Outbreaks of Gram-Negative Bacterial Bloodstream Infections Traced to Probable Contamination of Hemodialysis Machines—Canada, 1995; United States, 1997; and Israel, 1997 FREE

MMWR. 1998;47:55-59

1 figure omitted

DURING 1996, approximately 236,000 persons received hemodialysis in the United States; of these, an estimated 183,000 (78%) received chronic hemodialysis.¹ Patients who receive chronic hemodialysis are at increased risk for bloodstream infections (BSIs) because of the need for repeated vascular access. Reported BSI rates for hemodialysis patients have ranged from 8.4 to 16.8 episodes per 100 patient-years,² and BSI has been identified as the cause of death in 6%-18% of hemodialysis patients.² Outbreaks of BSIs in hemodialysis units usually have been caused by inadequate disinfection of (1) water treatment or distribution systems^3,4 and (2) reprocessed dialyzers.^5-8 This report summarizes the investigations of three clusters of gram-negative bacterial BSIs at hemodialysis centers in Canada, the United States, and Israel. The findings indicate that all three outbreaks probably resulted from contamination of the waste drain ports in the same model of hemodialysis machine.

From June 17 through November 15, 1995, nine adult patients at an ambulatory hemodialysis center in Montreal, Canada, had Enterobacter cloacae BSIs. All patients at the hemodialysis center were dialyzed on COBE® Centrysystem 3* (CS3, GAMBRO® Healthcare™, Lakewood, Colorado) hemodialysis machines. Each CS3 had a Centry® Waste Handling Option (WHO™), which is a waste port designed to dispose of the saline used to flush a dialyzer before the machine is used for a patient. The WHO waste drain line employs two one-way valves to prevent drain line waste from refluxing into the WHO. The investigation indicated that at least one of the two one-way valves in the WHO waste drain lines of seven of 11 machines were incompetent,† potentially allowing drain backflow and contamination of dialysis lines in contact with the WHO port.

An epidemiologic investigation demonstrated that case-patients (i.e., the nine patients at the hemodialysis center who had Enterobacter cloacae BSIs) were more likely than control-patients to have received dialysis on a machine that had at least one incompetent valve on the WHO waste drain line (all seven case-dialysis sessions versus 145 [53%] of 272 control-dialysis sessions; odds ratio: undefined; p=0.02). Case- and control-patients were otherwise similar in demographic characteristics, underlying renal disease, type of vascular access, and dialyzer type.Enterobacter cloacae isolated from all nine infected patients and from the WHOs of 10 of 11 dialysis machines were identical when examined by pulsed field-gel electrophoresis (PFGE).

From December 5, 1996, through January 25, 1997, a total of 10 adult patients at an ambulatory hemodialysis center in Maryland had gram-negative bacterial BSIs. Six BSIs were caused by Enterobacter cloacae, four by Pseudomonas aeruginosa, and two by Escherichia coli; two were polymicrobial BSIs. All patients at the hemodialysis center were dialyzed on CS3 hemodialysis machines that had WHOs. Results of a cohort study of all patients receiving dialysis at the center during the 2-month epidemic period indicated that the risk for gram-negative BSI was associated with exposure to any of three particular dialysis machines (seven BSIs in 20 patients who were exposed to one or more of the three machines versus three BSIs in 64 patients who were exposed to the other machines; relative risk=7.5; 95% confidence interval=2.1-26.2). Incompetent valves on WHO waste drain lines were present in eight of 26 dialysis machines and in two of the three implicated machines. Enterobacter cloacae was recovered from the WHOs of 14 of 26 machines, and P. aeruginosa was recovered from seven of 26. PFGE patterns of available Enterobacter cloacae isolates from the dialysis machines and from three patients were identical; none of the P. aeruginosa isolates obtained from patients were available for PFGE testing.

From February 9 through September 19, 1997, eight adult patients at an ambulatory hemodialysis center in Jerusalem, Israel, had gram-negative bacterial BSIs. BSIs in four patients were caused by Escherichia coli, three by P. aeruginosa, two by Enterobacter cloacae, and one by Stenotrophomonas maltophilia; two patients had polymicrobial BSIs. All patients at the hemodialysis center were dialyzed on CS3 hemodialysis machines that had WHOs. All eight patients who had BSIs had been dialyzed on three of 13 dialysis machines. Backflow was observed in the WHOs of the three implicated dialysis machines, and cultures obtained from the WHOs of six of 13 machines were positive for gram-negative organisms. Five of the eight patients, including all four with Escherichia coli BSIs, had been dialyzed on one machine that subsequently was culture-positive for Escherichia coli and P. aeruginosa. Both patients with Enterobacter cloacae BSIs had been dialyzed on a second machine that was culture-positive for Enterobacter cloacae and P. aeruginosa. Escherichia coli isolates obtained from three patients and the WHO of the implicated machine were identical by PFGE.

Daily quality-control testing of WHOs as specified by the manufacturer had not been performed at any of the three hemodialysis centers. The manufacturer specifies that preventive maintenance of the valves in the WHO waste drain line includes replacement of the two valves after every 2000 hours of use. However, personnel at the three hemodialysis centers were aware of the need to change only one valve in the WHO waste drain line, and personnel at two centers did not know a second WHO valve existed; schematic diagrams provided by the manufacturer to these two hemodialysis centers identified only one of the two valves. At one center, experimentally bending and twisting the main drain line of a machine that had incompetent valves in the WHO waste drain line demonstrated the ease with which backflow can occur in the WHO.

In one hemodialysis center, the outbreak was controlled after high-level WHO disinfection (i.e., disinfecting dialysis machines with formaldehyde on two occasions and increasing the dwell time for routine weekly machine disinfection). In the other two centers, the outbreaks were terminated by discontinuing use of the WHO. All three hemodialysis centers discontinued using the WHOs.

In June 1997, GAMBRO Healthcare sent a Medical Device Safety Alert letter to all hemodialysis centers of record that use the CS3. This letter informed users of the need to ensure proper functioning of the WHO and outlined procedures for proper disinfection and maintenance of the equipment.

C Frenette, MD, M Delorme, Hôpital Charles LeMoyne, Quebec; J Hockin, Health Canada, Ottawa, Ontario, Canada. FG Grillo, MD, T Killar, SJ Boyer, Maryland; DM Dwyer, MD, State Epidemiologist, Maryland Dept of Health & Mental Hygiene. C Block, MBBCh, R Backenroth, MD, M Shapiro, MD, Hadassah Univ Hospital, Jerusalem; B Lev, MD, Associate Director General, Israel Ministry of Health. Hospital Infections Program, National Center for Infectious Diseases; and EIS Officers, CDC.

Bacterial BSI is a potentially severe complication associated with hemodialysis vascular access. In the United States, complications associated with vascular access represent one of the most common sources of morbidity among patients undergoing end-stage renal dialysis, with associated costs exceeding an estimated $1 billion per year.⁹ This report links three outbreaks of gram-negative bacterial BSIs to a unique design feature of the CS3 hemodialysis machine. The results of these outbreak investigations demonstrated that the WHO, if not properly maintained and disinfected, may be a source of bacterial contamination leading to BSIs in hemodialysis patients. Because waste backflow can occur with incompetent valves and WHO contamination can occur easily, the design of the WHO creates a mechanism for possible cross-contamination of the patient dialysis line.

In addition to the problems associated with the WHO feature, insufficient training of hemodialysis personnel about the design and proper handling and maintenance of WHOs might contribute to transmission of BSIs to hemodialysis patients. In June 1996, GAMBRO Healthcare and CDC surveyed 595 U.S. dialysis centers that use CS3 machines to characterize the methods used to clean and disinfect the dialysis machines and to characterize quality-control procedures (GAMBRO Healthcare and CDC, unpublished data). The survey indicated that personnel at most (87%) of the responding dialysis centers reported weekly disinfection of their dialysis machines as specified by COBE guidelines, although most (62%) were not disinfecting dialysate and bicarbonate sampling ports as often as recommended. Of the 290 centers that reported using the WHO, only 42 (14%) performed the recommended daily quality-control assessment of the WHO valves to determine whether drain reflux was occurring. Of the 137 centers responding to the question "If fluid can be aspirated from the WHO, what is done?," 112 (82%) indicated the need for replacing WHO valves or taking the machine off-line for servicing.

This report underscores the importance of surveillance and infection control in the ambulatory health-care setting. The detection of these outbreaks and identification of the likely cause was aided by the brief time-frame during which multiple infections were identified. The limited availability of data about infection rates in ambulatory dialysis centers impedes the identification of small or prolonged low-level outbreaks. Because of the lack of such data, inappropriate infection-control or maintenance practices that were identified in the GAMBRO Healthcare/CDC survey could not be linked to adverse patient outcomes at the dialysis centers surveyed. Outbreaks of gram-negative bacterial BSIs in hemodialysis patients that appear to be associated with use of the WHO should be reported to state health departments and to CDC's Hospital Infections Program, National Center for Infectious Diseases; telephone (404) 639-6413.

References 9 available.

*Use of trade names and commercial sources is for identification only and does not imply endorsement by CDC or the U.S. Department of Health and Human Services.

†The manufacturer recommends daily testing of the competency of WHO valves by filling a 30 cc syringe with water, injecting the contents into the WHO drain port, and attempting to draw back fluid from the WHO. Competent valves should prevent backflow.

MMWR. 1998;47:27-30

AGRICULTURE HAS one of the highest occupational fatality rates of all U.S. industries. Since the mid-1970s, traditional small-bale balers have gradually been replaced by large-bale balers in the agriculture industry. Expanded use of these balers has resulted in worker exposure to new hazards not present during handling of traditional small bales; the larger size of the bales increases the potential for serious injury or death while workers handle them. During 1994-1996, seven persons in Minnesota died in separate incidents that involved large round hay bales (i.e., cylindrical bales approximately 5 feet in length with flat ends, diameters of approximately 6 feet, and weights ranging from 750 to 1500 lbs). The Minnesota Fatality Assessment and Control Evaluation program (MN FACE), a program sponsored by CDC's National Institute for Occupational Safety and Health (NIOSH),* was notified of these incidents by the Minnesota Extension Service, a newspaper clipping service, and/or by death-certificate review. This report describes three incidents that were reported to MN FACE during 1994-1996, summarizes national surveillance for bale-associated deaths during 1980-1995, and provides recommendations to prevent fatalities associated with large bales.

On January 23, 1994, a 38-year-old male farmer died from injuries sustained when a large round hay bale fell on him while the bale was being loaded onto a flatbed trailer. The worker was using a tractor and front-end loader to load hay bales onto the trailer. The loader bucket had been modified to lift hay bales by attaching two removable tines, which cradled the bale as it was lifted; the bucket was not equipped with a bale clamp or other bale-handling device specifically designed to secure the bale during lifting. The farmer was loading the second layer of bales, which required raising the bucket to near its maximum height above the tractor. The unsecured bale tumbled down the loader lift arms and struck the farmer while he was seated in the operator's seat of the tractor. He died 30 days later from severe head injuries.

During the evening of November 26, 1996, a 59-year-old male part-time farmer died of injuries sustained when the tractor he was driving overturned. As reconstructed by investigators, he was using a tractor and loader to move a large round hay bale into a cattle lot. The incident occurred after dark, and the farmer may have raised the loader and bale above the tractor hood so that the bale would not interfere with illumination provided by the tractor headlights. As the tractor was driven through the lot, it overturned to the right and came to rest upside down. The farmer was pinned to the ground beneath the loader and the left rear wheel of the tractor. The following morning, a passing motorist discovered the overturned tractor and called emergency medical personnel. The farmer was pronounced dead at the scene. The tractor was not equipped with a rollover protective structure (ROPS) and a seat belt.

On November 30, 1996, a 52-year-old man died from injuries sustained at his farm when he was crushed by a large hay bale. The bale fell from a parked trailer that was being loaded to transport bales that had been sold for cattle feed. The man was crushed by the unsecured bale and died at the scene.†

Since 1992, the Minnesota Department of Health has compiled surveillance and field investigation data about selected work-related agricultural fatalities through the FACE program. FACE collects epidemiologic data about occupational fatalities from multiple sources (including local law enforcement reports, on-site fatality investigations, and Minnesota Occupational Safety and Health Administration reports) and develops and disseminates safety recommendations to address identified risks and reduce the potential for the occurrence of similar incidents.

During 1994-1996, all seven persons in Minnesota who died in incidents involving large round hay bales were men; their ages ranged from 38 to 70 years (mean: 55 years). All of the incidents occurred on family-owned farms. Four incidents occurred when tractors being used to transport large bales overturned; two incidents occurred when a hay bale fell off the tractor loader and onto the tractor operator; and one incident occurred when a hay bale fell from a trailer that was being loaded to transport hay bales. The weights of the bales involved in these incidents ranged from 750 to 1500 lbs.

During 1980-1991, NIOSH's National Traumatic Occupational Fatalities (NTOF)‡ surveillance system identified 41 work-related fatalities resulting from hay bale-associated injuries in the United States. The Census of Fatal Occupational Injuries (CFOI)§ identified an additional 46 such cases during 1992-1995. Of the 87 persons who died, 86 were male; 37 (38%) were aged ≥65 years; and 72 (74%) were employed in the agriculture/forestry/fishing industries. Forty-two (43%) deaths occurred in the Midwest; 23 (24%), in the West; 20 (21%), in the South; and two (2%), in the Northeast.∥

Of the 46 deaths identified through CFOI, 20 (44%) occurred when a hay bale fell from a piece of equipment and struck a worker. Ten (22%) other deaths involved tractor rollovers. In some rollovers, the bale fell from the tractor, and the rollover occurred as the tractor struck the bale on the ground; in others, the narrative stated only that the tractor overturned as a hay bale was being transported. In eight (17%) incidents, the bale fell on a worker in a storage area or fell from a transport vehicle. Eight (17%) case narratives indicated only that the worker was struck by a falling hay bale. Narratives of cases identified by NTOF and CFOI contained varying levels of information; although some narratives specified shape and weight of the bale, others only stated that a hay bale was involved.

GL Wahl, MS, M Brown, MPH, DL Parker, MD, Minnesota Dept of Health. Div of Safety Research, National Institute for Occupational Safety and Health, CDC.

The findings in this report indicate that fatalities associated with large bales are a continuing source of preventable work-related deaths among workers in the agricultural industry. Although the cases in Minnesota involved large round bales, large square bales pose similar risks to agricultural workers.

In general, bales can be transported more safely by tractors equipped with rear attachments rather than front-end loaders. The likelihood of tractors rolling over sideways or tipping over backwards is reduced because bales are carried in a lower position than when hauled with front-end loaders. In addition, the rear tractor tires can accommodate the extra weight more effectively.¹ Bales transported at the rear of a tractor do not block the operator's forward vision and generally do not interfere with rearward vision.² When large bales cannot be transported by means of a rear attachment, front-end loader attachments specifically designed for transporting large bales should be used to prevent crush injuries. The potential for an unsecured bale to roll down the lift arms of a front-end loader and onto the tractor operator increases when the loader is raised.³ Loader attachments that securely hold bales include bale forks that have a tri-spear design, bale grapples with support arms that wrap around bales, and bale huggers that secure bales by squeezing them between two arms.

Preventing death and serious injury to tractor operators during tractor rollovers requires the use of a ROPS and a seat belt.¶ A ROPS may be either a roll-bar frame or an enclosed roll-protective cab and is designed to withstand the dynamic forces during a rollover; seat-belt use is necessary to ensure that the operator remains within the "zone of protection" provided by the ROPS.

The risk for a rollover can increase when a tractor is equipped with a front-end loader because a loader changes the tractor's center of gravity. The center of gravity rises as the loader is raised and as the weight of transported items increases. Raising the center of gravity increases the potential of a side rollover, especially if the tractor is driven across inclined terrain. When front-end loaders are used to transport large bales, appropriate counter weights should be added to the rear of the tractor. Counter weights increase tractor stability by counterbalancing items being transported and ensure that the rear tractor wheels remain in contact with the ground. Conversely, when bales are transported with rear attachments, appropriate counter weights should be added to the front of the tractor. Front-end counter weights enable the operator to maintain steering control of the tractor by ensuring that the front wheels remain in contact with the ground.

To reduce the risk for injuries and fatalities associated with transporting large bales, the following safety precautions are recommended:

References 4 available.

*Minnesota is one of 16 states (Alaska, California, Iowa, Kentucky, Maryland, Massachusetts, Minnesota, Missouri, Nebraska, New Jersey, Ohio, Oklahoma, Texas, Washington, West Virginia, and Wisconsin) that receives funding from NIOSH for state FACE programs.

†Although this farm work-related incident was reported to the MN FACE program, a detailed FACE report was not completed because local authorities and immediate family members declined to participate in a FACE investigation. General details concerning the incident were obtained from public information published in local news reports of the incident.

‡NTOF is based on death certificates for the 50 states, the District of Columbia, and New York City for persons aged ≥16 years for whom there was a work-related injury that was the cause of death.

§CFOI is a multiple-source reporting system for occupational fatalities implemented nationwide by the Bureau of Labor Statistics in 1992.

∥Northeast=Connecticut, Maine, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, and Vermont; Midwest=Illinois, Indiana, Iowa, Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin; South=Alabama, Arkansas, Delaware, District of Columbia, Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia; and West=Alaska, Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming.

¶Occupational Safety and Health Administration (OSHA) regulations⁴ require that all tractors built after October 25, 1976, and used by employees of a farm owner must be equipped with a ROPS and a seat belt. This standard is not actively enforced on farms with <11 employees, and family farms without other employees usually are exempt from enforcement of OSHA regulations.

MMWR. 1998;47:5-7

1 table omitted

SOME PUBLIC health policy goals in the United States have been expressed as increases in the number of years of healthy life (YHL) (i.e., quality-adjusted life years), a measure of health that combines the effects of mortality with information about morbidity and disability.¹ Data from national health surveys, in combination with life-table death rates and other information, have been used to calculate national estimates of the expected number of YHL at a given age.^2,3 This report summarizes an analysis of data from the Behavioral Risk Factor Surveillance System (BRFSS) using these methods to estimate YHL for state populations during 1993-1995. The findings indicate substantial variability among the participating states.

The BRFSS is a continuous, state-based, random-digit-dialed telephone survey of the U.S. adult, noninstitutionalized population that measures the prevalence of health-risk behaviors and preventive health-care practices in the population.^4-6 During 1993-1995, a total of 16 states* participating in BRFSS gathered the additional information required to estimate the expected YHL. Expected YHL was estimated using BRFSS interview data about limitations in activities of daily living and self-rated overall health status (categorized as excellent, very good, good, fair, or poor); preliminary, unofficial lifetable estimates for states for 1993; and national data about the institutionalized population. BRFSS estimates were weighted to provide representative estimates, and confidence intervals were computed using SUDAAN.

An index of health-related quality of life (HRQL) was computed.² YHL was calculated by first computing the index of HRQL for each respondent (HRQL ranged from 1.0 [for those in excellent health and with no limitations] to 0.1 [for those who were limited in self-care activities of daily living and who were in poor health]). Second, the HRQL index was combined with lifetable functions to compute age-group specific expected YHL. This computation was based on multiplying the age-group specific lifetable number of total person-years lived by the average HRQL (range: 0.1-1.0) within each age group. The number of healthy person-years lived was summed for each age group and divided by the number of persons at each age. These data were adjusted using data from previous national estimates of the relative size and HRQL of institutionalized persons.² Age-specific estimates of YHL represent the average number of YHL remaining to a person at a given age.

When averaged over all ages, state-specific estimates of HRQL ranged from 0.79 to 0.85. In most states, HRQL was higher for men (range: 0.79-0.87 across states) than for women (0.79-0.85).

The estimated YHL at age 25 years was 39-44 years and at age 65 years was 11-14 years. YHL was higher for women: at age 25 years, YHL was 41-47 years for women compared with 38-43 years for men; at age 65 years, YHL was 12-16 years for women and 10-13 years for men. State-specific HRQL varied directly with life expectancy. For example, expectation of life at age 25 years is correlated with average HRQL index (r=0.77, p <0.001).

J Cook, MBA, Alabama; B Bender, Arizona; M Leff, MSPH, Colorado; N Costello, MPA, Indiana; M Perry, Kansas; K Asher, Kentucky; R Meriwether, MD, Louisiana; H McGee, MPH, Michigan; T Murayi, PhD, Missouri; P Smith, Montana; S Huffman, Nebraska; K Zaso, MPH, New Hampshire; TA Melnik, DrPH, New York; J Grant-Worley, MS, Oregon; L Redman, Virginia; M Futa, MA, Wyoming. S Bland, MS, TRW, Inc, Atlanta. Behavioral Surveillance Br, Div of Adult and Community Health, National Center for Chronic Disease Prevention and Health Promotion; Mortality Statistics Br, Div of Vital Statistics, National Center for Health Statistics, CDC.

One of the national health objectives for 2000 is to increase the expectation of healthy life at age 65 years from 12 to 14 years (objective 17.1c).¹ The findings in this report indicate that, among the 16 states assessed, only one (Montana) had achieved that level during 1993-1995. State and local public health programs need local data to evaluate and guide their prevention efforts, especially because of jurisdiction-specific differences in health, demographic, and socioeconomic conditions. In addition, many states independently establish health objectives similar to the national objectives. The analysis described in this report demonstrates the feasibility of developing state-specific estimates for HRQL and expected YHL and indicates state-specific variations in these indicators.

The methods used in this analysis are subject to at least two limitations. First, in addition to survey data on health and disabilities, the methodology requires lifetable data specific for the populations covered. The use of age-specific rates requires that many computations be based on small numbers of observation, thereby limiting the ability to calculate estimates for population subgroups. Potential alternative approaches would not depend on age-specific data (e.g., multivariate individual-level analysis of the determinants of HRQL or components). Second, the same nationally based correction for the institutionalized population was used for all states and subgroups; however, rates for institutionalization and health status of the institutionalized vary among states and subgroups. Despite these limitations, the state estimates are in the same range as the national estimates of YHL.² Consistency of estimates between years for those states that collected the data for >1 year (Kansas, Nebraska, and New York) and the association between HRQL and mortality levels also support the quality of the estimates.

References 6 available.

*Alabama, Arizona, Colorado, Indiana, Kansas, Kentucky, Louisiana, Michigan, Missouri, Montana, Nebraska, New Hampshire, New York, Oregon, Virginia, and Wyoming.