Introduction

The world is complex. Human beings tend to manage the complexity by breaking up experiences into manageable components that have meaning. These components are stored as mental representations, or concepts, that permit thinking when the object is not present. Naming of concepts permits recognition and communication with others; grammatical rules for combining concepts permits thoughts to be shared through language, and concepts within a classification system permit organization of ideas.

A fundamental part of learning concerns concepts, categories, and classification systems. Saying "that is a dog and this is a cat" requires classifying sensory impressions. To know that both of these fall into the taxonomic category of animals is even higher level classification and to know that angry dogs bite involves accessing a network of concepts. This is similar to the situation in nursing practice when patients are deemed "difficult" or when a health problem is classified as preoperative fear. Recognition occurs when what is observed is placed into previously learned classes, or categories, on the basis of observed characteristics. Human beings think of ideas as related to each other and the world as a somewhat orderly place.

The relationship among concepts is the basis for a hierarchical classification system and the organization of knowledge in a discipline. Science develops formal concepts to classify animals, plants, chemicals, minerals, and other things of scientific interest. In a science when concepts are considered important they become formal, standardized classifications within a system. For example, classifying a person as a citizen of the United States is important for determining rights and responsibilities. Similarly classifying signs and symptoms as "High Risk for Pressure Ulcer" or "Risk for Other-Directed Violence" is important for identifying risk management activities in nursing. In the last half of the 20th century, nursing has recognized the importance of its clinical science and has constructed concepts, theories, and classification systems to further scientific development and patient care. These developments come nearly 100 years since Nightingale wrote that the "very elements of nursing are all but unknown" (Nightingale, 1859).

The purpose of this paper is to examine from a historical perspective the development of classification systems in nursing and the interdisciplinary and intradisciplinary issues that influence development. The focus will be limited to

1. three major classifications used in practice in the United States and
2. the developing international classification.

History of Classification

A brief examination of the history of classification will serve as a reference point for nursing classifications. It is sometimes stated that the idea of classification dates back to the book of Genesis. Out of chaos "God divided the light from darkness." Primitive cultures, as well as more advanced societies, have classified health-related ideas important to the culture. In his studies of world cultures Murdock (1980) classified theories of illness into theories of natural causation and theories of supernatural causation. (Information for Murdock's subcategories listed in Table 1 are based on information from pages 8-20 of his book.) The Hippocratic School of ancient Greece explained disease using the concept of "humors" rather than supernatural or magical forces.

TABLE 1.
Classification of Theories of Illness
Worldwide Theories of Illness
A. Natural Causation	B. Supernatural Causation
Type 1. Infection	Type 6. Fate
Type 2. Stress	Type 7. Ominous Sensations
Type 3. Organic deterioration	Type 8. Contagion
Type 4. Accident	Type 9. Mystical Retribution
Type 5. Overt Human Aggression

The first systematic attempt to classify disease is credited to Francois Bossier de Lacroix with the publication of Nosologia Methodica in the mid-1700s. Interest in the 18th and 19th century centered on the classification of causes of death, although many thought the classification should also include non-fatal conditions. Florence Nightingale delivered a paper titled Proposals for a Uniform Plan of Hospital Statistics at the fourth International Statistical Congress in London, 1860, urging the inclusion of non-fatal conditions.

By 1893 the International Classification of Diseases and Causes of Death was being used and in 1946 the World Health Organization (WHO) assumed responsibility for reviewing and revising this system every 10 years. The WHO recommended that countries establish national committees on vital and health statistics in 1948 to encourage international cooperation and to be used as a resource by WHO for future revisions of the classification (Zernott, 1982). This structure for international cooperation in classification system development has many similarities to that which is evolving in nursing.

Similarities exist also between nursing and medicine in the number of classification systems with different foci. In medicine morbidity and mortality statistics are compiled using the International Classification of Diseases (ICD). The ICD is adapted for the United States and has been further adapted for hospitals' statistical reports (H-ICDA). For example, in the United States the category "chronic ulcer of the skin" has been differentiated into "decubitus ulcer" and "leg ulcer." Other countries have other needs for specificity. In an international, nursing classification system, adaptations, such as these, will occur when a country wishes more or less specificity, relative to epidemiology or cultural needs.

Some of the other nomenclature systems used in medicine include Standard Diagnostic and Statistical Manual of Mental Disorders (DSM-IV), International Classification of Injuries, Disabilities, and Handicaps (ICIDH), Standard Nomenclature of Disease and Operations (SNODO), Systematized Nomenclature of Pathology (SNOP), and Systematized Nomenclature of Medicine (SNO MED III). As may be noted from the names, classification systems are designed for specific purposes to meet the needs of particular users. The idea of systems for specific purposes is also a trend in nursing classification; there are systems which are multipurpose and systems for particular settings, such as home care nursing.

Historical Perspective on Nursing Classification Systems

Historically, modern nursing based on the Nightingale model used disease entities from medical classifications to organize their thinking, speaking, and writing. These were the only concepts available in health care delivery up to the middle of the 2Oth Century. Only recently has there been an interest in the substantive structure of nursing knowledge (Tripp-Reimer, Woodworth, McCloskey, Bulechek, 1996) and, consequently, middle range theory development of diagnostic concepts (Eakes, Burke, and Hainsworth , 1998). With an increase in nursing research there has also been a search for useful classification systems for coding studies (Larson, Dear, and Keitkemper, 1991).

It is well to remember that classification system development parallels knowledge development in a discipline. Even basic structures for knowledge development in nursing had to await the introduction of theories and philosophies of nursing in the 1950s. Today many of the difficulties experienced in classification development are due to, but also are enriched by, the theoretical pluralism in nursing.

National nursing organizations have advanced the idea of classifications for nursing practice. Dr. Gertrude Torres (National League for Nursing) and Dr. Roberta Thiery (American Nurses Association-ANA) were liaisons from two organizations during the early decades of classification work in the United States. ANA has developed criteria for recognizing nursing classifications (Warren, 1997) and has recognized the work of NANDA and the Iowa projects discussed below, as well as the home health care (Saba, 1992) and community classifications (Martin and Sheet, 1992).

This permitted the nomenclatures (diagnostic labels) to be added to the literature search terminology of the National Library of Medicine, the Unified Medical Language System (UMLS). A Unified Nursing Language System (UNLS) "would allow linking or mapping of similar terms while maintaining the integrity and purpose of each classification system. The UNLS is the foundation on which the nursing profession develops, analyzes, and uses national data sets" (Warren, 1997). Hoskins (1997) has reported on the mapping of the aforementioned diagnostic classifications.

Early Classifications

Simply stated, classification is the ordering of entities into groups or classes on the basis of their similarity, minimizing within-group variance and maximizing between-group variance (Bailey, 1994). As nursing knowledge development increased and "entities" (diagnostic categories) were identified, interest in organizing knowledge for practice, education, and research also increased. A very, early reference to building knowledge in nursing is found in Bertha Harmer's Methods and Principles of Teaching and Principles and Practice of Nursing published in 1926 (Aydelotte & Peterson, 1987). She asked

1. whether "organized content of nursing knowledge could be built up as had resulted in medicine and other fields" and
2. "Should nurses not prescribe nursing care for each patient as doctors prescribe medical care?" and
3. she applied the term "social diagnosis" to nursing, a term possibly borrowed from social work (Richmond, 1917).

This appears to be the first query about the elements of a classification system for nursing diagnosis and interventions.

As nursing theories and nursing process problem-solving were introduced in mid-20th century, concepts of practice emphasized procedures, tasks, and functions. It was in this practice milieu that Abdellah (1959) reported a classification of nursing problems based on a survey of 40 schools of nursing. Consistent with the times, the 21 problems were therapeutic problems that described therapeutic goals of the nurse, rather than health problems of the patient or family. The classification of these goals of nursing served to organize curricula and practice for many years.

A second classification of basic, functional needs was developed by Henderson (1966). The components describe problem-areas; thus it is a conceptual classification into which empirical entities may be classified. At the time no entities, such as nursing diagnoses, existed. These two, early classification systems and the developing theories or philosophies of nursing were influential in setting the stage for the next phase of knowledge development: diagnostic, intervention, and outcome concepts.

The process of classification will be discussed in the context of diagnostic classification. When processes are similar, only differences will be considered in the discussion of intervention and outcome classifications.

Diagnostic Classification

Historical Perspectives: Diagnostic Classification

The North American Nursing Diagnosis Association (NANDA) is recognized in this and other countries as the pioneer in diagnostic classification in nursing. It began as a Task Force that was created at the First National Conference on Classification of Nursing Diagnoses, 1973, and evolved into an incorporated Association in 1982 to assist nurses in the United States and Canada in classification (Gordon, 1998). Initiation of work on classification for the nursing profession can be attributed to the foresight of two faculty at St. Louis University, Kristine Gebbie and Mary Ann Lavin who called the first conference on classification. Their belief that all nurses should have the opportunity to participate in the development of classifications used in nursing practice has influenced many decisions about the process of identification and classification of nursing.

Diagnostic Concept Development

Diagnoses are concepts that are given a word-label. Gordon observed that a diagnosis is a conceptual model for interpreting a set of observations and therefore provides a way of understanding and thinking about the set. The conceptual basis of a diagnostic concept is summarized in four dimensions: the definition, defining characteristics, and related (contributing, etiological) factors. Ideally, the conceptual base of each diagnostic concept is firmly grounded in studies of the phenomenon (Gordon, 1990). In 1973 when the first classification conference was held, research was minimal and substantive literature on concepts of this type, negligible.

After specifying the purpose of a proposed system, the first step in classification is to identify the phenomena of concern to be classified. It was clear at the First Conference in 1973 that the purpose of classification was to develop a classification system of use to all nurses in their practice, education, and research. Since that time the use in practice, alone, has been considerable. Examples of applications are quality assurance (McCourt,1986; Gordon, 1980), staffing (Halloran, 1985), nursing minimum data set (Werley and Lang, 1987; Mehmert & Delaney, 1991) identifying trends (Rantz and Miller, 1987), and information systems (Warren, 1997; Warren, Delaney, and Ryan, 1997).

An inductive approach was used initially by NANDA to begin to identify classes/categories. This is in contrast to deduction of elements from a nursing theory. Theoretical pluralism prevails and the choice of one theory would negate the others. The 100 participants (staff nurses, clinical specialists, educators, researchers, administrators, theorists, and consultants) at the 1973 conference generated a set of nursing diagnoses, definitions, and defining characteristics from their nursing practice expertise stored in memory. These diagnoses represented 29 conceptual areas with approximately 100 terms which were later condensed. This can be compared to the current classification system that contains 71 conceptual areas and 143 terms (North American Nursing Diagnosis Association, In Press).

Definition and Criteria for Classification. Certain questions are important and should be asked early in classification development: At an abstract level, what is to be classified? How does a concept (notion) become a diagnosis, accepted for classification?

The NANDA definition of a nursing diagnosis was accepted by the NANDA membership in 1988; it is adapted from a national, Delphi study by Dr. Joyce Shoemaker (1984):

Nursing diagnosis is a clinical judgment about individual, family, or community responses to actual or potential health problems/life processes. Nursing diagnoses provide the basis for selection of nursing interventions to achieve outcomes for which the nurse is accountable (North American Nursing Diagnosis Association, 1997).

Because the focus of nursing diagnosis is abstract, this definition is acceptable to nurses of different theoretical persuasions and colleagues in other professions. In nursing it is challenged by some because of the implications for the scope of nursing diagnosis. A discussion of this controversy over the last sentence is better left to a discussion of issues below.

A diagnostic concept, or category, is formed from a set of empirical indicators (signs/symptoms) observed together frequently enough to capture nurses' attention. The set of indicators is given a name and definition. Ideally, as in other professions, clinicians would report their observations in a journal to alert others to the phenomena. Further observation leads to identification of contributing factors which will be the focus of nursing intervention. Qualitative studies using grounded theory or other methods, should follow. Quantitative research methods are useful when there is a tentative formulation of the concept and contributing factors or when the category is ready for epidemiological studies. Norris (1982) has offered methods of concept clarification and Gordon (1990) has proposed a cycle of development for determining the conceptual base of a diagnostic category.

Recent work in cognitive theory suggests ways of looking at concepts that will be useful for studies of the developmental, cultural, gender, and illness-related variables which influence defining characteristics of a diagnostic concept. Research and development in this area has to be informed by the research on human cognitive architecture, concept learning, and diagnostic reasoning (Rosch and Lloyd, 1990; Van Mechelen, Hampton, Michalski, and Theuns, 1993; Gordon, 1994).

Research on the validity and reliability of diagnostic categories has increased considerably in the last decade but is still insufficient. This is a serious problem. Diagnostic reasoning and judgment require valid and reliable diagnostic concepts that facilitate, rather than inhibit, thinking. Large scale funding for the basic research that is needed to identify, develop, and validate diagnostic categories has not been available in grants or contracts in the USA or Canada. In contrast, in other regions of the world large scale validation projects have been funded (Mortensen, 1996). Although nursing diagnosis always has been one of the standards for professional practice (American Nurses Association, 1973) and part of the profession's contract with society (American Nurses Association, 1980; Canadian Nurses Association, 1993) leaders in nursing research have not always supported the need for a language and classification system that differs from medicine.

Source of Diagnoses. Now to further answer the question of how an idea becomes an accepted nursing diagnosis in the NANDA system. The NANDA is the only group in nursing classification that has a formal review process. Diagnoses for review and possible classification are accepted from any nurse or group of nurses. Submission and staging guidelines are published (North American Nursing Diagnosis Association, 1997, pp.89-94). Staging reflects multiple levels of development and research. For example, guidelines begin with a "received for development" category in order to offer consultation from the Diagnosis Review Committee to the submitter. Publication during the review of a submitted diagnosis in the Nursing Diagnosis journal offers opportunities for comments from other nurses.

The review of a submitted diagnosis is coordinated hy the NANDA Diagnosis Review Committee consisting of seven elected members. The reviewer uses

1. nurse-experts in the content of the diagnosis and
2. specialty organizations who wish to participate.

In addition, diagnoses go to the NANDA International Committee for recommendations regarding translation problems. After presentation at forums during biennial NANDA conferences, diagnoses go to the NANDA Board of Directors for acceptance and are published (North American Nursing Diagnosis Association, 1997; In press 1998). The process is designed for broad participation and multiple inputs.

Participation of nursing specialty organizations is important. Some organizations have submitted high frequency diagnoses in their specialty that are not already classified (Association of Rehabilitation Nurses, the former ANA Council of Psychiatric-Mental Health Nurses, Holistic Nurses, Association of Operating Room Nurses, etc.). The Association of Rehabilitation Nurses submitted new diagnoses from a research study that involved a national, random sample of members. This Association also funded a study that resulted in publication of 21 rehabilitation nursing diagnoses-intervention-outcome linkages (Rehabilitation Nursing Foundation, 1995).

A recent, but major source of diagnostic submissions is the collaborative project between NANDA and project directors, Drs. Martha Craft-Rosenberg and Connie Delaney, at University of Iowa. The Nursing Diagnosis and Extension Classification (NDEC) Project is designed to refine current diagnoses, reference the source of defining characteristics, and extend the classification (Craft-Rosenberg and Delaney, 1997). Teams working on the project review the literature, do concept analysis of existing diagnoses, and generate new concepts when suggested by the literature. This work sets the stage for clinical validation and epidemiological studies.

Enduring Controversies. NANDA has been criticized for the multiple levels of abstraction in the diagnostic categories. For example, the category "altered parenting" requires further specification of a taxonomy of "parenting" diagnoses. Fatigue and nausea are diagnoses at a concrete level of abstraction. Two other nagging issues that have followed developers down through the years are associated with the definition of a nursing diagnosis. These issues are the "physiological" diagnoses and "wellness" diagnoses. This controversy surrounding "what is a nursing diagnosis, and what is not," relates to the basic principle of classification: "... successful classification, then, is the ability to ascertain the key or fundamental characteristics on which the classification is to be based" (Bailey, 1994, p.2). Bailey goes on to say that there is no specific formula for identifying key characteristics.

Clinicians' comments regarding "physiological" diagnoses suggest an underlying need for a classification of physiological problems (Kim, 1984). These are treated in collaboration with the physician using both physician-initiated and nurse-initiated interventions. Many of the conditions are not diseases, but rather problems or the potential for complications that require frequent monitoring by the nurse. Decreased cardiac output and impaired gas exchange are examples. Yet, many nurses contend that these collaborative problems (Carpinito, 1995) are not nursing diagnoses, according to the NANDA definition that specifies accountability for the outcome(s).

A second issue concerns the diagnoses relating to "wellness" or health. On one side of the controversy the argument is that prevention involves potential problems or risk states that the nurse is expected, has a duty, and possibly is reimbursed, to treat. Risk states, for example "High Risk for Injury or Falls," describe conditions that require risk reduction through health promotion and preventive intervention. They ask: Why do we need "wellness" diagnoses? The argument continues, if a person does not have a problem, don't diagnose an "effective" state or "potential for enhancement."

On the other side of the controversy nurses argue that there is a need for categories such as "Effective Breastfeeding" and "Potential for Enhanced Community Coping." These guide health promotion and health protection interventions and the identification of strengths (Popkess, 1981). They are particularly important in community nursing, school nursing, and general programs emphasizing health (Lunney, Cavendish, Luise, Richardson, 1997). The counter argument that is given is: Do all persons have "potential for enhanced..." and thus, the diagnosis is a routine? Do some patients not have the potential? Would the term "Desire for Enhanced...," with a defining characteristic that the patient requested help to achieve higher levels of wellness capture the idea?

In regard to the second type of "wellness" diagnosis, Effective Breastfeeding, is this an outcome? Will there be reimbursement for "treatment" if the behavior is already effective? Similarly, will reimbursement be scant for "enhancing potential?" From the theoretical perspectives of some nurses, enhancing potential is the essence of the nurse-patient relationship. The wellness and physiological diagnoses would be excellent topics for a national consensus conference on conceptual (Gordon, 1990) and semantic issues (Nielsen, 1995).

It may be noted that controversies about inclusion and exclusion are not unique to nursing. They involve both nomenclature and taxonomy issues that are experienced by classifiers in medicine and the sciences. The criticism on the "wellness" side that the NANDA taxonomy is acute care- rather than community-care oriented has to be balanced with the criticism from some nurses on the "physiological" intensive care side that the taxonomy does not sufficiently represent this area of practice. These are some of the issues that make the classification of nursing practice phenomena so intellectually challenging for those involved.

Classification of Nursing Diagnoses

A classification system can be as simple as an alphabetical listing or as formal as a numerical taxonomy. The ideal classifications of mathematics and logic with mutually exclusive categories are seldom found in other disciplines (Bailey, 1994). Biology struggles to distinguish between the animate and inanimate and medicine finds obesity in its classification of diseases (Webster, 1984). Classification is not a simple task with the complexity of nursing, phenomena of concern.

How did the current North American classification system grow? It was nurtured through the efforts of NANDA Taxonomy Committee chairpersons, Drs. Phyllis Kritek, Joyce Fitzpatrick, and currently Kay Avant. Models and frameworks (Maslow, Abdellah) for organizing nursing diagnoses were suggested by participants at the First National Conference and later (Lunney, 1984; Loomis, 1987) but from 1973 to 1986 diagnoses were arranged in an alphabetical listing. In retrospect this was a wise decision for that time period. NANDA currently classifies nursing diagnoses into Taxonomy I, Revised using the structure of Human Response Patterns, as seen in Table 2. This structure for a classification system was accepted by participants in 1986 at the Seventh Conference.

TABLE 2.
HUMAN RESPONSE PATTERNS	DEFINITIONS OF HUMAN RESPONSE PATTERNS
CHOOSING.	To select between alternatives; the action of selecting or exercising preference in regard to a matter in which one is a free agent; to determine in favor of a course; to decide in accordance with inclinations.
COMMUNICATING.	to converse; to impart, confer, or transmit thoughts, feelings, or information, internally or externally, verbally or non-verbally.
EXCHANGING.	To give, replenish, or lose something while receiving something in return; the substitution of one element for another; the reciprocal act of giving and receiving.
FEELING.	To experience consciousness, sensation, apprehension, or sense: to be consciously or emotionally affected by a fact, event, or state.
KNOWING.	To recognize or acknowledge a thing or a person; to be familiar with by experience or through information or report; to be cognizant of something through observation, inquiry, or information; to be conversant with a body of facts, principles, or methods of action; to understand.
MOVING.	To change the place or position of a body or any member of a body; to put and/or keep in motion; to provoke an excretion or discharge; the urge to action or to do something; to take action.
PERCEIVING.	To apprehend with the mind; to become aware of by the senses; to apprehend what is not open or present to observation; to take in fully or adequately.
RELATING.	To connect; to establish a link between; to stand in some association to another thing, person, or place; to be born or thrust in between things.
VALUING.	To be concerned about; to care; the worth or worthiness; the relative status of a thing, or the esteem in which it is held; according to its real or supposed worth, usefulness, or importance; one's opinion of linking for a real person or thing; to equate in importance.
From: Fitzpatrick, J. J. (1991) Taxonomy II: Definitions and development '” Table I: Definitions of human response. Classification of nursing diagnoses: Proceedings of the ninth conference, p. 25. Reprinted with permission, Nursecom,Inc.

Taxonomy is a term used interchangeably with classification system, such as in biology and other disciplines. Yet it is actually defmed as the study of classification including its bases, principles, procedures and rules (Sneath & Sokal, 1973, p.3). Similar to the term, classification system, taxonomy can refer to both the process of classification and the end product. In the NANDA taxonomy, the roots of the term "human response" lie in the ANA Social Policy Statement (Kritek, 1989). In this document nursing is defmed as the diagnosis and treatment of human responses to actual or potential health problems. (The term "problems" is used in a very broad, generic sense, such as that which is of therapeutic concern to the nurse, the patient, or both.) Human responses are the indicators of patterns (actual or potential health problems).

Between 1977 and 1982 a group of 14 prominent nurse-theorists developed an organizing framework from which the patterns in Table 2 are taken (Roy, 1982a, 1982b). Sister Callista Roy, a member of the National Task Force that preceded NANDA, coordinated the work of this group at national conferences and by mail. It was the first time that these leaders in nursing theory

1. had met together,
2. worked with a group of clinical specialists to test their work, and
3. developed a common conceptual framework for nursing and for organizing nursing diagnoses.

They called it the Unitary Man Framework. The nine concepts in Table 2 were re-named Human Response Patterns in 1984 by the NANDA Taxonomy Committee.

Kerr (1991) outlined methods for taxonomy validation but the literature does not reflect the use of these methods in validating the NANDA Taxonomy or further developing the pattern-definitions in Table 2. Clinicians have attempted to use the patterns as an assessment tool or in information systems and, on this basis, many have rejected their abstract nature. In 1998, the Taxonomy Committee under the leadership of Dr. Kay Avant explored other typologies for organizing diagnoses that may be more useful (North American Nursing Diagnosis Association , In Press). A second factor for re-examining the nine key concepts was the difficulty classifying diagnoses that crossed patterns, such as syndromes (Avant, 1997; McCourt, 1991.). Comments and suggestions will be elicited from the profession prior to the Taxonomy Committee's report to the Fourteenth NANDA Conference in 2000. Similar to diagnosis review, taxonomy review is based on the philosophy that all nurses should have an opportunity to participate in the development of a classification system for the profession.

Outcome Classification

Health care providers have appreciated the importance of outcomes as requirements for measuring economic efficiency and system effectiveness in a cost-control environment. The concept of outcome has not been analyzed in nursing resulting in a multiple terms or variations in the basic term. In addition to the term "result", eight terms are commonly used in the literature to modify, an outcome: patient, nurse-sensitive, desired, effective, expected, predicted, projected, and actual outcome (van der Bruggen & Groen, 1995). Outcomes may defined as the end results of care, yet when quality of care is being measured outcomes are linked to diagnoses. From this perspective, outcomes are indicators of problem resolution or progress toward resolution.

Historical Perspective: Outcome Classification

Aydelotte (1962) was an early pioneer in the measurement of patient outcomes. She was the first in nursing to use changes in characteristics of patients to evaluate nursing care delivery. In 1988 Heater, Becker, and Olson completed a meta-analysis of studies that suggested a growing interest in nursing goals and outcomes during the previous decade. Outcomes at this time were general, such as "the patient's self care skills" (Hover and Zimmer, 1978).

Johnson and Maas summarized the multiple reports of outcome generation that were characteristic of the 1980s and early 1990s and concluded that identification of outcomes was based mainly on literature reviews and practical experience rather than research or conceptual frameworks" (1997, p.5). These authors also note that nursing outcomes differ in content from medical outcomes. Nursing includes client knowledge and behaviors, safety ,use of resources, home maintenance, and caregiver status (1997, p.5). In a large, funded project Johnson and Maas (1997) and a team of investigators at the University of Iowa developed a set of outcomes and proposed their linkages to nursing diagnoses. Outcomes and their indicators are the concepts, or elements, to be used in classification.

Outcome Concept Development

As previously discussed it is important in classification to clearly define that which is to be classified. Secondly, whether the classification is outcomes, diagnoses or interventions, each category at a particular level in a classification system should be at the same level of abstraction. In the Johnson- Maas study three characteristics of outcomes are identified: a variable state, a consequence, and a level of abstraction.

For the team's research, a nurse-sensitive patient outcome is defined as a variable patient or family caregiver state, behavior or perception that is responsive to a nursing intervention and conceptualized at middle levels of abstraction (e.g., mobility level, nutritional status and health beliefs). Nursing- sensitive outcome indicators are defined as variable patient or family caregiver states, behaviors, or perceptions at a low level of abstraction that are responsive to nursing interventions and used for determining a patient outcome (e.g., for the outcome Mobility Level, indicators include "joint movement, "transfer performance," and "ambulation: walking"(1997, p.21).

This shares some of the characteristics of van der Bruggen and Groen's (1996) definition contained in an interim report of an internafional, Delphi study of nurse-experts.

Further specification of outcome-concepts in the Johnson-Maas report includes measurability on a numerical scale (1997, p.22). This latter characteristic allows

1. measurement on a continuum and evaluation of progress toward resolution of a nursing diagnosis, rather than just resolution-no resolution and
2. a scale for projecting outcomes (e.g., prognosis).

Correspondence between a nursing diagnosis and its projected/desired outcome would be expected. The unhealthy behaviors that characterize a diagnosis are the healthy opposites in an outcome indicator. There is an indication that some nursing diagnoses are broader (more inclusive, e.g., altered thought processes) than a corresponding outcome and that some outcomes (child development) are broader than corresponding diagnoses in the literature. The former is of more concern than the latter. Diagnoses are the basis for outcome projection and evaluation. In addition, there are outcome concepts with no corresponding NANDA nursing diagnosis (e.g., social support). Meetings among developers of these two classifications would probably enrich both systems.

As required in classification development, methods for developing the elements for classification and guidelines for decisions are clearly delineated in the Johnson-Maas report (1997). Methods included review of the literature, information systems and other sources. Concept analysis and graduate student surveys were used to determine the validity of the concepts and their indicators. Field testing is planned in multiple settings which will evaluate "frequency of use and sensitivity to nursing interventions" (1997, p.26).

Outcomes in the Johnson-Maas study were limited to individuals and family-caregivers. Plans exist to expand the nomenclature to include family community, and organization. Also further work on linkages is planned (1997).

Classification of Outcomes

Nurses have been developing outcome typologies in various specialty areas of practice for at least three decades. Early typologies (conceptual groupings) used general terms. Quality assurance programs provided the major impetus to development in the 1970s and 1980s. Currently, diagnosis-specific standards/guidelines for practice, cost containment, and development of computerized patient records give further impetus. Additionally, large projects were funded in nursing; outcomes were not as controversial as the term, nursing diagnosis (Simmons, 1980; Horn & Swain, 1978, Johnson & Maas, 1997).

Johnson and Maas have used similarity ratings and hierarchical clustering techniques to develop a taxonomy with 24 classes and six domains. These were used to classify the more than 200 outcomes and outcome indicators developed in their project. The domains are

1. functional,
2. physiologic, and
3. psychosocial health,
4. health knowledge and behavior,
5. perceived health, and
6. family health

(Johnson and Maas, 1998).

It is claimed that nursing is invisible in health care delivery because we have not articulated our contribution. Although recognized as incomplete, it is impressive to see a list of nurse-sensitive outcomes and know that this list represents a visible "image" of nursing's contribution to the nation's health.

Intervention Classification

In the previous sections emphasis has been on descriptive concepts and their classification. Interventions, or prescriptive concepts, complete the diagnosis-intervention-outcome linkage that specifies the major elements and relationships that need to be developed for a nursing classification.

Historical Perspective: Intervention Classification

Nurses and administrators have always been interested in identifying what nurses do, perhaps for different reasons. At various points in time, even industrial task analysis was common. Assessing and monitoring medical treatments and technology, administering medications, and assisting the patient to carry out the physicians orders were high priority in the traditional typology of nursing interventions. This situation changed with the publication of nursing theories and philosophies beginning at mid-century and the introduction of nursing diagnosis and clinical judgment. "Independent" nursing interventions that were nurse-initiated were given increased emphasis in curricula and in practice. In the 1980s textbooks began to name nurse-initiated interventions and relate these to nursing diagnoses (Bulechek and McCloskey, 1992; 1985; Maas, Buckwalter, and Hardy, 1991; Snyder, 1985).

Intervention Concept Development

The first systematic approach to naming classes of interventions was reported by McCloskey and Bulechek in 1992 and updated in 1996. The nursing intervention classification (NIC) research team identified a set of intervention concepts through content analysis of the literature and other sources, project team focus groups, and graduate student ratings. Validation procedures included surveys of specialty organizations, individual nurses, nurse experts in theory development (taxonomy and content), and clinical field testing. Currently, over 433 intervention concepts and over 3000 nursing activities have been identified by the research team. Field testing continues in a variety of settings, an important factor when the claim is generalizability to all nursing areas (McCloskey and Bulechek, 1996).

The structure of an intervention includes the concept label, definition, and activities. The close involvement of one of the directors of both the NIC and NOC projects in the NANDA work may have influenced decisions about the structure of interventions and outcomes. All three classification elements consist of a

1. concept label,
2. definition, and
3. referents (defining characteristics, outcome indicators, or activities)

which provides consistency across classifications for users.

Three types of interventions have been classified by the McCloskey and Bulechek research team in the Iowa project (1992). They are defined as follows:

A nursing intervention is any direct care treatment that a nurse performs on behalf of a client. These treatments include nurse-initiated treatments resulting from nursing diagnoses, physician-initiated treatments resulting from medical diagnoses, and performance of the daily essential functions for the client {who} cannot do these (1992, p.21).

This definition provides clear direction for what is to be classified and, similar to the outcome project, procedures and criteria used in the classification project are clear. Critique of the intervention elements in the NIC includes:

1. various levels of abstraction in the intervention concepts,
2. unnecessary specification of populations or settings that may limit the use of interventions, and
3. the definition of nursing intervention as both autonomous and collaborative (Snyder, Egan, and Nojima, 1996).

Classification of Nursing Interventions

The number of interventions and activities (N=>3000) identified in the Iowa project required the use of a computer clustering. Similarity ratings and hierarchical clustering techniques were used to develop a taxonomy with 34 classes and six domains. The domains are physiological: basic, physiological: complex, behavioral, safety, family, and health system (McCloskey and Bulecheck, 1996, pp. 56-57). The use of nursing's tradifional terms may be necessary when classifications (intervention and outcome) contain both medical and nursing conditions. This is in contrast to the more abstract, but internally consistent, concepts, such as the nine Human Response Patterns (North American Nursing Diagnosis Association, 1996) or the 11 functional health patterns (Gordon, 1994). Perhaps with the start in 1998 of biennial NANDA-NIC-NOC Conferences there will be more contact among developers and a common way of looking at taxonomic structure may emerge.

Multidimensional scaling, factor analysis, and other procedures were used to analyze the dimensional structure of 26 NIC intervention classes. Results suggest that the embedded structure of interventions contains three components: complexity (urgency and skill and knowledge needed), intensity (acuity), and focus of care (target: individual to system) (Tripp-Reimer, Woodworth, McCloskey, and Bulechek, 1996). These are similar in nature to the taxonomic branches proposed for the NANDA classification: acuity, developmental level and individual-family-community.

Similar to a diagnostic and outcome classification, the intervention concepts within the classification represent first level, factor-isolating theory. This is the base for middle range theory development and the structure of nursing science

Summary of Major Classifications

In summary, it may be noted that these three systems have been developed separately but can be linked, provisionally (Daly, 1993). Outcomes are linked to the problem (nursing diagnosis) in a diagnostic statement. Interventions are linked to the related or contributing factors. Diagnosis-intervention linkages (McCloskey and Bulechek, 1996) assume that a nursing diagnosis is being used as a contributing factor for another nursing diagnosis and that the classification is similar to a dictionary of terms. Linkages are more easily seen when diagnosis, intervention, and outcomes are identified within one project, such as the Home Health Care Classification (Saba, 1992) which shares many concepts with the NANDA Taxonomy or the Omaha Classificafion System (Martin and Sheet, 1992) for community health, which also overlaps considerably with the NANDA

Similar difficulties have been encountered by the three groups of classifiers. These were

1. concept development: definition of elements, level of abstraction, issues of validity and reliability, and
2. classification: determining the key constructs for organization of a nursing taxonomy and decisions about classification of elements that are autonomous, collaborative, or both.

Large scale funding (interventions and outcomes) has made a difference in the specification of criteria, procedures, field testing, and use of a variety of statistical methods for classification.

In the 25 years since the national effort began in a volunteer organization to systematically classify nursing diagnoses, and later interventions and outcomes, great progress has been made. Perhaps now, the time nurses spend on documenting the care they give can result in documentation that is systematically organized to advance nursing knowledge, develop nursing practice, and improve patient care.

The International Perspective on Classification

The International Council Of Nurses (ICN) in Geneva is preparing an International Classification for Nursing Practice (ICNP) that eventually will be submitted to the World Health Organization for classification of conditions necessitating nursing care. WHO will be use this coding schema for statistical purposes. Dr. Margarita M. Styles, Chair of the ICN Professional Services Committee began the project in 1991 and supported it through her Presidency in the years that followed. Through the leadership of Dr. Styles, and Dr. Fadwa Affara, the work of consultants Drs. Norma Lang (USA), June Clark (UK), and Randi Mortensen (Denmark) and advisors, G. Neilsen (Denmark) and M. Murphy and M. Wake (USA) an alpha version of the ICNP is available (International Council of Nurses, 1996).

The ICN is very aware that "without a language, nursing is invisible in health care systems and its value and importance go unrecognized and unrewarded" (1993, p.2). They have emphasized the importance of the work to every country and Lang has said that "If we cannot name it, we cannot control it, finance it, research it, teach it, or put it into public policy" (International Council of Nurses, 1993, p.2). An ICNP requires the collaboration of all nations. International interest is high and regional groups, similar to NANDA, have formed to work on classification. For example, there is a European organization, including formal organizations of Francophone nurses (Europe and Quebec) and Spanish nurses (Spain and Cuba). For a number of years Japan has had a national organization for classification and plans to develop an all-Asia organization. Debate is lively among nurses from various nations at conferences (Japan Academy of Nursing Science, 1997), in unversities, and in health care agencies. The high participation of internationals in the biennial NANDA classification conferences and the participation of Americans in European and Asian conferences provide opportunities for sharing ideas.

Concept Development and Classification

The elements to be classified in the ICNP are diagnoses, interventions, and outcomes. This will foster hypotheses about linkages. Classifications from all member countries were collected and comprised the elements for the alpha classification (International Council of Nurses, 1993;1996). Principles of division, rules, and procedures have been developed for classification. These are described in the European Telenurse Project that has been funded by the European Union for Telematics in Health Care to study the validity of the elements and classification system in European nurses clinical practice (Mortensen, 1996). (A comparable validation project in North America will have to be done eventually.)

The architecture of the alpha version is based on classification principles but appears complex (ICN, 1995) in comparison with the alphabetical listing of the first report (ICN, 1993). Key concepts for organizing elements are neutral. The orginal definitions of diagnoses and interventions are concise and are available for all terms at all levels of the classification. The beta version of ICNP is due for publication in the near future. As the ICN has stated the ICNP "provides a vocabulary, a new classification for nursing and a framework into which existing vocabularies and classifications can be cross-mapped to enable comparison of nursing data collected using other recognised nursing vocabularies and classifications. For this reason the ICNP is referred to as A Unifying Framework" (International Council of Nurses, 1996, p.13).

Summary

Discussion has focused on historical perspectives in diagnosis, intervention and outcome classification in nursing and the current status of the major developers' projects. These classifications supplement medical classifications and will permit statistical coding of functional problems that are so common in an aging population.

Classifiers in North America have encountered the major types of problems identified by van Mechelen and his colleagues (1993, p.2): creating categories (clustering problems) and inducing general classification rules from descriptions of members and non-members (rule induction problems). Classification of the phenomena of concern in nursing is a complex task. It is complex because of the inherent complexity of persons, the phenomena of concern in nursing. It also is complex because it requires clarity of language; concepts and definitions that are understood in North America may be unknown or unclear in other countries and the opposite is also true. If regional groups around the world are to contribute to an international classification used by nurses in many language groups, attention has to be paid to the construction of clear, concise, translatable terms. Countries adopting nursing classifications from regions and cultures, other than their own, have to determine the cultural-sensitivity of the concepts. This has been evident as countries have adopted the North American work. Clarity of the work is also an important factor in developing information systems which are being developed world-wide.

Will historians say that in the last half of the 20th century the developing classifications revolutionized nursing practice? Perhaps so. Nursing diagnosis encouraged thinking to move from the notion of a work-task to a conceptualization of a patient's problem. It provided a language to communicate and a tool for critical thinking at a time when documentation was characterized by statements such as "appears to be bleeding" or "appears to be dead." The development of a language for interventions and outcomes revolutionized the "work-tasks" at a time when nurses thought of intervention as "provide emotional support" and outcomes as "slept well." The profession has come a long way in this past century. One wonders what the 21st century will bring.

Note: The name of the Home Health Care Classification (HHCC) System was changed to the Clinical Care Classification (CCC) System in 2003.

Author

Marjory Gordon, PhD, RN, FAAN
E-mail: gordon@bc.edu

Dr. Gordon is Professor Emeritus at Boston College and is engaged in studies on diagnostic reasoning and nursing diagnoses. She lectures internationally on these subjects and has published two books, Nursing Diagnosis: Process and Application and Manual of Nursing Diagnosis. Currently, she is co-authoring a book in Japanese on functional health patterns. She was President of NANDA in its early years and currently is a member of its Board of Directors and co-chairs its Diagnosis Review committee.

---