Methods

Top Methods Results Discussion Author & Article Info References

CDC Surveillance Project

The surveillance project has been described previously [6, 7]. In brief, workers in a group of participating health care institutions throughout the United States are voluntarily enrolled by cooperating investigators after an exposure to blood from a patient with documented HIV infection as a result of percutaneous injury (for example, a needlestick or a cut from a sharp object), contamination of mucous membranes, or contamination of nonintact skin. (Extensive or prolonged blood contact with intact skin may constitute an occupational exposure [4], but workers with intact skin exposures are not enrolled in this project.) Epidemiologic information and a blood specimen for HIV serologic testing are collected by the cooperating investigator at a baseline visit within 1 month after exposure and at follow-up visits 6 weeks, 3 months, 6 months, and 1 year after exposure. At enrollment, workers complete and mail directly to the CDC an anonymous questionnaire that includes information on nonoccupational risk factors for HIV infection. Workers who seroconvert are interviewed in person by an investigator using a standard CDC protocol for the interview of persons with no identified risk.

Human immunodeficiency virus serologic testing is done either at the CDC or at the cooperating institution's laboratory, with all positive results confirmed at the CDC.

Use of Postexposure Zidovudine

In October 1988, the surveillance project was expanded to collect additional information on the postexposure use and toxicity of antiviral agents, such as zidovudine. A standardized protocol for offering or using postexposure zidovudine was not provided by the CDC; in accordance with the Public Health Service statement [4], individual cooperating investigators and exposed workers determined whether postexposure zidovudine would be used, and, among those using zidovudine, the dosage and duration of prophylaxis.

Adverse Events

At the 6-week follow-up visit, the cooperating investigators were asked to record reported symptoms on a standardized checklist provided by the CDC, as well as the zidovudine regimen, if used. In this surveillance project, a standard protocol for laboratory testing for adverse effects (for example, anemia) of zidovudine was not provided by the CDC; such tests were done at the discretion of the cooperating investigator, who recorded the test date and any results outside the testing laboratory's normal range.

Statistical Analysis

Data were analyzed using PRODAS (Conceptual Software, Houston, Texas), Epi-Info (CDC, Atlanta, Georgia), and the Statistical Analysis System (SAS Institute, Inc., Cary, North Carolina). The upper bounds of the 95% CI were calculated using the binomial distribution. Because of missing data, totals for specific analyses may not equal the total number of enrolled workers.

Results

Top Methods Results Discussion Author & Article Info References

From August 1983 through June 1992, 1245 workers from 312 institutions were enrolled and tested for HIV antibody at baseline and at least 180 days after exposure. These workers comprised nurses (64%), physicians and dentists (12%), phlebotomists (8%), laboratory workers (6%), medical students (2%), housekeepers (1%), and other workers (such as technicians and physician and nursing assistants) (8%).

Exposures to HIV-infected Blood and Risk for HIV Infection

A total of 1245 workers were enrolled and tested for HIV antibody at baseline and at least 180 days after exposure. These workers had been exposed to blood from source patients who had AIDS as defined by the CDC surveillance case definition in use at the time of enrollment, (1012 [81%] exposures) [8], who were HIV-antibody positive and symptomatic but who did not have AIDS (68 [5%] exposures), or were HIV-antibody positive and asymptomatic (165 [13%] exposures). Exposure types were percutaneous injury (1103; 89%) and blood contact with mucous membranes (67; 5%) or nonintact skin (75; 6%). The 1103 percutaneous injuries were caused by syringe needles (375; 34%), intravenous needles (343; 31%), suture needles (41; 4%), and other needles (228; 21%); scalpels (41; 4%); lancets (27; 2%); and other sharp objects (48; 4%).

Among 1103 workers with percutaneous injuries who were HIV seronegative at baseline, 4 (0.36%; upper limit CI, 0.83%) seroconverted to HIV. Among workers with percutaneous injury, the seroconversion rate after exposure to blood from source patients with AIDS was 0.44% (4 of 899; upper limit CI, 1.01%); the seroconversion rate was not statistically different after exposure to blood from source patients who were HIV-antibody positive and symptomatic but who were without AIDS (0 of 57, P > 0.2) or source patients who were HIV-antibody positive and asymptomatic (0 of 147, P > 0.2). No seroconversions occurred among workers with blood exposure of mucous membranes (67 workers; upper limit CI, 4.31%) or skin (75 workers; upper limit CI, 3.87%).

An additional 497 workers enrolled in the early 1980s, before HIV-antibody testing was available, were tested for HIV antibody at least 180 days after exposure but not at baseline. One of these workers was HIV-antibody positive when first tested 10 months after exposure [7]. At that time, a sex partner of this worker was also tested and found to be HIV seropositive; the dates of HIV infection for this worker and partner are unknown. Including these 497 workers, the HIV seroprevalence rate was 0.32% (5 of 1532; upper limit CI, 0.68%) after percutaneous exposure, 0% (0 of 100; upper limit CI, 2.9%) after mucous membrane exposure, and 0% (0 of 110; upper limit CI, 2.7%) after skin exposure to HIV-infected blood.

Of the four workers enrolled in this project who have seroconverted after exposure to HIV, three have been previously described [6, 7]. The fourth worker, a female laboratory worker, was injured in 1992 by a 21-gauge syringe needle after doing a phlebotomy on a patient who was known to have AIDS. The worker reported no behavioral risk factors for HIV infection, except for having had sexual contact with a person at risk for HIV infection more than 1 year before her injury. The worker began taking zidovudine 2 hours after the injury, in a dose of 100 mg, five times a day, for the first day and then 200 mg, five times per day, for 16 additional days (Table 1, report 5). The worker discontinued use of zidovudine after 17 days because of nausea, fatigue, and myalgia. Thirty-eight days after exposure, the worker developed fever, malaise, fatigue, nausea, arthralgia, myalgia, and rash. Serologic test results for HIV were negative for specimens of the worker's blood collected on the day of the injury and 6 weeks afterward, were indeterminate by enzyme immunoassay and positive by Western blot 3 months after injury, and were positive by both enzyme immunoassay and Western blot 4 months after injury. The source patient was not known to have received zidovudine before the needlestick incident.

Use of Postexposure Zidovudine

From October 1988 to June 1992, the period when use of zidovudine was studied, 848 workers were enrolled. Postexposure zidovudine was used by 265 (31%) of these workers. Of 200 cooperating investigators who enrolled workers during this period, 110 (55%) reported that at least one worker used zidovudine. Zidovudine was prescribed in doses ranging from 200 to 1800 mg/d (median, 1000 mg/d) and for periods of 1 to 180 days (median, 42 days). The interval from exposure to first dose of zidovudine ranged from less than 5 minutes to 17 days (median, 4 hours). No clinically significant changes in prescribed regimens of zidovudine were seen during the study period (data not shown).

The proportion of enrolled workers using zidovudine increased from 5% in the fourth quarter of 1988 to 50% in the third quarter of 1990 and has been stable subsequently, averaging 43% during January to June 1992 (Figure 1). Physicians, dentists, and medical students were more likely to use zidovudine than other health professionals (59 of 135 [44%] compared with 206 of 713 [29%], P = 0.001), as were workers with percutaneous injury compared with contamination of mucous membranes or skin (244 of 750 [33%] compared with 21 of 98 [21%], P = 0.03).

View larger version (36K): [in this window] [in a new window]   Figure 1. Proportion of enrolled health care workers using zidovudine by quarter year. Public Health Service Statement refers to reference 4. P < 0.001 (chi-square for trend). *Number of enrolled workers.

Among 219 workers for whom information was available, the source of payment for zidovudine was the employer (69%), workers compensation (17%), private insurance (8%), the workers themselves (2%), and other or multiple sources (9%). Sources of payment for office visits and laboratory studies were similar. Median charges, excluding the cost of the time spent by hospital employee health care and infection control personnel, were reported for some workers who took zidovudine for at least 21 days: $408 for zidovudine (n = 60 workers), $132 for office visits (n = 31 workers), and $132 for laboratory testing (n = 41 workers).

During October 1988 to June 1992, 444 workers with percutaneous injury were enrolled and tested for HIV antibody at baseline and at least 180 days after exposure; no seroconversions occurred among 301 workers not using zidovudine, and 1 (0.70%) seroconversion occurred among 143 workers using zidovudine.

Adverse Events

Three workers who planned to use zidovudine for only 1 day were excluded from the tabulation of adverse events. Information on symptoms that occurred during the 6 weeks after HIV exposure was available for 236 (89%) workers using zidovudine; 176 (75%) reported one or more symptoms, most commonly nausea, malaise or fatigue, or headache (Table 2). Symptoms were reported less frequently among workers who did not use zidovudine (26% of 439 workers, P< 0.01; Table 2). Cough or acute respiratory illness was the most common symptom among those not using zidovudine.

Limited information was available on hemoglobin or hematocrit values of workers using zidovudine. Of 175 workers who completed 21 or more days of prophylaxis, 51 (29%) had paired hemograms at least 21 days apart; among these 51 workers, 7 (14%) had a 10% or greater reduction in hemoglobin or hematocrit values.

One or more symptoms occurred in a similar proportion of workers taking less than 1000 mg per day (75% of 56 workers), 1000 mg per day (73% of 105), and more than 1000 mg per day (77% of 73). Seventy-four (31%) workers did not complete their planned regimen of zidovudine because of adverse symptoms (73 workers) or reduction in hemoglobin level (1 worker). Of these 74 workers, 43 (58%) stopped taking zidovudine, 14 (19%) used a lower dose, and 13 (18%) used a lower dose and later stopped taking zidovudine (information missing on 4 workers). For an additional 15 workers, zidovudine was discontinued or reduced in dosage for reasons not related to toxicity (for example, the workers changed their minds). Twenty-eight (12%) workers were absent from work for periods ranging from 1 to 49 days (median, 2 days) because of adverse events attributed to zidovudine.

Discussion

Top Methods Results Discussion Author & Article Info References

Based on data from this ongoing surveillance project, the risk for HIV infection after percutaneous exposure to HIV-infected blood is 0.36%, similar to earlier reports from this and other studies [2, 6, 13]. Although mucocutaneous blood contact has resulted in HIV transmission [14, 15], none of the workers enrolled in this project with mucous membrane or skin exposures tested positive for HIV. One worker enrolled in this project seroconverted despite taking zidovudine for 17 days after exposure; however, the number of enrolled workers is insufficient to assess zidovudine efficacy.

Failure of Postexposure Zidovudine Prophylaxis

Including the worker described in this report, the CDC is aware of eight reported instances of failure of postexposure zidovudine prophylaxis after percutaneous injuries to health care workers (see Table 1). In these cases, zidovudine was begun 30 minutes to 12 hours (median, 1.75 hours) after exposure and was used in doses of 800 to 1200 mg/d (median, 1000 mg/d) for 8 to 54 days (median, 21 days). The workers in reports 4 and 8 (see Table 1) stopped zidovudine because of evidence that they were seroconverting; for the other six workers, it is uncertain whether longer periods of prophylaxis would have been more effective. An additional five instances of failure of postexposure zidovudine have been reported after exposures in which the quantity of HIV-infected blood injected was larger than would be expected from a needlestick. These five include two instances of accidental intravenous inoculation of blood or other material from patients with HIV infection during nuclear medicine procedures [16, 17]. They also include one instance each of a blood transfusion [4], suicidal self-inoculation [18], and an assault on a prison guard with a needle syringe [19]. In most instances, information is not available on whether the source patient was taking zidovudine, which might have led to development of a zidovudine-resistant strain of HIV. In our case, the source patient was not known to have been taking zidovudine before the needlestick exposure, and no zidovudine-resistant HIV strains were detected by genetic sequencing or PCR techniques. These data suggest that the worker became infected with a zidovudine-sensitive strain of HIV, despite taking zidovudine after exposure. Collectively, these case reports indicate that if zidovudine is protective, any protection afforded is not absolute.

Efficacy of Postexposure Zidovudine Prophylaxis

Little additional information exists with which to assess the efficacy of postexposure zidovudine. Animal studies using murine and feline retroviruses are encouraging but of limited value because these viruses have pathogenic mechanisms different from HIV [4]. Zidovudine did not prevent simian immunodeficiency virus infection in monkeys injected with moderate or high inocula or with a rapidly lethal variant of simian immunodeficiency virus [20-22]. When lower challenge doses of simian immunodeficiency virus were used and zidovudine was begun before exposure, two of two monkeys in one study and one of six monkeys in the highest initial-dose treatment group in another study did not become infected [23, 24]. In a study in SCID-hu mice (immunodeficient mice transplanted with human hematolymphoid cells), HIV infection was suppressed during drug use, but infection was detected in all animals after discontinuation of zidovudine [25]. Because the mice were exposed to HIV using intrathymic injection of a sizable virus inoculum, the relevance of this experiment to needlestick exposure in health care workers is unclear. In a subsequent study [26] in SCID-hu mice, HIV infection suppression was most effective when zidovudine treatment was begun within 2 hours after exposure, but the investigators did not determine whether infection was present after zidovudine was discontinued.

Evaluation of the efficacy of postexposure zidovudine in humans is difficult because the HIV seroconversion rate after an occupational exposure necessitates a large sample size to show a prophylactic effect. For example, calculations based on a seroconversion rate of 0.36% after exposure, 75% drug efficacy, 80% power, and a 0.05% one-tailed significance level indicate that more than 2101 workers using postexposure zidovudine would be required to evaluate efficacy. The Burroughs-Wellcome Company (the manufacturer of zidovudine) sponsored a double-blind, placebo-controlled trial to evaluate postexposure zidovudine. The trial was terminated in June 1989, when it became apparent that an insufficient number of exposed workers could be recruited for a randomized trial [27]. Because of uncertainty about efficacy and safety, the Public Health Service concluded in January 1990 that a recommendation for or against the use of postexposure zidovudine could not be made [4].

In this surveillance project, the proportion of health care workers using zidovudine increased from 5% in the last quarter of 1988 to 40% to 50% after the January 1990 Public Health Service statement on postexposure zidovudine use, and was more common among physicians, dentists, and medical students than among other workers. The proportion of exposed workers who used postexposure zidovudine was 75% for the University of California San Francisco-San Francisco General Hospital (UCSF-SFGH) and 76% for the National Institutes of Health Clinical Center (NIH-CC) [28]. Rates of postexposure zidovudine use may differ among studies because of differences in the types of exposures studied, in the institutional policies about zidovudine availability, and in the knowledge as well as opinions of those who counsel exposed workers.

In this study, 75% of workers using zidovudine reported symptoms, and 31% did not complete their planned regimen because of adverse events. Although only limited data were available to assess the hematologic effects of postexposure zidovudine, no severe sequelae of postexposure zidovudine were reported. Similar results have been reported by other researchers. In an Italian multicenter study [5], adverse effects (most commonly nausea, asthenia, and vomiting) were reported in 53%, and postexposure zidovudine was discontinued in 25%. In studies at UCSF-SFGH and NIH-CC, "substantial side effects" were reported by 33% and 24% of workers, respectively; the drug was discontinued in 17% and 50% of workers, respectively [28]. In the only placebo-controlled trial (the Burroughs-Wellcome study), nausea, vomiting, and arthralgia were reported more commonly among workers taking zidovudine than among those taking placebo [27].

Limitations of this surveillance project include the possibility that workers or exposures voluntarily reported by workers and cooperating institutions may not be representative of all occupational exposures in the United States. Assessment of short-term adverse effects of zidovudine is complicated by the lack of a standardized postexposure regimen and a placebo control group; also, ascertainment of symptoms may have been more active among zidovudine users than nonusers. Assessment of long-term toxicity is not feasible because of the 1-year duration of follow-up. Detailed information on the nature and frequency of the toxicity of postexposure zidovudine will be determined optimally from carefully supervised clinical trials in selected institutions. The strength of surveillance data include descriptions of the current experience of various hospitals nationwide as well as enhanced ability to monitor trends and detect rare events.

Continued assessment is needed of the risk for infection after occupational exposures to HIV and of the risks and benefits of postexposure use of antiviral agents. Additional information about participation in the CDC surveillance project can be obtained from the Hospital Infections Program, National Center for Infectious Diseases, CDC, Mail Stop A-07, Atlanta, Georgia 30333; telephone, (404) 639-1547.

Appendix

The CDC Cooperative Needlestick Surveillance Group consists of cooperating investigators from 312 institutions nationwide. The following investigators enrolled more than 20 health care workers who participated in this study:

Kathleen S. Bean, RN, Fairfax Hospital, Falls Church, VA; Jessamyn Bound, CRNP, Temple University Hospital, Philadelphia, PA; Mary Braun, RN, Abbott Northwestern Hospital, Minneapolis, MN; Pat Brennan, RN, Medical Center at Princeton, Princeton, NJ; Anthony Burton, MD, Henry Ford Hospital, Detroit, MI; Michela T. Catalano, MD, Montefiore Hospital & Medical Center, Bronx, NY; Mary Cooney, MD, Overlook Hospital, Summit, NJ; Mary Fitzpatrick, RN, Episcopal Hospital, Philadelphia, PA; John Gaeuman, MD, Ohio State University Hospital, Columbus, OH; Michael Grieco, MD, St. Lukes-Roosevelt Hospital Center, New York, NY; Barbara Grignol, RN, University of Rochester Medical Center, Rochester, NY; Connie Grout, RN, Mercy Hospital Medical Center, Des Moines, IA; Robert Harrison, MD, University of California-San Francisco, San Francisco, CA; Vicki Heitzer, RN, South Miami Hospital, South Miami, FL; William L. Hoppes, MD, Timken Mercy Medical Center, Canton, OH; Lawrence Kaplan, MD, Robert Wood Johnson Medical Center, Camden, NJ; Lenelle Kwong, MD, MPH, Queen's Medical Center, Honolulu, HI; Justine Larado, RN, Atlantic City Medical Center, Atlantic City, NJ; Dolores LeFever, Holy Cross Hospital, Ft. Lauderdale, FL; Susan Lienesch, RN, Monmouth Medical Center, Long Branch, NJ; John R. Middleton, MD, Raritan Bay Medical Center, Perth Amboy, NJ; Dorothy Minucci, RN, Staten Island University Hospital, Staten Island, NY; Roberta Popp, RN, University of Utah Medical Center, Salt Lake City, UT; Karen Putnam, PA, Albany Medical Center Hospital, Albany, NY; Steve Sanford, MD, University Hospitals of Cleveland, Cleveland, OH; Dee Sluder, RN, Memorial Mission Hospital, Asheville, NC; Lou Smith, MD, Veterans Administration Medical Center, Albany, NY; Phyllis Spechko, RN, University of California-San Diego Medical Center, San Diego, CA; Rachel Stricof, MPH, New York State Health Department, Albany, NY; Mary Tarver, RN, Georgia Baptist Medical Center, Atlanta, GA; Georgia Thomas, MD, University of Texas M.D. Anderson Hospital, Houston, TX; Sally B. Turner, RN, MPH, Boston City Hospital, Boston, MA; Vickie Wescott, RN, Park Plaza Hospital, Houston, TX.