Health care workers (HCWs) are at risk for HIV and other infectious pathogens by means of blood and body fluid exposures, but the true burden of occupational exposures is unclear. According to estimates from the Centers for Disease Control and Prevention (CDC) and the Exposure Prevention Information Network (EPINet), approximately 384,325 percutaneous injuries occur annually among hospital-based HCWs in the United States.1 These national figures do not incorporate other types of blood or body fluid exposures, do not include nonhospital HCWs, and do not describe the incidence of exposures by health care occupation. Two statewide studies from Massachusetts and California offer a broad comparative analysis of exposure frequency by health care occupational groups, but they are also limited to percutaneous injuries.2,3 A surveillance study from Duke University Medical Center (Durham, NC) compares the incidence of percutaneous and other blood or body fluid exposures by occupational groups but is restricted to hospital-based HCWs from a single hospital network.4 A recent study by Chen and Jenkins5 attempts to estimate the frequency of blood or body fluid exposures evaluated at US emergency departments (EDs) using data from 1998 to 2000 from the National Electronic Injury Surveillance System (NEISS). Because of the complex sampling scheme used for these types of surveillance systems and the data collected for these types of studies, the estimates are limited to national-level data instead of state- or regional-level data and to selected occupations.
One of the major concerns about blood or body fluid exposures is the risk of acquiring an HIV infection. Since 1996, the CDC has recommended the use of HIV postexposure prophylaxis (PEP) to reduce the transmission of HIV after certain HCW exposures.6-9 According to CDC guidelines, HIV PEP use should be dictated by the characteristics of the exposure and by the HIV status of the exposure source. According to Cardo et al,10 the risk of acquiring an HIV infection is greater for percutaneous injuries that involve hollow-bore needles that have been in contact with an artery or vein, when blood is visible on the device, and when the source patient has AIDS. Studies of HIV PEP have shown that use is not always in accordance with recommendations, however.9,11 In real-world clinical situations, details of the exposure and the status of the source may be limited. Therefore, other influences might prompt clinicians to prescribe HIV PEP.
Many HCWs first seek medical attention for their blood or body fluid exposures from the ED. Examining HCW visits to EDs could reveal the frequency for which HCWs are evaluated for percutaneous and nonpercutaneous blood or body fluid exposures, indicate when the visits are more likely to occur, and demonstrate how often HCWs are offered HIV PEP in the ED. In addition, analyzing ED visits could help to determine how often non-hospital-based HCWs are evaluated for these exposures. Furthermore, the incidence of exposures by hospital- and non-hospital-based HCWs by occupational group can be estimated from ED data. An advantage of focusing on actual ED visits is that estimates for exposures and HIV PEP use can be based on patient visits after the exposure rather than on summary reports derived from follow-up visits to employee or occupational health clinics. Data from follow-up visits may underreport the frequency of visits because many patients may seek treatment only once for their injury or might seek follow-up at non-hospital-based locations.
We sought to determine the incidence rates (IRs) of ED visits for blood or body fluid exposures by health care occupation from an entire state. We chose to study the exposures from 1 year before the introduction of national guidelines on occupational HIV PEP (1995) until the second revision of these guidelines (June 2001). We also compared the frequency of blood or body fluid exposure ED visits sustained by HCWs according to their occupations. In addition, we looked for temporal trends in ED visits for exposures by year, month, day of the week, and hour of the day. We further examined how often HIV PEP was offered and how often HIV PEP was accepted when it was offered by various factors: significance and type of exposure, HIV status of the source, time elapsed after the exposure, HCW gender, health care occupational group, type of hospital (community vs. teaching hospital), and year of exposure. We investigated whether or not these clinical and nonclinical factors influenced HIV PEP utilization. We hypothesized that HIV PEP utilization would be greater for percutaneous injuries, for exposures to known HIV-infected sources, and for the latter years of the study when awareness of the national guidelines was likely more widespread.
Study Setting and Population
The study included HCWs who presented for medical care after a blood or body fluid exposure to the 10 civilian adult general hospital EDs in Rhode Island for the 6.5-year period of January 1995 to June 2001. These exposures were percutaneous injuries from needles or other sharps and blood or body fluid splashes. The study population comprised hospital- and non-hospital-based HCWs who presented to an ED for a blood or body fluid exposure. These HCWs may or may not have been employees of the hospital to which they presented. The study included prehospital workers and clinic or private practice HCWs.
Of the 10 EDs in Rhode Island, 4 are general teaching hospitals (affiliated with a medical school and sponsor undergraduate and graduate medical education programs) and 6 are nonteaching (community) general hospitals. The institutional review boards of all hospitals approved the study.
We searched hospital and ED provider billing databases for blood or body fluid exposure visits using 6 International Classification of Disease, ninth revision (ICD-9), Clinical Modification (Department of Health and Human Services, sixth edition, 2001) codes. These codes were E920.5 (needlestick), V01.7 (exposure to other viral diseases), V01.8 (exposure to other communicable diseases), V07.8 (other specified prophylactic measure), V07.9 (unspecified prophylactic measure), and V15.85 (exposure to potentially hazardous body fluids). The codes were determined from a pilot study to be those most likely to identify ED visits for blood or body fluid exposures.11 Human bites were not included in the study, because there were no practical means of identifying all these visits for the study period using ICD-9 codes and because HIV PEP is generally not recommended for those exposures.
Three EDs had separate ED provider and hospital billing databases. These separate billing databases were searched independently to maximize capture of patient visits. For these 3 EDs, the 2 databases were merged, the duplicates were removed, and a single list of patients was generated. For all other hospitals, the sole source for cases was the hospital's billing database.
Data Collection and Processing
The authors searched for medical records of all patient visits identified by the ICD-9 code-directed database query. Each medical record was reviewed; those visits that were for an HCW blood or body fluid exposure were included in the study. HCWs were identified by their occupation listed in the medical record. When the occupation was unclear, the hospital was asked for clarification. HCWs whose occupation could not be determined were classified as “unknown” and constituted HCWs who worked in clinics or other facilities that were not affiliated with the hospital. Repeat or follow-up visits for the same exposure were excluded. For the visits included in the study, age and gender of the exposed HCW, occupation of the HCW, time of exposure and ED presentation, type of the exposure (percutaneous or blood or body fluid splash), body location of the exposure (skin or mucous membrane), HIV status of the source patient, and HIV PEP utilization were extracted and recorded on a standardized form by the primary author and trained research assistants. The primary author instructed the research assistants on the study protocol and directly observed their data extraction. The primary author also reviewed all data forms and provided interpretation for any data that were unclear. Exposures were considered “significant” if the exposure was a percutaneous injury or a mucosal exposure to blood. HCW occupations were collapsed into 7 groups (eg, nursing staff, medical staff, technicians). HIV PEP utilization was separated into 2 components: the offering of HIV PEP and the acceptance of HIV PEP when it was offered. HIV PEP was considered to be accepted when patients received it through the dispensing of antiretroviral medications in the ED or the provision of a prescription for these medications. Two trained research assistants independently entered each form into an Epi Info 2002 (CDC, Atlanta, GA) database and performed a data comparison analysis to verify that all forms were entered correctly. Incorrect entries were corrected, and subsequent analyses were performed on this verified database. Missing data were not imputed.
The data were analyzed using STATA 9.2 (Stata Corporation, College Station, TX). The analysis included calculating summary statistics for the demographics (age, gender, and occupation) of the exposed HCWs and the characteristics of their exposures (exposure type, body location, HIV source status, and time elapsed from exposure to ED presentation). The types of exposures were stratified by categories (percutaneous, blood splash to mucous membranes, blood splash to skin and other body fluid splash to mucous membranes, and other body fluid splash to skin) by occupational group.
The temporal patterns of ED visits for blood or body fluid exposures were examined by year, month, day of the week, and hour of the day using the seasonality test of Walter and Elwood.12 The average increase in ED visits by year was evaluated using simple linear regression.
The incidence of blood or body fluid exposures by occupation and occupational group was calculated using estimates of the average number of HCWs practicing in the state of Rhode Island during the study period by occupation. To calculate rates for custodial staff, clerks and medical secretaries, medical students, nursing students, and security guards, all the hospitals included in the study, the state's medical school, and all the state's nursing schools were contacted for estimates on the average number of people in these groups for the year 2000. The number of group home workers was obtained from the State Department of Children, Youth, and Families. For all other occupations, 2000 US Census estimates were used. The average IRs and corresponding 95% confidence intervals (CIs) by occupation were calculated using Poisson regression. The incidence was expressed as the average number of visits by occupation in the 6.5-year period per 10,000 HCWs in that occupation. The relative ranks of IRs by occupation were estimated using a parametric bootstrap method, under the assumption that the number of ED visits has a Poisson distribution. For each bootstrap resample, a Poisson count of ED visits corresponding to the 6.5-year study period was generated by using the estimated IRs for blood or body fluid exposures. The estimated 95% CIs for the ranks were based on the relative ranks of IRs across occupations for 10,000 bootstrap resamples.
The proportion of patients who were offered and who accepted HIV PEP when it was offered to them were compared by the following factors: significance and category of the exposure, HIV status of the source, time elapsed from exposure to ED presentation, gender, occupational group, type of hospital ED to which the HCW presented (academic vs. community hospital ED), and year of exposure. This comparison was limited to HCWs exposed to known HIV-infected and unknown HIV status sources. HCWs exposed to known HIV-negative sources were excluded from the HIV PEP analyses because HIV PEP was not indicated for these patients. We compared whether or not these factors were related to the offering and the acceptance of HIV PEP using the Pearson χ2 test for categoric values and 2-sample tests of binomial proportions for bivariate values. Differences were considered significant at the α = 0.05 level.
We created univariable and multivariable logistic regression models to compare factors potentially associated with the offering of HIV PEP and the acceptance of HIV PEP when it was offered. The factors examined were the significance of the exposure, HIV status of the source, type of hospital, whether or not the HCWs were medical staff, year of exposure, HCW gender, and time elapsed from exposure to ED evaluation. Significance of the exposure, HIV status of the source, and time elapsed were considered as factors because these are recognized determinants in HIV PEP decision making.6-9 The other factors were factors hypothesized as potentially important by the study authors. The factors were first evaluated in a univariable logistic analysis. All factors were considered in the multivariable models. No interactions were statistically significant; they are not shown. Separate multivariable models were formed for the offering of HIV PEP and the acceptance and receipt of HIV PEP when it was offered. The models were limited to HCWs who presented for evaluation of their exposures from January 1996 through June 2001, HCWs exposed to known HIV-infected or unknown HIV status sources, and exposures for which the time elapsed from exposure to ED presentation could be calculated. Exposures sustained in 1995 were excluded from the HIV PEP models because HIV PEP was not recommended for HCWs by the CDC until January 1996. For the odds ratio (OR) estimates, 95% CIs were created.
Results of the ICD-9 Code Search
A total of 2659 patients were identified through the ICD-9 code search, and 2554 (96.1%) medical records for these patients were available for review. Of these, 1994 visits involved a blood or body fluid exposure and there were 1551 visits by HCWs. The rest were visits by non-HCWs or were not blood or body fluid exposure visits (ie, they were miscoded visits).
Of the 1551 HCWs, the median age was 35 years (range: 15 to 76 years, interquartile range: 29 to 43 years), 69.7% were female, and 30.3% were male. Of all HCWs, 50.4% presented to an academic hospital and 49.6% to a community hospital.
Most (72.5%) HCWs sustained percutaneous injuries: 6.9% were splashed with blood to mucous membranes, 8.2% were splashed with blood to the skin, 7.9% were splashed with other body fluids to mucous membranes, and 4.6% were splashed with other body fluids to the skin. Of all exposures, 67.8% were needlesticks and 15.0% were blood splashes. Because of the large percentage of percutaneous injuries, most of the HCWs incurred exposures to skin (85.2%) instead of to mucous membranes (14.8%). Of the 14.8% with mucosal exposures, 82.5% were splashes to the eye and 17.5% were splashes to the mouth. Among the HCWs with mucous membrane exposures, most (46.7%) were splashed with blood.
At the time of the ED visit, 2.5% of HCWs were exposed to a known HIV-infected source, 1.9% to a known HIV-negative source, and 95.6% to source of unknown HIV status. Of the 38 exposures to known HIV-infected sources, 17 were percutaneous exposures, 4 were blood splashes to mucous membranes, 7 were blood splashes to skin, 6 were splashes from other body fluids to mucous membranes, and 4 were splashes from other body fluids to skin. Sixteen of the exposures to known HIV-infected sources were sustained by nursing staff, 7 by technicians, 7 by medical staff, 6 by emergency service (EMS)/firefighters, and 2 by other HCWs.
Temporal Trends for ED Visits for Exposures
Of the 1551 HCW visits for blood or body fluid exposures, 1384 had medical record documentation of the time elapsed from exposure to ED presentation. Of these 1384 HCWs, 98.2% presented within 72 hours and 91.2% within 24 hours of their exposure. Patients sustaining percutaneous injuries or blood splashes to mucous membranes (OR = 2.29, 95% CI: 1.00 to 5.24) were more likely to present within 72 hours than patients with other exposures. All HCWs exposed to known HIV-infected sources presented within 72 hours, and all but 1 presented within 24 hours. With the exception of the dental workers, 2.8% or fewer of each occupational group presented >72 hours after their exposure; 17.4% of dental workers presented >72 hours after exposure.
Figure 1 diagrams the temporal trends for HCW visits to the ED during the study period. We performed the test of seasonality by Walter and Elwood12 to evaluate cyclic patterns for ED visits by year, month, day of the week, and hour of the day. The results of the test by Walter and Elwood12 for the presence of a cyclic pattern were statistically significant for month (P ≤ 0.01), day of the week (P ≤ 0.00), and hour of the day (P ≤ 0.00). The results of this test show that the patterns of visits had a harmonic cycle, which signified that there were definite trends of peaks and nadirs over each time period. Blood or body fluid exposure visits peaked during March, on weekends, and around 5:00 pm and were at a nadir in September, on Tuesdays, and around 5:00 am. The test by Walter and Elwood12 for harmonic cycles for year was not significant because there was a clear increase in visits over the study period and no cyclic changes in visits. Simple linear regression demonstrated that there was an average estimated increase of 50 additional cases (95% CI: 27 to 73) each year.
Exposures by Occupation
Approximately half (50.2%) of the 1551 HCWs who sustained a blood or body fluid exposure and presented to the ED during the study period were nursing staff, 14.9% were hospital or laboratory technicians, 12.9% were EMS/firefighters, 9.7% were medical staff, 4.9% were support staff, 2.7% were allied health staff, and 1.7% were dental workers. An additional 3% were HCWs whose occupation could not be determined from the medical records or through hospital human resources databases.
Table 1 displays the frequency of exposures by occupational group and exposure category. For each occupational group, percutaneous injuries were the most frequent, with the exception of EMS/firefighters. Among EMS/firefighters, the percentage of other exposures was relatively greater than for most other occupational groups. Allied health staff, which includes respiratory therapists, sustained a relatively larger percentage of nonblood exposures to mucous membranes than other HCWs because of the larger number of spit/saliva/sputum exposures sustained by this group.
Table 2 provides the IRs per 10,000 persons and the ranking of IRs of ED visits for blood or body fluid exposures by occupation and by category of exposure. The IRs are also shown by occupational group. Hospital security guards, EMS/firefighters, and respiratory therapists had the highest IRs for all exposures. For percutaneous exposures, hospital custodial staff, nurse practitioners/physician assistants, and respiratory therapists had the highest rates. The high IRs for EMS/firefighters and security guards were attributable to other blood or body fluid exposures.
HIV PEP Utilization
Of the 1551 HCWs, 30.2% were offered HIV PEP and 48.9% of those accepted it. Table 3 shows a comparison of the percentages of HCWs exposed to known HIV-infected and unknown HIV status sources who were offered HIV PEP and the percentages of those who accepted HIV PEP when it was offered to them. More HCWs were offered HIV PEP if they sustained a significant exposure, if the source was known to be HIV-infected, and if they presented to an academic hospital ED. There was a nonstatistically significant trend for patients who presented within 72 hours to be offered HIV PEP compared with those who presented later. There was no difference in the offering of HIV PEP between women and men, but more men than women received HIV PEP. HIV PEP offering did vary by occupational group. Medical staff were more likely to be offered (54.1% vs. 28.2%; P ≤ 0.00) HIV PEP than all other occupational groups. Offering of HIV PEP increased steadily over the 6.5 years of the study (logistic regression test of trend: OR = 1.23, 95% CI: 1.15 to 1.32).
Of those who were offered HIV PEP, more HCWs accepted it if they were exposed to a known HIV-infected source, if they were male, and if they presented to an academic hospital ED. There was a nonstatistically significant trend for HCWs sustaining a significant exposure and who presented within 72 hours of their exposure to accept HIV PEP when it was offered to them. Medical staff were more likely to accept HIV PEP when offered it, compared with all other occupational groups (65.0% vs. 45.4%; P ≤ 0.00).
In 1995, the year before the release of CDC recommendations on HCW HIV PEP, 2 of 36 HCWs presenting to EDs for blood or body fluid exposures were offered HIV PEP. One HCW offered HIV PEP was a nurse who sustained a blood splash to a mucous membrane, and the other was a physician who had a percutaneous injury. The nurse was exposed to an unknown HIV status source and the physician to a known HIV-infected source. The physician accepted HIV PEP but the nurse did not.
Predictors of HIV PEP Offering of and Acceptance and Receipt of HIV PEP
Table 4 depicts the logistic regression analyses evaluating potential predictors of HCWs being offered and then accepting HIV PEP when it was offered to them. For the univariable and multivariable analyses, the study sample was restricted to HCWs who were exposed to sources known to be infected with HIV or of unknown HIV status, to the years 1996 through 2001, and for those whose time elapsed from exposure to ED presentation was known. Patients exposed to known HIV-negative sources were excluded. Model 1 is a multivariable logistic regression model in which the time elapsed from exposure to presentation for medical care in the ED was included as a covariate. Model 2 excludes this covariate. Time elapsed since exposure was excluded as a covariate because it did not seem to be a significant factor in the offering and acceptance and receipt of HIV PEP in the initial evaluation of this factor (see Table 4). Models are presented with and without this factor, however, because time elapsed from exposure is one of the CDC-recommended criteria for deciding if HIV PEP should be prescribed. Even though time elapsed since exposure was removed from model 2, HCWs whose time elapsed since exposure was not recorded were still excluded from the logistic regression analyses so that the models would be comparable. Additional multivariable models that included HCWs whose time elapsed since exposure was not recorded verified that the conclusions reached in the earlier analyses did not change (data not shown).
In regard to the offering of HIV PEP, the univariable and multivariable analyses were consistent in showing that HIV PEP was offered more to HCWs who sustained a significant exposure, were exposed to a known HIV-infected source, were medical staff, presented to an academic hospital ED, and presented after an exposure in the latter years since the initial release of the CDC occupational HIV PEP guidelines. The impact of year of exposure on the offering of HIV PEP was statistically significant for 1998 and later.
With respect to the acceptance of HIV PEP when it was offered, in the univariable analysis, HCWs were more likely to accept HIV PEP if they were exposed to a known HIV-infected source, were male, were medical staff, and presented to an academic hospital ED. There was a nonsignificant trend for decreased acceptance of HIV PEP over the years of the study. In the multivariable analyses, however, only male gender and presentation to an academic hospital ED remained statistically significant factors.
Although nurses comprised the largest group of HCWs presenting to EDs after blood or body fluid exposures, hospital security guards had the greatest overall incidence of these exposures. Because of the nature of their work responsibilities, it is not surprising that they had the greatest incidence. Most of their exposures were attributable to cutaneous or mucosal blood or body fluid exposures, however, exposures for which prophylaxis is not recommended. Hospital protocols or heightened concern by security guards about exposures, even when minor, might have prompted them to seek medical care even for minor exposures. In contrast, hospital custodial staff had the highest relative incidence of percutaneous injuries, which are exposures for which prophylaxis might be indicated.
Chen and Jenkins5 estimated the incidence of ED visits for blood or body fluid exposures by occupational groups using NEISS data from 1998 to 2000. The database is a probability sample from 67 EDs in the United States and aims to document nonfatal work-related injuries evaluated in US EDs. By the estimates of these investigators, nurses had the highest rate of exposures, followed by clinical laboratory technologists/technicians, physicians, health technologists/technicians, and maids/housemen. Their ranking of occupations by incidence of exposures differed from ours for several reasons, including the use of full-time equivalents to calculate rates, the use of current population survey data instead of the US Census and direct inquiry of health care facilities for estimates of the population by occupation, and the definitions and groupings of occupations. It is also possible that the exposure incidence patterns in Rhode Island are different than for the EDs sampled through the NEISS database.
Two statewide studies have reported the frequency of percutaneous injuries by health care occupation.2,3 A California study surveyed health care facilities in that state, whereas a Massachusetts study relied on mandatory reporting of HCW percutaneous injuries. Both found that most of the reported percutaneous injuries occurred in nurses. These studies did not assess the incidence of these events; thus, they do not adjust for the proportion of HCWs who were nurses. They also do not provide estimates on the frequency or incidence of other types of exposures. EPINet also reports the frequency of percutaneous and other blood or body fluid exposures by occupation.13 EPINet also provides composite incidence estimates for these exposures for all HCWs using the number of occupied hospital beds as a reference standard. As a result, these estimates are not comparable to those of the current study. The National Surveillance System for Hospital Healthcare Workers (NaSH) of the CDC reports frequencies of blood or body fluid exposures by HCW occupation, however, but does not calculate the incidence of these exposures by occupational group (Adelisa L. Panlilio, MD, MPH, personal communication, 2006).
As also found in the EPINet and NaSH studies, most HCWs presenting for an evaluation of an occupational exposure sustained a percutaneous injury.13 The higher prevalence of percutaneous injuries compared with other types of exposures is probably because HCWs frequently engage in work-related activities involving needles and other sharp objects and because they would naturally view such injuries as requiring medical attention. It is important to also note that as in this study and the EPINet studies, the eyes are most commonly affected during splashes of blood or body fluids to mucous membranes. This finding supports the value of using protective eyewear during health care duties that could involve splashes to blood or body fluids.
Although most exposures for all occupations were percutaneous, the spectrum of exposures varied by occupational group. Medical and nursing staff and technicians displayed a relatively similar distribution of types of exposures, probably because of an overlap of patient care responsibilities and work environment. The spectrum was different for allied health workers, dental workers, and EMS/firefighters, however. Each of these groups has unique work environments and opportunities for exposures specific to its role. These differences among occupational groups dictate the need for occupation-specific prevention strategies to reduce the occurrence of these exposures. For example, the importance of wearing gloves should be reinforced to EMS workers, given their high likelihood of blood splashes to the skin. No other studies have compared the spectrum of exposures by HCW occupation, although some have reported the types of exposures for selected occupations.14
Definite temporal trends regarding when HCWs present to EDs with these exposures were observed in this study. It is reassuring to report that most HCWs present within the recommended 72-hour window after an exposure when HIV PEP can be prescribed. Not surprisingly, HCWs were more likely to present in the evening hours and on weekends when most hospital- or clinic-based employee health facilities are closed. It is unclear why HCWs are more likely to present during the winter months, but it is likely because of the larger number of patients admitted to hospitals or evaluated in other health care facilities during the winter. These observations should help EDs to anticipate when these visits occur so that staff can be trained accordingly and mechanisms to facilitate evaluation and treatment of HCWs with these exposures can be optimized. We also observed an increasing trend of ED visits for blood or body fluid exposures over the years of the study. This increase could be attributable to the proliferation of guidelines and occupational safety programs and protocols on the management of these exposures that prompt HCWs to seek an evaluation for their exposure. EDs should be mindful when designing protocols in anticipation of a potential continued increase in the frequency of visits for these exposures.
In concurrence with CDC guidelines on HIV PEP, more patients who had significant exposures and were exposed to a known HIV-infected source were offered HIV PEP than those with nonsignificant exposures and those exposed to sources of unknown HIV status. Only a few patients with significant exposures were offered HIV PEP, probably because of the influence of additional features of the exposure, such as HIV status of the source and depth of the injury, that could be not evaluated in this study. We observed some deviations from CDC recommendations on the offering of HIV PEP. For example, although some HCWs had nonsignificant exposures, approximately 10% were offered HIV PEP. There might have been some features of the exposure that prompted clinicians to prescribe HIV PEP for these HCWs, such as exposure to a known HIV-infected source. In addition, it is important to emphasize that most HCWs were exposed to sources of unknown HIV status. As a result, most of those offered HIV PEP were exposed to an unknown HIV status source. It is in these situations that the CDC guidelines are less explicit and clinicians are forced to make case-by-case decisions regarding HIV PEP. It is difficult to determine the appropriateness of HIV PEP use under those circumstances.
Although determining the time elapsed from exposure to presentation for evaluation is a CDC-recommended component of assessing if HIV PEP should be offered after a blood or body fluid exposure, this factor was not statistically significant in our analyses. This apparent discrepancy is likely attributable to the small number of patients who presented more than 72 hours after their exposure. We would probably detect a difference with an even greater sample size.
We found several nonclinical factors that seemed to affect the offering of HIV PEP. As expected, HIV PEP offering was influenced most by the HIV status of the source and significance of the exposure. This predictable observation has been reported in other studies.15-18 Also, as anticipated, HIV PEP offering was greater as the years progressed from the introduction of the CDC occupational HIV PEP guidelines in 1996. HIV PEP was offered more frequently at academic EDs than at community hospital EDs. This could be attributed to differences in patient populations, experience and level of training of the HCWs at those institutions, and frequency of invasive procedures performed. Academic hospitals may have more patients who are at an increased risk for HIV infection, would have more HCW trainees, and would be more likely to conduct specialized procedures that bring HCWs into contact with blood or body fluids. It is possible that academic hospitals are more likely to have protocols in place that support the prescription of HIV PEP, however.
Medical staff were offered HIV PEP more often than other occupational groups. This is a surprising finding, given that medical staff did not have a higher percentage of significant exposures or exposures to known HIV-infected sources. There might have been features of their exposures that we could not identify, such as the nature of the significant exposure, as in the depth of the percutaneous injury. This trend has been observed elsewhere. Russi et al16 examined predictors of HIV PEP utilization at 2 Yale University-affiliated hospitals in New Haven. Using a multivariate model, they found that physicians were more likely to be prescribed HIV PEP than other HCWs even when controlling for HIV status of the source and the type of exposure. It is potentially concerning that medical staff may have been preferentially offered HIV PEP because of the nature of their hierarchical position and not because of the characteristics of the exposure.
In terms of the acceptance of HIV PEP when it was offered, approximately half of the HCWs offered HIV PEP accepted it. This finding indicates that HCW choice-and perhaps ED clinician influence-affects HIV PEP acceptance and receipt. This reduction may be attributable to a perception by the HCW that his or her risk of infection is minimal, fear of the potential adverse side effects of HIV PEP, influence from the clinician that HIV PEP is unnecessary for her or him, or an inability to pay for the medications. We were unable to assess which, if any, of these reasons influenced the choice for HIV PEP. Further, we found that HCW gender and type of hospital ED were associated with acceptance and receipt of HIV PEP. Male HCWs and HCWs presenting to academic hospital EDs were more likely to accept HIV PEP when it was offered. We suspect that the same reasons for the greater offering of HIV PEP at academic hospital EDs are applicable here. We cannot explain why gender was a factor in HIV PEP acceptance, however. We do not know if female patients are less likely to accept HIV PEP because they are not encouraged to take it or if some other unmeasured factor is confounding this relation. ED providers should be aware of this concern when counseling patients about HIV PEP and be certain that female patients have the same opportunities as male patients to accept HIV PEP when it is indicated and offered.
Our study had several limitations. First, our study does not include all HCW visits for blood or body fluid exposures in the state because an unknown percentage presented to employee or occupational health centers, to urgent care centers, or to primary care clinician offices or clinics. If visits to other settings were included, the incidence of exposures by occupation might be different. Our goal was to examine ED visits and HIV PEP utilization and not other settings, however. Further, we wanted to include prehospital and nonhospital HCWs, who have previously been excluded from other studies. In addition, apart from EMS/firefighters, the frequency, incidence, and spectrum of exposures by occupation would likely be similar if visits to all health care settings were included, because HCWs are not more likely to seek medical care preferentially from the ED. Second, because of the retrospective nature of the study, detailed information on the nature and extent of the exposures was unavailable, which limited our analyses of the significance of the exposures. Nevertheless, we were able to examine key factors that influenced HIV PEP use. Third, the use of ICD-9 codes to identify cases relies on proper coding and diagnosis, which is not always possible. As a result, some cases were missed. Given the large number of cases in the study, we do not believe that our findings were adversely affected by missing cases. Fourth, although we conducted a statewide analysis on this topic, our findings may not be generalizable to other states. We would expect similar trends nationwide, however, given our large sample. Fifth, given the introduction of rapid HIV testing in the clinical setting, HIV PEP protocols are undoubtedly different now. As a result, the study findings might not be applicable to current practice. Future studies can investigate whether or not the same patterns exist because of these advances in medical care.
Our study presents the incidence and temporal patterns of ED visits for HCW blood and body fluid exposures and describes the types and characteristics of these exposures. Incidence varied by occupational group and by type of exposure. Percutaneous injuries were the most common type of exposure, although the spectrum of exposures differed by occupational group. Tailored prevention strategies could be instituted to reduce the incidence of exposures specific to those groups. HCWs more likely to be offered HIV PEP were exposed to a known HIV-infected source, sustained a significant exposure, presented to an academic hospital ED, were a member of the medical staff, and were exposed during the latter years of the study. HCWs were more likely to accept HIV PEP when it was offered to them if they were male or presented to an academic hospital ED. The apparent preferential utilization of HIV PEP by medical staff, men, and HCWs presenting to academic hospitals suggests the need for uniform adherence to suggested guidelines on the use of HIV PEP.