Joumal of Advanced Nursmg, 1993,18,1118-1124

Insights horn latent partition analysis intocategories inherent in wellness-illness

Louise Jensen PhDAssocmte Professor

and Manon Allen PhD RNFaculty of Nurstng, Untverstty of Alberta, Edmonton, Alberta, Canada

Accepted for pubbcahon 26 October 1992

JENSEN L & ALLEN M (1993) Journal of Advanced Nursing 18,111&-1124Insights from latent partition analysis into categories inherent inwellness—illnessHov*r health-disease is perceived or conceptualized is important for nursmgresearch There is increasmg evidence that mdividual representations areimportant m construchng the expenence of health-disease What is the personalsabency of health-disease for the mdividual? To explore the pattems of mearunginherent m health-disease, a card sort was undertaken among 15 healthymdividuals and 15 mdividuals with chroruc renal disease Both groups vietegiven IS cards to sort twice once for when they felt 'well' and agam for whenthey felt 'ill' The theoretical basis underlymg the items of the card sort was amodel of wellness-illness bemg developed Latent partition analysis was used tocluster the concepts from each data set followed by multi-dimensional scalmg toanalyse the structure of the lntercategory probabihty estimates A possibleunidimensional pattem of mearung (harmony) emerged for the 'weU' data and atwo-dimensional pattem (disharmony and optimism) for the 'lU' data Thisrepresents a prehmmary step in the development of a theorehcal model thatwould permit assessment of the meanmg of health-disease for the mdividual

CENTRAL TO NURSING consensus on what the relahonsbp is and, m fad, there isbttle agreement on what these concepts actually mean

Health, disease, weUness and lUness have long been recog- p^^^^ ̂ ^^^^^^ f̂ empincal and theoretical bterahire,mzed as central to nursmg Numerous researdi shidies m j^^^jj^ ^^^ ^^^^^ ^ j ^^^^^ ^^^ ^ t comparedvanous disapbnes, with diffenng concephial models, and contrasted m an effort to seek concephial clanficahonpopulahons and methods, have been conduded to identify ^^ ^^^^^^ subsequently led to the fonnulahon of a pre-the fadors that influence or predetennme an mdividual's ^^^^ ^ ^ ^ j ^f ^ellness-ilbiess The present shidyresponse to health-^sease The results of these shidies ^presents the first phase of validahng pattems inherent mhave shown no consistent predidors of wellness-iUness, wellness-illnessalthough there is mcreasmg evidence that individualrepresentahons of the expenence of health-disease areimportant m conshuding pattems of weUness-iUness \ MODEL OF WELLNESS-ILLNESSAlthough it IS readily recognized that a relationsbp existsbetween health, disease, weUness and iUness, there is lack of The imfoldmg expenence is depided as a genenc paradigm

Correspondence DrL Jensen Assocutte Professor, FacuUy of Nursm,.Un.,^ty of ^ ^ ^ health, disease, we lbeSS and llbeSS eXlStmg m aAn>erbt,Edmonbm Alberta T6G2G3, Canada dialectic relahonsbp Tbs relationsbp symbobzes the

1118

Perceptions erf weUness and tllnex

synthesis of subjechve and objechve perspechves In tbsparadtgm, health, an objechve process, ts charactenzed bystabtltty, balance and tntegnty of funchomng Diseaseembodies health and is typified as breakdown, disturbanceor some alteration m funchonmg

Wellness, in contrast, is a subjechve expenence and isperceived as congruence between one's own possibilitiesand one's actual prachces and bved meanmgs (Bermer &Wrubel 1989) Similarly, lUness is the human expenence ofdisease Illness mcorporates life meamng and soaal par-hapahon, makmg health and welkess inherent aspects(Ruffing-Rahal 1985) Therefore, the process of weUness-lllness mvolves creatmg a pattem of meamng m everydaybfe

WeUness-^ilbess anses out of mtrapersonal, mterper-sonal, health-disease related, and extrapersonal factors(Lipowsb 1983) Intrapersonal factors lndude such aspersonality, past expenences and emotional state Inter-personal factors include such as soaal support andrelahonsbps Health-cbsease related factors mdude suchas health promotion onentation, funchonal status, visi-bibty of health-disease, and seventy and prognosis ofdisease Extrapersonal factors mdude such as soaoculturaland economic elements

Wellness-ilbiess is seen as a human expenence of actualor perceived funchon-dysfunchon, through the mter-action of cogmtive-affechve themes Cogmhve-affectivethemes afifed the pattemmg of perceptions regardmghealth-disease The umque configuration that results fromthese mteractions is reflected m vanous pattems of mean-ing that influence the lived expenence and subsequentbehaviours exbbited Cc^mhve themes ('what I think') havebeen identified as comprehensible-mcomprehensible,manageable-unmanageable and meanmgful-meaiungless(Antonovsky 1987) Affechve themes ('what I feel') wbchmay play a role m the mdividual's percephon of health-disease mdude joy-sadness, acceptance-disgust, fear-anger and anhapation-surpnse (Plutciiik 1980)

D E V E L O P I N G THE M O D E L

Method

The prelimmary step m model development was to venfythe proposed cogmtive and affechve themes These pattemsof meamng inherent m weUness-iIlness were exploredtbough the use of a card sort The items developed for thecard sort ciepided the cogmhve-affedive themes proposedm the model One theme was prmted on each of the 28 cardsTwelve of the cards represented items from the eogmhvethanes wble tbe remainmg 16 cards represented items fromthe affechve themes (Table I)

Table 1 Cogmtive and affechve themes

CognitiveManageable

Meamngful

Comprehensible

Unmanageable

Meamngless

Incomprehensible

AffectiveJoyful

Acceptance

Fear

Antiapahon

Sadness

Disgust

Anger

Surpnse

1 Life offers choices2 Everythmg IS under control3 Life has meamng4 Life has purpose5 Life IS predictable6 Life makes sense7 Life IS unfair8 Life IS overwhelmmg9 Life IS bonng

10 Life IS painful11 Day-to-day life IS chaohc12 Life IS a guessing game

13 I feel happy14 I enjoy life15 I feel loved16 I receive recogmtion17 I feel apprehension18 I feel helpless19 I have hope20 I look forward to tomonow21 I feel alone22 I feel sad23 I hate bfe24 I resent life25 I feel angry26 I feel hoshle27 I feel bewildered28 I feel suspiaous

The card sort was undertaken among 15 healthy indi-viduals and 15 mdividuals with chrome renal disease Thegroup of healthy mdividuals mcluded I I females and fourmales, with a mean age of 35 20 years (range 20 to 59years) Twelve members of the healthy group wereemployed m a wide range of occupations The group ofmdividuals with chrome renal disease mduded six malesand mne females, with a mean age of 50 30 years (range 20to 83 years) Eleven of the 15 mdividuals with chrome renaldisease were not employed

All members of the healthy group and the group withchrome renal disease were given tbe cards to sort twiceonce for when they would descnbe themselves as 'weU' andagam for when they would descnbe themselves as 'lU' Themstrudions for the card sort mduded

On these cards are vanous phrases Use these cards to descnbehow you think and feel when you are well Sort the cards intothree piles In one pile put those cards that best descnbe you

1119

L JensenandM Allen

when you are well, m the second pile put those that sranewhatdescnbe you, and m the tbrd pile put those that least descnbeyou when you are well Now, sort each pile of cards m theorder of what you think and feel most often at the top to leastoften at the bottom

Now, take the same cards and use them to descnbe howyou tbnk and feel when you are ill Sort the cards mto threepiles agam In the first pile put those cards that best descnbeyou when you are ill, m the second pile put those that some-what descnbe you, and m the third pile put those that leastdescnbe you when you are ill Now, sort each pile of cards morder of what you tbnk and feel most often at the top to leastoften at the bottom

Analysis of dataFor the purposes of analysis, data were uuhally dividedmto two groups 'well' data, wbch represented the com-bmed (chroruc disease and healthy) group card sort con-ducted m the 'well' state (M = 29), and 'ill' data, whichrepresented the combmed (chronic disease and healthy)group card sort conduded m the 'ill' state (n = 30) Onemember of the group with chronic renal disease completedthe card sort task for only the 'lU' state Another senes ofanalyses were conducted on data divided mto four groupshealthy group dunng the 'well' state (n = 15), healthygroup dunng the 'ill' state (M = 15), chroruc disease groupdurmg the 'well' state (n = 14), and chroruc disease groupdunng the 'ill' state (n = 15)

Wiley's (1967) latent parhtion analysis was used toduster the concepts from each of these six data sets Latentpartition analysis enables an examinahon of the modalpopulahon categonzation of a set of items Latent partihonanalysis used for evaluation purposes can best be descnliedas searchmg for sms of commission and for sms ofomission

In the first mstance, one wants to make sure conceptsclustered together by subjects really belong together on aconceptual basis Second, one looks for concepts thatshould be placed together that have not been connededTo venfy the categonzahon, the confusion matnces of theSIX analyses usmg latent parhhon analysis were then runthrough mulh-dunensional scabng analysis with a rruru-mum and maximum dimension of 1 and 4, respechvelyMulh-dimensional scaling and latent partihon analysis canbe integrated by usmg latent partihon analysis to deter-mine the qualitahve latent parhhon and then usmgmulti-dimensional scabng to analyse the stmcture of themtercategory probability estimates (Miller ei al 1986)

RESULTS AND DISCUSSIONIn the mterest of brevity, a detailed analysis of results ispresented for only the combined 'well' and 'dl' data Tbs is

followed by a more general discussion of the results of theremainmg four subgroup analyses

Analysis of combined 'well' data

The f̂ ^pbcahon of latent parhhon analysis to the combmed'weir data yielded four categones wbch appeared to fit thedata and provide meanmgful results for evaluahon (Table2) The column 'l-delta'*2' m the latent parhtion matnx isan estimate of the probabibty that a parhcular item and anyother item from the same latent category will he sorted mtothe same observed category The first category representsdustermg of both cogruhve and afifective posihve itemswhich appear to fit with a notion of 'exuberance' Thesecond category appears to represeit a nohon of 'tolerableconfusion' The tbrd category dusters negahve items rep-resentmg a nohon of 'unhappmess' The fourth categoryalso dusters items that provoke negahve feebngs,represenhng a nohon of 'resentment'

The confusion matnx (Table 3) is the probabibty that anitem fi-om one latent category is sorted with an item from adifiFerent latent category It is a measure of the distmctnessof the latent categones from each other An analysis of theconfusion matnx mdicates that there was a tendency toconfuse the second with the other three categones Thismakes soise m view of the apparent dustenng of positiveand negahve thoughts and feebngs witbn category 2The negative connotation depicted in categones 3 and 4IS distmct from the nohon of exuberance repre^nted mcategory 1

The result of the mulh-dimensional scabng analysis(MDSCALE 0 and ALSCAL) supports the nohon that thereIS possibly one dimension refleded m the data (Table 4)The stress for one dunension is 0169 with an RSQ of0-856 Although a sense of 'unhappmess' and 'resentment'emerged, tbs may not necessanly reflect a distmchon, butrather a natural tempermg of 'exuberance' for a number ofmdividuals The four latent categones of tbs dimensionwere felt to represent a contmuum of thoughts and feelmgsassociated with 'harmony'

Analysis of combined 'ill' data

The ̂ pbcahon of latent parhhon analysis to the combmed'ill' data yielded six categones (Table 5) The first categoryrepresents the nohon of 'despai/ The second categoryrepresents a dustenng of negahve feelmgs wbch appear tofit with a nohon of 'resentful disruption', while the thirdcategory dusters negahve thoughts representmg a notionof 'bearable ad^^^5Ity' The fourth and fifth categonesduster posihve thoughts and feelmgs representmg the

1120

Percephons ofwelbiess and illness

Table 2 Latent partitionmatnx combmed 'well' data(n = 29)

Stimulus l-delta**2

2 manageable 2 0 4536 comprehensible 2 0 639

16 acceptance 2 0 55219 anticipation I 0 665I manageable 1 0 7024 meanmgful 2 0 725

15 acceptance 1 0 74713 joy 1 0 7733 meaningful 1 0 790

20 anticipahon 2 0 89914 joy 2 0 869

27 surprise 1 066412 mcomprehensible 2 0 45921 sad 1 0 6719 meanmgless 1 0 7255 comprehensible 1 0 528

17 fearl 0 69611 incomprehensible 1 0615

22 sad 2 064618 fear 2 0 68510 meanmgless 2 0 76625 angry 1 0 861

7 unmanageable 1 0 853

26 angry 2 0 81428 surpnse 2 0 5918 unmanageable 2 0 656

24 disgust 2 090323 disgust 1 0 958

0 6040 707

0 782

0 9090 927

0 956

0 993

1007

1028

1109

1111

0 082

0 002

0 081

01610154

01160107

0 068

0 051

0047

0 025

0 053

0 079

0 015

0003

0053

0 032

05110-661

0 241

0141

0025

0016

-0129

-0018

- 0 061

-0117- 0 284

0 5290 584

0607

I 0131214

1247

1326

0 355

0 266

-0148

-0105

- 0 226

0 401

0135

0 237

-0168

- 0 025

-0058- 0 091-0090

-0103

0182

0 097

0184

0 008

-0008

-0103

0016

0472

0 470

0 203

-0117-0171

-0196

- 0 050

0657

0 739

1065

1236

1300

0358

0 288

0 241

-0047

-0166

-0124- 0 222

0073

0015-0112

-0032

-0054

0 018

0049

0170

0139

0107

-0-150

03110 301

— 0214

0113

- 0 228

0 0390051

0044

-0112

-0119

0 422

0 510

0 527

1047

1148

Table 3 Confusion matnx combined 'well' data (n = 29)

1

2

3

4

1

0 765

0172

0086

0024

2

0172

0531

0480

0462

3

0 086

0480

0 743

0629

4

0 024

0462

0629

0 909

notions of 'contentment' and 'hope', respechvely Thesixth eategory is one of 'apprehension'

An analysis of the eonfusion matnx mdieates a tendeneyto eonfuse eategones 2,3, 4 and 6 (Table 6) The results ofthe multi-dimensional seabng analysis suggests the nohonof possibly two dimensions refleeted m the data (Table 4)The stress for two dimensions is 0 245 with an RSQ of

0 510 One dunension is represented by categones I and2, and categones 3 and 6 Tbs appears to represent aconhnuum of thoughts and feebngs related to illness assoa-ated with a sense of 'disharmony' The second dimension isrepresented by categones 4 and 5 suggestmg a notion of'optimism'

Overview of analysis of other data sets

It appeared that results indicate stmctural differences mpattems of meamng between the healthy mdividuals andthe mdividuals with chrome renal disease, as weU as withmgroup between different states (weU v lU) Tbs wasdemonstrated m both the latent partihon analysis and theresults of processing the confusion matnces through multi-dimensional scaling (Table 4) As shown, the remainmgdata sets resulted m dififerent numbers of categones as weUas different dusters of items witbn the categones It is

1121

LJensmcmdM Allen

Table 4 Multi-dimensicmalscabng analysis

Combmed well(« = 29)

Combmed ill(n = 30)

Healthy well (n= 15)

Healthy ill(n=15)

Cbonic well (« = 14)

Cbonic ill(n=15)

Number ofcategones

4

6

4

7

6

7

Dimensions

1

21

4321

4321

4321

4321

MDSCALE 0

Stress

0-0080O0800070289

000900980-1790415

0-00800880-1940374

0-07101310-2160437

ALSCAL

Stress

0169

02450-452

0-000

00950 3130478

0-2290-530

0-1920 263

RSQ

0-856

05700385

1000

0 8520-4470443

04940 369

0 7340 737

recognized that analyses of these speafic groupmgs isbmited due to the sample size However, of mterest is thatthe data analysis assoaated with 'well' data consistentlyresulted m less dusters of items Chealthy—well', 4, 'chronic-well', 6) than the 'lU' data Chealthy-ill', 7, 'chrom^ll', 7)

Also, the 'ill' data tended to yield more dimensionsChealthy-^ll', 3, 'chronic-ill', 2) than the 'well' dataChealthy—well', I, 'chronic—well', 2), when the respechveconfusion matnces were processed through mulh-dimensional scabng The dunensicms ebated fi'om the 'well'data appear to represent thoughts and feebngs assoaatedwith nohons of tiarmony' and 'optimism' On the otherhand, the dimensions ebated fi-om the 'ill' data seem torepresent nohons of 'disharmony' ten^>ered with 'guardedoptimism' (the liealthy—weU' group — liarmony', the'chroruc-well' group — 'disharmtmy' and 'optmuan', the'healthy-ill' group—'disharmony', 'ophmian' and 'dis-may', the 'chronIC-̂ ll' group — 'dishannony' and 'guardedoptimism') The overview of categones sid dimermons iar\he vanous groupmgs is shown m Tdbk 7.

In addihon, the fi-equency distribuhon of how iteaiswere ranked withm each pile was assess«i Among theccHnbmed 'well' individuals, the items most fi^quoitly

sorted m the 'best descnbes you' category were 'life haspurpose' (item 4), 1 enjoy bfe' (item 14) and 'I feel loved'(item 15), the items most frequently sorted m the 'leastdescnbes you' category were 1 hate life' (item 7), 'I feelsuspiaous' (item 28) and 1 feel hostile' (item 26), whereas,the items frequently failmg m the 'somewhat descnbesyou' category were 1 resent bfe' (item 24) and 1 feel angry'(item 25)

The combmed 'ill' mdividuals most frequently sorteditems 1 feel loved' (item 15), 1 feel alone' (item 21), 'bfe haspurpose' (item 4) and 'Life is painful' (item 10) mto thecategory wbch Iwst descnt>es you', items 'I hate bfe' (item23), 'I feel happy' (item 13) and 1 resent life' (item 24) mtothe category which least descnbes you', and items 'every-thmg IS under control' (item 2), 1 feel suspiaous' (item 28)and 1 have hope' (item 19) mto the 'somewhat descnbesyou' category

Self-reports of healtii-disease status

FoUowmg the card sort, subjects were asked to rate then-health status CHI a scale from I (worst) to 10 (best) Inadditk»i, they were asked to descnbe their health-disease

1122

Percephons of wellness and tllness

Table 5 Latent parhhon matnx combmed 'ill' data (n = 30)

Stimulus l-delta"2

82122

11

2528

726272423

1895

1210

161

2014132

6193

154

17

urunanageable 2sad Isad 2

mcomprehensible 1angry 1surpnse 2unmanageable 1angry 2surpnse 1disgust 2disgust 1

fear 2meanmgless 1comprehensible 1mcomprehensible 2meanmgless 2

acceptance 2manageable 1antiapation 2joy 2)oylmanageable 2

comprehensible 2anticipation Imeamngful Iacceptance 1mearungful 2

fearl

0 5320 7490 690

0 5400572064706050 59605750 7600 856

0-6020549045906400 785

055005330 6740 7110 7010618

0 7560 5930 78006390899

0845

048506460 713

01370144

- 0 5010088

- 0 22601400023

-0012

04750 219

-0-301-0165

0118

0-0490 015

-0141-0049-0106

0151

0 055-0012-0018

00230 020

0021

0286-0045

0097

04200 50306020 9210928096411481480

Olio0 2910 243

- 0 092- 0 075

0379-0003-0064

0 5100 246

-0231

-0324-0121-0187

00120 052

-0163

-01240378

- 0 040

0 313- 0 227

0 436- 0 092

0084- a 4 7 5

0 376- 0 098

0 4850 5340 6300 7731182

-0107-0-086

0201- 0 339

00740035

- 0 0260301OIOI

-0048- 0 079

-0069

01800134

- 0 004

-00200 2360053

- 0 45301340-094

- 0 1 4 8-0067

- 0 0 7 800720373

-0076- 0 035

0 6960 7090 8120 84710241115

12720 2330129

- 0 255- 0 0 6 9

0040

0 036-0095

0074

- 0 060- 0 1 0 8-0025

0 249-0132- 0 0 1 0- 0 002- 0 0 6 9

0103- 0 1 4 0

0 053- 0 0 0 9- 0 0 1 0

021801870 2520 254

- 0 0 1 7- 0 202

- 0 1 9 806610 94710051186

0056

0-147- 0 0 1 3

0-112

0 2370 50104320 2530 2180 255

-0-367- 0 203

- 0 1 1 10-007

- 0 1 2 1- 0 564- 0 1 1 0

- 0 2 3 90202

- 0 0 6 9- 0 2 1 8- 0 210

0108

0 220- 0 0 6 0

00270 224

-0O73

1130

Table 6 Confusion matnx combmed 'ill' data (n = 30)

123456

I

10770-4080-4680-18401360469

2

04080-5550-39202640-1730-382

3

0-4680-3920-6020-30002450399

4

0-1840 2640-3000-6180-4770-247

5

0-1360173024504770 7290242

6

046903820 3990-2470 2420 746

as well as outbne the words they would use to descnbetheir health-<lisease The combmed mean rating of healthstatus for both groups was 6 70 (range 3 to 10), with

healthy mdividuals havmg a mean rating of 8 10 (range 5 to10) and mdividuals with cbomc renal disease of 5 40(range 3 to 7), respedively

It IS mterestmg to note that although most mdividualsdid not rate themselves as expenencmg their 'best' health,they did not respond to the ratmg m terms of 'worst' healtheither Many said they had 'nuisances to health', 'didn't feelweir or 'were sick' rather than Iwmg ill' One healthysubject stated 'I was m the hospital and had my gallbladder out, but I wasn't ill Illness is when you have allsorts of tubes and stuff, somethmg senous that hospitalizesyou for a long hme'

Individuals with cbomc renal disease often admittedthat they were not ill ather, even though they knew theirkidney fadurc was bfe-tbeatemng The dialysisthem to feel weU

1123

L Jensen and M Alien

Table 7 Overview of categones and dimensions

Categones

Combined WellExuberanceTolerable confusionUnhappmessResentment

Combined III

DespairResentful disruptionBearable adversityContentmentHop>eApprehension

Healthy WellConfusionResentmentGuarded certamtyExuberance

Healthy IIISadnessCertamtyResentful disruptionUneaseContentmentHopeBearable adversity

Chronic Well

UncertamtyOverpowered

HopeContentmentFrustrationResentment

Chronic IIIManageableResentful uncertamtyOverwhelmedConfusion

CertamtyHopeUnease

Dimensions

Harmony

DisharmonyOptimism

Harmony

DismayOptimismDisharmony

OptirmsmDisharmony

Guarded ophmismDisharmony

Individuals had difficulty drawmg a distmct bnebetween 'well' and 'ill' They commented, 'How can I tbnkof bemg well and til as separate? Eacb situahon changes sofrequently it is really bard to do' The feelings that they

held with tbs changmg situahon of wellness also duoigedOne mdividual with chrome renal disease stated 'Wellnesshas many stages wake up m the mommg and you feelgreat, 10 a m feel sick, 4pm exhausted When I'm verysick, I don't give a damn, chuck it When I'm m the middle Ican put up with thmgs, and when I am not so sick, I havehope'

Common themes appeanng in the self-desenptions ofillness were ones of 'prevention from partiapahon machvihes' or 'restnction' and a sense of bemg 'over-whelmed' Frequent descnptors mduded 'I feel frustrated','useless', 'I can't do what I want to do' or 1 ean't do whatnormal people do'

A eonunon theme prevalent in the self-desenptions ofwellness was one of 'havmg energy' Tbe most frequentdescnptors meluded 1 feel bappy', 'I ean do wbat I want'and 1 am glad to be alive'

CONCLUSION

From the preliminary analysis of the eard sort data, evideneehas been provided to suggest a possible unidimensionalpattem of meanmg mberent m wellness-illness, eonsistentwith the proposed eoneeptual model

Latent eategones beginmng to emerge from tbs analysiswill be further refined to reflect tbe eogmhve and affectivethemes inherent m the expenenee of health-disease

Determinmg the personal salieney of health-disease forthe mdividual will be useful m future development andtestmg of mtervenhon strategies to promote wellness-illness behaviours

References

Antonovsky A (1987) Unravelling the Mystery of Health HowPeople Manage Strex and Stay Well Jossey-Bass, San Franasco

Benner P & Wrubel J (1989) The Pnmacy of Canng Stress andCoping in Health and Illness Addison-Wesley, Don MiUs,Ontano

lapowski Z J (1983) Psychoscxaal reachon to physical lUnessCanadian Medtcal Association Joumal 128(9), 1069-1072

Miller D , Wiley D & Wolfe R (1986) Categorization method-ology an approach to the collechon and analysis of certainclasses of qualitahve information Mulhvanate BehavioralResearch 21,135-167

PlutchikR (1980) Emotion A Psychoevoluttonary Synthesis Harper& Row, New York.

Rufi&ng-Rahal M A (1985) Well-bemg and chroniaty bemgsingular rather than sick Health Values 9(5), 17-26

Wiley D (1967) Latent parhhon analysis Psychomeinka 32(2),183-193

1124