INTRODUCTION
Intravenous (IV) infusion is a medical procedure used to administer fluids, medications, or nutrients directly into patients’ veins. In clinical settings, this technique is essential for maintaining fluid balance, prompt medication delivery, and providing nutrition when oral intake is not feasible. IV fluid therapy is widely used owing to its critical role in modern medicine [1]. Accurate infusion rates are crucial for optimal patient outcomes; nevertheless, maintaining a precise flow rate can be challenging [2,3]. Reportedly, up to 84% of patients experienced complications caused by incorrect IV therapy [4-6].
IV infusion flow regulators (IIFRs) are popular tools that are used to control infusion rates without complex machinery. These devices are typically small, portable, and cost-effective, making them accessible across various healthcare settings, including those with limited resources. However, as gravity-based products, IIFRs have certain drawbacks. The known causes of inaccuracy in these systems include fluid viscosity, fluid bag height, changes in patient position, and resistance in patient blood vessels [7-11]. Inaccuracies in flow regulation can result in improper fluid administration, a common error in IV therapy. These errors tend to increase over long infusion durations, which increases the likelihood of complications. Consequently, continuous monitoring and adjustment of infusion rates are required to reduce the risks associated with incorrect fluid administration; nonetheless, the task is labor-intensive and often relies on medical personnel, introducing potential human error.
The use of automated infusion control devices can help enhance accuracy, significantly reducing fluid administration errors. However, these devices are typically large and expensive, which limits their applicability across patient populations. Therefore, a fluid infusion device that is simpler and more intuitive than an electric pump is required. This device should also be more accurate than gravity-based systems such as IIFRs. In previous studies, gravity-based IV sets have been modified, attempting to achieve greater accuracy in fluid infusion; however, the solutions have not been widely adopted [12-15]. In previous studies, drop-size variations caused by hydrostatic pressure changes inherent in gravity-based IV sets were not accounted for in the technologies that were introduced. To improve the accuracy of gravity-based IV sets, the authors aimed to address this issue by applying the single drop weight, calculated from the most recent 100 drops to the real-time drop rate, thereby accounting for these factors. Furthermore, single-drop weight measurements to drop counts per min measured using a light source have not been used in previous studies. Therefore, in this study, the aim was to compare the accuracy of IIFRs with a measurement system using pulse oximetry as a light source and to evaluate the effect of single-drop weight measurements in enhancing infusion accuracy.
MATERIALS AND METHODS
Study products
The gravity-based IV sets used in this study had IIFR, commonly used in domestic hospitals, and were obtained from three different companies (Table 1). In all experiments, a single fluid type, 1,000 ml of 0.9% normal saline (HK inno.N), was used without medication. A 100 ml solution of 0.9% normal saline (HK inno.N) without medication was used to remove air from the IIFR set tubing. A pulse oximeter (PO) (Nellcor™ SpO2 Adhesive Sensors; Neonatal/Adult < 3 kg or > 40 kg, Medtronic plc), which served as a light source, was used to measure the number of fluid drops in the drip chamber. It was connected to an IntelliVue MP70 monitor (Philips) to record the number of drops per min. Fluid from each experiment was collected in a beaker (Samduk-Lab), and its weight was measured using an electronic scale (CAS MWⅡ-H, CASSCALE). This study did not involve human participants or identifiable personal data, and therefore did not require Institutional Review Board approval under institutional and international guidelines. The study was limited to non-clinical evaluation of medical devices.
Common study design
The manufacturer-recommended hanging height of the IV set was 100 cm. Hence, the saline bag was hung such that the distance from the air-fluid level in the drip chamber to the outlet, where the fluid dripped into the beaker, was 100 cm (Fig. 1). The drip chamber was filled to the levels indicated by the manufacturer. The PO was positioned in the center of the air space above the air-fluid level of the drip chamber and wrapped to ensure its light source was oriented perpendicular to the falling fluid drops (Fig. 1A). Furthermore, it was connected to the IntelliVue MP70 monitor to record the drop count per min (Fig. 1B). The beaker opening was covered with plastic wrap to prevent evaporation.
The roller clamp was locked before connecting the IIFR set to a fluid bag with a specified residual volume. Subsequently, the prepared fluid bag with the specific residual volume was connected to the IIFR set, with the tubing tip fixed to the beaker. For each experiment, the fluid injection order was as follows: setting the IIFR dial to a specified flow rate, fully opening the roller clamp to initiate the fluid injection, and then locking the roller clamp when the experimental conditions (100 drops, 10 min, or 30 s) were met. The alignment of the IIFR flow rate scale was verified using a prefabricated plastic plate slit (Fig. 1C). All experiments were conducted by one researcher (SM).
Fluid bags with specified remaining volumes (900, 500, and 100 ml) were prepared by removing fluid from the injection port (not the IV set connecting port) of the 1,000 ml fluid bag using a 50 ml syringe (WEGO, Shandong WeiGao Group Medical Polymer Co., Ltd.). The fluid bag was prepared by a single researcher (DO). The drop count readings were confirmed jointly by all researchers, except for the two who conducted the experiment and prepared the fluid. Each fluid volume in the beaker was indirectly calculated using the relative density of the fluid (kg/L), previously measured by recording the weight of 100 ml of the fluid at 24°C. All experiments and measurements were conducted in the operating room of the Inje University Haeundae Paik Hospital.
Measurement of the weight of 100 drops
The weight of 100 drops was measured to calculate the weight of each drop. A pre-filled fluid bag with a specified remaining volume (900, 500, or 100 ml) was connected to the IIFR set, and the roller clamp was opened after setting the flow rate. The clamp was closed, and the collected fluid was weighed after exactly 100 drops were delivered. This process was repeated when measuring flow rates using all the IIFRs.
Measurement of drop counts per min and fluid volume collected over 10 min
The same IIFR set used to measure the weight of 100 drops of fluid was used to collect fluid for 10 min. A prepared fluid bag was connected to the IIFR set, and the fluid was collected in a beaker for 10 min at a specific flow rate. While fluid was collected, a PO was used to simultaneously measure drop counts per min. The drop counts per min maintained for more than 30 s were recorded; multiple values were observed; and their average was recorded. The drop counts for all flow rates were measured, as well as for the prepared fluid bags (remaining volumes of 900, 500, and 100 ml). The fluid’s weight collected over 10 min was measured using an electronic scale. All experiments were conducted using two identical IIFR sets of the same model from each manufacturer.
Measurement of drop counts per min based on remaining fluid volume and IIFR flow rate in a single fluid bag
An IIFR set with air removed was placed in a 1,000 ml bag of 0.9% normal saline, and a PO was wrapped around the drip chamber. The scale was adjusted to the highest flow rate of the IIFR set, and the roller clamp was fully opened. The drop counts per min were measured for 30 s using a PO, and the average of the values maintained for more than 5 s was recorded. After 30 s, the roller clamp was locked, and the dial was adjusted to the next lower flow scale on the IIFR set. In this manner, drop counts were measured down to the lowest IIFR flow rate. After measuring at the lowest IIFR flow rate, the roller clamp was reopened and closed once 100 ml of fluid was collected in the beaker. Subsequently, the scale was reset to the highest flow rate, and the drop-count measurement procedure was repeated down to the lowest flow rate. The drop counts per min were recorded from the highest to the lowest IIFR flow rate until the remaining fluid volume in the bag reached 100 ml. All experiments were conducted using three sets of IIFRs of the same model from each manufacturer.
Formula for predicting fluid collected over 10 min
The predicted volume after 10 min was calculated as follows:
Predicted volume (ml)=drops per minute (count) × single drop volume (ml) ×10 (min)
The group in which the volume was predicted after 10 min using the manufacturer-provided volume per drop (20 drops = 1 ml) was designated manufacturer’s drop volume (C) group. The group in which the actual measured weight of one drop was used was designated as the PO group. The group in which drop counts per min were applied based on the IIFR flow rate was the IIFR group. Thus, the predicted volume formulas for each group over 10 min were as follows:
Predicted volume of group C (ml) = drops per minute (PO) × 0.05 ml × 10 min
Predicted volume of group PO (ml) = drops per minute (PO) × the measured weight of 1 drop (mg) × 10 min
Predicted volume of group IIFR (ml) = drops per minute (IIFR flow rate) × the measured weight of 1 drop (mg) × 10 min
Single drop volume was indirectly calculated using the fluid’s relative density (kg/L), previously measured by recording the weight of 100 ml at 24°C.
Primary and secondary endpoints
For the primary outcome, the differences between the actual fluid collected over 10 min and the predicted values of the three groups (C, PO, and IIFR) were compared using the mean absolute percentage error (MAPE). The closer the MAPE was to zero, the closer the predicted value was to the actual measurement. For the secondary outcome, the correlation between the residual fluid volume and drop rate was determined.
Statistical analyses
Power analysis for an independent sample t-test was conducted using the G*power 3.1.9.4 to determine a sufficient sample size with an alpha of 0.05, a power of 0.90, a large effect size (d = 0.8), and two tails. An equal allocation ratio was used. Based on the aforementioned assumptions, the desired sample size was 68 (34 per group). Using a drop-out rate of 10%, the total sample size was 76 (38 patients per group). The data are presented as mean ± standard deviation for continuous variables. For numeric data, group differences were tested using the independent t-test or the Mann-Whitney U test, and analysis of variance or the Kruskal-Wallis test, as appropriate. To check if its distribution is normal, we used the Shapiro-Wilk test. A linear regression line was used to estimate the relationship between the drop rate measured using the PO and the remaining fluid. A scatter plot with a fitted line was constructed for data visualization.
All statistical analyses were performed using the IBM SPSS Statistics for Windows ver. 26.0 (IBM Co.). Statistical significance was set at P < 0.05.
RESULTS
Comparison of MAPE between the PO and C groups
The predicted value obtained in the pulse oximeter (PO) group was statistically closer to those of the actual measurements across all manufacturers’ products (P < 0.001), as shown in Table 2 and Fig. 2. For all residual fluid volumes used in the experiment, the PO group’s predictions were statistically closer to those of the actual measured values (P < 0.001). In the MAPE comparison by infusion rate, the PO group’s prediction was significantly closer to that of the actual value, except at an infusion rate of 300 ml/h (P = 0.939). Detailed MAPE results for each manufacturer’s IIFR are presented in Supplementary Tables 1-3, which show that for all products used in the experiment, the value in the PO group was statistically closer to that of the measured value than that in the C group (P = 0.002, P < 0.001, and P < 0.001, respectively). A statistically significant difference in the MAPE was also observed across all manufacturers’ IIFR at the same flow rates, as shown in Table 3 (P < 0.001).
Comparison of MAPE between PO and IIFR groups
The predicted values from the PO group were statistically closer to the measured fluid weight than were the values from the IIFR group (P < 0.001), as shown in Table 4 and Fig. 2. The MAPE for the residual fluid volumes also indicated that the PO group value was closer to the measured values regardless of the amount of fluid (P < 0.001). Across all IIFR flow rates, those from the PO group were statistically closer to the measured values than were the values from the IIFR group (Table 4, Fig. 2). A comparison of the MAPE values for each manufacturer is presented in Supplementary Tables 4-6. For all products used in the experiment, those from the PO group showed a statistically significant closeness to the measured values compared with those from the IIFR group (P < 0.001). Differences in MAPE across all manufacturers’ IIFR at the same flow rate are shown in Table 5; no statistically significant differences were observed (P = 0.160).
Correlation between residual volume and drop rate
The correlation analysis between residual fluid volume and drop rate for products A, B, and C is depicted in Fig. 3, respectively. All the tested products exhibited a strong correlation, with correlation coefficients (r = R 2
DISCUSSION
In this study, the accuracy of the predicted infusion volume was significantly improved using the actual measured weight of the single drop method, which was statistically closer to the actual measured volume compared with those predicted using the IIFR flow rate (Table 4, Fig. 2). Additionally, even when the number of drops per min measured with a PO was the same across groups, using the actual measured weight of a single drop. This method yielded results that were statistically closer to the measured values than those obtained when using the manufacturer-provided volume of drops (20 drops = 1 ml; Table 2, Fig. 2). Furthermore, a strong correlation between the remaining fluid volume and the drop rate, as measured by the PO was confirmed, suggesting that these correlations can influence the accuracy of the predicted infusion volumes (Fig. 2).
Thus, it was demonstrated in the present study that real-time measurement of drops using a light source, the weight of a single drop, and the remaining fluid volume are crucial variables affecting the accuracy of gravity-based infusion sets. All tested products exhibited a droplet weight error range of < 10% for 20 drops of 1 ml at a flow rate of 150 ml/h (50 drops/min), which meets the International Organization for Standardization (ISO) 8536-4 requirement (Supplementary Tables 7-9). However, for certain products, deviations exceeding the 10% error range were observed at flow rates other than 150 ml/h. This large deviation in droplet weight suggests that it may affect the accuracy of the fluid administration predictions. Furthermore, a 10% margin of error indicates that the total range of deviation in fluid administration predictions could be up to 20%. The clinical impact of infusion errors is likely to vary depending on the patient’s clinical condition and underlying disease. Critically ill patients with impaired cardiopulmonary function or older patients are more susceptible to fluid overload or depletion, which can significantly affect hemodynamic stability and organ function. Furthermore, in pediatric and neonatal patients, even small infusion errors can lead to dehydration or overhydration, posing a greater risk owing to their limited physiological reserves. Given the potential clinical significance of precise fluid administration, particularly in high-risk patient populations, the effects of accurate fluid infusion in critical conditions should be further investigated, and the threshold at which error reduction impacts clinical outcomes should be determined. These findings underscore the need for precise monitoring of gravity-based IV administration, especially in critical care settings where strict fluid control is paramount.
IIFRs are widely used in medical settings because of their affordability, simple design, and ease of use. While electric pumps provide more accurate fluid management, their large size, heavy weight, and high-cost limit their widespread use. Operational complexities have also been highlighted with their use. These problems include different usage methods across different electric pump manufacturers, requiring specific training for electric pump users, and guidelines for standardized designs [16,17]. In this study, the aim was to validate the accuracy and predictability of fluid administration using pulse oximetry across a range of products, building on a previous study, which showed that pulse oximetry can be used as a light source to measure infusion rates in real-time [18]. As shown in Supplementary Tables 7-9, slight variations were found in the weight of 100 fluid droplets depending on the remaining fluid volume and the infusion rate; however, statistical verification of this relationship was not conducted.
Various factors affecting the flow rate accuracy of gravity-based infusion sets, including fluid viscosity, bag height, and patient position, have been identified [9-11]. Light sources, ultrasound, and weight measurement technologies were incorporated into drop detection procedures to improve the accuracy of fluid infusion [12,19]. Light sources, such as infrared or light-emitting diode, have been used for drip detection in several patents and products; nonetheless, broad commercial adoption has not yet occurred [13-15,19-21]. Notably, drop-size variations or the impact of remaining fluid volume were not addressed in these patents and studies. If these variations and the remaining fluid are not accounted for, it will be difficult to provide an accurate real-time infusion rate or predicted volume, which could be a significant barrier to clinical adoption. Therefore, the authors aimed to address this issue by applying the single drop weight, calculated from the most recent 100 drops to the real-time drop rate, thereby accounting for the effects of drop size variations and the remaining volume. Specifically, the average weight of the most recent 100 drops was used to reflect the dynamic variability in drop size caused by changes in the remaining fluid volume. As direct measurement of a single drop was not feasible with the electronic scale, the average of 100 drops was applied to maintain measurement sensitivity and experimental practicality. This study is the first study where the weight of individual fluid drops and the drop rate were measured using a light source to predict the volume of fluid injected. The results showed that more accurate predictions of fluid volume are possible than with current gravity-based sets.
Findings from a previous study indicated discrepancies between the flow rate set on IIFR devices and the actual drop rate [18]. The flow rate of a gravity-based set can vary because of several factors, meaning that even if a healthcare provider monitors it visually, the actual infusion rate can fluctuate over time and in different environments [5,22]. Our findings show a strong correlation between residual fluid volume and drop rate, indicating that as the fluid volume decreases, the infusion rate also declines (Fig. 2). This trend is likely driven by reduced hydrostatic pressure, even when the fluid bag height and initial flow rate settings remain unchanged. These findings reveal the need for careful monitoring of gravity-based IV infusions, particularly in critical care settings where precise fluid administration is important. Furthermore, the IIFR sets from different manufacturers delivered statistically different fluid volumes under identical flow rates (Table 3). Therefore, real-time monitoring of the fluid drop is essential for ensuring the accuracy of infusion with gravity-based sets. The monitoring and adjusting of infusion rates heavily depend on the nursing staff, who are frequently understaffed in practice [23,24]. Using a light source to measure the infusion rate allows for more accurate, real-time detection of changes due to environmental factors, which may help reduce the burden on healthcare workers.
Advances in optical and electronic technologies have enabled drop rate measurements across various simple tools such as POs, as demonstrated in this study [13,18,19]. This capability, supported by prior research on various light sources, allows for accurate real-time drop-rate monitoring [12]. Thus, using light sources enables relatively accurate, real-time drop rate measurements and allows for the simultaneous monitoring of multiple patients through interconnected devices [12,21]. Additionally, the need for direct patient contact, particularly for those suffering from infectious diseases, could be reduced if drop rate measurement using light sources and weight measurement using an electronic scale could be remotely conducted. Finally, our findings show that gravity administration sets can be easily verified using an accessible light source, such as a PO.
The authors propose several criteria for selecting superior gravity administration sets based on the study’s findings. These include minimal variations in infusion rate across products from the same manufacturer, consistent single-drop weight, and minimum variation in infusion rate with residual fluid volume changes. We showed that POs and electronic scales can be used to evaluate these criteria, enabling healthcare providers to directly verify products using such light sources. Manufacturers can also incorporate light sources and an electronic scale into quality control to efficiently measure infusion rates, thereby verifying their products.
Our findings reveal that variations in single-drop weight due to infusion rate can significantly affect the predicted infused volume in gravity administration sets. Based on these findings, the authors propose developing a device comprising a scale and light source, which allows real-time recognition of the drop weight and rate. Feeding the drop weight measured in real-time and the drop count from a scale and light source, respectively, into an electronic device and automatically integrating the calculations addresses two key variables in gravity-based IV infusion sets. This approach enables more accurate measurement of infusion rates and volume predictions. Such a device can be used to significantly improve infusion rate measurements and volume predictions for gravity-based sets, offering a simpler and more accurate real-time alternative to electronic pumps. While our approach primarily serves as a proof of concept rather than a direct clinical application, it highlights the potential for cost-effective, real-time monitoring of gravity-based IV infusions. In future advancements, similar light-based sensors and electric scales could be integrated into IV administration systems, enhancing automation and reducing the burden on healthcare providers.
Nevertheless, several precautions were identified when using light sources for drop detection. The fluid chamber should remain clean to prevent measurement errors from light refraction or obstruction. The light emitter and receiver should be aligned perpendicular to the falling drops to ensure accuracy, with the drip chamber vertically oriented to ensure drops intersect with the light beam.
This study has some limitations. First, this study was conducted in a controlled laboratory setting and not on actual patients; therefore, factors such as venous resistance, movement of the patient, and temperature fluctuations, which could influence infusion accuracy, were not considered. While real-time detection of changing drop rates is possible using a light source, additional clinical trials or observational studies in real patient settings are necessary to validate these findings and assess potential real-world variables affecting fluid infusion. The focus of future research should be on evaluating the clinical applicability of this method in diverse patient populations and hospital environments. Second, only products available in South Korea and a specific PO and monitoring equipment currently used in hospitals were utilized in this study (Table 1). While all gravity-based sets used in this study met the ISO 8536-4 standards, similar inaccuracies associated with these sets have been reported in many studies. Products used in other countries were not included in our investigation. Therefore, further research is required to assess these results using products from international manufacturers and other brands of POs and monitoring equipment. Third, at lower IIFR flow rates (80 and 83 ml/h), multiple drop rate values were observed, and the authors used an average value. It is unclear whether this variability was caused by the characteristics of the gravity administration set or the limitations of the light source. Further assessment with more precise equipment is required.
Finally, the cost of the devices used for a single measurement in this study was less than one-twentieth of the lowest-priced infusion pump currently used in the authors’ hospital, suggesting a favorable cost-performance ratio. However, the commercialization of this technology would require addressing various challenges, including legal regulations and medical device approvals. More intuitive usability and error reduction would be essential to use this technology in clinical settings, necessitating further research on integrating electronic scales with computational units and implementing machine learning for automatic error correction. Additionally, alternative methods that are simpler and more accurate than estimating single-drop volume from 100 drops should be explored.
In this study, the findings show that using the measured single drop weight and real-time drop rate improves the accuracy of the predicted infusion volume. Additionally, using a light source, it was confirmed that the drop rate changes relative to the remaining fluid volume, and a strong correlation exists between the two. Rather than focusing on the mechanical accuracy of IIFR devices, we propose a novel monitoring model where electronically measured drop rate is integrated with real-time weight feedback. These findings reveal that gravity-based administration sets can be verified using accessible light sources, potentially reducing the burden on healthcare providers while improving their accuracy.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement1
