  |
|
 |
|
FDA Home Page | Federal-State | Import Program | Compliance | Inspection | Science | ORA Search  | ||
RESEARCH GUIDELINES
10.1 INTRODUCTION 10.2 SELECTING THE PROBLEM 10.3 PREPARATION OF RESEARCH PROPOSAL 10.3.1 Defining the Problem 10.3.2 Preliminary Literature Review 10.3.3 Plan of Work 10.3.4 Budget 10.4 LITERATURE REVIEW 10.4.1 Literature Search Records 10.4.2 Computer-Assisted Literature Searching 10.4.2.1 System Use Charges 10.4.2.2 Accessing DIALOG 10.4.2.3 Search Definition 10.5 EXPERIMENTAL PLAN 10.5.1 Experiment Outline 10.5.2 Designing Experiments 10.5.2.1 Variables 10.5.2.2 Sampling 10.5.3 Execution of Experiments 10.5.4 Evaluating Experimental Data 10.5.4.1 Statistical Analysis 10.5.5 Method Validation 10.5.6 Research Notebook 10.6 PUBLISHING GUIDANCE 10.6.1 General Format 10.6.2 Organization of Manuscript 10.6.2.1 Title 10.6.2.2 Author(s) 10.6.2.3 Abstract 10.6.2.4 Text 10.6.2.5 Acknowledgments 10.6.2.6 References 10.7 BIBLIOGRAPHY
Exhibits
EXHIBIT 10-A Research Project Evaluator
EXHIBIT 10-B Evaluation Factors
EXHIBIT 10-C Research Project Record
Back of Exhibit 10-C
This chapter provides general guidelines that the analyst should follow when conducting research. Topics include selecting the research problem, preparing the proposal, searching the literature, planning the experiment, good laboratory procedures for conducting experiments, evaluating experimental data, and publishing results. References appear at the end of this chapter; the researcher should study them and follow their suggestions.
Research may be defined as a systematic and comprehensive investigation of a phenomenon or problem to seek an explanation or solution. Generally, research performed in FDA field laboratories is directed toward the development of analytical methods or procedures that can be used to obtain accurate and precise values in the evaluation of regulated products. The analyst should keep this in mind when submitting a project for consideration. Each year, the Centers publish a list of research topics that they wish investigated. This list is distributed to field laboratories when the Division of Field Science issues the call for research proposals for the next fiscal year. This is a good source of ideas, either as listed, or as a basis for a similar project. The researcher should also confer with headquarters and field personnel to discuss analytical problems that have arisen or new analytes or products that FDA is considering for regulation. Keeping abreast of the scientific literature can provide new approaches for determining FDA-regulated analytes.
Research can be specific to an analyte, an analyte-product combination, an analyte-class and product-class combination, or can address generalized analytical procedures. Regardless of the category, success in applied research depends on the following principles:
* Clearly defining the problem
* Understanding the current state-of-the-art procedures in the field
under consideration
* Designing necessary and sufficient tests
* Drawing correct inferences from the data
10.3 Preparation of Research Proposal
The merit of any research idea must be proved to others. For the Office of Regulatory Affairs (ORA), this is the Research and Technology Committee in conjunction with the Centers. Exhibits 10-A and 10-B show the evaluation factors used by the Committee to rank research proposals. These factors should be kept in mind when considering what information to include. The proposal's quality will play an important part in its rating. A poorly described project might not be considered at all and certainly cannot be considered fairly.
A well-written proposal not only will supply information about the research objectives but will give solid reasons for its importance to FDA. Proposals should be complete and self-explanatory so that a reviewer with no previous knowledge of the research field can judge the project's merit fairly.
FDA form 1609 is essentially a cover sheet, so one or more continuation pages may be necessary to present a complete proposal. There is no limit on the number of additional pages, but they should be organized and relevant. If a project is a continuation of work already in progress, the proposal should contain a progress report to show that previous time was well used and to indicate the prospects for success. Because item 10 of FDA form 1609, Summary of Proposed Work, contains the main body of the proposal, it will be elaborated in the following sections. Exhibit 10-C is an example of a completed FDA form 1609 research proposal. For SARAP research proposals, refer to the current SARAP Guidelinesfor instructions and examples.
Defining the research problem requires separating or differentiating the specific problem from other problems in the field of interest. A clear, concise statement of the problem is necessary to its solution. The definition should specify the purpose of the research, that is, what problem it is intended to solve, and its significance to the accomplishment of FDA's mission and other related work.
An exact definition aids in the literature search by narrowing the search area. It also helps to convince the review committee that the project has a satisfactory "cost/benefit" ratio and can be completed satisfactorily in a reasonable length of time.
10.3.2 Preliminary Literature Review
Refer to section 10.4 for detailed information on conducting a literature review. The preliminary review should cover the research area in sufficient depth to give the researcher a good grasp of the state of the art in the problem area and should provide enough information to design a research protocol.
The research protocol should contain sufficient detail to convince a reviewer that the researcher understands the problem and the steps necessary to solve it. The protocol should include an explanation of the state of the art in the problem area, other researchers' approaches to solving similar problems, and reasons for selecting the specified approach and rejecting others. Related research done by others should be described and their publications cited. The steps for solving the problem should be outlined, including plans for any collaborative studies and publication.
To ensure that adequate funds are allocated to purchase equipment and supplies to conduct research experiments, the researcher should carefully consider and investigate all reagents, supplies, and equipment already available and identify what must be purchased. Catalogs and company representatives are good sources for exact prices. It is easier to include all expenses initially than to justify additional expenses later. Each budget item and its cost should be listed along with justification for its purchase.
There are two important reasons for undertaking a literature search prior to beginning research: The first is to ascertain if the problem under study has already been solved, and the second is to acquire an up-to-date understanding of progress in the relevant field to see how similar problems have been resolved.
The literature search should not end before the research paper has been published, if then. Every new issue of relevant journals available to the researcher should be scanned for information that may contribute to the project or that must be acknowledged in the final paper. When the researcher is assured that the research problem has not already been solved, proposed solutions can be developed and experimental work can begin. The literature search should continue but not delay action toward a practical solution.
Before starting the search, the researcher should decide exactly what information to seek. It is a good idea to write down the specific key words and general topics, and to review, correct, and update these periodically during the project. Key words are particularly important when performing a computerized literature search
The search should be conducted broadly and should be sufficiently comprehensive to fill any gaps in the researcher~s knowledge of the specific problem. For example, a literature review in conjunction with the development of an improved analytical method for a pharmaceutical drug should include a search for information on the drug itself (physical and chemical characteristics), all available alternative methods for its analysis, favored analytical methods and techniques for related drugs, and any criticisms of presently available methods of analysis.
Literature review includes reading as well as looking for articles of relevance. Each paper found will include a list of references. These constitute a useful source for additional information on the subject and should not be overlooked. The length of the literature search can vary according to the magnitude of the research project. For sample-analysis research, a short search may be all that is necessary. It will usually consist of the following:
* Compendial methods, latest editions
* Journal of AOAC INTERNATIONAL
* Laboratory Information Bulletins (LIB)
* Consultations with FDA personnel with expertise in the analytical area
For a research project that involves modification of an existing method, a search through the past 10 years of the journals will probably be required. Projects that entail development of an entirely new method or process will necessitate a comprehensive search, which should begin with a manual or computerized search of the chemical and biological abstracts. If the laboratory does not subscribe to abstracting services, then the researcher should use the library facilities of a local university.
10.4.1 Literature Search Records
The literature review is a personal activity, and the researcher should feel free to invent or adopt any procedures that meet these objectives: (I) recording information accurately and quickly, and (2) tracking which sources have been searched and which key words in those sources have been searched. If a photocopy or reprint of an article is obtainable, note-taking can be minimized.
Index cards are sometimes used to record notes during a literature review. Their advantage is that they can be rearranged; their disadvantages are that they can be easily lost, and that their small size encourages illegible, abbreviated note-taking.
The bound research notebook can also be used for collecting the literature review information, with index cards employed only when rearranging those references into the order in which they will be cited. Each notebook entry should include the name of the journal (or American Chemical Society abbreviation), volume~ page number(s), author(s), and complete title. For journals not readily available, the full author citation with the correct order of names should be recorded, instead of listing "Somebody, et al." This should be followed by whatever notes the researcher feels are important. If the entries are numbered at the outside edge of the notebook page, index cards can later be marked with the numbers of those articles to be cited in the References section, the cards rearranged in the order in which the references will appear in the paper, and the cards renumbered in that order. Using the cards as a guide and the notebook as the source of complete information, the References section can be written easily.
The first few pages of the notebook can be reserved for listing references taken from annual indexes and abstracting journals. As these references are checked, and either rejected or entered as a full-description item in the following pages, they can be marked off the list but still provide a record of sources searched and what was found.
10.4.2 Computer-Assisted Literature Searching
The scientific literature has become so extensive that conducting a thorough search without using a computer is almost impossible. Many FDA units now subscribe to DIALOG database searching service. DIALOG provides access to scientific databases such as Chemical Abstracts Service (CAS) and MEDLINE, as well as business and general-interest databases. DIALOG searching provides the researcher with a good starting reference list that should be used as a guide but not considered as a complete review. Each reference must be checked and read just as with a manual search. The expansiveness of the sources searched is the main advantage of using DIALOG.
DIALOG provides many manuals of instructions for using the search service. Because these instructions are so lengthy, no attempt is made to cover them in detail here. The researcher should consult these manuals for specific instructions. The following sections provide an overview of how the system works.
DIALOG's fee structure is based on the time spent on-line in the search and the amount of information printed. The user should be advised that designing search strategy while on-line can be a costly proposition. To reduce costs, the search strategy should be planned before accessing DIALOG.
The exact instructions for accessing the database service will depend on the modem and communication program installed on the researcher's computer. After the phone connection is made, a prompt will appear on the CRT asking for the name of the service desired. The user must enter "DIALOG." The prompt will then ask for the user number and password. The user must enter the number and password assigned to the user's particular work unit. When the prompt asks for the search definition, the user should enter the preplanned search statement.
The database search is based on words that occur in the title of the article or Definition key words listed by the authors or abstractors. If the word or word combination selected is not among these, references will be missed. However, the initial search statement should be as definitive as possible to prevent excessive "hits" (references containing words and/or word combinations in the search statement). If the number of hits is insufficient, the search statement can always be expanded to search broader categories.
Searching protocol is based on Boolean logic with the use of the connecting words AND, OR, NOT, and WITH and parentheses to define the priority of the search order. For example, the following statement would define the search order for the category "HPLC determination of sulfamethazine in feeds":HPLC AND SULFAMETHAZINE AND FEEDS
When the initial search is completed, it can be modified to further restrict or expand the search field, depending on the number of hits. When the number of hits is narrowed to less than 50 (as a rough guide), a printout of the titles of the articles should provide an idea of the suitability of the search statement and what modifications, if any, need to be made.
10.5 Experimental Plan A researcher should never start a course of experiments to solve a problem without a plan for the process. The plan will help prevent many false starts and unrewarding sidetracks. Prior to planning, the researcher should have a good basic understanding of the nature of the problem and any relevant theory associated with it. It is essential that any experiment be designed on the basis of one or more hypotheses.
The researcher should break down the problem to its simplest elements. This enables a researcher to answer separate questions rather than managing the whole complex problem. Also, obtaining a clear-cut idea in advance of what will be tested aids in designing experiments. A suggested question sequence to assist the researcher in planning the experiments and accomplishing the objectives of the project is as follows:
* Title?
* Objective?
* Background?
* Specific method?
* Anticipated journal of publication?
Questions of the type "Why am I doing this particular thing?" and "Will it really tell me what I want to know?" should preface the design of any experiment.
Experiments should be planned in detail in advance of execution to ensure that the necessary data will be obtained to solve the problem and to support any conclusions. The initial planning also will provide an overview of the experimental program. As the program progresses, it can be reevaluated with regard to the following questions:
* Should the experimental design be altered?
* What experimental series should be done next?
* Which experiments can be omitted because their type of data is no
longer needed?
In some cases, it is possible to set up a single experiment whose outcome largely determines the fate of a given hypothesis. However, it is rare that a single experiment is decisive. "Shotgun" experiments, although they appear promising in accelerating data gathering, often create the need for more experimentation and should be regarded as a risky approach to solving the problem.
A variable is a factor that can affect the outcome of an experiment. Background information on a research topic may suggest which variables can be considered controlling, that is, most likely to cause differences in analytical results. Variables can be divided into two types:
1. Those that can be regulated during the experiment (such as composition of an extracting solvent)
2. Those that cannot be regulated during the experiment (such as the amount of sugar in different types of apples)
The unregulated variables can be further classified as those whose values can be ascertained at the time of the experiment and those that must remain unknown. Most experiments are designed to show conditions resulting from the variation of regulated factors. It is a safe axiom to change only one variable at a time, although with mathematical analysis of the results, many new experimental design methods provide procedures for changing several variables simultaneously. Multivariant designs are more complicated but can maximize the experimental data obtained from each run and show the overall influence that the variables exert together. Reference (7) is an excellent source of information on multivariant analysis.
Whether the single-variant or multivariant method is used, one should always plan to run sufficient determinations for valid statistical evaluation.
A collection of things that have some defined quality in common is called a class. A portion of a class having common individual characteristics is called a sample of that class, and the process whereby the sample is selected is known as sampling. In every scientific investigation, sampling is one of the first problems to be solved. It is important that the method of selecting the sample not correlate with the attribute under study. A researcher with a vested interest can never be trusted to select the members of a sample for testing.
Bias will always occur. Experienced researchers are well aware of this difficulty and carefully select a sampling process to eliminate personal bias.
10.5.3 Execution of Experiments
After the experiments have been planned, the researcher must perform the actual "bench work." The success of the experiments will depend on the skill and efficiency of the researcher. The following are general suggestions regarding the use of experimental apparatus:
1. The researcher should understand the theory on which the apparatus operates. This will assist in understanding unexpected quirks in its operation as well as unusual data.
2. A checklist should be prepared before using an apparatus requiring a number of operations.
3. The equipment should be checked for proper operation. The researcher should make a practice of performing individual calibrations prior to experiments and at regular intervals.
4. Before an important experiment, no untested changes should be made in any part of the apparatus.
5. Trial runs should be made before crucial experiments to detect any failure in the apparatus.
10.5.4 Evaluating Experimental Data
In general, the scientific method implies a logical derivation of conclusions by means of inference correctly drawn from reliable data. Observations are meaningless until they are interpreted. The analysis of experimental data is, therefore, a critical stage in every research study. Valid conclusions are predicated only on reliable data and must be based on all the available evidence. Usually, the conclusions will not fit all the known facts. The researcher then has the difficult task of deciding whether error has caused these alleged facts or whether they were due to chance effects.
In the analysis of experimental data, unexpected difficulties often arise. Therefore, analysis should be performed immediately after each experiment so the research plan can be revised at once if needed. This will preclude the repetition of whole phases of experimentation or performance of unnecessary experiments. It is often useful to prepare a first draft of the paper for publication as soon as all planned experiments for the project are completed. This often points out missing data more readily than just trying to decipher firsthand what they are. Missing data can then be obtained before the researcher forgets the interpretation of the previous data.
Statistics need not be applied to all experimental designs nor must all data be interpreted statistically, but most methods intended to determine an analyte in a matrix will require some statistical evaluation. As stated in section 10.5.2 Designing Experiments, the researcher should plan the experiments to allow valid statistical interpretation. One should keep in mind that statistics provide a technique for extracting maximum amounts of information from experimental data, but the use of statistics will not improve poor, inconclusive data.
References (7) through (9) provide a thorough description of statistical methods, their application, and interpretation. The following concepts are basic to statistical studies:
l. The population is the set of things one has defined to study.
2. The sample is a representative part of the population. It is used to answer questions about the characteristics of the population. To give valid results, qualifications for selecting the sample from the population must be specified, for example, the size of the sample. A randomsample is one that is chosen in such a way that every possible sample of the same size is equally likely to be chosen. The main advantage of a random sample is that it can be used to derive mathematical conclusions regarding the population. It also helps eliminate bias, in that the researcher is not given the opportunity to select the sample most likely to support the proposed theory. The larger the random sample, the closer the sample average is to the true average of the original population. Another term used to describe this type of sample is representative.
3. An array is a rearrangement of the data to enable further study. Another way of describing a population is the analysis of numbers that may describe a set of data. These numbers are called statistical parameters.
4. The mean is the number that describes the center of distributionof results. The standard deviation and its square, the variance, measure the scatter of a set of data. The standard deviation and mean are important statistical parameters. The range and relative standard deviation are used in conjunction with the standard deviation. The range is defined as the difference between the largest and smallest numbers in a set of data. The relative standard deviation tells how far "on the average" the individual observations differ from the mean of the set.
5. Testing significance--In testing the hypothesis that a variable used in an experiment has a negligible effect on the result, the level of significance must be selected in advance. These tests let the researcher investigate the truth of a hypothesis made about the population, based on the sample selected. Tests of significance are defined in terms of probability. Some of the more common tests used are the t-test, analysis of variance, and chi-square.
When gathering information to support regulatory decisions, FDA does not use methods that have not been validated. The validation study can be as simple as a recovery test run on a single product, or as complex as an interlaboratory collaborative study on a family of related products. An analyst can make a significant contribution to the Agency by conducting a collaborative study on a method that has been developed and reported by someone else.
References (10) through (12) provide the guidelines for performing a collaborative study. Also refer to Chapter 9, section 9.5.3, for ORA policy regarding the conduct of collaborative studies.
As a guide, the following parameters should be checked, if applicable, for method validation:
* Linearity of detection response
* Minimum detection level O Recoveries at various analyte levels
* Method repeatability
* Freedom from interference in blank samples
Many journals will not accept papers for publication without the above validation checks.
The researcher should keep a notebook for each project and use it from the very beginning to record ideas, experimental data, and evaluations. The notebook should be sewn-bound, and entries should be made in ink. The major reason for the notebook is obvious: Neither the memory nor unorganized slips of paper are satisfactory for storing information accurately and completely. Because research notebooks can be treated as evidence in a challenge over the precedence of discovery, certain requirements are imposed to ensure the integrity of the record. These same requirements preserve the continuity and permanence of the researcher's records and reinforce the importance of working in an orderly and organized way.
The notebook is, in effect, the researcher's "worksheet." It is the place, and the only place, to record original data. As with any worksheet, the notebook must be orderly. Data from recorder charts should be reported, preferably in tabular form, whether or not the charts themselves are included in the notebook. Charts not included in the notebook should be kept in an easily accessible file. The plan and progress of the research pro ject must be clear from this record.
The notebook should be used as a planning book for the research project. A few pages near the beginning of the book should be used for recording the experimental outline. As the plan develops and the researcher has new ideas on how to proceed, these ideas and the research progress should be recorded in the notebook. Discussions about the project with others, such as the science advisor, supervisor, or other researchers, should be noted and any advice, instructions, or changes in approach should be recorded. Writing down plans and ideas often helps to clarify them and may prevent the loss of an important alternative approach.
A list of suggestions on the use of the research notebook follows:
1. A few pages at the beginning should be reserved for a table of contents. This can be filled in as important stages of the project are completed.
2. Writing should be legible so entries can be understood months after they are made. Unusual names (people, chemicals) that might be difficult to read should be printed.
3. Pages should be numbered if not preprinted.
4. Writing should be in ink. Writing on both sides of each page is acceptable as long as each side remains legible. Instead of erasing, the researcher should line through suspect data and errors and then date and initial them.
5. Charts should be included, and graphs and tables should be constructed wherever they will enhance or clarify information. Instrument charts can be filed separately, using a system that is clearly referenced to specific sections in the notebook, rather than pasted in the notebook.
6. Similar information should be recorded in a similar format so it will be easy to recognize.
7. The brand, source, grade, and treatment of chemical reagents, solvents, column materials, TLC plates, etc., should be recorded.
8. The make, model, and FDA number of equipment should be recorded to identify each specific instrument used.
9. All standardizations and calibrations should be recorded.
10. All experimental conditions should be recorded. This includes not only controlled conditions, such as column or water bath temperatures, but also environmental conditions that might influence analytical results. Room temperature and humidity are two examples.
11. Each experiment should include a statement of purpose and a conclusion of what was proved or disproved. This summary will aid the interpretation of experimental results for the project in its entirety.
When the experimental work is completed, a carefully kept notebook will quickly repay the time and trouble to prepare it. All the material necessary to write the research paper will be organized in the notebook. The remaining work will consist of arranging the material in proper format for the paper, selecting tables and charts to incorporate into the paper, and adding references. If a reviewer has a question or criticism about the project, the material for the response will likely be found in the notebook.
A method that has not been published is essentially nonexistent and cannot be used (or validated) except by the few people aware of it. For this reason, publication of analytical methods and procedures is encouraged. Publication in journals is often a slow process and can take up to a year before the paper appears in the journal. The LIB can be used as an initial publication to rapidly disseminate the information to FDA personnel. Although the LIB report may be an abbreviated version of the journal paper, it must still be carefully reviewed for technical accuracy, organization, completeness, and, in some cases, policy. This is a local responsibility and ordinarily does not prevent expeditious publication.
Scientific journals usually stipulate rules governing the format for papers. Because each journal's rules may be different, they should be studied carefully before beginning the manuscript preparation. Papers published previously in the journal serve as good examples. The format rules will generally cover four main categories as follows:
1. Form--Margin widths; line spacing; systems for numbering, titling, figures, and tables; style for reference citation; etc.
2. Organization--Order and relationship of the parts of the manuscript.
3. Referencing--Means used to summarize and find the contents, list of figures, tables, and indexes.
4. Style--Grammatical style preference such as voice, sentence structure,
10.6.2 Organization of Manuscript Journal manuscripts usually run from 2 to 20 printed pages, or 2,000 to 20,000 words. The manuscript is usually divided into title, name(s) and affiliation(s) of author(s), abstract, text, acknowledgments, and list of references. The psychology of the expected reader must be taken into account when preparing the article. Two concepts regarding this psychology must be considered: (l) capturing and holding the reader's interest, and (2) presenting adequate evidence to support conclusions derived from the analysis. The following arrangement of material is one that journals often use and provides guidance for accomplishing the aforementioned objectives.
The title should be carefully chosen. Readers usually decide whether or not to read the rest of the article on the basis of their impression from the title. Index and/or key words should be placed in the title. A title should not claim too much or too little.
The decision as to whose names to include in the paper and in what order requires fair-mindedness and objectivity. The senior author, whose name goes first, is usually the one who takes responsibility for the content and conclusions derived from experimental data. Many researchers make it a rule not to accept co-authorization unless they have gone through the paper in detail, checked the data, and are satisfied with its contents.
The abstract's purpose is to give the reader a description of the study in condensed form. After the title, the abstract exerts the next influence on the reader's decision to continue reading the article or not. Index and/or key words should be included in the abstract. The main objectives of the study and its results should be addressed but not in excessive detail.
The first few paragraphs should introduce the topic under study, orient the reader, and associate the new work with the current situation in the area of study. Sufficient references should always be given to allow the history of past work to be traced. The introduction should supplement the title and abstract.
Usually the next portion of text will consist of method used to conduct the study. It should include a complete description of the reagents, apparatus, equipment and procedures including any figures and tables needed in the application of the method. This description should enable the reader to conduct the same method exactly as written and obtain similar results.
The next section should discuss observations about the method, any cautions for conducting it, conclusions drawn from the experimental data, and advantages or disadvantages as compared to other procedures. Experimental data that support the conclusions should be included, preferably in tabular form. Where numerous measurements have been made it is usually preferable to give a statistical summary rather than individual values. Graphs often best present the relationship of variables to the response sought. Figures and/or photographs can also be used to present observations that numerical values alone would not adequately describe. Figures and graphs should be understandable without the reader having to refer to the text. All graphs, tables, and figures should be cited in the text to explain their relationship and significance, but textual discussions should not duplicate data shown in tables and figures.
It is extremely important that proper credit be given for assistance from other scientists or organizations. Their further cooperation and assistance on research projects can depend on remembering this courtesy.
Regardless of the style specified by the journal for presenting references, they should be double checked for accuracy. It is the responsibility of the author(s), not the journal, to ensure that the names and titles are cited correctly. The text of the article should also be reviewed to verify that the references are cited properly.
RESEARCH GUIDES
1. Bunge, M.A. "Scientific Research," Vol 3; Springer-Verlag: Berlin, 1967.
2. Weimer, W.B. "Notes on the Methodology of Scientific Research;" Halsted Press Division of Wiley: New York, 1979.
3. de Vleeschauwer, H.J. "Introduction to Scientific Research: A Guide for Students;" Pretoria:Pretoria, South Africa, 1965.
SCIENTIFIC WRITING
4. Tichy, H.J., Fourdrinier, S. "Effective Writing for Engineers, Managers, Scientists," 2nd ed.; Wiley Interscience: New York, 1988.
5. Harkins, C., Plung, D.L., (Eds.). "A Guide for Writing Better Technical Papers;" IEEE Press: New York, 1982.
6. Day, R.A. "How to Write and Publish a Scientific Paper," 3rd ed.; Oryx Press: Phoeniz, AZ, 1988.
EXPERIMENTAL DESIGN
7. Box, G.E.F, Hunter, W.G., Hunter, J.S. "Statistics for Experimenters: An Introduction to Design, Data Analysis, and Model Building;" John Wiley & Sons: New York, 1978.
8. Koopmans, L.H. "Introduction to Contemporary Statistical Methods," 2nd ed.; Duxbury Press: Boston, 1987.
9. Wernimont, G.T. "Use of Statistics to Develop and Evaluate Analytical Methods;" AOAC: Arlington, VA, 1985.
10. Youden, W.J., Steiner, E.H. "Statistical Manual of the AOAC;" AOAC: Arlington, VA, 1975.
11. Horwitz, W., et al. "Guidelines for Collaborative Study Procedure to Validate Characteristics of a Method of Analysis," J. Assoc. Off. Anal. Chem. 1988, 71(1), 161-72.
12. "Handbook for AOAC Members," 5th ed.; AOAC: Arlington, VA, 1982.
(continuation) SPECIATION OF ARSENIC IN FOODS & FOOD SUPPLEMENTS OBJECTIVE To answer the agency's need for information on the speciation of toxic contaminants in foods: develop a method for the separation and characterization of inorganic and organic arsenic compounds. SIGNIFICANCE Arsenic is widely used as a fungicide, herbicide and pesticide, and is produced as a by-product of many industrial activities. Organic and inorganic arsenic has been found in water, sediments, soils, marine life, animal feeds, vegetables, and other foods. As arsenic contamination is ubiquitous in the food chain, and as consumers are encouraged to consume more fish and dietary supplements, information about the speciation and toxicity of arsenic in these products becomes essential. Environmental pollution deposits all forms of arsenic, and it is well known that living organisms can convert arsenic [1]. Waters, soils and sediments contain predominately inorganic arsenic [2-6], but methylated species can be present [7]. Vegetable products contain inorganic and organic arsenicals [8], depending on specie and closeness to the soil. Fish contain mostly higher molecular weight organoarsenics [9], but the load of both inorganic and organic arsenicals varies widely by specie and location. Fresh water fish and bottom-eaters may contain more inorganic arsenic. Other marine organisms contain significant proportions of low molecular-weight arsenics [7] as does seaweed [10]. Arsenic toxicity varies by the molecular form [11)] Arsenite [As (III)] is highly toxic, more so than arsenate [As (V)). Inorganic arsenic is more toxic than organic forms, which roughly decrease in toxicity as the molecular weight increases. Forms commonly of interest for their toxicity are: As (III), As (V), methylarsonic acid [MAA], dimethylarsinic acid [DMA]. Other, less toxic, compounds of interest are: trimethylarsine oxide, phenylarsonic acid, phenylarsine oxide, arsenobetaine and arsenocholine. PLAN OF WORK Speciation of toxic metals is a very active field of investigation. The first step in this project should be a through literature review and summary. Use of the literature information database, DIALOG, is invaluable in summarizing and up-dating references. The Chicago Laboratory also has excellent access to university library collections. Selection of foods of interest in the next step which would be accomplished with the advice of the technical contact. Of possible interest are marine products including salmon and concentrated fish oil, and vegetable products grown in close soil contact such as rice and potatoes. (continuation) SPECIATION OF ARSENIC IN FOODS & FOOD SUPPLEMENTS The initial laboratory work should involve accurate and reproducible characterization of the total arsenic content of some foods of interest. Determination of total arsenic concentrations is well established. The major caution is to use a method which ensures that all the very stable, higher molecular weight arsenicals are totally decomposed. The overall method must be rigorously checked for reproducibility. Dry ash, wet ash and pressure digestions are documented. With a perchloric acid hood available to me, I prefer three-acid digestions, with modifications as needed for the samples' alkali metal content. Detection is generally by AAS hydride. AAS is applicable to ppm concentration levels or ppb with graphite furnace. Methodology exists for arsenic graphite furnace detection with and without hydride generation. (The Chicago Laboratory Facility has flame, hydride and graphite furnace Atomic Absorption and Direct Current Plasma (DCP) Emission spectrometers.) Speciation method development falls into three, interrelated stages: • Extraction, • Separation, • Detection. If initial samples are well characterized for total arsenic levels, and if the total-arsenic determination is reliable, then each stage can be verified by recovery analysis. In addition, GC/MS can be used for further validation [9]. Extraction from the sample matrix must not affect speciation and must work efficiently and reproducibly. Efficient extraction requires that all the arsenic is released from the matrix and that a minimum of matrix is transferred. Initial steps usually involved extracting the sample with a basic [7] or acidic [12] solution followed by organic [7-9, 12] solvent isolation of the arsenic species. Extractions can also be used to cleanup the sample. Additional cleanup steps may be needed for complex food samples. Lawrence, et al. [9] used an alumina column to clean fish extract. Separation of the arsenic species can be done by extraction schemes [8, 12] or chromatography [7, 9]. Selective generation of arsines is also possible [13] but generally not favored [3, 7, 10]. Again, use of recovery data is needed to establish the method. Initial speciation work should utilize standards, but behavior of the separation when food matrix is present is essential. Extraction schemes are a minority in the literature [3, 8, 12] but offer great flexibility. The more common methodology involves coupling LC separation with graphite furnace AA [4, 7-9, 14-18], ICP [10, 19-22] detection. Liquid chromatography columns used include cation exchange, anion exchange and reverse-phase LC separation will allow for a number of arsenic species to be separated, allowing discrimination between the more and less toxic species. (continuation) SPECIATION OF ARSENIC IN FOODS & FOOD SUPPLEMENTS Detection of speciated arsenic compounds is frequently done by hydride AA [4, 7, 8, 12], but other techniques are also used [10, 14-23]. Hydride-flame AA has the advantage of being widely accepted. It also has the disadvantages of only moderate sensitivity and possible hydride interferences. Depending on the levels of arsenic detected, this technique may be useful with in situ hydride generation of the LC effluent. However, I feel that the use of the graphite furnace is necessary for increased sensitivity. Graphite furnace technology has progressed greatly in recent years, such that methodology exists for the determination of arsenic without the need for hydride generation. Schlemmer and Weiz [24] have shown that arsenic responds well to the use of palladium and magnesium nitrate matrix modifiers. Another interesting possibility for detection is the use of the DCP. Since DCP nebulizer rates approximate HPLC flow rates, a direct interface would give a continuous chromatogram of the arsenic speciation. The advantages of continuous sampling, versus discontinuous for the graphite furnace, would make up for the decrease in sensitivity. Methods developed for the IC? should transfer to the DCP. In addition, spectral interferences seen in some ICP analysis may be minimized in DC? due to the highly accurate Eschelle grating. SUMMARY In conclusion, the following will be attempted to develop methodology for the speciation of arsenic: • Literature search • Selection of some foods of interest • Total Arsenic characterization Selection of Detection method, beginning with graphite furnace AA • Development of an extraction/cleanup stage, with verification by total arsenic determination LC speciation method, first using standards and then examining the effects of the food matrix. Optimization of best detection method to satisfy agency needs and abilities. PUBLICATION As the method is developed, communications within the agency will be used. Upon completion, the method and results will be published in an appropriate journal. BUDGET All necessary major equipment, (AAS, graphite furnace, hydride generator, DC? and GC/MS) is available at the Chicago District Laboratory Facility. For incidental reagent and column costs, $1000 is requested. (continuation) SPECIATION OF ARSENIC IN FOODS & FOOD SUPPLEMENTS 1. Marafante, E.; Vahter, M.; Dencker, L. "Metabolism of Arsenocholine in Mice, Rats and Rabbits" Sci. Total. Environ. (1984) , 34, 223-240. 2. Aggett, J.' Kriegman, M.R. "Preservation of Arsenic (III) and Arsenic (V) in Samples of Sediment Interstitial Water" Analyst (1987), 112, 153-157. 3. Chakraborti, D.; Irgolic, K.J.; Adams, F. "AAS Determination of Arsenite in Water Samples by Graphite Furnace After Extraction with Ammonium sec-Butyldithiophosphate" JAOAC (1984), 67, 277-284. 4. Ricci, G.R.; Shepard, L.S.; Colovos, G.; Hester, N.E. "Ion Chromatography with AAS Detection for Determination of Organic and Inorganic Arsenic Species" Anal. Chem. (1981), 53, 610-613. 5. Haswell, S.J.; O'Neill, P.; Bancroft, K.C.C. "Arsenic Speciation In Soil-pore Waters from Mineralized and Unmineralized Area of South-West England" Talanta (1985), 32, 69-72. 6. Takamatsu, T.; Aoki, H.; Yoshide, T. "Determination of Arsenate, Arsenite, Monomethylarsonate, and Dimethylarsinite in Soil Polluted with Arsenic" Soil Science (1982), 133, 239-246. 7. Maher, W.A. "Determination of Inorganic and Methylated Arsenic Species in Marine Organisms and Sediments" Anal. Chim. Acta (1981), 126, 157-165. 8. Pyles, RA.; Woolson, E.A. "Quantitation and Characterization of Arsenic Compounds in Vegetables Grown in Arsenic Treated Soil" J. Agric. Food Chem. (1982), 30, 866-870. 9. Lawrence, J.F.; Michalik, P.; Tam, G.; Conacher, H.B.S. "Identification of Arsenobetaine and Arsenocholine in Canadian Fish and Shellfish by HPLC with AA Detection and Confirmation by Fast Atom Bombardment Mass Spectrometry" J. Agric. Food Chem. (1986), 34, 315-319. 10. Morita, M.; Uehiro, T.; Fuwa, K. "Determination of Arsenic Compounds in Biological Samples by LC with ICAP-AES Detection" Anal. Chem. (1981), 53, 1806-1808. 11. Underwood, E.J. "Trace Elements in Human and Animal Nutrition," 3rd Edition; Academic Press: New York, 1971. 12. Munz, H.; Lorenzen, W. "Selective Determination of Inorganic and Organic Arsenic in Foods by AAS" Fresenius Z. Anal. Chem. (1984), 319, 395-398. 13. Anderson, R.K.; Thompson, M.; Culbard, E. "Selective Reduction of Arsenic Species by Continuous Hydride Generation Part 1. Reaction Media" Analyst (1986) , 111, 1143-1152. I (continuation) SPECIATION OF ARSENIC IN FOODS & FOOD SUPPLEMENTS 14. Woodson, E A.; Aharonson, N. "Separation and Detection of Arsenical Pesticide Residues and Some of Their Metabolites by HPLC - Graphite Furnace AAS" JAOAC (1980), 63, 523-528. 15. Iadevaia, R.; Aharonson, N.; Woodson, E.A. "Extraction and Cleanup of Soil Arsenical Residues for Analysis by HPLC - Graphite Furnace AA" JAOAC (1980), 63, 742-746. 16. Foa, V.,; Colombi, A.,; Maroni, M.; Buratti, M.; Calzaferri, G. "The Speciation of the Chemical Forms of Arsenic in the Biological Monitoring of Exposure to Inorganic Arsenic" Sci. Total. Environ. (1984) , 34, 241-259. 17. Fish, RH.; Brinckman, F.E.; Jewett1 K.L. "Fingerprinting Inorganic Arsenic and Organoarsenic Compounds in Situ Oil Shale Retort and Process Waters Using a Liquid Chromatograph Coupled with an Atomic Absorption Spectrometer as a Detector" Environ. Sd. Technol. (1982), 16, 174-179. 18. Brinckman, F.E.; Jewett, K.L.; Inverson, W. P.; Irgolic, K.J.; Ehrhardt, K.C.; Stockton, RA. "Graphite Furnace AA Spectrophotometers an Automated Element-specific Detector for HPLC: The Determination of Arsenite, Arsenate, Methylarsonic Acid and Dimethylarsinic Acid" J. Chromatogr. (1980), 191, 31-46. 19. Spall, W.D.; Lynn, J.G.; Anderson, J.L.; Valdez, J.G.; Gurley, L.R. "HPLC Separation of Biologically Important Arsenic Species Utilizing On-line Inductively Coupled Argon Plasma Atomic Emission Spectrometric Detection" Anal. Chem. (1986), 58, 1340-1344. 20. Mccarthy, J.P.; Caruso, JA.; Fricke, F.L. "Speciation of Arsenic and Selenium via Anion-Exchange HPLC with Sequential Plasma Emission Detection” J. Chrom. Sci. (1983), 21, 389-393. 21. Gast, C.H.; Kraak, J. C.; Poppe, H.; Maessen, F.J.M.J. "Capabilities of On-Line Element-specific Detection in HPLC Using an ICAP Emission Source Detector" J. Chromatogr. (1979), 185, 549- 561. 22. Irgolic, K.J.; Stockton, RA.; Chakraborti, D.; Beyer, W. "Simultaneous ICAP - Emission Spectrometer as a Multi-element- specific Detector for HPLC: The Determination of Arsenic, Selenium and Phosphorous Compounds" Spectrochim. Acta, Part B (1983), 38, 436-445. 23. Chakraborti, S.; Hillman, D.C.J.; Irgolic K.; Zingaro, RA. "Hitachi Zeeman Graphite Furnace AAS as a Selenium-Specific Detector for Ion Chromatography" J. Chromato. (1982), 249, 81-92. 24. Schlemmer, G.; Welz, B. "Palladium and Magnesium Nitrate, A More Universal Modifier for Graphite Furnace AAS" Spectrochim. Acta (1986), 41B, 1157-1165.
