138

Social Support and Health: A TheoreticalFormulation Derived from King’s Conceptual

FrameworkMAUREEN A. FREY, RN; PHD*

* University of Michigan, Ann Arbor, MI.The study on which this article is based was partially

funded by the American Nurses’ Foundation, MichiganNurses’ Association, and a predoctoral fellowship fromthe American Diabetes Association-Michigan Affiliate,Inc.The author wishes to acknowledge Jacqueline Camp-

bell, R.N.; Ph.D., for her critique of the article.

This article describes the development and initial empirical testing ofa theoretical formulation of social support, family, health, and childhealth derived from Imogene King’s conceptual framework for nursing.A correlational design was used to test the formulation with 103 familieswho have children with diabetes mellitus. Three hypotheses were sup-ported : parents’ social support had a direct and positive effect on familyhealth, parents’ social support and child’s social support were positivelyrelated, and illness factors had a direct and negative effect on childhealth. Both the supported and unsupported hypotheses are discussedin terms of the present substantive knowledge base and evidence ofvalidity for King’s framework. Direction for further theory developmentand research are identified.

Key Words: Social Support, Health, King’s ConceptualFramework

Received September 6, 1988.Accepted October 25, 1988.

The 1980s may well be known as the decadeof social support. Nursing, like other healthsciences, has moved from debating the directversus indirect effect of social support onhealth to incorporating the concept of socialsupport into existing and new theoretical for-mulations. As a result, an impressive body ofknowledge of the mechanisms by which socialsupport influences health is developing insuch areas as transition to parenthood, teen-age pregnancy outcome, fetal attachment, in-fant attachment, maternal discipline, and par-enting of developmentally delayed infants(Barnard, Brandt, Raff, & Carroll, 1984).-

This article describes the development andinitial empirical testing of a theoretical for-mulation of social support, family health, andchild health in families who have childrenwith insulin dependent diabetes mellitus(IDDM). The proposed formulation makes sev-eral significant contributions to the existingknowledge base of social support and health

for nursing practice. First, few theory or re-search models specifically incorporate familyand child variables. Although the effect of so-cial support on the family has been increas-ingly addressed, very little is known aboutsocial support and its effects on child healthor the relationship between parent supportand child support. Second, the context for theresearch is families with children with achronic illness. Social support has been exten-sively investigated in relation to the onset ofvarious diseases, cumulative life events, andrecovery from acute illness episodes. However,less attention has been given to the role ofsupport in chronic illness either for individualadults, children, or family units.

In addition to addressing an important sub-stantive area for nursing knowledge, an addi-tional aim of the research was to determine ortest the underlying validity of King’s concep-tual framework. Empirical testing of middlerange theories derived from conceptual frame-works contributes indirect evidence as to the

credibility of the conceptual framework (Faw-cett, 1989; Silva, 1986). Further, this processof theory development is likely to result inknowledge unique to the discipline of nursing(Phillips, 1988; Whall & Fawcett, 1988).

Development of the Theoretical Formulation

The theoretical formulation was expllcitlyguided by King’s conceptual framework.King’s framework served as a basis for con-ceptual definitions of concepts, identifyingempirical indicators of concepts, and propos-ing relationships among concepts. Since con-

139

ceptual frameworks are broad, abstract, andnot specific to populations and practice set-tings, additional theoretical and empirical evi-dence was also used in development of theformulation.

King’s conceptual framework consists ofthree interacting, open systems: individuals aspersonal systems, two or more individualsforming interpersonal systems, and largergroups with common interests and goals form-ing social systems. Many concepts, borrowedfrom other disciplines and redefined by King,are used to identify relevant knowledge forunderstanding each system. Definitions andrelationships in this formulation are derivedfrom King’s concepts of health and interac-tion.

Child and family healthIn the present study, child health is concep-

tualized from King’s ( 1981 ) concept of healthin relation to the personal system. King de-fines health as a process of growth and devel-opment and the ability to function in socialroles. Health has biological, psychological, andsocial dimensions. From King’s perspective,health and illness are both dimensions in thelife events of human beings. The distinctionbetween health and illness is often arbitraryand influenced by such factors as one’s cul-ture, definition of health, perception of health,and environment. Illness indicates interfer-ence in the life cycle and is defined as animbalance in a person’s biological structure,psychological make-up, or social relations.For this formulation, child health is theoret-

ically defined as developmentally appropriatephysical, biological, psychological, and socialfunctioning. This definition is consistent withKing’s view of health. Recognition of develop-mental level was identified by King as an at-tribute of health, and is included in the defi-nition in order to differentiate child healthfrom adult health.

Family health is conceptualized from King’s( 1981, 1983) concept of health in relation tosocial systems. The family is identified as abasic structural and functional unit of societyand the primary social environment for theindividual. Family health is also viewed interms of functional ability. A functional orhealthy family is able to adjust to stressors inthe internal and external environment andcope with maturational and situational crises

(King, 1983).Although King does not offer a conceptual

definition of family health, she does indicatethat the concept of health can be applied tothe family as a social system. For this formu-lation, family health is theoretically defined asthe adaptive potential and functional ability ofthe family in terms of performance of socialroles. From King’s perspective, this would in-

clude structural and functional dimensions ofthe family system. Other indicators of familyhealth could be extrapolated from the addi-tional concepts that King identified as impor-tant for understanding social systems: orga-nization, authority, power, status, decision

making, and control (King, 1981). King’s useof general systems and structural-functionalconcepts to evaluate family health, and theinterchangeable use of the terms family healthand family functioning, is consistent with thework of other family scholars (Barnhill, 1979;Feetham, 1984; Minuchin, 1974; Olson,Sprenkle, & Russell, 1979).

Social supportSocial support is conceptualized from King’s

( 1981 ) concept of interaction for interpersonalsystems. King placed the concept of interac-tion within the interpersonal system becauseinteraction occurs between two or more per-sons. Each interaction involves verbal andnonverbal communication and is character-ized by values. Communication is defined byKing as an exchange of information. Interac-tions also include exchange of either materialgoods or services. According to King, interac-tion is essential in developing relationships. Itis through relationships that growth, change,and personal development take place. There-fore, the quality of interactions may have apositive or negative influence on health.The multiple definitions of social support in

the literature are consistent with what King( 1981 ) describes as the content of interaction.For example, Cobb (1976) defined social sup-port as information leading one to believe sheor he is cared for or loved, esteemed or valued,and part of a network of communication andmutual aid. Kahn and Antonucci (1980) iden-tified social support as interpersonal transac-tions that included the expression of positiveaffect, affirmation of behavior, or material aidto another person. Weiss (1974) proposed thatsocial support included the provision of at-tachment, social integration, nurturance, re-assurance of worth, a sense of reliable alli-ance, and obtaining guidance through socialrelationships. House ( 1981 ) defined social sup-port as interpersonal transactions involvingemotional concern, information, appraisal, orinstrumental aid.Several nurse scientists have addressed the

definitional diversity surrounding the con-

struct of social support. In reviewing multipledefinitions of social support, MacElveen-Hoehn and Eyres (1984) found that House’s(1981) four categories tended to account formost of the descriptive terms used by othersin relation to the content of social support.Diamond and Jones (1983) attempted an in-tegrative approach to a theoretical definitionof social support. Four points on which various

140

definitions converged were identified: com-munication of positive affect, a sense of socialintegration, reciprocity of directedness as afactor in the continuance of support and sat-isfaction in interaction, and the provision oftangible aid.Despite definitional diversity, it seems fairly

well established that social support is concep-tualized as a component of social interactionwith family, friends, neighbors, and otherswith whom one has personal contact. The ele-ments of exchange of communication and aid,reciprocity, and mutuality are consistent withKing’s (1981) view of the content and charac-ter of interactions.

Social support is theoretically defined as anexchange of positive affect, a sense of socialintegration, emotional concern, and/or directaid or services between two persons. Since,according to King, the quality of interactionsinfluences health, the quality of social sup-port, as a component of social interaction, alsowould be expected to influence health.

Social support and family healthIn specifying the relationship between inter-

action (social support in this formulation) andhealth, King did not distinguish between thehealth of personal systems (individuals) andsocial systems (families). Extensive empiricalevidence supports the generally positive effectof social support on individual health andfunctioning (Berkman & Syme, 1979; Gottlieb,1983; Hyman & Woog, 1982). Perhaps becauseof the evidence indicating the generally posi-tive effects of social support on the health ofindividual adults, increased attention hasbeen directed toward social support and thefamily unit. In general, support of the nuclearfamily from immediate or extended kin as wellzaps from friends, work associates, neighbors,and the community has been increasingly rec-ognized as a factor that influences family well-being and behavior (Anderson, 1982; Pilisuk& Parks, 1983; Unger & Powell, 1980). Thus,it is postulated that support for the family(parents’ support) has a direct and positiveeffect on family health.

Social support and child health

King’s framework does not explicitly differ-entiate between children and adults. As a re-sult, the framework offered little direction asto the nature of the relationships between par-ents’ support and child’s support and child’ssupport and child’s health. However, Cochranand Brassard’s (1979) conceptual model of re-lationships between child development andthe personal social networks of immediatefamily members provides a perspective forviewing these relationships. Although focusedon network analysis, which is different fromsocial support, social support was clearly iden-

tified as a major function of the network bythese authors.

According to Cochran and Brassard (1979)direct network influences come to the child

through the variety and range of persons withwhom the child has contact on a recurringbasis, either independently or with other fam-ily members. It is through the parents’ socialnetwork that children are exposed and pro-vided access to opportunities for interaction,different environmental settings, and newstimuli. From a developmental perspective, theyounger the child, the more likely this is to betrue.Other influences come to the child indirectly

via the family system, more specificallythrough the impact of network members onadult functioning in the parenting role (Coch-ran & Brassard, 1979). Several studies inves-tigated social support for young children andinfants via various family and parent behav-iors (Brandt, 1984; Pascoe & Earp, 1984; Zel-kowski, 1987). However, there has been verylittle investigation of social support for olderchildren or adolescents.One additional outcome for children as a

result of direct and indirect influences of theparents’ social network is the development ofthe child’s own network building skills (Coch-ran & Brassard, 1979). The notion of recipro-cal exchange is learned through social inter-action. Cochran and Brassard suggest that thenature of roles and power structures withinfamilies is such that the opportunity to prac-tice and observe these exchanges is limited.Therefore, development of the concept of re-ciprocity is dependent on observation of par-ent exchanges with network members and ac-cess to an arena in which to practice ex-changes. This link between the parents’ andchild’s social network would have added sig-nificance in later developmental stages whenpeer relationships increasingly influence childbehavior.

In the present formulation, it is hypothe-sized that a child’s social support directly andpositively influences child health and that par-ents’ support and child’s support are positivelyrelated. These relationships are extrapolatedfrom adult models of social support for individ-uals, the theoretical perspective of Cochranand Brassard (1979), and in consideration ofthe older child and adolescent’s developmentallevel.To summarize, the concepts of health and

interaction from King’s conceptual frameworkfor nursing were used as the basis for theoret-ical definitions of child health, family health,and social support. The linkages or- relation-ships among these concepts were derived fromKing’s framework as well as additional theo-retical and empirical literature. The new, the-oretical formulation of social support andhealth is shown in Figure 1. The hypotheses

141

Figure 1. Hypothesized relationships among concepts.

tested were as follows:

1. There is a positive and reciprocal relation-ship between family health and child health.

2. Parents’ support has a direct and positiveeffect on family health.

3. Child’s support has a direct and positiveeffect on child health.

4. There is a positive and reciprocal relation-ship between parents’ support and child’ssupport.

MethodologyThe target populationFamilies with children with IDDM were se-

lected as the target population because IDDMis an example of a chronic condition that haslong term, but not fatal, implications for thehealth of the family and child. For the family,there is monitoring of physiological symptoms,management of the medical regime, and re-sponsibility for day to day decision making.For the child, there is daily insulin injections,urine or blood glucose testing, and dietary re-strictions. The goals of treatment for diabeticchildren are threefold: the children should feel

good, function normally, and have stable glu-cose metabolism or metabolic control. Al-

though not curable, administration of exoge-nous insulin should correct the metabolic def-icit. However, multiple psychosocial factors,including structural and functional character-istics of the family system, have been associ-ated with adverse child health outcome in this

population (Anderson, Auslander, Miller, &

Santiago, 1981; Minuchin et al. 1975; Orr,Golden, Myers, & Marrero, 1983; White, Kohl-man, Wexler, Polin, & Winter, 1984).Interpretation of and generalizations from

empirical findings have been limited by meth-odological and theoretical issues (Anderson &Auslander, 1980; Johnson, 1980). With fewexceptions (see Anderson et al. 1981, Minu-chin et al. 1975; Newbrough, Simpkins, &Maurer, 1985), family-child relationships hadnot been approached from a theoretical per-spective. Metabolic control was the most fre-quent and often the only indicator of childhealth. A major limitation of diabetes relatedresearch has been the lack of attention toother variables that might influence or explainthe relationship between family factors, child

factors, and metabolic control, especially fam-ily interactions or linkages beyond the familysystem.

Social support for diabetic youths had notbeen extensively examined. However, diffi-culty with social relations with peers, atschool, and in the home had been reported inthe literature (Orr et al. 1983). The relation-ship with the health care team was identifiedas a concern of parents of diabetic youths(Hodges & Parker, 1987). Overall, there hadbeen little systematic examination of socialsupport for the child or family unit in thispopulation despite an identified need to do so.King’s conceptual model offers a useful

structure for examining the relationshipsamong family, child, and environmental vari-ables in families with children with IDDM. Thegeneral systems perspective directs attentionto factors outside the family that potentiallyhave an impact on family and child behavior.Additionally, King stresses the relationshipsbetween interaction with the environment andhealth as an important focus for nursing re-search and practice. King ( 1981 ) states:An understanding of the ways that humanbeings interact with their environments tomaintain health is essential for nurses; thisenables these professionals to promote health,to prevent disease, and to care for ill or disabledpeople (p. 2).

Overall, King’s framework is both useful andappropriate for examining social support andhealth in this population.

SampleFor this descriptive, correlational design a

convenience sample of 103 families with dia-betic children between the ages of 10 to 16was used. Additional criteria for entering thestudy were that (a) the child had been diag-nosed for at least 6 months, (b) the parentsand child were able to read at a 5th grade level,(c) the parents and child had no obvious devel-opmental or physical handicap, and (d) thechild had no other chronic illness. All familieswere English speaking. The sample was ob-tained from the medical practice of severalpediatricians and pediatric endocrinologistsand the diabetic clinic of a university hospitalin the midwest. Selected demographic char-acteristics of the sample are shown in Ta-ble 1.

OperationalizationTwo indicators of family health (functional

and structural) were selected for this studybased on King’s structural-functional and gen-eral systems perspective of the family. Thefunctional indicator was family functioning.Family functioning was operationalized as the&dquo;Family Distance from the Center&dquo; score deter-mined by parent and child scores on the Fam-

142

Table 1

Demographic Characteristics of the Sample

I Females n = 62, males n = 41.b Mothers n = 102, fathers n = 70.

ily Adaptability and Cohesion EvaluationScales (FACES) III (Olson, Portner, & Lavee,1985). Scoring is such that lower scores indi-cate more favorable family functioning.FACES III is a 20 item scale designed to

assess family adaptability and cohesion. Thesubconcepts that make up the adaptabilityand cohesion dimensions are power structure,role relationships, negotiation styles, relation-ship rules, feedback, emotional bonding,boundaries, coalitions, decision making, time,space, and friends (Olson et al., 1985). Thesesubconcepts are consistent with several of theconcepts King ( 1981 ) identified as importantconcepts for understanding social systems.The psychometric properties of the scale

have been addressed in several large studies._ Internal consistency, using Cronbach’s Alpha,for the adaptability and cohesion subscalesand scale total were reported as 0.62, 0.77,and 0.68, respectively (Olson et al., 1985).Test-retest reliability was reported as equal toor greater than 0.80. In the present study,alpha coefficients for cohesion, adaptability,and scale total ranged from 0.66 to 0.79, 0.70to 0.86, and 0.71 to 0.?6 for children, mothers,and fathers, respectively.Family composition was an indicator of fam-

ily structure which took into consideration thenumber, role relationship, and participation ofadults in the home. Family composition wasdetermined by parents’ self-report of maritalstatus and an open-ended interview questionon persons residing in the household. Severalcategories of family composition were identi-fied. The following categories were assignedthe numbers 1 to 4 in order to operationalizefamily composition: single parent; two parent,biological or nonbiological, one parent partici-pated ; two parent nonbiological, both parentsparticipated; and, two parent, biological, both

parents participated. (There were 22, 11, 7,and 63 families in the four family compositiongroups, respectively.) Higher scores are as-sumed to have a more favorable effect on fam-

ily and child outcome as has been suggestedin the literature (Egbuonu & Starfield, 1982;Kellam, Ensminger, & Turner, 1977). -

Parents’ social support was operationalizedas the total functional support score on theNorbeck Social Support Questionnaire (NSSQ)(Norbeck, 1984). The NSSQ was selected be-cause it taps content dimensions of supportthat are consistent with King’s perspective ofthe content of interaction. The scale alsomeasures perceived support available whichis appropriate, given King’s emphasis on per-ceptions. The one limitation of the scale, how-ever, is that it does not tap the reciprocity ofsupport.Child’s social support was likewise opera-

tionalized using the child’s scores. Use of theNSSQ with children has not been reported inthe literature but seemed feasible given the5th grade reading level of the scale (Fry, 1977).The functional support subscale consists of

six items, two each for aid, affirmation, andaffect. Subjects list all the important people intheir life and rate each person on each item

using a 4 or 5 point response scale. Scoringwas done by summing the ratings for eachitem and adding item scores together.Children, mothers, and fathers each com-

pleted the NSSQ. The following method fordetermining a parent support score was devel-oped for this study. Scoring was done jointlyfor mothers and fathers in two-parent fami-lies. A person listed by both the mother andfather was counted only once. The score forthat person was the average of the numberassigned by the mother and father. Therefore,in two-parent families, parent support meas-ures were determined from mother and fatherscores; and in single parent families, parentsupport measures were the scores of the singleparent. Higher scores indicated more supportperceived to be available to the parents orchild.

Psychometric properties of the NSSQ wereinvestigated in several large studies (Norbeck,Lindsey, & Carrieri, 1983). Internal consist-ency measures for the NSSQ were reported (foradult respondents) to be from 0.69 to 0.97 forthe functional support subscale. In the presentstudy alpha coefficients for the functional

support subscale for mothers, fathers, andchildren ranged from 0.78 to 0.84.There were seven indicators of child health.

These indicators tapped psychosocial, physi-cal, and biological dimensions of health. Thepsychosocial indicators were scholastic com-petence, social acceptance, athletic compe-tence, physical appearance, and self-worth op-erationalized as subscale mean scores on theSelf-perception Profile for Children (SPPC).

143

The SPPC (Harter, 1985) is designed to meas-ure competence in age appropriate social rolesand tasks in several domains of psychologicaland social development. As such, it is a verygood fit with King’s perspective of health.

Higher scores indicate a more adequate self-evaluation in each domain. Reliability, usingCronbach’s Alpha for the six subscales is re-ported as 0.71 to 0.84. In the present studyalpha coefficients ranged from 0.76 to 0.85.

Physical symptoms were operationalized asthe summative score on the Diabetic SymptomSurvey (DSS). The DSS was a modification ofa symptom checklist used in a previous studyby Eastman, Johnson, Silverstein, Spillar, andMcCallum (1983). The tool included symptomsassociated with hypoglycemia and hypergly-cemia identified by diabetic children. The childrated the frequency of 32 symptoms on a 5-point response scale. Higher scores indicatedmore frequent occurrences of physical symp-toms. Face validity was addressed and the toolwas pretested prior to use in the study. Thealpha coefficient in the present study was .90.Metabolic control was measured by the value

of the child’s hemoglobin Alc. Hemoglobin Alc(HgbAlc) is an indirect and retrospective meas-ure of glucose metabolism over the previous 2month period. It is considered a valid and re-liable indicator of metabolic control and is ex-

tensively used for research purposes (Cohen,1986). In general, a HgbAlc value of less than10% is considered good metabolic control, 10to 13% is considered fair metabolic control,and greater than 13% is considered poor met-abolic control (Anderson et al. 1981 ).

Procedures

All families meeting eligibility criteria werecontacted by this author. If the family waswilling to participate, an appointment to com-plete the interview and questionnaires wasmade at the family’s convenience. Data werecollected during in-home interviews by traineddata collectors or in the clinic if all familymembers designated to participate were avail-able. Parents signed informed consent papersand children provided verbal assent at the timeof data collection. Demographic data were ob-tained by interview. HgbAlc was obtained fromthe medical records. All other data were ob-tained from the self administered scales.

Analysis of dataThe relationship among the various indica-

tors of child health were addressed by corre-lational analysis and exploratory factor analy-sis. LISREL VI (Joreskog & Sorbom, 1984) sta-tistical program was used to analyze the causalmodel. The LISREL analysis is particularly ap-propriate for testing complex relationshipsamong variables using multiple indicators

(Boyd, Frey, & Aaronson, 1988). All analyseswere done with subprograms of SPSSX (1986).

Preliminary Analysis and Model Revisions

Several revisions were made in the causalmodel due to a combination of technical, meth-

odological, and theoretical limitations and con-straints. First, negative correlations amongthe physical, biological, and psychosocial di-mensions of health suggested both conceptualand measurement concerns about the indica-tors. That is, the indicators were not measur-

ing a single underlying construct, a necessaryrequirement of LISREL analysis. Conse-quently, a principal component factor analysiswith varimax rotation was performed with theseven child health indicators. In addition, theduration of illness and the child’s age wereentered into the analysis in order to determineif they were appropriate indicators of, or insome way related to, child health.The factor analysis resulted in a three factor

solution (eigenvalues > 1 ) which accounted for58% of the variance in child health. The psy-chosocial indicators loaded on the first twofactors and accounted for 45% of the variance.Duration of illness, metabolic control, andchild’s age loaded together on the third factor.This last factor appeared to be somethingother than health and was subsequently la-beled as illness factors. The physical symptomscale loaded on more than one factor. Becauseit was unclear whether the frequency of phys-ical symptoms reported by children was ahealth or illness factor, it was dropped fromsubsequent analysis.Based on those preliminary analyses, a ma-

jor change was made in the proposed model.The psychosocial indicators remained as in-dicators of child health. Age, metabolic con-trol, and duration of illness were identified asindicators of a different construct labeled ill-ness factors. Illness factors are hypothesizedto have a direct and negative effect on childhealth. Table 2 contains the correlations,means, and standard deviations of indicatorsused in the LISREL analysis.

Results

LISREL analysis provides information aboutthe overall fit of the hypothesized model to thedata, the strength and direction of the hypoth-esized relationships in the model, and the ad-equacy of the measures as indicators of con-structs (Boyd, Frey, & Aaronson, 1988). Sev-eral measures of goodness-of-fit were used toassess the extent to which the proposed modelfit the data. The Relative Likelihood Ratio

(RLR) is a ratio of the chi-square value to itsdegrees of freedom (Joreskog & Sorbom, 1984).It may be viewed as an indicator of how themodel being tested compares to an alternative,

144

Table 2

Correlations, Means, and Standard Deviations of Indicators in the Causal Model

I Scoring is such that lower scores are more positive outcome; r ~ 0.17, p <0.05; r ~ 0.23, p <0.01.

unconstrained model. Ratios in the range of 2to 1 or 3 to 1 are considered indicative of goodfit (Carmines & McIver, 1981).Joreskog and Sorbom (1984) also developed

a Goodness-of-fit Index (GFI) which measuresthe relative amount of variance and covari-ance jointly accounted for by the model. A GFIin the 0.90s is considered to indicate good fit.The proposed model had a x~ of 106.23 with

51 df for a RLR of 2.08 to 1. The GFI for theproposed model was 0.87. The GFI indicatedadequate fit, and the RLR indicated good fit ofthe model to the data. The coefficient of deter-mination, a measure of the strength of allstructural relationships jointly (Joreskog &Sorbom, 1984), was 0.42. Although accepta-ble, this suggests that a moderate amount ofthe variation in family health and child healthwas due to factors outside of the model.Figure 2 shows the results of the LISREL

analysis. The causal and noncausal relation-ships between the concepts are standardizedmaximum likelihood (ML) estimates and areanalogous to path coefficients and correla-tions. The structural parameters, representinghypothesized relationships, are tested with thet statistic (Joreskog & Sorbom, 1984).Three of the five hypotheses were supported:

parents’ social support had a direct and posi-tive effect on family health (hypothesis 2); par-ents’ and child’s social support were signifi-cantly related (hypothesis 4); and, illness fac-tors had a direct and negative effect on childhealth (hypothesis 5, added after model revi-sion). The reciprocal relationship betweenfamily health and child health (hypothesis 1)was not supported, nor was the direct effect ofchild’s support on child health (hypothesis 3).

Discussion

The findings provide some empirical supportfor the theoretical formulation of social sup-port and health in families with children with

IDDM and, therefore, indirect validation of

King’s conceptual framework for nursing. Inaddition, the findings provide direction for fur-ther theory development and research withthis, and perhaps other, clinical populations.

Social support and healthTwo hypotheses were tested in relation to

social support and health: social support forparents in relation to family health, and socialsupport for children in relation to child’shealth. It was predicted and supported thatparents who perceived higher amounts of-available support had significantly (t = 4.87,p <0.05) higher levels of family health. This isconsistent with findings identified in both thesocial support and family health literature (Pil-isuk & Parks, 1983; Pratt, 1976; Unger & Pow-

ell, 1980). Since this relationships was derivedfrom the propositions of King’s conceptualframework, the findings provide partial empir-ical support for the relationship between in-teraction and health. The findings are alsoevidence of the interacting relationship be-tween the interpersonal system (social sup-port) and social systems (the family) as pre-dicted by King’s framework.The hypothesized relationship between

child’s social support and child’s health wasnot supported by the study findings. King didnot differentiate between adults and childrenin conceptualizing the relationship betweeninteraction and health. Although there is am-ple evidence of the generally positive effect ofsocial support on adult health, there was adearth of theoretical or empirical knowledgeabout social support in older children and howsocial support influenced child health. As aresult, this hypothesis was tentative, at best.

In retrospect, there were several factors thatmay have accounted for the lack of relation-

ship between child’s support and child’shealth. From a theoretical perspective, it maynot have been appropriate to infer from adult

145

Figure 2. Causal model with maximum likelihood estimates.

models of social support and health to a childmodel. It could also be that support for theparents directly influenced child health, a re-lationship that was not tested in this formu-lation. The importance of supportive actionsby parents was described in the IDDM litera-ture which does suggest a direct effect (Bob-row, AvRuskin, & Siller, 1985; Schafer, Glas-gow, McCaul, & Dreher, 1983). However, thelink between support for parents, supportiveactions by parents, and child health was dif-ficult to interpret from previous research, dueto inconsistent conceptualization and meas-urement of support and child health. Clearly,this is an important area for further research.

Parent’s support and child’s supportA significant (t = 3.13, p <0.05) relationship

also occurred between parents’ support andchild’s support as hypothesized. Higher levelsof general support available to parents’ wereassociated with higher levels of general sup-port available to children.Again, King ( 1981 ) did not differentiate be-

tween interaction (social support in the for-mulation) for children as oppo’sed to adults.The hypothesized relationship was based onthe theoretical model proposed by Cochran andBrassard (1979) and consideration of the de-velopmental level of the youths in the popula-tion being studied. No studies were found thatdescribed relationships between support forparents’, and support for children in school-aged or older youths so empirical evidence forthe hypothesized relationship was also lack-ing.

The findings relating parents’ social supportand child’s social support have both substan-tive and theoretical import. First, this is animportant extension of the knowledge base ofsocial support for nursing and other disci-plines. The findings also suggest that the dif-ficulties with social relationships reported inthe literature for some diabetic youths may bemore related to the social skills learned withinthe family environment than to the illness!

Family health and child healthThe hypothesized relationship between fam-

ily health and child health also was not sup-ported by the study findings. The lack of asignificant relationship between family healthand child health was unexpected because ithad strong theoretical support. This relation-ship was predicted based on King’s ( 1981 ) gen-eral systems perspective, particularly the fam-ily serving as the primary environment for thedeveloping child. The lack of empirical supportfor the relationship in this formulation prob-ably was due to a combination of factors.One potential difficulty may have been with

selection of indicators of the constructs. Al-

though the use of multiple indicators to mea-sure constructs is a strength of LISREL analy-sis, selecting an adequate number of appropri-ate indicators is somewhat of an art (Bentler,1980). The measurement properties of familyhealth might be improved by using more thantwo indicators. Some of the difficulty couldalso have been due to an incomplete concep-tualization of child health. After dropping thephysical symptom measure, the remaining in-

146

dicators tapped only the psychosocial dimen-sions of health. Additional indicators of otherdimensions of health need to be identified. Inaddition, evaluation of self-perceived healthstatus would be especially relevant givenKing’s emphasis on perceptions.Aside from the measurement and concep-

tual issues, it is very possible that the relation-ship between family health and child healthwas not a direct effect as hypothesized. It couldbe the case that family health has an indirecteffect on child health. The formulation, astested, did not permit examination of indirecteffects. Clearly, there is a need to expand theparameters of the formulation. Again, basedon reexamination of the literature, self-care inrelation to diabetic management may be animportant variable linking family health, par-ents’ support, child’s support, and child healthin this population.

Illness and health

Although serendipitous, the negative andsignificant (t = 2.53, p <0.05) effect of illnessfactors on child health also provides supportfor King’s conceptual framework. King (1981)says that illness was one factor that influenceshealth. For this sample, youths who had IDDMlonger, were older, and had poorer metaboliccontrol had less adequate psychosocial health.

In addition to supporting King’s framework,the findings reinforce the postulate that nurs-ing knowledge of health depends on a clearconceptualization of health and cannot be in-ferred from the absence of overt pathology.

. From review of the IDDM literature, healthwas generally defined by medical standards.In view of nursing’s multidimensional view ofhealth (Reynolds, 1988; Smith, 1983), knowl-edge generated to date may not be sufficientas a basis for nursing practice. However, as-pects of disease and illness are of concern tonursing because they influence health. Al-

though inclusion of both health and illnessindicators is a strength of this formulation andKing’s framework, additional attention needsto be directed toward illness indicators at boththe conceptual and operational level.

Summary and Conclusions

Social support has emerged as an importantdeterminant of the health and functioning offamilies and individuals. Interest in social sup-port is shared by nursing and other healthsciences. In order to incorporate concepts ofsocial support into nursing practice, a soundknowledge base of how and under what con-ditions social support influences health isneeded.The purpose of the study described here was

to generate and test a theoretical formulationof social support, family health, and child

health in families with children with IDDM.An additional aim of the study was to validateKing’s ( 1981 ) conceptual framework for nurs-ing. King’s conceptual framework was used asthe basis for defining, selecting indicators, andproposing relationships among the concepts.Additional theoretical perspectives and theempirical findings from the IDDM literaturewere considered in development of the for-mulation. This process of developing middlerange theories from nursing’s broader concep-tual frameworks has particular utility for gen-erating knowledge unique to the discipline ofnursing.

Initial testing of the formulation demon-strated the positive effect of parents’ socialsupport on family health, the negative effectof illness factors on child health, and the pos-itive and reciprocal relationship between par-ents’ support and child’s support. These find-ings provide empirical evidence which sup-ports King’s conceptual framework in relationto the effect of interaction on health, at leastfor adults. In addition, the findings extend theknowledge base of social support and healthfor nursing in a context that had not beenextensively addressed in the literature.Several of the hypothesized relationships

were not supported. The lack of a significantrelationship between family health and childhealth, and child’s support and child health inthe present study seemed primarily related toselection and measurement of indicators andto several missing variables that might explainor mediate the relationships. This lack of re-lationship seemed less related to the theoreti-cal relationships predicted by King’s (1981)conceptual framework. Based on the findingsof this study, new directions for expanding andrefining the research model were identified. Atthe present time, this author is testing a modelthat includes child’s self-care as an interven-

ing variable between family health and childhealth and child’s social support and childhealth. Additional indicators of child healthand illness are being used in order to addressthe limitations identified as a result of prelim-inary testing of the formulation.

In conclusion, development and testing of amiddle range theory of social support, familyhealth, and child health has made a contribu-tion to the knowledge base of professionalnursing. Several concepts which were of inter-est to other disciplines were investigated froma nursing perspective. The proposed formula-tion should be viewed as a preliminary approx-imation of a more complex process. However,simplification of complex relationships is bothuseful and necessary especially in the earlystages of theory building and testing. Althoughadditional studies are needed, nursing is closerto understanding the complex nature of therelationship between the social environmentand health.

.47

References

Anderson, C. (1982). The community connection: The im-pact of social networks on family and individual func-tioning. In F. Walsh (Ed.), Normal Family Process, (pp.425-445). New York: Guilford Press.

Anderson, B. J., & Auslander, W. F. (1980). Research ondiabetics management and the family: A critique. Dia-betes Care, 3, 696-702.

Anderson, B. J., Miller, J. P., Auslander, W. F., & Santiago,J. V. (1981). Family characteristics of diabetic adoles-cents: Relationship to metabolic control. Diabetes Care,4, 586-594.

Barnhill, L. R. (1979). Healthy family systems. The FamilyCoordinator, 28, 94-100.

Barnard, K. E., Brandt, P. A., Raff, B. S., & Carroll, P.(Eds.), (1984). Social support and families of vulnerableinfants. New York: March of Dimes Birth Defects Foun-dation.

Bentler, P. M. (1980). Multivariate analysis with latentvariables: Causal modeling. Annual Review of Psychol-ogy, 31, 419-456.

Berkman, L. F., & Syme, S. L. (1979). Social networks,host resistance, and mortality: A nine-year follow-upstudy of Alameda County residents. American Journalof Epidemiology, 109, 186-204.

Bobrow, E. S., AvRuskin, T. W., & Siller, J. (1985). Mother-daughter interaction and adherence to diabetes regimes.Diabetes Care, 8, 146-151.

Boyd, C. J., Frey, M. A., & Aaronson, L. S. (1988). Struc-tural equation models and nursing research: Part I.

Nursing Research, 37, 249-252.Brandt, P. A. (1984). Stress-buffering effects of socialsupport on maternal discipline. Nursing Research, 33,229-234.

Carmines, E. G., & Mclver, J. P. (1981). Analyzing modelswith unobserved variables: Analysis of covariance

structures. In G. H. Bohrnstedt & D. C. Borgatta (Eds.),Social measurement: Current issues, (pp. 65-115).Beverly Hills, CA: Sage.

Cobb, S. (1976). Social support as a moderator of lifestress. Psychosomatic Medicine, 39, 300-313.

Cochran, M. M., & Brassard, J. A. (1979). Child develop-ment and personal social networks. Child Development,50,601-616.

Cohen, M. (1986). Diabetes and protein glycosylation:Measurement and biological relevance. New York:Springer Verlag.

Diamond, M., & Jones, S. L. (1983). Social support: Areview and theoretical integration. In P. L. Chinn (Ed.),Advances in nursing theory development (pp. 235-249). Rockville, MD: Aspen.

Eastman, B. G., Johnson, S. B., Silverstein, J., Spillar,R. P., & McCallum, M. M. (1983). Understanding hypo-and hyperglycemia by youngsters with diabetes andtheir parents. Journal of Pediatric Psychology, 8, 229-243.

Egbuonu, L., & Starfield, B. (1982). Child health and socialstatus. Pediatrics, 69, 550-557.

Fawcett, J. (1989). Analysis and evaluation of concep-tual models of nursing (2nd ed.). Philadelphia: F. A.Davis.

Feetham, S. L. (1984). Family research: Issues and direc-tions for nursing research. In H. H. Werley, & J. J.

Fitzpatrick (Eds.), Annual review of nursing research(Vol. 2, pp. 2-25). New York: Springer.’

Fry, E. B. (1977). Fry’s readability graph: Clarification,validity, and extensions to level 17. Journal of Reading,249.

Gottlieb, B. H. (1983). Social support strategies: Guide-lines for mental health practice. Beverly Hills, CA:Sage.

Harter, S. (1985). Manualfor the self-perception profilefor children. Colorado: University of Denver.

Hodges, L. C., & Parker, J. (1987). Concerns of parentswith diabetic children, Pediatric Nursing, 13, (1) 22-28.

House, J. S. (1981). Work stress and social support.

Reading. MA: Addison-Wesley.Hyman, R., & Woog, P. (1982). Stressful life events and

illness onset: A review of crucial variables. Research in

Nursing and Health, 5, 155-163.Johnson, S. B. (1980). Psychosocial factors in juvenile

diabetes: A review. Journal of Behavioral Medicine, 3,95-116.

Joreskog, K. G., & Sorbom, D. (1984). LISREL VI analysisof linear structural relationships by the method ofmaximum likelihood. Mooresville, IN: Scientific Soft-ware.

Kahn, R. L., & Antonucci, T. C. (1980). Convoys over thelife course: Attachment, roles and social support. In P.Baltes & O. Brim (Eds.), Life-span development andbehavior, (pp. 251-286). New York: Academic Press.

Kellam, S. G., Ensminger, M. E., & Turner, R. J. (1977).Family structure and the mental health of children.Archives of General Psychiatry, 34, 1012-1022.

King, I. (1981). A theoryfor nursing: Systems, concepts,process. New York: John Wiley & Sons.

King, I. (1983). King’s theory of nursing. In I. Clements &F. Roberts (Eds.), Family health: A theoretical ap-proach to nursing care, (pp. 177-188). New York: JohnWiley & Sons.

MacElveen-Hoehn, P., & Eyres, S. J. (1984). Social supportand vulnerability: State of the art in relation to familiesand children. In K. E. Barnard, P. A. Brandt, B. S. Raff,& P. Carroll (Eds.), Social support and families of vul-nerable infants. New York: March of Dimes Birth De-fects Foundation.

Minuchin, S. (1974). Families and family therapy: Astructural approach. Cambridge, MA: Harvard Univ.Press.

Minuchin, S., Baker, L., Rosman, B. L., Liebman, R.,Milman, L., & Todd, T. C. (1975). A conceptual model ofpsychosomatic illness in children. Archives of GeneralPsychiatry, 32, 1031-1038.

Newbrough, J. R., Simpkins, C. G., & Maurer, H. (1985).A family developmental approach to studying factors inthe management and control of childhood diabetes. Di-abetic Care, 8, 83-92.

Norbeck, J. S. (1984). The Norbeck social support ques-tionnaire. In K. E. Barnard, P. A. Brandt, B. S. Raff, &P. Carroll (Eds.), Social support andfamilies of vulner-able infants (pp. 45-57). New York: March of DimesBirth Defects Foundation.

Norbeck, J. S., Lindsey, A. M., & Carrieri, V. L. (1983).Further development of the Norbeck social support ques-tionnaire : Normative data and validity testing. NursingResearch, 32, 4-9.

Olson, D. H., Partner, J., & Lavee, Y. (1985). FACES III.St. Paul: University of Minnesota Press.

Olson, D. H., Sprenkle, D., & Russell, C. (1979). Circum-plex model of marital and family systems I: Cohesionand adaptability dimensions, family types and clinicalapplication. Family Process, 18, 3-28.

Orr, D. P., Golden, M. P., Myers, G., & Marrero, D. G.

(1983). Characteristics of adolescents with poorly con-trolled diabetes referred to a tertiary care center. Dia-betes Care, 6, 170-175.

Pascoe, J. M., & Earp, J. A. (1984). The effect of mother’ssocial support and life changes on the stimulation oftheir children in the home. Public Health Briefs, 74,358-362.

Phillips, J. R. (1988). The reality of nursing research.Nursing Science Quarterly, 1, 48-49.

Pilisuk, M., & Parks, S. H. (1983). Social support andfamily stress. Marriage and Family Review, 6, 137-156.

Pratt, L. (1976). Family structure and effective healthbehavior: The energized family. Boston: Houghton Mif-flin.

Reynolds, C. L. (1988). The measurement of health innursing research. Advances in Nursing Science, 10(4),23-31.

Schafer, L. C., Glasgow, R. E., McCaul, K. D., & Dreher,M. (1983). Adherence to IDDM regimes: Relationship topsychosocial variables and metabolic control. Diabetes

148

Care, 6, 493-498.Silva, M. (1986). Research testing nursing theory. Ad-vances in Nursing Science, 9(1), 1-11.

Smith, J. (1983). The idea of health: Implications for thenursing profession. New York: Teachers College Press.

SPSSX users guide (2nd ed.). (1986). New York: McGraw-Hill.

Unger, D. G., & Powell, D. R. (1980). Supporting familiesunder stress: The role of social networks. Family Rela-tions, 29, 566-574.

Weiss, R. S. (1974). The provisions of social relationships.In Z. Rubin (Ed.), Doing unto others (pp. 23-24). Engle-

wood Cliffs, NJ: Prentice-Hall.Whall, A., & Fawcett, J. (1988. May). Family theory de-velopment in nursing: State of the act and science.Paper presented at the International Family NursingConference, Calgary, Alberta, Canada.

White, K., Kohlman, M. L., Wexler, P., Polin, G., & Winter,R. J. (1984). Unstable diabetes and unstable families: Apsychosocial evaluation of diabetic children with recur-rent ketoacideosis. Pediatrics, 99, 41-45.

Zelkowski, P. (1987). Social support and aggressive behav-ior in young children. Family Relations, 36(2). 129-134.