Secondary Logo

Journal Logo

Technology, Computing, and Simulation: Research Reports

An In Vitro Analysis of Central Venous Drug Delivery by Continuous Infusion: The Effect of Manifold Design and Port Selection

Moss, David R. MD*; Bartels, Karsten MD*; Peterfreund, Gregory L. BA*; Lovich, Mark A. MD, PhD; Sims, Nathaniel M. MD*; Peterfreund, Robert A. MD, PhD*

Author Information
doi: 10.1213/ANE.0b013e3181b7c359
  • Free


Central venous catheters (CVCs) are used extensively in the operating room and intensive care unit for continuous infusion of vasoactive, inotropic, antidysrhythmic, sedative, and analgesic drugs driven by pumps. Precise control of drug delivery is crucial for patient safety and proper treatment.

If a drug infuses through a CVC lumen dedicated solely to that solution, with no carrier flow, then any change in pump settings for the drug delivery rate propagates instantly to the patient’s bloodstream, thereby conferring the desired control of delivery. However, some patients require multiple infusions. The number of infusions often exceeds the number of available lumens. A multiport manifold may be adjoined to a CVC lumen allowing simultaneous administration of multiple drug infusions in 1 fluid stream. The drugs are often transported by a carrier fluid through the infusion system dead volume (V), which we have previously defined as the volume of the fluid path from the point where a drug infusion meets the carrier flow to the tip of the intravascular catheter where the drug enters the patient’s bloodstream.1 When a clinician adjusts the pump settings for an infused drug that is transported by a carrier, changes in the drug’s rate of delivery do not instantly propagate to the point of entry into the patient’s bloodstream.1 Because critically ill patients are frequently intolerant of volume administration, clinicians often infuse drugs prepared in a concentrated form, which are then transported by a slow-flowing carrier fluid. Consequently, the ratio of total fluid flow rate (drugs plus carrier) to infusion system V becomes an important factor for determining the speed with which the drug enters the circulation and ultimately exerts clinical effects.2

Prior work2 comparing different CVC lumens demonstrated that drug delivery kinetics were significantly affected by the V of the selected CVC lumen. With the addition of a manifold to the fluid path, V includes contributions from the catheter and the manifold. For an individual infusion, the V further depends on the location of the port in the manifold through which the infusion enters the fluid path. Using our established in vitro model, we tested the hypothesis that the V of a manifold quantitatively affects infused drug delivery. We measured delivery kinetics when different ports of a traditional 4-stopcock manifold were chosen for the entry point of a drug infusion into the common fluid stream. To further test the hypothesis, we compared the temporal responsiveness of the traditional manifold with a manifold specifically designed to minimize V.


Methylene blue (MB), USP, 10 mg/mL (American Reagent, Shirley, NY), served as the model drug for these experimental drug delivery studies. The stock solution was diluted to a concentration of 0.1 mg/mL with normal saline (NS) and loaded into a 60-mL syringe. Microbore tubing connected the syringe containing MB to the port being evaluated on one of the manifolds. NS served as the carrier flow to the inlet of the manifold and a dual-channel syringe pump (Harvard 2 Clinical Pump) controlled both the NS and MB flow rates. Two manifolds were evaluated (Fig. 1): a traditional 4-stopcock Hi-Flo manifold (Arrow International, Reading, PA,, Product No. W21122) and a 6-cm, 6-port Multi Line Extension Set designed for microinfusions (Summit Medical Products, Worcester, MA,, Product No. MC8001). The 4-stopcock manifold is a commercially prepared union of four 3-way stopcocks in series, whereas the 6-line microinfusion manifold consists of 2 ports in series, each accommodating up to 3 lines in parallel. The manifold outlet was connected to the hub of a 16-gauge lumen of a 16-cm, 7F triple-lumen catheter (Arrow International, model number MSO-12703-PHS) suspended vertically over collection tubes in a turntable fraction collector.

Figure 1
Figure 1:
Figure 1.

The manifold and CVC were carefully primed with NS to eliminate air bubbles before starting an experimental infusion. The NS carrier flow rate was set to 10 mL/h, and the MB infusion rate was set to 3 mL/h. Samples of fluid leaving the CVC were collected every 60 s. The volumes of each sample were calculated from the known flow rates and diluted to 1 mL using saline. Aliquots (200 μL) were transferred to a microtiter plate for quantitative spectrophotometric absorbance assay at 668 nm. The absorbance data of the experimental samples were compared with a standard curve prepared with known concentrations of MB. Experimental infusions for each condition were repeated 3 times. Values for the time to reach 50% of the steady-state drug delivery after initiation of the MB infusion, t50 (on), and from steady state, the time to reach 50% of the previous drug delivery after cessation of the MB infusion, t50 (off), were calculated through linear interpolation of drug delivery rates between successive time points. Results were compared by unpaired, 2-tailed, t-test using Microsoft Excel or by analysis of variance with a Bonferroni posttest comparison using Prism 2.0® (GraphPad Software, San Diego, CA).

The size of the V was determined for each port by measuring, with a calibrated syringe, the volume of fluid required to load the manifold and adjoined CVC lumen. The measurement was repeated 3 times per port.


The dynamics of drug delivery in this model were quantified during initiation, steady state, and cessation of the infusion of the model drug MB. A carrier flow was established through the manifold-CVC fluid pathway before initiation of MB infusion. Figure 2 depicts the kinetics of drug delivery onset from the traditional 4-stopcock manifold (Ports 1, 2, and 4) and from the 2 ports of the microinfusion manifold. Figure 3 shows the corresponding graphs for offset kinetics. The measured values for t50 (on), t50 (off), and V are summarized in Table 1.

Figure 2
Figure 2:
Figure 2.
Figure 3
Figure 3:
Figure 3.
Table 1
Table 1:
Effect of Port Selection on Onset and Offset Kinetics Using Two Different Manifolds

Comparing the 2 manifolds, we found that t50 values for the first port of the stopcock manifold were similar to t50 values for both ports of the microinfusion manifold. The onset and offset kinetics for both upstream ports of the stopcock manifold were significantly prolonged compared with either port of the microinfusion manifold. Figure 4 plots t50 (on) and t50 (off) as a function of each port’s V. In general, half-times increased with increasing V.

Figure 4
Figure 4:
Figure 4.

We evaluated whether predictions of the previously described plug flow or well-mixed models1 apply to the data. We calculated the time constant, τ, for the system at Ports 1, 2, and 4 of the stopcock manifold and also the time constant for the microinfusion manifold where

and Q is defined as the total volumetric flow rate through the fluid path including contributions from the carrier and the drug (Table 2). We then calculated the value of twice the half-time for infusion onset to reach steady state (2 × t50 [on]) as a surrogate for the time to steady state. Next, we divided these values by τ, which will reveal the number of time constants needed to reach steady state. Averaging across all ports, we found that 2 × t50 (on) = 3.21 ± 0.29 × τ (Table 2, last row). This closely matches the predictions of the well-mixed model, where delivery to 95% of the steady-state rate is achieved in approximately 3 time constants (3 τ), and delivery to 99% of steady state is reached in 4 time constants (4 τ).1,3

Table 2
Table 2:
Time to Achieve Steady-State: Data Compared with Prediction from Well-Mixed Model

We hypothesized that increasing the carrier flow rate would decrease t50 for a drug infusion entering the fluid path at stopcock Port 4. We used Eq. 1 to estimate the increased carrier flow rate required for accelerating stopcock #4 delivery kinetics to approximate stopcock #1 delivery kinetics. For Port 4 of the stopcock manifold, the V is 2.39 times that of Port 1. To achieve the onset kinetics observed at Port 1 (with carrier rate of 10 mL/h and drug infusion rate of 3 mL/h), we predicted that the carrier flow for an infusion entering the fluid path at Port 4 would have to be 28 mL/h (Appendix). This hypothesis was tested experimentally. In Figure 5, data averaged from 3 runs at Port 4 using a carrier flow of 28 mL/h are superimposed on the data from Ports 1 and 4 at lower carrier flows (10 mL/h). As predicted, there was no statistical difference between the t50 (on) at Port 1 using low flow versus Port 4 using the higher carrier flow. Thus, the effect of a difference in V can be overcome with a higher carrier flow predicted by a mathematical model.

Figure 5
Figure 5:
Figure 5.


Continuous IV infusions are a principal means of drug delivery in anesthetized and critically ill patient populations. Clinicians frequently administer these infusions via CVCs in the operating room or intensive care environments. A manifold connected to 1 port of a CVC allows simultaneous infusion of medications through an individual lumen. There are no published studies assessing whether drug delivery kinetics depend on the design of the manifold or which manifold port is selected for connecting a drug infusion to the delivery system.

This study extends our previous work1–3 to demonstrate that, with a traditional stopcock manifold, port selection significantly affects drug delivery kinetics. With the fourth port of a traditional 4-stopcock manifold, onset half-times were approximately 8.72 min with the carrier (10 mL/h) and drug flows (3 mL/h) used in this study. This means that a steady-state delivery of the desired medication dose entering the fluid path at Position 4 would not be achieved for more than 17 min, potentially a clinically significant prolongation compared with the onset half-time of about 5 min (approximately 10 min to achieve steady-state delivery) for a medication entering the fluid path at the first port with the same flow rates. Selection of a more upstream port results in a larger V that must be traversed by the infusion resulting in slower onset times. If the carrier flow is increased to hasten the achievement of steady-state delivery of a new infusion, or increase the dose of an existing infusion, other infusions in the same line may be at least transiently affected, resulting in unplanned overdoses and possible fluctuations in the patient’s physiologic condition. The magnitude of this bolus delivery will depend on the V, total flow rates, and concentrations of the drug solutions.2,4,5

When a drug infusion ceases, there is a small, abrupt reduction in drug delivery (Fig. 3). This step-off is attributable to the acute reduction of total fluid flow in the delivery system with the existing dilute steady-state concentration still within the V. Delivery offset (or dose reduction) thus appears to be a 2-phase process, in contrast to onset or increases of drug delivery. The first phase of drug offset is a function of the difference between total flow at steady state (when the drug infusion is running) and the reduced total flow after cessation of the drug infusion. The second phase depends on V, and thus on port selection and manifold design. The in vitro data suggest a long lag time (up to 18 min) to a new steady state when trying to completely eliminate delivery of a particular medication and by extension when attempting to decrease infusion to a lower dose. The magnitude of this lag depends on the port. Temporarily increasing the carrier flow rate may reduce the lag time to achieve the intended steady-state delivery of a medication at a lower dose. However, this might be clinically counterproductive because the patient may receive an unintended bolus of the undesired drug.2 The subsequent decreasing of the carrier flow rate would then result in a transient underdosing that may be prolonged.1

In previous work,1 we described 2 mathematical models to illustrate the behavior of these systems. In the plug-flow model, the drug traverses the V as a fluid plug without any mixing. Steady state is reached immediately after 1 time constant (τ) (see equation).

In contrast, the well-mixed model posits that from the moment the drug enters the carrier stream it mixes instantaneously and completely with the remainder of the fluid in the V. This yields an exponential relationship between concentration and time.

We found that the infusion system onset values appear to behave closely to the predictions of the well-mixed model (2 × t50 approximates 3 × τ).

This finding provides the basis to estimate the increase in carrier flow required to have Port 4 of the stopcock manifold exhibit delivery kinetics for onset comparable with Port 1. The data suggest that by using the desired time constant along with the V of the fluid path, clinicians can calculate the increase in carrier flow required to provide comparable kinetics for onset between different ports of a stopcock manifold. A clinical consequence is that the enhanced rate of onset comes at the expense of an obligate delivery of additional fluid, which may not be appropriate for individual patients intolerant of volume.

The dependence of t50 (onset, offset) on V is not strictly linear (Fig. 4), suggesting that other factors, such as extent of mixing at flow convergence, tubing diameter, or connection angles, may be relevant. Port selection was less important for the microinfusion manifold because the dead volumes were nearly identical, and kinetics of drug delivery were faster when compared with the second and fourth (upstream) ports of the traditional stopcock manifold at the carrier flow rate used in this study (10 mL/h). This carrier flow rate was chosen because critically ill patients are often sensitive to volume, and efforts must be made to limit excessive fluid administration, leading to the practice of infusing drugs prepared as concentrated stock solutions with low flow rates for carriers.6 In addition, syringe pump startup behavior will influence the net delivery profile,7,8 particularly in the pediatric setting.9 However, the flow rates used for adult microinfusions are likely to be significantly higher than for pediatric microinfusions (e.g., 13 vs 2 mL/h total flow, respectively). The relative impact of time lags attributable to syringe pump startup behavior for onset kinetics in the adult system will likely be smaller than for pediatric infusions. This justifies using only V and Q in time constant calculations for adult infusion systems to estimate drug infusion behavior.

As exemplified by the findings with the microinfusion manifold, small dead volumes minimize delays in onset and offset at low carrier flows, while providing the opportunity to infuse multiple medications simultaneously. The effect of manifold design, and how the manifold is used in clinical settings, may be underappreciated and potentially plays a substantial role in optimizing drug delivery when it is most essential.10 Our findings have implications for safe, controlled management of powerful medications administered to anesthetized or critically ill patients. With a mathematical model to predict the clinical significance of carrier flow rate as well as manifold and catheter dead volumes on drug delivery kinetics, central venous drug infusions can be effectively used in a controlled and predictable manner.


The authors thank Dr. Michael Parker for careful commentary on the manuscript.


The equation for calculating total flow for a given time constant is Q = V/τ, where Q is the total system flow (carrier plus drug) and V is the dead volume of the fluid pathway. For drug entering the fluid pathway at stopcock #4, the dead volume, V, is 1.46 mL with this configuration of components. To approach the time constant for entry of the drug at stopcock #1 (dead volume 0.61 mL, time constant 2.81 min at a total flow of 13 mL/h), the total flow must be approximately 31 mL/h. Assuming a drug flow rate fixed at 3 mL/h, the carrier rate needs to be 28 mL/h.


1. Lovich MA, Doles J, Peterfreund RA. The impact of carrier flow rate and infusion set dead-volume on the dynamics of intravenous drug delivery. Anesth Analg 2005;100:1048–55
2. Lovich MA, Peterfreund GL, Sims NM, Peterfreund RA. Central venous catheter infusions: a laboratory model shows large differences in drug delivery dynamics related to catheter dead volume. Crit Care Med 2007;35:2792–8
3. Lovich MA, Kinnealley ME, Sims NM, Peterfreund RA. The delivery of drugs to patients by continuous intravenous infusion: modeling predicts potential dose fluctuations depending on flow rates and infusion system dead volume. Anesth Analg 2006;102:1147–53
4. Geater RE, Leff RD, Roberts RJ. Factors affecting drug delivery from a syringe-pump infusion set. Am J Hosp Pharm 1985;42:2510–3
5. Kubajak CA, Leff RD, Roberts RJ. Influence of physical characteristics of intravenous systems on drug delivery. Dev Pharmacol Ther 1988;11:189–95
6. Fadi A, Khamiees M, DeGirolamo A, Amoateng-Adjepong Y, Manthous CA. Negative fluid balance predicts survival in patients with septic shock. Chest 2000;117:1749–54
7. Neff T, Fischer J, Fehr S, Baenziger O, Weiss M. Start-up delays of infusion syringe pumps. Paediatr Anaesth 2001;11:561–5
8. Rooke GA, Bowdle TA. Syringe pumps for infusion of vasoactive drugs: mechanical idiosyncrasies and recommended operating procedures. Anesth Analg 1994;78:150–6
9. Bartels K, Moss D, Peterfreund RA. An analysis of drug delivery dynamics via a pediatric central venous infusion system: quantification of delays in achieving intended doses. Anesth Analg 2009;109:1156–61
10. Kohn LT, Corrigan J, Donaldson MS. Building a safer health system. Washington, DC: National Academies Press, 2000
© 2009 International Anesthesia Research Society