- Email: [email protected]
Three-phase study of phlebitis in patients receiving peripheral intravenous hyperalimentation
Three-Phase Study of Phlebitis in Patients Receiving Peripheral Intravenous Hyperalimentation Eric B. Rypins, MD, Beverly H. Johnson, RN, Sr. Barbara Reder, RD, I. James Sarfeh, MD, Kathi Shimoda, RN, Irvine, California, Long Beach, California We found clinical phlebitis in 57 of 88 patients with peripheral hyperalimentation ( 6 5 % ) . To determine if this was a problem common to all intravenous fluid therapy at our hospital, we performed a point-prevalence study. The prevalence of phlebitis in nonhyperalimentation intravenous patients was 18% (84 of 456 patients). We then performed a randomized, prospective, double-blind trial of sham versus standard in-line filters to determine if bacteria or filterable particulate matter was responsible for phlebitis in the peripheral hyperalimentation group. The standard-filter group had a phlebitis rate o f 74% compared with 64% in the sham-filter group. We then eliminated in-line filters and replaced the standard glucose-based solution with a glycerol-based peripheral hyperalimentation solution. The phlebitis rate decreased from 68% to 27% (p < 0 . 0 0 1 ) . In conclusion, phlebitis in peripheral hyperalimentation patients was probably due to chemical properties of the peripheral hyperalimentation solution rather than bacteria or particulates.
is a frequent complication of intravenous fluid p hlebitis administration, its incidence varying between 11% and 70% of hospitalized patients [1-9]. The problem is particularly common in patients receiving peripheral hyperalimentation intravenously. The catheter material, duration of therapy, sterile technique, composition of the solution (pH, osmolality), location of the site, decreased blood flow, and infusion rates may be responsible for it. Other related aspects may be age, blood abnormalities, diabetes mellitus, debilitation, level of activity, and the quality of patient care. The infusion system can be contaminated when tubing or filters are inserted or when dressings are changed. From the Nutritional Support Team, Veterans Administration Medical Center, Long Beach, California, and the Department of Surgery, University of California, Irvine, California. Requests for reprints should be addressed to Eric B. Rypins, MD, University of California, Irvine Medical Center, Department of Surgery, Orange, California 92668. Manuscript submitted November 1, 1988, revised January 9, 1989, and accepted January 20, 1989.
222
THE AMERICAN JOURNAL OF SURGERY
In-line millipore filters may remove particulate matter as well as microorganisms and have been shown to significantly reduce phlebitis rates in patients receiving intravenous therapy [7,10,11 ]. However, unlike routine intravenous lines, hyperalimentation lines are usually restricted to the solution alone, and the efficacy of filters in preventing phlebitis in this setting is questionable. In our hospital, regulations forbid nurses to inject antibiotics and other medications into hyperalimentation lines. Currently, there is no consensus regarding the usefulness of in-line filters for peripheral hyperalimentation. Various uncontrolled studies have found that filters do not prevent phlebitis and increase the cost of therapy [12-14]. To reduce phlebitis rates in peripheral hyperalimentation, one group adds chemical buffers and anti-inflammatory agents to the solution [15]. We conducted this study in three phases to determine the sources of phlebitis in our peripheral hyperalimentation patients and identify methods for preventing it. In the first phase, we determined the rate of phlebitis in patients with peripheral hyperalimentation over the 30 months that we monitored this complication. In addition, we did a point-prevalence study to determine the prevalence of phlebitis in patients with intravenous catheters but not receiving peripheral hyperalimentation. In the second phase, we evaluated the efficacy of in-line filters for preventing phlebitis by conducting a double-blind, randomized, prospective trial. In the third phase, we eliminated filters and replaced our standard 10% glucose-2.75% amino acid solution with a 3% glycerol-3% amino acid solution to determine the effect of different peripheral hyperalimentation solutions on phlebitis rates. MATERIAL AND METHODS The nutritional support team first screened and approved all requests for hyperalimentation (including peripheral hyperalimentation) before the pharmacy dispensed the solution. Once hyperalimentation was begun, the nutritional support team nurse inspected the intravenous infusion sites daily. Criteria for the presence of phlebitis consisted of three or more of the following findings: erythema, tenderness, induration, or palpable venous cord greater than 2.5 cm. Bacterial phlebitis was diagnosed if the catheter tip grew microorganisms in culture. We monitored patients for fever and chills on a daily basis and did a complete blood count at least once a week. We examined the records of all patients receiving peripheral hyperalimentation over the 30 months of the study and determined the incidence of phlebitis. During this time, we followed our previous policy of using in-line filters for peripheral hyperalirnentation.
VOLUME 159 FEBRUARY 1990
PHLEBITIS WITH PERIPHERAL HYPERALIMENTATION
We did a point-prevalence study to determine the alimentation solution with 3% glycerol-3% amino acid phlebitis rate in hospitalized inpatients receiving nonhy- solution with electrolytes, 735 mOsm/L (Procalamine, peralimentation intravenous therapy. Ward nurses con- KendaU-McGaw, Irvine, CA). We compared a consecuducting the survey were instructed regarding the specific tive series of 41 patient s with the 41 patients in the rancriteria for diagnosing phlebitis as described above. On a domized trial. Other than replacing the solution and elimsingle day, all inpatients were surveyed and the preva- inating in-line filters, there were no changes in the protocol. lence of phlebitis was determined. In the second phase of the study, we randomized 50 consecutive patients approved for peripheral hyperali- RESULTS Retrospective review and point-prevalence study mentation into either standard filter or sham filter groups based upon a computer-generated list of random num- (Phase 1 ): We began carefully monitoring for phlebitis bers. The Healthtek Corporation (Nevada City, CA) 30 months before the randomized trial. During that perimanufactured, coded, and packaged the filter devices, od, 88 patients received peripheral hyperalimentation. and they appeared identical. The standard device was a All patients receiving hyperalimentation had in-line fil0.22-/~ millipore filter, and the sham device had a 50-/~ ters during this part of the protocol. Clinical phlebitis pore size with no significant filtering capabilities. At the occurred in 57 patients, or 65% of the entire group. The point-prevalence study identified patients receivend of the study, we broke the code and determined the differences in the incidence and cause of phlebitis using ing nonhyperalimentation intravenous therapy on the day chi-square statistical comparisons. Comparisons between of the survey. By the criteria listed above, phlebitis was group characteristics were by chi-square and Student's t present in 18% of 456 patients on active medical and test. The Human Investigation Committee approved the surgical wards. Randomized trial of standard versus sham filters protocol without the need for written informed consent. The Research and Development Committee approved (Phase 2): Fifty patients receiving peripheral hyperalimentation were randomized for standard or sham filters, the double-blind, randomized, prospective design. but 9 were excluded for major protocol violations The The cannula insertion technique was standardized. The site was prepared by brisk scrubbing with povidone- reasons for exclusion were nurses using an unapprovexl, iodine solution and allowed to dry. A 20-gauge plastic non-study filter from old ward stock (seven patients) and cannula was used. One venipuncture attempt per cannula not having the site evaluated by the nutritional support was allowed with a new cannula used for each attempt to team nurse before ending the infusion (two patients). The enter a vein. Once inserted, the cannula was secured with data below represents our findings in the 41 evaluable silk tape, Neosporin ointment was applied to the site, and patients. The standard filter group consisted of 20 patients and a gauze dressing was used to cover it. All tubing, filters, and dress!ngs were changed every 48 hours unless re- the sham filter group of 2i patients. All subjects were quired sooner. No additional solutions or medications men with a wide variety of diagnoses. Surgical service were added to the intravenous lines. Cannulation, care of patients represented 68% of the total and medical service the site, and tubing/filter changes were the responsibility patients, 32%. Six medical patients and four surgical paof ward nurses according to established procedure and tients had more than one episode of phlebitis. There were protocol. Before the cannula was removed, the site was no significant differences between the sham filter group prepared with povidone-iodine solution, and any excess and the standard filter group with respect to age, total was removed with a sterile 2 • 2 gauze sponge. A random treatment days, average treatment days, or distribution sampling of cannula tips from patients with clinical signs between medical and surgical services (Figure 1). None of phlebitis was sent to the microbiology laboratory for of the patients in the study showed systemic signs of culture and sensitivity. Cultures of the catheter tips were sepsis. The phlebitis rate in study patients was 68%, not done according to the poured-plate technique described significantly different from the 65% rate found in the retrospective review. Phlebitis occurred in 74% of patients by Snydman et al [16]. The peripheral hyperalimentation solution used in this phase of the study was 10% glucose- in the standard-filter group and 64% of patients in the 2.75% amino acid and 10% soybean emulsion. Electro- sham-filter group (Figure 2). The differences were not lytes, multivitamins, and trace elements were added per significant. We sent a random sampling of 25 cannulas from the orders of the attending physicians. The osmolarity of the glucose-amino acid solution without electrolytes was patients with clinical phlebitis for culture and sensitivity. 792 mOsm/L and with electrolytes was 930 mOsm/L. Only one culture was positive. The organism was identiMost of the time (>95%), the solution was orderedwith fied as Staphylococcus epidermidis. Random bacterial electrolytes. The peripheral hyperalimentation solution cultures of the solutions done by the pharmacy service was always infused concurrently with 10% soybean emul- were consistently negative. Comparison of peripheral hyperalimentation sosion (300 mOsm/L) via a Y-connection to buffer the osmolarity of the glucose-amino acid solution. Random lutions (Phase 3) : After the randomized trial, the use of samples of prepared solutions were routinely cultured by in-line filters was abandoned, and 3% glycerol-3% amino acid injection (Procalamine) and 10% fat emulsion were pharmacy personnel for infection control purposes. Upon completing the randomized trial, we eliminated used as the peripheral hyperalimentation solution in a in-line filters and replaced our standard peripheral hyper- consecutive series of 41 patients. We compared the phleTHE AMERICANJOURNALOF SURGERY VOLUME159 FEBRUARY1990
223
RYPINS ET AL
160
144
1401 120 100 8O
65 71
60 4O 21 2 0 2O 0
I NUMBER PATIENTS
i
AVERAGE AGE
AVERAGE %SURGICAL %MEDICAL DAYS PATIENTS PATIENTS
TOTAL DAYS
S H A M FILTER
m
STANDARD FILTER
Figure 1. Comparison of' sham versus standard filter. The two groups are compared for differences in number of patients, average ages, total days on hyperalimentation, average number of days on hyperalfmentation per patient, and the percentages of patients on surgical and medical services. There were no SignifiCant differences between the groups on any of the characteristics examined.
INCIDENCE (%)
68
STANDARD FILTER
SHAM FILTER
GLUCOSE pHAL
GLYC=:RmE pHAL
ly
Figure 2. Comparison of phlebitis rates between groups: The two bars on the left represent the results of the randomized trial. The standard-filter group had a higher rate of phlebitis than the shamfilter group, although the difference was not statistically significant. The three bars on the right represent differences between groups Of patients based on the type of solution being infused. The bar labeled glucose pHAL (peripheral hyperalimentation) represents the phlebitis rate of the 41 patients in the randomized trial (phase 2)receiving the standard 10% glucose-2.75% amino acid solution. The bar labeled glycerine pHAL represents the phlebitis rate of 41 consecutive patients receiving 3% glycerol3% amino acid solution. The bar on the far right labelled IV represent's the preva[enca of phlebitis in456 patients receiving nonhyperalimentation intravenous therapy.
bids rate in these patients with that of the 41 patients in the randomized group that received the glucose-based solution. The phlebitis rate decreased to 27% in the 41 patients receiving the glycerol-based solution compared with 68% in the 41 patients receiving the glucose-based solution (Figure 2), The difference between the two groups was highly significant (p <0.001). The rate of phlebitis in patients receiving glycerol-based peripheral hyperalimentation was not significantly different from that in patients receiving nonhyperalimentation intravenous therapy (p = 0.23). 224
COMMENTS There is little discussion in the literature regarding phlebitis occurring in patients receiving peripheral hyperalimentation, leading some physicians to assume that it is not a significant problem, However, we were distressed by the magnitude of this problem in our patients, leading to this study. A retrospective review indicated that the rate of phlebitis in patients receiving peripheral hyperalimentation was 65%. This compared most unfavorably wit h the 18% prevalence of phlebitis in patients receiving nonhyperalimentation intravenous therapy on the medical and surgical services. Therefore, we investigated the reasons for the increased phlebitis rates in patients receiving Peripheral hyperalimentation. One factor implicated in causing phlebitis is particulate matter in the solution [4,9-12]. Our results did not support this. If particulate matter were responsible, we would have expected to find a higher phlebitis rate in the sham-filter group than the standard-filter group. However, no significant differences were found (Figure 1). :Some investigators have implicated bacterial contamination as the main cause of phlebitis in hospitalized patients requiring intravenous therapy [2,13-19]. We could not confirm this either. In a random sampling of cannulas from patients with clinical phlebitis, only one positive culture (S. epidermidis) was obtained. Moreover, a standard 0.22-~t filter should be effective in removing bacteria from intravenous infusions. We found that the 10% glucose-2.75% amino acid solution resulted in phlebitis rates of 68% regardless of whether a standard or sham filter was used. In the third phase of our study, we eliminated in-line filters and used 3% glycerol-3% amino acid injection with electrolytes. The phlebitis rate decreased to 27%, comparable with that found in nonhyperalimentation patients receiving intravenous therapy in our hospital. There are some possible explanations for our results. One is that filters are responsible for causing phlebitis and that eliminating them decreased phlebitis rates. Another explanation relates to the osmolarity of the solution. The osmolarity of our 10% glucose-2.75% amino acid solution without electrolytes was 792 mOsm/L, similar to the osmolarity of the 3% glycer01-3% amino acid solution with electrolytes (735 mOsm/L). However, most of the time, physicians added electrolytes to the glucosecontaining solution, increasing the 0smolarity to an average of 930 mOsm/L. We stress that the above explanation is speculative because we did not specifically design this study to test osmolarity as a cause of phlebitis, and we did not have a control group without filters in the randomized part of the study. Our data suggest that the solution itself may be a major etiologic factor in phlebitis occurring in patients receiving peripheral hyperaiimentation. Infection does not appear to be an important cause of phlebitis if careful sterile technique is applied to cannu!a insertion and maintenance of the intravenous system. Particulate material contained in antibiotics and other medications contributes to phlebitis in patients receiving nonhyperalimentation intravenous therapy [7]. In-line filters have been
THE AMERICAN JOURNAL OFSURGERY VOLUME159 FEBRUARY1990
PHLEBITIS WITH PERIPHERAL HYPERALIMENTATION
shown to prevent phlebitis in these settings [3,5,7], but our study in hyperalimentation patients showed that filters had no effect and perhaps even a deleterious effect. The results of our study should be interpreted and applied to other medical centers cautiously. O u r nutritional support team has tight control over all parenteral nutrition consults and maintains careful surveillance of all patients on peripheral hyperalimentation using a fulltime, fully trained hyperalimentation nurse. Moreover, this study was conducted under strict protocol conditions. However, we feel justified in concluding that routine inline filters for peripheral hyperalimentation are unnecessary provided that the line is not used for concurrent administration of medications.
Phlebitis is a frequent but underestimated complication of peripheral intravenous nutrition; these trials point out the primary role of osmolarity in this process. ACKNOWLEDGMENT The authors wish to acknowledge the cooperation of the Nursing Service and the Office of Quality Assurance for doing the pointprevalence study. REFERENCES 1. Adams S, Killien M, Larson E. In-line filtration and infusion phlebitis. Heart Lung 1986; 15: 134-40. 2. Consentino F. Personnel-induced infusion phlebitis. Bull Parent Drug Assn 1977; 31: 288-93. 3. DeMonaco H J, Cronin CM, Dempsey M J, Tulley S. Comparison of phlebitis rates with buretrol and Harvard mini-infusor for administration of IV piggyback medications. NITA 1986; 9: 475-9.
4. Ervin SM. The association of potassium chloride and particulate matter with the development of phlebitis. NITA 1987; 10: 145-9. 5. Hershey CO, Tomford JW, McLaren CE, Porter DK, Cohen DI. The natural history of intravenous catheter associated phlebitis. Arch Intern Med 1984; 144: 1373-5. 6. Hessov I. Prevention of infusion thrombophlebitis. Acta Anesthesiol Scand [Suppl] 1985; 29: 33-7. 7. Falchuk KH, Peterson L, McNeil BJ. Microparticulate-induced phlebitis: its prevention by in-line filtration. N Engl J Med 1985; 312: 78-82. 8. Crossley K, Matsen JM. The scalp vein needle: a prospective study of complications. JAMA 1972; 220: 985-7. 9. Maki D. Preventing infection in intravenous therapy. Hosp Pract 1976; 11: 95-104. 10. Harrigan CA. Care and cost-justification of final filtration. NITA 1985; 8: 426-30. 11. Rapp RP, Bivins B. Cost effectiveness of in-line filters in the era of DRG's. Am J Hosp Pharm 1984; 41: 2634-41. 12. Kaye W. Catheter and infusion-related sepsis: The nature of the problem and its prevention. Heart Lung 1982; 11: 221-8. 13. Murphy L, Lipman T. Preventing infection from total parenteral nutrition. Drug Ther Bull 1981; 1I: 51-8. 14. Turnidge J. Hazards of peripheral intravenous lines. Med J Aust 1984; 14: 37-40. 15. Makarewicz PA, Freeman JB, Fairfull-Smith R. Prevention of superficial phlebitis during peripheral parenteral nutrition. Am J Surg 1986; 151: 126-9. 16. Snydman DR, Murray SA, Kornfield SJ, Majka JA, Ellis CA. Total parenteral nutrition-related infections. Am J Med 1982; 73: 695-9. 17. Sanders RA, Sheldon GF. Septic complications of total parenteral nutrition, a five year experience. Am J Surg 1976; 132: 21420. 18. Stillman RM, Soliman F, Garcia L, Sawyer PN. Etiology of catheter-associated sepsis. Arch Surg 1977; 112: 1497-9. 19. Simmons B. Alternatives to IV filter usage. Infect Control 1985; 6: 342-4.
THE AMERICAN JOURNAL OF SURGERY VOLUME 159 FEBRUARY 1990 225
is a frequent complication of intravenous fluid p hlebitis administration, its incidence varying between 11% and 70% of hospitalized patients [1-9]. The problem is particularly common in patients receiving peripheral hyperalimentation intravenously. The catheter material, duration of therapy, sterile technique, composition of the solution (pH, osmolality), location of the site, decreased blood flow, and infusion rates may be responsible for it. Other related aspects may be age, blood abnormalities, diabetes mellitus, debilitation, level of activity, and the quality of patient care. The infusion system can be contaminated when tubing or filters are inserted or when dressings are changed. From the Nutritional Support Team, Veterans Administration Medical Center, Long Beach, California, and the Department of Surgery, University of California, Irvine, California. Requests for reprints should be addressed to Eric B. Rypins, MD, University of California, Irvine Medical Center, Department of Surgery, Orange, California 92668. Manuscript submitted November 1, 1988, revised January 9, 1989, and accepted January 20, 1989.
222
THE AMERICAN JOURNAL OF SURGERY
In-line millipore filters may remove particulate matter as well as microorganisms and have been shown to significantly reduce phlebitis rates in patients receiving intravenous therapy [7,10,11 ]. However, unlike routine intravenous lines, hyperalimentation lines are usually restricted to the solution alone, and the efficacy of filters in preventing phlebitis in this setting is questionable. In our hospital, regulations forbid nurses to inject antibiotics and other medications into hyperalimentation lines. Currently, there is no consensus regarding the usefulness of in-line filters for peripheral hyperalimentation. Various uncontrolled studies have found that filters do not prevent phlebitis and increase the cost of therapy [12-14]. To reduce phlebitis rates in peripheral hyperalimentation, one group adds chemical buffers and anti-inflammatory agents to the solution [15]. We conducted this study in three phases to determine the sources of phlebitis in our peripheral hyperalimentation patients and identify methods for preventing it. In the first phase, we determined the rate of phlebitis in patients with peripheral hyperalimentation over the 30 months that we monitored this complication. In addition, we did a point-prevalence study to determine the prevalence of phlebitis in patients with intravenous catheters but not receiving peripheral hyperalimentation. In the second phase, we evaluated the efficacy of in-line filters for preventing phlebitis by conducting a double-blind, randomized, prospective trial. In the third phase, we eliminated filters and replaced our standard 10% glucose-2.75% amino acid solution with a 3% glycerol-3% amino acid solution to determine the effect of different peripheral hyperalimentation solutions on phlebitis rates. MATERIAL AND METHODS The nutritional support team first screened and approved all requests for hyperalimentation (including peripheral hyperalimentation) before the pharmacy dispensed the solution. Once hyperalimentation was begun, the nutritional support team nurse inspected the intravenous infusion sites daily. Criteria for the presence of phlebitis consisted of three or more of the following findings: erythema, tenderness, induration, or palpable venous cord greater than 2.5 cm. Bacterial phlebitis was diagnosed if the catheter tip grew microorganisms in culture. We monitored patients for fever and chills on a daily basis and did a complete blood count at least once a week. We examined the records of all patients receiving peripheral hyperalimentation over the 30 months of the study and determined the incidence of phlebitis. During this time, we followed our previous policy of using in-line filters for peripheral hyperalirnentation.
VOLUME 159 FEBRUARY 1990
PHLEBITIS WITH PERIPHERAL HYPERALIMENTATION
We did a point-prevalence study to determine the alimentation solution with 3% glycerol-3% amino acid phlebitis rate in hospitalized inpatients receiving nonhy- solution with electrolytes, 735 mOsm/L (Procalamine, peralimentation intravenous therapy. Ward nurses con- KendaU-McGaw, Irvine, CA). We compared a consecuducting the survey were instructed regarding the specific tive series of 41 patient s with the 41 patients in the rancriteria for diagnosing phlebitis as described above. On a domized trial. Other than replacing the solution and elimsingle day, all inpatients were surveyed and the preva- inating in-line filters, there were no changes in the protocol. lence of phlebitis was determined. In the second phase of the study, we randomized 50 consecutive patients approved for peripheral hyperali- RESULTS Retrospective review and point-prevalence study mentation into either standard filter or sham filter groups based upon a computer-generated list of random num- (Phase 1 ): We began carefully monitoring for phlebitis bers. The Healthtek Corporation (Nevada City, CA) 30 months before the randomized trial. During that perimanufactured, coded, and packaged the filter devices, od, 88 patients received peripheral hyperalimentation. and they appeared identical. The standard device was a All patients receiving hyperalimentation had in-line fil0.22-/~ millipore filter, and the sham device had a 50-/~ ters during this part of the protocol. Clinical phlebitis pore size with no significant filtering capabilities. At the occurred in 57 patients, or 65% of the entire group. The point-prevalence study identified patients receivend of the study, we broke the code and determined the differences in the incidence and cause of phlebitis using ing nonhyperalimentation intravenous therapy on the day chi-square statistical comparisons. Comparisons between of the survey. By the criteria listed above, phlebitis was group characteristics were by chi-square and Student's t present in 18% of 456 patients on active medical and test. The Human Investigation Committee approved the surgical wards. Randomized trial of standard versus sham filters protocol without the need for written informed consent. The Research and Development Committee approved (Phase 2): Fifty patients receiving peripheral hyperalimentation were randomized for standard or sham filters, the double-blind, randomized, prospective design. but 9 were excluded for major protocol violations The The cannula insertion technique was standardized. The site was prepared by brisk scrubbing with povidone- reasons for exclusion were nurses using an unapprovexl, iodine solution and allowed to dry. A 20-gauge plastic non-study filter from old ward stock (seven patients) and cannula was used. One venipuncture attempt per cannula not having the site evaluated by the nutritional support was allowed with a new cannula used for each attempt to team nurse before ending the infusion (two patients). The enter a vein. Once inserted, the cannula was secured with data below represents our findings in the 41 evaluable silk tape, Neosporin ointment was applied to the site, and patients. The standard filter group consisted of 20 patients and a gauze dressing was used to cover it. All tubing, filters, and dress!ngs were changed every 48 hours unless re- the sham filter group of 2i patients. All subjects were quired sooner. No additional solutions or medications men with a wide variety of diagnoses. Surgical service were added to the intravenous lines. Cannulation, care of patients represented 68% of the total and medical service the site, and tubing/filter changes were the responsibility patients, 32%. Six medical patients and four surgical paof ward nurses according to established procedure and tients had more than one episode of phlebitis. There were protocol. Before the cannula was removed, the site was no significant differences between the sham filter group prepared with povidone-iodine solution, and any excess and the standard filter group with respect to age, total was removed with a sterile 2 • 2 gauze sponge. A random treatment days, average treatment days, or distribution sampling of cannula tips from patients with clinical signs between medical and surgical services (Figure 1). None of phlebitis was sent to the microbiology laboratory for of the patients in the study showed systemic signs of culture and sensitivity. Cultures of the catheter tips were sepsis. The phlebitis rate in study patients was 68%, not done according to the poured-plate technique described significantly different from the 65% rate found in the retrospective review. Phlebitis occurred in 74% of patients by Snydman et al [16]. The peripheral hyperalimentation solution used in this phase of the study was 10% glucose- in the standard-filter group and 64% of patients in the 2.75% amino acid and 10% soybean emulsion. Electro- sham-filter group (Figure 2). The differences were not lytes, multivitamins, and trace elements were added per significant. We sent a random sampling of 25 cannulas from the orders of the attending physicians. The osmolarity of the glucose-amino acid solution without electrolytes was patients with clinical phlebitis for culture and sensitivity. 792 mOsm/L and with electrolytes was 930 mOsm/L. Only one culture was positive. The organism was identiMost of the time (>95%), the solution was orderedwith fied as Staphylococcus epidermidis. Random bacterial electrolytes. The peripheral hyperalimentation solution cultures of the solutions done by the pharmacy service was always infused concurrently with 10% soybean emul- were consistently negative. Comparison of peripheral hyperalimentation sosion (300 mOsm/L) via a Y-connection to buffer the osmolarity of the glucose-amino acid solution. Random lutions (Phase 3) : After the randomized trial, the use of samples of prepared solutions were routinely cultured by in-line filters was abandoned, and 3% glycerol-3% amino acid injection (Procalamine) and 10% fat emulsion were pharmacy personnel for infection control purposes. Upon completing the randomized trial, we eliminated used as the peripheral hyperalimentation solution in a in-line filters and replaced our standard peripheral hyper- consecutive series of 41 patients. We compared the phleTHE AMERICANJOURNALOF SURGERY VOLUME159 FEBRUARY1990
223
RYPINS ET AL
160
144
1401 120 100 8O
65 71
60 4O 21 2 0 2O 0
I NUMBER PATIENTS
i
AVERAGE AGE
AVERAGE %SURGICAL %MEDICAL DAYS PATIENTS PATIENTS
TOTAL DAYS
S H A M FILTER
m
STANDARD FILTER
Figure 1. Comparison of' sham versus standard filter. The two groups are compared for differences in number of patients, average ages, total days on hyperalimentation, average number of days on hyperalfmentation per patient, and the percentages of patients on surgical and medical services. There were no SignifiCant differences between the groups on any of the characteristics examined.
INCIDENCE (%)
68
STANDARD FILTER
SHAM FILTER
GLUCOSE pHAL
GLYC=:RmE pHAL
ly
Figure 2. Comparison of phlebitis rates between groups: The two bars on the left represent the results of the randomized trial. The standard-filter group had a higher rate of phlebitis than the shamfilter group, although the difference was not statistically significant. The three bars on the right represent differences between groups Of patients based on the type of solution being infused. The bar labeled glucose pHAL (peripheral hyperalimentation) represents the phlebitis rate of the 41 patients in the randomized trial (phase 2)receiving the standard 10% glucose-2.75% amino acid solution. The bar labeled glycerine pHAL represents the phlebitis rate of 41 consecutive patients receiving 3% glycerol3% amino acid solution. The bar on the far right labelled IV represent's the preva[enca of phlebitis in456 patients receiving nonhyperalimentation intravenous therapy.
bids rate in these patients with that of the 41 patients in the randomized group that received the glucose-based solution. The phlebitis rate decreased to 27% in the 41 patients receiving the glycerol-based solution compared with 68% in the 41 patients receiving the glucose-based solution (Figure 2), The difference between the two groups was highly significant (p <0.001). The rate of phlebitis in patients receiving glycerol-based peripheral hyperalimentation was not significantly different from that in patients receiving nonhyperalimentation intravenous therapy (p = 0.23). 224
COMMENTS There is little discussion in the literature regarding phlebitis occurring in patients receiving peripheral hyperalimentation, leading some physicians to assume that it is not a significant problem, However, we were distressed by the magnitude of this problem in our patients, leading to this study. A retrospective review indicated that the rate of phlebitis in patients receiving peripheral hyperalimentation was 65%. This compared most unfavorably wit h the 18% prevalence of phlebitis in patients receiving nonhyperalimentation intravenous therapy on the medical and surgical services. Therefore, we investigated the reasons for the increased phlebitis rates in patients receiving Peripheral hyperalimentation. One factor implicated in causing phlebitis is particulate matter in the solution [4,9-12]. Our results did not support this. If particulate matter were responsible, we would have expected to find a higher phlebitis rate in the sham-filter group than the standard-filter group. However, no significant differences were found (Figure 1). :Some investigators have implicated bacterial contamination as the main cause of phlebitis in hospitalized patients requiring intravenous therapy [2,13-19]. We could not confirm this either. In a random sampling of cannulas from patients with clinical phlebitis, only one positive culture (S. epidermidis) was obtained. Moreover, a standard 0.22-~t filter should be effective in removing bacteria from intravenous infusions. We found that the 10% glucose-2.75% amino acid solution resulted in phlebitis rates of 68% regardless of whether a standard or sham filter was used. In the third phase of our study, we eliminated in-line filters and used 3% glycerol-3% amino acid injection with electrolytes. The phlebitis rate decreased to 27%, comparable with that found in nonhyperalimentation patients receiving intravenous therapy in our hospital. There are some possible explanations for our results. One is that filters are responsible for causing phlebitis and that eliminating them decreased phlebitis rates. Another explanation relates to the osmolarity of the solution. The osmolarity of our 10% glucose-2.75% amino acid solution without electrolytes was 792 mOsm/L, similar to the osmolarity of the 3% glycer01-3% amino acid solution with electrolytes (735 mOsm/L). However, most of the time, physicians added electrolytes to the glucosecontaining solution, increasing the 0smolarity to an average of 930 mOsm/L. We stress that the above explanation is speculative because we did not specifically design this study to test osmolarity as a cause of phlebitis, and we did not have a control group without filters in the randomized part of the study. Our data suggest that the solution itself may be a major etiologic factor in phlebitis occurring in patients receiving peripheral hyperaiimentation. Infection does not appear to be an important cause of phlebitis if careful sterile technique is applied to cannu!a insertion and maintenance of the intravenous system. Particulate material contained in antibiotics and other medications contributes to phlebitis in patients receiving nonhyperalimentation intravenous therapy [7]. In-line filters have been
THE AMERICAN JOURNAL OFSURGERY VOLUME159 FEBRUARY1990
PHLEBITIS WITH PERIPHERAL HYPERALIMENTATION
shown to prevent phlebitis in these settings [3,5,7], but our study in hyperalimentation patients showed that filters had no effect and perhaps even a deleterious effect. The results of our study should be interpreted and applied to other medical centers cautiously. O u r nutritional support team has tight control over all parenteral nutrition consults and maintains careful surveillance of all patients on peripheral hyperalimentation using a fulltime, fully trained hyperalimentation nurse. Moreover, this study was conducted under strict protocol conditions. However, we feel justified in concluding that routine inline filters for peripheral hyperalimentation are unnecessary provided that the line is not used for concurrent administration of medications.
Phlebitis is a frequent but underestimated complication of peripheral intravenous nutrition; these trials point out the primary role of osmolarity in this process. ACKNOWLEDGMENT The authors wish to acknowledge the cooperation of the Nursing Service and the Office of Quality Assurance for doing the pointprevalence study. REFERENCES 1. Adams S, Killien M, Larson E. In-line filtration and infusion phlebitis. Heart Lung 1986; 15: 134-40. 2. Consentino F. Personnel-induced infusion phlebitis. Bull Parent Drug Assn 1977; 31: 288-93. 3. DeMonaco H J, Cronin CM, Dempsey M J, Tulley S. Comparison of phlebitis rates with buretrol and Harvard mini-infusor for administration of IV piggyback medications. NITA 1986; 9: 475-9.
4. Ervin SM. The association of potassium chloride and particulate matter with the development of phlebitis. NITA 1987; 10: 145-9. 5. Hershey CO, Tomford JW, McLaren CE, Porter DK, Cohen DI. The natural history of intravenous catheter associated phlebitis. Arch Intern Med 1984; 144: 1373-5. 6. Hessov I. Prevention of infusion thrombophlebitis. Acta Anesthesiol Scand [Suppl] 1985; 29: 33-7. 7. Falchuk KH, Peterson L, McNeil BJ. Microparticulate-induced phlebitis: its prevention by in-line filtration. N Engl J Med 1985; 312: 78-82. 8. Crossley K, Matsen JM. The scalp vein needle: a prospective study of complications. JAMA 1972; 220: 985-7. 9. Maki D. Preventing infection in intravenous therapy. Hosp Pract 1976; 11: 95-104. 10. Harrigan CA. Care and cost-justification of final filtration. NITA 1985; 8: 426-30. 11. Rapp RP, Bivins B. Cost effectiveness of in-line filters in the era of DRG's. Am J Hosp Pharm 1984; 41: 2634-41. 12. Kaye W. Catheter and infusion-related sepsis: The nature of the problem and its prevention. Heart Lung 1982; 11: 221-8. 13. Murphy L, Lipman T. Preventing infection from total parenteral nutrition. Drug Ther Bull 1981; 1I: 51-8. 14. Turnidge J. Hazards of peripheral intravenous lines. Med J Aust 1984; 14: 37-40. 15. Makarewicz PA, Freeman JB, Fairfull-Smith R. Prevention of superficial phlebitis during peripheral parenteral nutrition. Am J Surg 1986; 151: 126-9. 16. Snydman DR, Murray SA, Kornfield SJ, Majka JA, Ellis CA. Total parenteral nutrition-related infections. Am J Med 1982; 73: 695-9. 17. Sanders RA, Sheldon GF. Septic complications of total parenteral nutrition, a five year experience. Am J Surg 1976; 132: 21420. 18. Stillman RM, Soliman F, Garcia L, Sawyer PN. Etiology of catheter-associated sepsis. Arch Surg 1977; 112: 1497-9. 19. Simmons B. Alternatives to IV filter usage. Infect Control 1985; 6: 342-4.
THE AMERICAN JOURNAL OF SURGERY VOLUME 159 FEBRUARY 1990 225