Evaluation of selected publications on reference values for lead in blood

Evaluation on As a part of the global Tracy project, whose aim is to define metal concentrations in tissues and body fluids of reference populations, more than 1000 papers published from 1980 to 1994 were scrutinized that presented tentative reference values for lead in blood in occupationally unexposed adult populations. Ten studies exemplifying criteria for proper sampling, analysis and data treatment are presented and discussed. Levels of lead in blood are influenced by numerous factors. Accordingly, a wide variation in blood lead concentrations was observed. As an example, in a global study in 1983 of nonsmoking female schoolteachers, the geometric mean value for lead in blood varied from 52 yg . I-' in Tolcyo, Japan, up to 193 yg . I-' in Mexico City. The Tracy survey de~uonstrates the importance of factors such as age, gender, ethnicity, food, drinking and smoking habits, hobbies, season and year of sampling, residential area, and geographic location. Lead in blood was shown to be both time and area specific. Thus it was not possible to establish a general reference value for lead in blood.

Lead has many uses in society. World-wide emission of lead occurs through the use of leaded gasoline (1). Industrial emissions may also contribute, in some areas, to increased lead levels in air, food, and water. Lead has been a constituent in house paint in some countries (2). Furthermore, the use of lead pipes in fresh water systems and the use of lead-glazed household ceramics may substantially increase lead intake (3).
The absorption of lead from the gastroi~ltestinal tract is about 10% in adults, but can be considerably higher in children (4). It increases when a person's diet is deficient in calcium and iron. In occupational settings, the main intake is through inhalation. Absorption through the respiratory tract is influenced by the size and solubility of the inhaled particles.
After absorption into blood plasma, lead is quicltly taken up by the red blood cells, which contain about 99% of the lead in whole blood (5). Thus far, lead in whole blood (B-Pb) has been the main indicator medium used for the biological ~nonitoring of humans. As an indicator of exposure and of internal dose, B-Pb is a mandatory requirement for the biological monitoring of lead-exposed worl<ers. This index has several advantages. Blood samples are easy to obtain, and the analysis is not partic-ularly complicated for experienced laboratories. The risk of contamination is negligible when evacuated tubes are used for sampling.
However, there are several limitations which must be kept in mind when B-Pb values are interpreted. First, there is a nonlinear relationship between B-Pb and lead exposure and uptake, irrespective of the route of uptake (6). Second, there is a nonlinear relationship between B-Pb, and lead in other indicator media, such as plasma, urine, and milk. Third, there is also a nonlinear relationship between different metabolic and toxic effects such as disturbed heme synthesis and B-Pb (6). From the nonlinear relationship it follows that B-Pb is a more sensitive index at low exposure levels as compared with high ones. 65 years and randornly selected from a larger population. Recently, the scope of the project has been widened to include the most essential and nonessential elements. In the course of the work by the Tracy group, tentative reference intervals have been presented, for example, for mercury (7) and chromium (8). A thorough description of the principles for the work of this expert group has recently been given by Vesterberg et a1 (9).
Several factors are of importance when reference intervals are evaluated for a certain element. According to the principles agreed upon by the Tracy group, each publication is graded and allocated into one of four categories given (zero to three stars) as to the quality of the sampling, analysis, and statistical treatment, including the presentation of the data. However, the grade given applies only to its usefulness in Tracy. Thus the paper evaluated could be of high quality from other points of view, even if it has been given a low grade in the Tracy evaluation (9).
Several factors are of great inlportance when reference values of B-Pb in adults are studied. In the following discussion, these factors are further explained and exemplified under dirrerent subheadings related to sampling and analysis and in the format proposed as an aid to the standardization of evaluation (9).

Sampling
Numerous factors can affect B-Pb. Thus the country and district and grade of urbanization must be specified. In urban areas, B-Pb is usually higher than in rural ones due to pollution from industries and traffic (4). The year of sampling can be related to the use of leaded or unleaded gasoline in a region.
B-Pb is affected by factors such as age, gender, and ethnic origin. Levels in men are usually somewhat higher than in women (10). B-Pb also norn~ally increases with age (1 1,12). For proper statistical treatment of the data, it is preferable for the sample size to exceed about 40 subjects (9).
Lead intake from food and water has a great impact on B-Pb in the general population (4,13). Alcoholic beverages (eg, wine) may contain lead. The health status of the study population may also be of importance for B-Pb, since some herbal medicines, taken orally in some Asian countries, may contain lead.
Furthermore, smoking may be of importance, as tobacco contains lead. The lead content in a cigarette is 3-12 pg, and it has been estimated that about 2% of this amount is inhaled by the smoker (13). Smoking may also have an indirect effect by negatively affecting the mucociliary escalator.
ICnowledge about possible occupational exposure is important as any such exposure is an exclusion criterion for blood lead reference values. Lead may be transferred to the home from lead-contaminated workplaces and give rise to elevated B-Pb in household members who are not occupationally exposed (14). Certain hobbies, for example, indoor shooting, which often takes place during the winter period, may lead to raised B-Pb concentrations (1 5). Other hobbies that must be taken into consideratio~i include tin soldier molding (16), ceranlic work with leadcontaining glazes (17), frequent long-distance jogging or cycling in areas with high environmental lead exposure (18), and n~otor sports connected to worlc with exhaust systems.
Procedures for contamination control and the transport and storage of sanlples before a~ialysis are highly relevant for proper evaluation (19). Contamination can occur at every step from the sampling to the analysis. Contamination sources such as the subject's skin, sample containers, additives, reagents, and the analytical equipment must be considered (20). Even minor contamination from the slcin or sanlpling equipment can lead to gross overestimation of the true B-Pb (21). Venous blood sampling from the antecubital vein is preferred for trace ele~nent analysis as compared with capillary blood sampling from the fingertip or ear.
Long-term storage of tissue samples, including reference materials, may lead to gradually decreasing metal concentrations. Therefore, regular checlting for possible losses has to be undertalcen. The storage temperature is i~nportant in this context. Thus a decrease in B-Pb by about 20% in samples stored at -20°C for four years has been claimed (22).

Analysis
A thorough description of the analytical method used to determine B-Pb is important for a proper interpretation of the results. The term "quality assurance" refers to all steps undertaken to ensure that data are reliable; it includes the collection, transport and storage of samples, the laboratory analysis, and the recording, reporting and interpretation of the results. It also includes training and management designed to irnprove the reliability of the lneasurelnents (22). By keeping all sources of ersor on the lowest possible level, the total sum of errors for a data set may fall within previously established narrow acceptance limits. Some of these sources of errors may be discovered by means of external and internal quality control procedures using certified reference materials, interlaboratory comparisons, or an analysis of the same material by two independent analytical techniques (9).
The quality of the analytical results depends on the precision and accuracy of the method. The detection limit of the method also has to be clearly below the lowest concentrations that have to be determined.
Reference samples, or what is called external quality control samples, are used to test the accuracy of the results. These samples are available from several organizations, for example, NIST (National Institute of Stand-ards and Technology, Gaithersburg, Maryland, USA). The reference samples have to be of the same type or matrix and in the same concentration range as the measured samples.
It is important that reference samples be stable over long time periods and also at different temperature ranges. The results may also be checked by regularly including internal quality control samples with known B-Pb concentrations.

Data treatment
Information about the underlying distribution of the data set (normal, log-normal or skewed), arithmetic or geometric mean, standard deviation or geometric standard deviation, or median value and range is highly relevant for proper evaluation of the values presented (9). The regression method has been used in the UNEPIWHO study on the assessment of human exposure to lead and cadmium (23,24). This statistical method can discover syste~natic errors in the analytical process.
It is important that laboratories participating in quality assessment schemes present data on current method control, acceptance, and rejection criteria, together with the results of the B-Pb determinations. It is not enough to refer to a particular method that has produced satisfactory results in an emlier study, perhaps at another laboratory several years ago (22). It is also important that reference samples be analyzed in parallel with the study samples to avoid problems with time trends.

Material and methods
Over 1000 papers published during the period 1980 to 1994 and presenting B-Pb values of occupationally unexposed subjects have been identified using the Medline data base and personal contacts in the field of trace element analysis. Many of these papers refer to studies carried out in lead-contaminated environments, as, for example, in the vicinity of lead, zinc and copper smelters, and they include children with reference to, for example, their mental development. These studies show a wide range of B-Pb values, particularly values related to time and sampling area.
The evaluation and grading of publications on B-Pb values reported from studies on occupationally unexposed adult populations was performed by three independent investigators (LG, GK and AS). The publications were graded and categorized according to the Tracy criteria for sampling, analysis, and data treatment. With regard to their relevance for B-Pb levels these criteria have been listed and described in the preceding sections. For the Tracy project, evaluators are required to use standard scoring procedures, to dichotomize the evalua-tion criteria into sampling (S) and analytical (A) categories, and to express the overall rating in terins of stars as indicators of the usefulness of publications for Tracy, where 0 indicates unsuitable, 1 acceptable, 2 good quality, and 3 excellent. It was stressed at the Tracy expert meeting in May 1992 that the rating could be of high quality from other points of view even if only a few stars are allocated for Tracy (9).
In this survey, 10 papcrs have been selected as exainples of the criteria required for B-Pb reference values. The selected papers have been given grades of 1 and 1 or higher for sampling and analysis, respectively, and are presented in table 1

Results
In a global study of 200 school teachers (occupationally unexposed) from an urban area in each of 10 participating countries, the geometric mean B-Pb levels varied from 60 pg . I-' in Tokyo, Japan, to 225 pg . 1-I in Mexico City (24). In this study, the geometric mean B-Pb level for nonsmoking men in Tokyo was 65 pg .   (28) showed a progressive downward trend for the B-Pb concentration over the period 1984-1987 by 4-5% per year in all age groups, all social classes, and all categories of smoking and drinking habits, age of dwelling, and length of residence. Grandjean et al (29) examined the B-Pb of I00 Danish men and 100 women in relation to age, smoking, and alcohol intake. In both groups, the B-Pb increased with age, but for the women the tendency was also related to menstruation status, with a mean concentration of 33 pg .I-' in premenopausal woinen compared with 62 pg . 1-I in postmenopausal women. For the total material, a significant increase in B-Pb was related to alcohol consumption and to cigarette smoking. In this study, a higher B-Pb was seen for urban residents than for rural ones. Multiple regression analysis showed alcohol consumption to be a significant predictor of the B-Pb level of men, while lnenstruation status and cigarette smoking were significant predictors for woinen.
In a study of the general population in Belgium (30), the B-Pb increased with age and was higher for the men than for the woman. For both genders, the B-Pb increased with smoking and alcohol consumption. In this study, the B-Pb showed regional differences and was lower in rural areas than in urban ones. A 10-fold increase in B-Pb was associated with a reduction of 10-13 ml . min-I in creatinine clearance. It was concluded that, while lead exposure may impair renal function, the alternative hypothesis that renal impairment may lead to an increase in B-Pb could not be excluded. The study illustrated the importance of ascertaining health status when B-Pb reference levels are being determined.
Similm findings, with higher B-Pb levels in urban than in rural areas, have also been reported by Orlando et a1 (18) in a study of noncompetitive runners in northern Italy. The mean B-Pb of a group of 28 runners, who trained in a large town (Genoa) with heavy traffic and high atmospheric lead levels, was 259 pg . I-'. Lower values, mean 205 pg . I-', were found for a group of 10 runners from the same town, who trained mostly in a rural environment, and even lower values were found in a comparable group of 182 nonrunners, mean 95 pg . 1-I.
The impact of industrial emissions was exemplified in a study from Taiwan by Chao et a1 (14). In this study, the B-Pb level of 41 workers (both genders) in a forging factory next to a lead recycling plant was significantly higher (mean 204 pg . I-') than that of a group of 51 workers (mean 59 pg . 1-I) from another forging factory, situated about 20 km from the lead plant. At the reference factory, the B-Pb level was not related to gender. As potential lead sources, such as household activities and transportation, did not affect the B-Pb at the forging factory next to the lead recycling plant, the major contribution would have been airborne emission from the latter. To reiterate, this study has been included to emphasize that it is necessary to investigate carefully the possibility of the presence of any lead emission source that may affect the B-Pb of the persons intended for inclusion in reference groups. In many studies, reference groups are selected from the same workplace as the workers under investigation. Since they may have neighborhood exposure, they would be inappropriate as referents.
In a recently published comprehensive study froin Taiwan (31), B-Pb was associated with several personal characteristics. The B-Pb concentration was higher for the men than the women, and it increased with age, alcohol consumption and smoking, as well as with the degree of urbanization. On the other hand, B-Pb was negatively associated with level of education and residential distance fro111 roads. A clear difference was observed for the B-Pb of the three ethnic groups in Taiwan. The highest concentrations were found for members of the Halcka group, followed by the Mainlanders and Fukien-Taiwanese (values for nonsmolters given in table I). These three ethnic groups have different food habits, which may explain the observed difference. Haklta Chinese, for example, typically consume salted vegetables prepared in lead glazed containers. Use of salt and acidic conditions may cause leakage of lead from such containers and therefore probably increases the peroral intake in this population. In this study, B-Pb was also influenced by the intake of water. Persons drinking bottled water, well water, or spring water had higher levels than persons drinking tap water. This study indicates that ground water, especially in the southern part of Taiwan, may have been polluted, for example, by industrial emissions of lead.

Discussion
The studies sumlnarized in table 1 illustrate the outcome when criteria specified by the Tracy group as being a necessary requirement for reference levels of B-Pb (9) are applied to several reports on this subject. While in all the scrutinized studies an arithmetic or geometric mean or median and standard deviation or range were presented, the distribution type was in most cases not adequately reported.
At the ~neeting of the Tracy group held in Nice, in 1993, it was agreed that for tentative reference values for B-Pb, two essential criteria with regard to the analytical procedure were participation in an interlaboratory cornparison program and the use of certified reference materials of the same matrix and of a similar concentration range as the real samples. For the time being, however, no certified reference blood samples with certified lead conce~ltrations at "normal levels" are available. Accordingly, these two criteria could not be fulfilled.
Nowadays, no absolute analytical method is available that guarantees accurate results under all circumstances. Without information about the use of proper quality assurance programs, the quality of the data can be questionable even if the concentration ranges presented fall within an interval that seems reasonable.
Because B-Pb is influenced by numerous factors, varying with the place and time of sampling, it is not possible to present general international reference values for B-Pb. Due to geographic and ethnic differences, such values are area specific, as is exemplified in this survey. Sometimes these values may be valid for a minor geographic region only and sometimes for a larger region, such as a country. As the lead exposure sources are constantly changing, for example, the decreasing use of leaded gasoline in many countries, reference values are also restricted to certain time periods. In southern Sweden, the B-Pb (geometric mean) of schoolchildren has decreased from about 60 p g . 1-' in 1978 to about 25 pg . 1-I during a period of about 15 years (1,32). The change from leaded to unleaded gasoline has probably played a dominant role in this context. During the same period, however, the use of lead-soldered cans and leadglazed ceramics has also decreased in Sweden. A parallel decrease of airborne lead exposure from other countries is another factor that can have influenced the blood lead levels.
The studies presented in this survey represent only a very small part of all recent papers on B-Pb in reference populations. However, they exemplify some of the factors that are important when trying to define reference values for B-Pb in reference populations. In the future, evaluation criteria such as those adopted by the Tracy group must be more generally used. Scientific journals have a responsibility to encourage authors to increase the quality of published data in this direction.