The main objective of this program was to offer to all participant laboratories means to evaluate their technical performances on determination of blood lead and foster technical cooperation among Latin American Countries, thus facilitating the establishment of a regional intercalibration network on occupational and environmental health. QA/QC activities are essential to assure the quality of results in chemical analysis since these activities allow, when regularly carried out, identification of errors and their possible causes, contributing for improvement of the analytical procedures used.

This Program was initiated in 1994, when laboratories of several Latin American Countries were invited by PAHO to participate in a workshop at FIOCRUZ, Rio de Janeiro, Brazil in order to discuss the basic ideas, the main subjects and implementation of such Program. It was agreed that lead and mercury should be the initial analyses to be studied since they are important pollutants in Latin America.

Due to its unique properties, lead has been used largely in industry (battery factories, construction material, pigments, alloys, bearing metal, etc) and up to a few years ago, as a gasoline additive, representing a serious risk for the environment and for human health. The effects of lead exposure on human health are very well described in the scientific and medical literature.

Currently, it is very well known that lead is most harmful to children under 6 years of age, where it can cause developmental delays, behavioral disorders and speech impairment, mainly. Blood lead level is the most useful screening and diagnostic test for lead exposure. Some data reveal that in the USA the level of Pb in general population dropped sharply from 1976 to 1991 after several measures to avoid exposure has been implemented. According to the National Health and Nutrition Examination Survey (NHANES), the mean of Pb-blood is 28 m g/L. The consensus level of concern for children is 100 to 140 m g/L although some health effects has been reported at levels as low as 40 m g/L. According to CDC, medical evaluation and environmental investigation and remediation should be done for all children with blood lead levels equal or greater than 200 m g/L. Medical treatment may be necessary in children if lead concentration in blood is higher than 450 m g/L. For occupationally exposed adults, when the blood level is higher than 500 m g/L (this value can change slightly among countries), medical removal from exposure is mandatory.

Levels of blood lead much higher than 600 m g/L have been identified in a significative number of Brazilian workers (> 50% in workers from small battery factories). Significative levels of environmental contamination have also been found associated with this industrial use in Rio de Janeiro urban area).

A part of its industrial, laboratorial/medical uses, mercury is an important pollutant mainly for the Amazon region where it has been extensively used for goldmining purposes. It has been estimated that from 40 to 160 ton of mercury/year have been introduced into the amazonic environment during this gold rush. As expected, it is a significative cause of contamination of the environment, food and humans in this region. High levels of mercury have been reported in sediments, water, air, food chain and humans.

For technical reasons lead was chosen to initiate this Program.

At this time, Dr Daniel Marcuello, coordinator of the Programa Interlaboratorios de Control de Calidad (Gabinete de Seguridad e Higiene en el Trabajo, Spain), was invited to present their experience in conducting a similar program and it was discussed the possibility of supply the necessary samples for this exercise. This agreement was later confirmed but for several reasons it never happened creating great difficulties in achieving our objectives.

In parallel, as the number of participants were considered small, we decided increase this number inviting and encouraging other laboratories to participate. New identifications and registration (document 1 in the Annex) were made and finally 23 laboratories from 9 Latin American countries, took part in this Program. From the questionaires (Annex 1) answered by the participants, we found that:

5 of them analises just biological material (blood, urine, etc); 2 of them are specialised on enviromental analysis and 16 can perform both types. Just 1 laboratory reported do not have any kind of internal quality control and 3 do not participate in any intercalibration program.

Except the laboratory 106, all the others use atomic absorption as analytical tool for metal determination: 11 reported the use of flame (labs: 101, 103, 105 ,110, 111, 112, 119, 122,123, 124, 125); labs 102, 108, 109, 113, 116, 118, 120 and 126 can use both, flame or graphite furnace. The laboratory 106 uses spectrophotometric method (dithizon method). Its result was not acceptable, according the requirements used in this work. The laboratory 126 sent us results obtained by both methodologies, being reported as (126/1 and 126/2).

The implementation of analytical activities occurred in 1999 after identification of a new supplier for the necessary samples (Laboratoire de Toxicologie, Institute national de sante publique du Quebec) and redefinition of our schedule. In September/99 a first set of samples was sent to the participants. The results were evaluated and in March/00 the second round took place. In both rounds, each laboratory received 2 samples for analysis. Together with the samples for the second round (in some cases even before it) 2 reference samples were sent to the participants to evaluate their analytical methodology and start internal QC programs.

Several professionals had important participation in this Program, such as:

BSc Ana Cristina Simões (CESTEH/ENSP/FIOCRUZ)
MSc Ana Rosa Moreno (ECO/PAHO, Mexico)
MSc Armando Meyer (CESTEH/ENSP/FIOCRUZ)
MSc Claudio Bustamante de Sá (CESTEH/ENSP/FIOCRUZ)
Dr Jean-Philippe Weber (Laboratoire de Toxicologie, INSPQ, Canada)
Dr Joachim Feltes (GTZ, Germany)
Dr Josino Costa Moreira (CESTEH/ENSP/FIOCRUZ)
MSc Maria de Fatima Ramos Moreira (CESTEH/ENSP/FIOCRUZ)
Dr Maria Luiza Sparza (CEPIS/PAHO, Peru)

Laboratories participants on the PAHO/GTZ QA/QC Program on Blood Lead

All samples used in this Program were supplied by the Laboratoire de Toxicologie du Quebec, Canada.

Two samples used in each round were sent directly to the participants by means of the Federal Express.

The reported results (average) shown in Table 1.

A simple comparison of these resuts with those from Quebec shows that in the first round (samples 101 and 102) both means were higher in "our" study. On the second round, the agreeement of means obtained from our results and the target values was almost complete indicating a much better performance of the participant laboratories (it is important to take in mind that there ais small difference on the participants in both rounds which can limitate but not invalidate this comparison).

Graphic representaions of these values are shown on the Annex 3.

The first step of the statistical analysis used to evaluate the data was the elimination the outliers. Although our initial idea was to use Cochran's and Grubbs' tests in this study, it was not possible since some the laboratories haven't attended our request, i.e., send us 3 results for each sample.

So, this step was carried out by Dixon's test, as folows:

X₃ – X₁/X_(n-2) – X₁ or X_(n) – X_{(n –2)}/X_(n) – X₃
Critical values for
n = 20 0,489; n =19 0,501; n =18 0,514

Dixon's test is a bilateral one. It is based on the normal distribution curve and rejects outliers.

For a group of data X(H), H = 1,2......, h, organised in crescent order, this test uses the following criteria:

        H                    if X_h is suspect                 if X₁ is suspect
   3£ h £ 7            (X_h - X_h-1)/(X_h - X₁)         (X₂ - X₁)/(X_h - X₁)
  8£ h £ 10           (X_h - X_h-1)/(X_h - X₂)         (X₂ - X₁)/(X_h-1 - X₁)
11 £ h £ 13          (X_h - X_h-2)/(X_h - X2)         (X3 - X₁)/(X_h-1 - X₁)
14 £ h £ 40          (X_h - X_h-2)/(X_h - X₃)         (X₂ - X₁)/(X_{h- 2} - X₁)

The performance of the participants were evaluated according internationally accepted procedures (Score "Z" and Normalised Youden plot).

The Score "Z" was calculated as follows:

Z = (x_i - x_ref)/ s_L

Where: x_i represents the average value of each laboratorie

X _ref = target value
S_L = standard deviation, estimated as: [S (x_i - x_average)² /n - 1]^1/2.

As comparison two values of S_L were used: one given by Quebec and the second estimated from from the interlaboratory set of results after elimination of the outliers ("our" results).

It is the first step to evaluate how far is a given result from the reference value: X₁ - X_ref.

An analytical system is considered to be "under control" if "z" presents a normal distribution with an mean value of zero and standard deviation equal to 1.Under these conditions, a value ½ z ½ > 3 should be very rare and indicates an non satisfactory result.. The great majority of results should have ½ z ½ < 2. So,

½ z ½ £ 2 - satisfactory 2 < ½ z ½ < 3 - questionable ½ z ½ ³ 3 - unsatisfactory 

Calculated "Z" Scores:  

"Z" Score

Graphical representation of these calculated values are shown in Figures 1 to 8.

Figure 1: Results of the Sample 101 (first round) estimated from " our" results.

Figure 3: Results of the Sample 104 (second round) estimated from " our" results 

Figure 4: Results of the Sample 103 (second round) estimated from " our" results 

Figure 5: Results of the Sample 101 (first round) estimated from Quebec´s results

Figure 6: Results of the Sample 102 (first round) estimated from Quebec´s results 

Figure 7: Results of the Sample 104 (second round) estimated from Quebec´s results

Figure 8: Results of the Sample 103 (second round) estimated from Quebec´s results 

Normalised Youden plots

(WHO, EuroReports "Quality Assessment in Health Laboratories" , 1981).

The coordinates used in the normalized Youden plot were calculated for samples A and B, as follows:

x_Ai - x_A/ s_A and x_Bi - x_B/s_B

Evaluation criteria:

a) Points within 3s circle: (permissible interval) subject to random errors.
b) Points within tangents but outside the circle: systematic errors exist for these results.
c) Points near the 45° line and within the circle: results very precise.
d) Points near the 45° line but outside the circle: results are precise but subject to constant systematic errors;
e) Points outside tangents: gross errors.

Figures 9 to 12 shows the comparison of these results. As a comparison two different treatments were made: the first, took into consideration the standard deviation reported by Quebec and the second that found considering the results of this study (represented as "plot according to laboratories).

 Fig 11 - Normalized Youden plot calculated taking into account the results from the participants

Fig 12 - Normalized Youden plot calculated taking into account the results from Quebec 

It is clearly seen that, as the standard deviations reported by Quebec were smaller than those obteined from our results, the "permissible interval" is smaller when used Quebec's values. In this case, many laboratoires situated inside the permissible interval in our studies are outside it.

" Overall Comments and suggestions"

This Program was initially organized to take place in 4 rounds which would have made possible a closer evaluation of the participant performances, providing enough time to intervene in those with non-satisfactory performances. By some practical reasons, it was changed and carried out in 2 rounds decreasing considerably the time for this kind of help. Eventhough, 2 reference samples were sent to those laboratories and a comparison of the results of both rounds shows an improvement on the general performance.

It is also clear that some laboratories have not implemented any kind of internal quality control program. This is essential to keep the quality of results under control and should be encouraged. Some participants for which results were not so accurate on the first round did not participate on the second one, showing some "fear" of having their results evaluated. This behavior must be changed. It is also very important also to make possible/facilitate comparisons between results produced in different Countries of this Region.

Comparing the results obtained from this group of Laboratories (called our results) with those obtained by the Laboratoire de Toxicologie du Quebec, the following considerations should be taken into account: On the first round, the means calculated from "our results" are higher, but still inside the acceptable values used by Quebec; on the second round these differences were negligible. Also the standard deviation was better on the second round of samples.

The standard deviations are also much higher than those presented by Quebec, showing a greater dispersion of "our results". As a consequence, the Youden plot prepared using the Quebec data is better for understanding the quality of results produced by the participants. It shows much more clearly "clusters"of laboratories outside the acceptable range.

Statistical tests used in this exercise showed systematic and other errors in some results, reinforcing the necessity of an internal quality assurance program.

Taking into consideration the Proficiency Testing Program of the US Clinical Laboratories Improvements Amendments (1988) which establishes that the Laboratory result must be within +/- 40 m g/L of the target value below 400 m g/L and within +/- 10% above that level, the performance of great part of the participants (45% in the first round and about 70% on the second round produced acceptable results for both samples) were satisfactory, but results produced in other laboratories are still non satisfactory (35 % in the first round and 12% in the second). The others produced changeable results (acceptable and non acceptable) showing the necessity of urgent implementation of quality control programs.

A part of the changeable and non satisfactory results, some facts, such as results presented in unities different of that required, etc, shows the lack of Quality Assurance System which must be promoted in this region even for those laboratories which prodecued acceptable ones. It is important to point out that this System involves the overall Institution (administrative and technical aspects). On the analytical point of view, it requires trained professionals, adequate installations and calibration/ maintenance of equipments, waste disposal, use of validated analytical methodologies and internal and external quality control programs.

The results obtaine in tis Program shows that it has contributed for improvement and integration of all participants, and its continuation is important. Two suggestions can be formulated:

1. it should be maintained, initially, to promote the implementation of internal quality control in Latin American laboratories. In this case CESTEH's laboratory can prepare a pool of blood and distribute it among some selected participant laboratories considered for analysis (selection will be based on their performances in this exercise). The results will be collected and evaluated statistically and part of this pool will be sent to all laboratories as secondary standard. For those, with difficulties to obtain acceptable results new analytical methodologies and training opportunities will be offered.
2. Increase the number of analytes, adding Hg and Cd to this program. Mercury exposure is a potential problem for many countries in the region due to its use in gold production, and the consequent environmental pollution. For mercury analysis, analysis using a cold vapor system is preferable. Mercury in blood presents an additional difficulty: the presence of two distinct species of mercury, organic and inorganic. In the case of organic mercury exposure, usually through dietary sources (fish and seafood), it is important for the laboratory to be able to distinguish the mercury species. Cadmium may also be a problem as it is found in the environment associated with zinc. Techniques for measuring cadmium in blood or urine are very similar to those used for lead, i.e. graphite furnace atomic absorption spectrophotometry. Thus, laboratories possessing this equipment should be able to measure cadmium as well. However, because cadmium levels are lower, contamination control is more important.

Costs of extending the program to mercury and cadmium

Assuming that 30 laboratories participate in 4 exchanges per year for a duration of two years, that two samples each for cadmium and mercury in urine and that samples would be prepared by and sent from the Laboratoire de toxicology/INSPQ, Canada by FedEx to participants, the cost could be as follows:

Cost of samples : Because the CTQ is part of a PAHO/WHO collaborating center for environmental health, samples could be supplied to participants at production costs, i.e. $8 per sample. The cost per shipment per laboratory would therefore be $ 32.

Cost of shipping: Assuming that FedEx is used as the carrier (no other reliable alternative appears feasible presently), the average cost per shipment would be around $40. It is hoped that slightly better rates can be negotiated however.

Total costs: The total cost per laboratory per shipment would be $72, thus an annual cost of $2160 for 30 laboratories or a total of $4320 for 2 years.