Marchant, Gary E. PhD, JD; Crane, Angus JD
Voluntary or cooperative programs between government and industry have become a popular alternative or supplement to traditional regulations. Voluntary programs will likely continue to flourish because the high cost of formal rulemaking hinders government agencies charged with addressing an ever-growing number of regulatory targets and priorities. The North American Insulation Manufacturers Association (NAIMA), a trade association of companies manufacturing fiberglass, rock wool, and slag wool insulation products, recently completed an 8-year voluntary occupational safety program for synthetic vitreous fibers (SVFs) in partnership with the Occupational Safety and Health Administration (OSHA). The centerpiece of this program was the creation of an SVF occupational exposure database.
In May 1999, NAIMA began implementing a comprehensive voluntary work practice partnership with OSHA in response to OSHA's Priority Planning Process. This NAIMA–OSHA partnership program, known as the Health and Safety Partnership Program (HSPP), promoted the safe handling and use of insulation material and incorporated education and training for workers involved in the manufacture, fabrication, installation, and removal of fiberglass, rock wool, and slag wool insulation products. As a result of the HSPP, a voluntary permissible exposure limit (PEL) of one fiber per cubic centimeter (1 f/cm3) was established and, most relevant for this article, an extensive worker exposure database was created.
This article describes the experience of creating, implementing, and completing an exposure database for SVFs. The lessons learned and benefits achieved through this experience may be instructive to government and industry when assessing the need and utility of an occupational exposure database for substances such as nanomaterials. This article summarizes the background and creation of the SVF exposure database; describes the benefits achieved to date, followed by discussion of the challenges involved in the creation of any industry-wide exposure database, and how those challenges were addressed. This article concludes that the SVF database provides a good case study to illustrate the potential benefits of an exposure database as well as the potential challenges and pitfalls in creating such a database.
BACKGROUND AND HISTORY OF THE SVF EXPOSURE DATABASE
Synthetic vitreous fibers are a class of inorganic fibrous materials including glass wool or fiberglass, mineral wool (also known as rock and slag wool), textile glass fibers, and refractory ceramic fibers. Historically, this class of fibers has also been described as man-made mineral fibers, man-made vitreous fibers, and manufactured vitreous fibers. Fiberglass and rock and slag wool fibers are used primarily in a variety of thermal and acoustic insulation products, but also have numerous filtration, fireproofing, and other applications.
Human exposure to SVFs occurs almost exclusively in the occupational context, because installed product usually do not result in exposure to airborne fibers.1 Synthetic vitreous fibers are used in a variety of applications. Insulating homes, other buildings, and industrial processes against heat loss and heat gain represents the largest single use for glass and rock and slag wools; up to 70% of industry output is for these applications. These wools can be blown into structural spaces, such as in walls and attics. Rock wool and glass fiber are also incorporated into ceiling tiles to provide fire resistance and thermal and sound insulation. Batts, blankets, and semirigid boards made of glass, rock wool, or slag wool fibers are used in both residential and commercial buildings. Pipe and board insulations are used extensively in industrial processes. In addition, glass, rock wool, or slag wool can be used to insulate cold and hot pipes both indoors and outdoors and in many climates. They are also used on sheet-metal ducts and plenums for thermal and acoustic insulation, resulting in quieter and more energy-efficient heating and air conditioning systems. Glass, rock, and slag wools are effective thermal and acoustic insulators and improve energy efficiency in many electrical appliances and other types of machinery. Vehicles or carriers (cars, ships, aircraft, and spacecraft) are fitted with glass wool insulation to enhance their performance and provide the appropriate thermal and acoustic conditions for the goods or passengers being transported. Glass and rock wools are also used in sound-absorbent barrier panels alongside motorways and railways. Glass and rock wools are used as growing media and for soil conditioning in agriculture. Rock wool mats are used for insulation of railway and tramway tracks against vibration. The unique properties of special-purpose fibers make them ideal for use in battery separator media and as filtration medium.
THE SVF HSPP
Before 1995, occupational exposure to SVFs was regulated primarily as a nuisance dust. In 1995, OSHA published the results of its Priority Planning Process, a multiyear process to develop a list of 18 occupational safety and health issues that the agency deemed needed additional attention because of either the seriousness of the hazard or the number of workers potentially exposed. Recognizing that it lacked the resources to conduct formal rulemaking on all 18 substances or issues on its list, OSHA prioritized five of the issues for rulemaking and announced its intention to address the other 13 issues through voluntary or other measures. Synthetic vitreous fibers were among the list of 18 work-related issues identified by OSHA as a priority, but were designated for voluntary measures rather than rulemaking. The Occupational Safety and Health Administration listed SVFs as a priority largely because OSHA estimated that more than 225,000 workers were exposed to SVFs, and that projections indicated that the total number of workers handling SVFs in the coming years would increase.
In early 1996, NAIMA approached OSHA to discuss a voluntary worker protection program in response to the agency's announcement of the Priority Planning Process listing SVFs as a nonregulatory priority. The North American Insulation Manufacturers Association and its member companies had already instituted their own product stewardship program, and were eager to share the results of these efforts with OSHA, in hope of resolving agency concerns about workplace safety for SVFs. For example, NAIMA member companies had funded tens of millions of dollars of health research on SVFs at leading independent laboratories and universities in the United States and abroad. In addition, NAIMA and its member companies had developed safe work practices to protect workers against exposures to SVFs, including an internal recommended worker 8-hour time-weighted average (TWA) exposure limit of 1 f/cm3.
From 1996 to 1999, NAIMA negotiated with OSHA to create and implement a voluntary program for SVFs, known as the HSPP (http://www.naima.org/pages/benefits/hspp/hspp.html). The HSPP was formally adopted by OSHA on May 18, 1999, and applied to the manufacture, fabrication, installation, and removal of fiberglass, rock wool, and slag wool insulation products. The HSPP contemplated a 3-year implementation period (1999 to 2002), followed by a 5-year compliance period (2002 to 2007), which has now been successfully completed. The HSPP included a number of specific commitments imposed upon NAIMA and its member companies that were designed to educate and encourage compliance with the HSPP guidelines by other employers and their workers.
SPECIFIC PROVISIONS OF THE HSPP
A significant feature of the HSPP was the establishment of a voluntary 1 f/cm3 PEL for fiberglass and rock and slag wools fibers. The HSPP committed NAIMA member companies to use product design, engineering controls, work practices, respiratory protection, or a combination of any or all of these measures to bring fiber exposure to the voluntary 1 f/cm3 PEL. To strengthen these control measures, the HSPP specified comprehensive work practices for those working with fiberglass, rock wool, and slag wool insulation. NAIMA also undertook sponsorship of training sessions to help educate workers and employers about the consolidated work practices. To do so, NAIMA gave its members and other employers educational tools such as video tapes and literature to further explain the recommended work practices.
A fundamental aspect of the recommended work practices dealt with when and where to use respiratory protection. The HSPP recommended respiratory protection whenever exposures on a job exceeded the 1 f/cm3 8-hour TWA PEL. The N95 series dust respirators certified by NIOSH were the approved type of respirators recommended by the HSPP.
Most important, the HSPP committed NAIMA to provide an exposure database to help contractors and workers determine the level of potential exposure to fiberglass, rock wool, or slag wool for a given task. NAIMA also committed to supplement the database with additional exposure data collected from various sources and studies. Exposure monitoring and an exposure database are closely related to the respiratory protection guidelines, thereby offering contractors standardized methods for determining whether respiratory protection is needed for a particular task. This helps contractors reduce the burden of compliance under the OSHA Respiratory Protection Standard.
When OSHA endorsed the HSPP, OSHA supported the ability of contractors to rely on the NAIMA exposure database as the means for determining exposure levels. Specifically, the preamble to OSHA's 1998 Respiratory Protection Rule states that “OSHA recognizes that there are many instances in which it may not be possible or necessary to take personal exposure measurements to determine whether respiratory protection is needed.”2 In addition, OSHA's rule preamble states that the “Final rule permits employers to use other approaches for estimating worker exposures.” Consistent with this incentive for voluntary compliance in the OSHA regulations, the agency approved the use of “[d]ata from industry-wide surveys by trade associations” and noted that such information is “often useful in assisting employers… to obtain information on employee exposures in their workplaces.” In fact, OSHA specifically cited NAIMA's database in the preamble as an example of industry data that could be relied upon by employers.
NAIMA met or exceeded all of the commitments set forth in the HSPP. First, NAIMA organized an HSPP committee to oversee the program's implementation, and soon thereafter formed an Occupational Health and Safety Subcommittee to govern the development of an exposure database and establish a quality assurance/quality control (QA/QC) auditing team to oversee population of the exposure database. The QA/QC auditing team included a faculty member from Arizona State University, certified industrial hygienists from several NAIMA member companies, a corporate officer of NAIMA, and a third-party computer expert. Professor Marchant of Arizona State University manages the database.
An important aspect to the creation and maintenance of the exposure database was the establishment of QA/QCprocedures for data submittal. The QA/QC procedure describes the steps that must be taken in approving data for the database and identifies the specific information that must be available for a data point. Specifically, the following details are required about all data points: (1) sample identifier; (2) sample date; (3) SVF type (fiberglass, rock wool, or slag wool insulation); (4) product type; (5) type of manufacturing/use (primary manufacturing, fabrication, etc); (6) job description (packer, installer, feeder, etc); (7) sample type (personal or area); (8) number of samples for TWA; (9) TWA quantifier; (10) results to two decimal places; (11) sampling and analytical methods employed; and (12) sample duration.
NAIMA committed to format and categorize by product and task 3000 to 5000 samples of pre-1990 exposure measurements. In the first year of implementation, the database had 4200 exposure samples. By the end of the second year (2003), the exposure database had expanded to include over 7000 exposure samples. According to the HSPP, 400 data points were to be added to the database each year after the first 2 years. This target of 400 additional exposure samples was exceeded each year. All new data entered into the database was first reviewed and approved by the QA/QC auditing team pursuant to its written procedures. By the end of the HSPP, the database had in excess of 14,000 exposure measurements, including exposure data on more than 35 different products and more than 60 different jobs. NAIMA is committed to maintaining the database beyond the HSPP, and thus the database is expected to continue to grow over the years.
BENEFITS OF SVP EXPOSURE DATABASE
The HSPP has officially been completed. In general, the HSPP successfully accomplished its goals. Creating the extensive exposure database was the element of the HSPP that required the most time, effort, and resources. Although the industry had abundant exposure data, those data had not been archived in one location and it was therefore not easily accessible. The HSPP created an ideal opportunity to assemble a robust and high-quality database that could be relied upon by workers, agency staff, and independent researchers. In that regard, the SVF database has proven to be very successful and useful.3,4
Contractors can rely on NAIMA's database without conducting their own exposure monitoring. It helps contractors and workers determine the level of potential exposure to fiberglass, rock wool, or slag wool for a given task. The exposure database contains sample data about exposure levels categorized by product type and specific work task. Furthermore, NAIMA has analyzed exposure data involving typical exposure levels for many common jobs, and documented that most of these jobs currently can be completed without exceeding the exposure limit of 1 f/cm3 for an 8-hour TWA.
The SVF database is clearly valuable, as demonstrated by various government agencies and other entities who have relied upon it, such as the 2002 International Agency for Research on Cancer monograph on SVFs cites both the HSPP exposure database and the HSPP itself.5 Similarly, when the Agency for Toxic Substances and Disease Registry of the U.S. Department of Health and Human Services created a toxicological profile on SVFs, that agency also relied upon the SVF database.1 Data from the SVF database have also been summarized in several user-friendly formats for use by employers, workers, and other interested parties. An example of such a data summary is shown in Table 1, showing that most exposures in the industry are below the 1 f/cm3 voluntary PEL.
TABLE 1-a. Fiberglas...Image Tools
CHALLENGES FACING EXPOSURE DATABASES
TABLE 1-b. Fiberglas...Image Tools
Exposure databases such as the SVF database provide important benefits and applications, but creation of such databases do present a number of challenges and potential problems. The experience with the SVF database provides some insight on some of these issues and how they have been addressed in this context.
A major challenge involves ensuring the consistency of the data submitted to and accepted into the database. It is important that individual data points be consistent and comparable because the strength of the database is found in its ability to provide information and estimates based on collective results. There are several potential problems that can adversely impact the consistency of data in any database.
One of the biggest problems encountered by the SVF database was that data submitted to the database had sometimes been collected using different analytical methods to measure SVF exposure. Some exposure samples used mass-based exposure measurements (eg, μg/m3) whereas more recent exposure data use a fiber-counting analytical method. Although North American industry has now agreed on the NIOSH 7400B analytical method, some data points were submitted to the database using other fiber counting rules (eg, NIOSH 7400A, phase contrast optical microscopy). Data submitted from other countries (eg, Australia) were often collected using a different analytical method than NIOSH 7400B. For the SVF database, exposure data collected using different analytical methods were entered into the database while preserving information about the analytical method recorded in the relevant data field. This maximized the amount of relevant data entered into the database but made it impossible to undertake analyses using all the data in the database because of the incompatibilities between the different methods (“apples to oranges” comparisons). This problem of diverse analytical methods may become significant for a nanotechnology exposure database, in the absence of a standard analytical method for nanomaterials.
The inherent differences in product types pose another problem. Many SVF products may present different exposure profiles because of various factors such as product use and quantities, fiber characteristics, and application environments. An exposure database, such as the SVF database, can provide greater resolution and hence greater utility if it therefore classifies exposure data points by product type. This presents challenges, however, that may also be anticipated for nanotechnology. Specifically, there are many different products in the SVF industries (as there is in the nanotechnology field), with subtle differences within product lines and across companies. For example, two similar product types made by different manufacturers might use a slightly different binder formula that could affect the likelihood and duration that fibers stay aloft and potentially inhaled, thus potentially affecting exposure levels.
More generally, different companies might define and categorize similar products differently, and those definitions might change over time. For example, one category of fibers in the SVF database is “special application fibers,” a category requested by OSHA and potentially subject to different interpretations by different companies if not clearly defined. Moreover, the nature of SVF products has changed over time, as many of the fibers have been reformulated, often to reduce any potential health concerns. Thus, comparing exposure to levels of a specific category such as glass batt insulation in different time periods may once again involve products with different exposure characteristics and potential risks (despite having the same nomenclature).
A final consistency issue concerns tasks assigned and their actual job description. In the SVF industry, and possibly when applying nanotechnology, exposure levels can vary significantly across different job categories. Accordingly, stratifying data points by job type is important for making the database useful and relevant. Such categories, however, raise questions about definition and consistency. Each facility is configured differently, and it may use different products or input materials, creating diverse sets of working conditions. All of these variables can impact exposure levels for workers assigned to the same job category. In addition, each company has the right to their own definition of their job types. For example, during the course of developing the SVF database, it was discovered that many companies defined the “general room” job category differently. Because the database managers kept a coded source list for all data points in the database, they were able to go back to the original data sources and ask for clarification, in order to correct the data entered.
Data Quality Issues
Data quality may remain a problem even when definitions are clear and ambiguous. Data in the SVF database were collected from a wide variety of different sources, with different assurances and reliability. A vigorous QA/QC process helped to screen out, correct, or resolve uncertainties about many of the questionable data points submitted to the database (eg, samples with overloaded filters). Nevertheless, some data points were so problematic that they could not be resolved by the QA/QC committee. Some data points submitted to the database were missing mandatory data fields. In such cases, the QA/QC committee would follow up with the original source of the data to determine if the missing data fields could be completed. If those missing data fields cannot be completed (often because the original records could not be located or did not contain the required information), the data points were not entered in the SVF database, but rather were maintained in a separate file called “Valid Data Not Otherwise Meeting Database Criteria.” Data in this file could potentially be used for other research purposes, but are kept separate from and not included in database analyses.
Other quality-related problems include reliability of the data, confidentiality and public access to the data, and representativeness of information in the database. In one case, for example, a set of exposure data points was submitted with exactly 480-minute duration of sampling time. This uniformity of exposure time did not seem consistent with the normal variation observed in the field, and thus follow-up by the QA/QC committee and database administrator were necessary to resolve this issue.
The SVF database operates by entering all data points available that meet the database criteria and QA/QC review. Thus, data cannot be representative of the industry as a whole. Samples are not taken randomly across workplaces but tend to concentrate on job tasks or product types where exposures are known or suspected to be the highest in the facility. Thus, exposure sampling tends to occur where the exposures occur (or at least believed to occur). This factor would tend to inflate average exposure levels in the database relative to the real-world average levels industry-wide but omits gaps in the analysis where exposures therefore remain unknown. In addition, companies differ in how often and when they sample exposure. Larger companies tend to collect and submit more data than smaller companies, so the data may disproportionately represent exposure levels in larger rather than smaller companies. If larger companies with more resources and expertise tend to control exposures better than smaller companies, this factor would tend to underestimate overall exposure levels, thereby skewing the data to reflect their experience and needs.
The exposure data also inevitably include some gaps or limited samples for some occupational contexts. To address these gaps, NAIMA commissioned and paid for outside consultants to obtain data points necessary to fill in these gaps. This can be an expensive undertaking, however, and unless there is an entity associated with the database prepared to make such investments, the gaps in exposure data may remain unfilled.
Several additional issues are raised by the experience from the SVF database. First, database issues can raise competitiveness issues. For example, the definition of certain categories such as special application fibers can create competitive advantages or disadvantages for certain companies. If these concerns are not handled effectively, they can create controversies associated with the database. In the case of the SVF database, these potential concerns were largely addressed, and resolved when they did arise, by a QA/QC committee that included respected experts from most of the major companies or industry sectors involved. These individuals were capable of identifying potential competitiveness issues early and taking proactive actions to resolve such issues before they became a significant problem.
Another potential problem concerns data confidentiality. Assurances of confidentiality of the identity of the company submitting the data were essential to ensure submission of the data (and attendant legal issues are beyond the scope of this article). This issue was resolved in the SVF database by coding the company name and facility submitting the data in the database. Only the database can decode this information. The confidentiality of this code key is strictly protected, but there is always a possibility that future litigants may request the key in a third-party subpoena. This risk could be eliminated by destroying the code key, but maintaining the key has been critical for the database manager to go back to original submitters when questions or ambiguities arise about some of the definitions used in a data submission.
Another issue is whether and how the database is made publicly available. Some databases are made accessible on the web, whereas others are kept more proprietary. The SVF database is not made publicly available in raw form, but summaries of the data are prepared and made widely available to stakeholders and other interested parties. Requests for more complete access to the raw data are considered on a case-by-case basis.
A final problem faced by the SVF database is the lack of participation by some entities with available data. The SVF database enjoys strong participation by all companies within the sponsoring trade association (NAIMA); those companies outside the trade association likely had relevant exposure data but chose not to participate. Beyond requests to such entities, there does not seem to be any incentive that could be used to bring them into the fold.
CONCLUSION: KEYS TO SUCCESS
The SVF database provides a good case study to illustrate the potential benefits of an exposure database as well as the potential challenges and pitfalls in creating such a database. Despite many challenges, the SVF database has generally been successful. In retrospect, some critical factors can be identified as key contributors to this success. First, and probably foremost, the existence of an active QA/QC team staffed with committed and highly respected experts from the industry greatly enhanced the technical accuracy and external credibility of the database. Second, the development and regular updating of a clear and thorough database dictionary to guide data submitters and which carefully defines all database fields and criteria helped to maximize the accuracy and consistency of the data submissions. Third, maintaining careful records of all data submissions, including the coded identity of the company and location associated with each data point, permitted follow-up, correction, or verification of any data point for which subsequent issues or questions may have arisen. Finally, the success of the database was made possible by the commitment and funding of the trade association and its member companies, from the top of each organization down. This combination created a high-quality, objective, and comprehensive exposure database that may serve as a model for nanotechnology and may continue in its own right.