs are mixed in a 1:1 ratio with the PN solutions. The original test results were reported in particles/mL, after which the particles per theoretical container volume were calculated. We applied Test 1.B for all mixed solutions, which sets a limit of 6000 particles per container volume for particles ≥10 µm and 600 particles per container volume for particles ≥25 µm. This interpretation of Test 1.B has been used in previous compatibility studies.19 Due to limitations in the subvisible particle testing capacity, we focused on the compatibility of PN 1–0% w/v glucose and PN 1–30% w/v glucose, considering them ‘worst-case’ solutions due to their higher electrolyte concentration than PN 2, potentially increasing susceptibility to incompatibilities. Subvisible particle test: Subvisible particle testing was conducted at the LNA, with sample preparation in a laminar air flow cabinet to prevent particle contamination. The laboratory conditions were controlled at a consistent ambient temperature of 21±3°C. We used the AccuSizer A2000 Particle Sizing System manufactured by Entegris/PSS. To ensure proper mixing, a magnetic stirrer was used with a magnet placed in the solution. The magnet was cleaned prior to each use to prevent contamination. We followed the ‘Light Obscuration Particle Count Test’ (Ph. Eur. method 2.9.19) to assess particle contamination (aligning with USP <788> Particulate Matter in Injections).17 18 The test criteria differ based on the nominal volume of the drug container: Test 1.A for volumes over 100 mL and Test 1.B for volumes under 100 mL. For our compatibility testing, we defined the ‘container’ as the maximum volume of the mixed PN and IV drug solution that a neonate would receive: 24 mL for mixed solutions with insulin and alprostadil, 10 mL for vancomycin, and 52 mL for lidocaine. These volumes are twice the values shown in table 3, as the drugs are mixed in a 1:1 ratio with the PN solutions. The original test results were reported in particles/mL, after which the particles per theoretical container volume were calculated. We applied Test 1.B for all mixed solutions, which sets a limit of 6000 particles per container volume for particles ≥10 µm and 600 particles per container volume for particles ≥25 µm. This interpretation of Test 1.B has been used in previous compatibility studies.19 Due to limitations in the subvisible particle testing capacity, we focused on the compatibility of PN 1–0% w/v glucose and PN 1–30% w/v glucose, considering them ‘worst-case’ solutions due to their higher electrolyte concentration than PN 2, potentially increasing susceptibility to incompatibilities. We considered the solutions physically compatible if there were no visible particles and if they complied with Test 1.B of Ph. Eur. method 2.9.19 at all time points. Results Visible particle test Table 4 shows the results of the visible particle test. Apart from foam formation at T=0 with insulin and PN 1–0% w/v glucose, none of the tested combinations showed any visible particles at either T=0, T=1 or T=4 hours. In the false negative controls, ceftriaxone showed visible particles after combination with PN 1–0% w/v glucose and PN 2–0% w/v glucose at T=1 and T=4 and furosemide showed visible particles after combination with PN 2–30% w/v glucose at T=1 and T=4. None of the false positive tests showed any visible particles (see online supplemental table S1 and S2–S4 for all results of the false negative and positive controls and for photos of all the test samples, respectively). Table 4Results of the visible particle test (the insulin solutions also contain 1 mg/mL of human albumin)DrugPNGlucose % w/vT=0T=1T=4Insulin10% w/vF, NoNoNoInsulin130% w/vNoNoNoInsulin20% w/vNoNoNoInsulin230% w/vNoNoNoLidocaine10% w/vNoNoNoLidocaine130% w/vNoNoNoLidocaine20% w/vNoNoNoLidocaine230% w/vNoNoNoAlprostadil10% w/vNoNoNoAlprostadil130% w/vNoNoNoAlprostadil20% w/vNoNoNoAlprostadil230% w/vNoNoNoVancomycin10% w/vNoNoNoVancomycin130% w/vNoNoNoVancomycin20% w/vNoNoNoVancomycin230% w/vNoNoNoF, foam formation; No, no visible particles; PN, parenteral nutrition. Results of the visible particle test (the insulin solutions also contain 1 mg/mL of human albumin) F, foam formation; No, no visible particles; PN, parenteral nutrition. pH test Table 5 shows the pH values following the mixing of drug and PN solutions at T=0 and T=4, including the difference in pH between the two time points, and the difference in pH at T=4 of the combination compared with the pH of the respective stock solutions. The combinations containing vancomycin and insulin showed the most significant changes in pH when compared with the pH of the drug stock solutions. In contrast, the change in pH relative to the four PN stock solutions was minimal across all tested mixtures. Table 5pH of the different combinations at T=0 and T=4, including the difference in pH between the two time points, and the difference in pH at T=4 of the combinations compared with the pH of the respective stock solutions (the insulin solutions also contain 1 mg/mL of human albumin)DrugPNGlucose % w/vT=0T=4Δ T=0 and T=4Δ stock (drug) at T=4Δ stock (PN) at T=4Insulin10% w/v5.405.410.01−1.080.01Insulin130% w/v5.095.130.04−1.360.08Insulin20% w/v5.135.160.03−1.330.07Insulin230% w/v4.844.890.05−1.600.14Lidocaine10% w/v5.395.420.03−0.420.02Lidocaine130% w/v5.085.120.04−0.720.07Lidocaine20% w/v5.125.130.01−0.710.04Lidocaine230% w/v4.804.840.04−1.000.09Alprostadil10% w/v5.425.420.000.960.02Alprostadil130% w/v5.135.140.010.680.09Alprostadil20% w/v5.135.130.000.670.04Alprostadil230% w/v4.854.850.000.390.10Vancomycin10% w/v5.365.360.002.07−0.04Vancomycin130% w/v5.025.040.021.75−0.01Vancomycin20% w/v5.075.080.011.79−0.01Vancomycin230% w/v4.724.750.031.460.00PN, parenteral nutrition. pH of the different combinations at T=0 and T=4, including the difference in pH between the two time points, and the difference in pH at T=4 of the combinations compared with the pH of the respective stock solutions (the insulin solutions also contain 1 mg/mL of human albumin) PN, parenteral nutrition. Subvisible particle test Table 6 shows the number of particles ≥10 µm and ≥25 µm for the stock solutions and combinations at T=0 and T=4 calculated per container volume. All combinations with insulin and lidocaine exceeded the threshold of 6000 particles ≥10 µm per container volume at T=0. Insulin mixed with PN 1–0% w/v glucose also exceeded the threshold of 6000 particles ≥10 µm per container volume at T=4. In the online supplemental table S5 shows the original test results in subvisible particles per mL. Table 6Subvisible particles of stock solutions and mixed solutions per specified container volume at T=0 and T=4 (the insulin solutions also contain 1 mg/mL of human albumin)T=0T=4DrugPNGlucose % w/vContainer volume≥10 µm/ container≥25 µm/ container≥10 µm/ container≥25 µm/ containerInsulin––12 mL1023.637.21258.825.2Lidocaine––26 mL806.07.8210.60.0Alprostadil––12 mL70.81.257.60.0Vancomycin––5 mL1129.51.5486.50.5–10% w/v5 mL1054.58.0162.00.512 mL2530.819.2388.81.226 mL5483.441.6842.42.6–130% w/v5 mL47.00.511.50.012 mL112.81.227.60.026 mL244.42.659.80.0Insulin10% w/v24 mL7468.8151.26108.0117.6Insulin130% w/v24 mL7852.896.02647.226.4Lidocaine10% w/v52 mL8330.436.43416.45.2Lidocaine130% w/v52 mL7701.25.25725.2Alprostadil10% w/v24 mL1656.00.02160.00.0Alprostadil130% w/v24 mL2839.24.8441.62.4Vancomycin10% w/v10 mL1105.02.01325.02.0Vancomycin130% w/v10 mL1079.01.052.00.0Highlighted cells do not comply with test 1.B of the Light Obscuration Particle Count Test (Ph. Eur. method 2.9.19). Subvisible particles of stock solutions and mixed solutions per specified container volume at T=0 and T=4 (the insulin solutions also contain 1 mg/mL of human albumin) Highlighted cells do not comply with test 1.B of the Light Obscuration Particle Count Test (Ph. Eur. method 2.9.19). Discussion This study investigated the physical compatibility of four commonly used IV drugs in the NICU (insulin aspart, lidocaine, alprostadil and vancomycin) with individualised two-in-one neonatal PN solutions. All four drugs formed no visible particles when mixed with our locally compounded neonatal PN solutions. However, some combinations with insulin and lidocaine exceeded the threshold of 6000 particles per container for particles ≥10 µm, as per the Light Obscuration Particle Count Test (Ph. Eur. method 2.9.19). These findings contribute valuable data on the physical compatibility of IV drugs with neonatal PN. Based on our results, combinations with vancomycin and alprostadil appear physically compatible with PN, whereas combinations with insulin and lidocaine may pose risks due to subvisible particle formation and should be used with caution in NICU settings. Visible particle test Foam was observed after mixing insulin aspart with PN 1–0% w/v glucose. The mixture had a pH of 5.4, representing a drop of 1.09 units from the initial insulin solution. While pH shifts can cause chemical instability in insulin formulations, physical instability such as aggregation is more commonly triggered by mechanical stress or temperature changes.20 Stability studies of insulin formulations diluted in normal saline to 1 IU/mL and 0.1 IU/mL have shown stability for at least 24 hours at room temperature, even with pH values below 5.21 22 Based on this evidence, we believe the transient foam formation was likely due to the swirling motion used during preparation and the surfactant-like properties of insulin. This swirling method was intended to homogenise the mixture and does not replicate clinical conditions of Y-site administration. Therefore, the transient foam is not considered clinically relevant. pH test Our pH test showed that the pH of the combinations of PN with IV drugs shifted towards the pH of the PN stock solutions, indicating a buffering capacity of the PN. For vancomycin and insulin drug formulations, this meant a change in pH of more than one point after combination. Compatibility studies with neonatal PN have proposed a pH shift of 1 or more, indicative of physical incompatibility.23 24 However, we believe a pH shift of 1 or more does not imply physical incompatibility if the IV drugs are still in a soluble and ionised form at that specific pH.25 Both vancomycin and insulin have been shown to be stable at the pH of the combined solutions in stability studies; in the case of insulin, stability was shown for at least 24 hours at room temperature.26 27 Subvisible particle test We found that combinations of lidocaine and insulin with PN did not comply with test 1.B (Ph. Eur. method 2.9.19) for particles ≥10 µm at T=0. At T=4, both lidocaine combinations complied with the test, possibly explained by a transient particle formation likely due to initial non-uniform concentration layering, which subsequently homogenised.3 The pH range of maximum stability for lidocaine was determined to be 3–6, in which lidocaine is present in its soluble ionised base form (HB+).28 Base precipitation of lidocaine has been shown to occur at a pH of 7.5–7.6.29 As the pH of the combination is around 5, theoretically, lidocaine would be primarily present in its soluble form. The observation of lidocaine not complying with the test can thus be best explained by the large theoretical container volume of the combination (52 mL) and a high initial particle count of PN 1–0% w/v glucose. Insulin combinations showed a more consistent trend over time with subvisible particle count, with the combination exceeding the test threshold for particles at both T=0 and T=4 for the combination with PN 1–0% w/v glucose. These particles might have been formed by insulin aggregation due to a pH shift or chemical reactions after mixing, a hypothesis that has been proposed before in stability studies of insulin aspart, but that was not tested in this study.30 The results of the subvisible particle tests should be considered in a clinical context. Lidocaine, used for neonatal seizures, has a total treatment duration of 28 hours and a total administered volume of 26 mL, which makes the absolute exposure to subvisible particles relatively low.31 Clinical complications of particles containing IV solutions, such as pulmonary artery granulomas, are more likely to occur in patients after long-term treatment.7 With the extra safety of a 0.2 µm filter, the benefits of therapy may outweigh the risks of co-infusion. For insulin therapy, however, the risk of prolonged infusion and total volume administered should be carefully taken into account with the results of our study. In clinical scenarios where there is no other option but to administer lidocaine or insulin through the same catheter lumen as PN using a Y-site connector, the use of an in-line filter is advised. Existing evidence Our results are in line with other studies investigating physical compatibility of drugs with neonatal PN. Watson et al, Veltri and Lee and Nezvalova et al showed physical compatibility of vancomycin (50 mg/mL, 5 mg/mL and 5 mg/mL, respectively) with neonatal solutions.24 32 33 Dice et al and Ross et al showed physical compatibility of alprostadil (15 µg/mL and 20 µg/mL in glucose 5% w/v, respectively) with neonatal PN solutions.34 35 To the best of our knowledge, this is the first study to test lidocaine and insulin aspart with a neonatal PN solution. Methodological considerations There is currently no standardised method for testing the compatibility of different IV drugs, and significant heterogeneity exists across studies in this field.36 Variations are observed in both the types of tests applied and the mixing ratios used. Most studies investigating compatibility with neonatal PN include visible particle inspection, pH measurement and a combination of subvisible particle testing and/or turbidity testing.2324 3235 37 We chose not to conduct turbidity testing in addition to subvisible particle testing as we believe the latter provides sufficient and clinically relevant information on particle formation. We acknowledge that the 1:1 mixing ratio used in this study does not replicate the dynamic conditions of true Y-site administration, where infusion rates of IV drugs and PN solutions can vary significantly. More recent compatibility studies have started to adopt clinically relevant ratios based on actual flow rates, proposing new approaches to compatibility testing that better reflect clinical practice.24 39 40 However, we believe the 1:1 ratio remains valuable for identifying key risks such as particle formation. This is particularly relevant for PN solutions, which have significant buffering capacity. Regardless of the mixing ratio, the pH of the final combination would likely shift towards the pH of the PN solution due to this buffering effect, minimising the impact of the mixing ratio on pH-dependent stability in drug–PN combinations. Limitations This study has several limitations. First, physical compatibility testing with lipid emulsions, which are frequently used in the NICU, was not conducted. Second, chemical stability was not assessed by measuring drug concentrations post-mixing; however, this limitation is less critical in the context of Y-site administration, where short contact times make chemical degradation unlikely during clinical use. Subvisible particle testing was limited to single preparations due to capacity constraints. Ideally, all combinations would have been prepared and tested in triplicate to rule out factors like unintended particle contamination. Glucose concentrations between 5% and 15%, which are most commonly used in clinical practice, were not tested in this study. However, by testing two extremes (0% and 30% w/v), we were able to comprehensively assess the impact of glucose on stability and compatibility. Finally, differences in PN formulations, particularly in the neonatal population, limit the generalisability of our findings. Extrapolation to other PN formulations or drug combinations should be approached with caution, taking into account factors such as total volume, concentration, diluent and infusion duration. Conclusion Alprostadil and vancomycin IV solutions are physically compatible with two individualised neonatal PN solutions, with high as well as low glucose concentrations. Combinations of PN with lidocaine or insulin form subvisible particles, which could have clinical implications if administered in large volumes over extended periods of time. In clinical scenarios where there is no other option but to administer lidocaine or insulin through the same catheter lumen as PN using a Y-site connector, the use of an in-line filter is advised. Our study adds important compatibility data that can guide clinical practice in NICU settings. However, the broader application of these results requires careful consideration of the unique characteristics of each neonatal PN solution and drug combination. Supplementary material