66ournal of Epidemiology and Conmmunity Health 1997;51:676-685

Do disease specific characteristics add to theexplanation of mobility limitations in patients withdifferent chronic diseases? A study in theNetherlands

Didi M W Kriegsman, Dorly J H Deeg, Jacques Th M van Eijk, Brenda W J H Penninx,A Joan P Boeke

AbstractStudy objectives-To determine whetherdisease specific characteristics, reflectingclinical disease severity, add to the expla-nation of mobility limitations in patientswith specific chronic diseases.Design and setting-Cross sectional studyof survey data from community dwellingelderly people, aged 55-85 years, in theNetherlands.Participants and methods-The addi-tional explanation of mobility limitationsby disease specific characteristics wasexamined by logistic regression analyseson data from 2830 community dwellingelderly people.Main results-In the total sample, chronicnon-specific lung disease, cardiac disease,peripheral atherosclerosis, diabetes melli-tus, stroke, arthritis and cancer (the indexdiseases), were all independently associ-ated with mobility limitations. Adjustedfor age, sex, comorbidity, and medicaltreatment disease specific characteristicsthat explain the association between dis-ease and mobility mostly reflect decreasedendurance capacity (shortness of breathand disturbed night rest in chronic non-specific lung disease, angina pectoris andcongestive heart failure in cardiac dis-ease), or are directly related to mobilityfunction (stiffness and lower body com-plaints in arthritis). For atherosclerosisand diabetes mellitus, disease specificcharacteristics did not add to the explana-tion of mobility limitations.Conclusions-The results provide evi-dence that, to obtain more detailed infor-mation about the differential impact ofchronic diseases on mobility, disease spe-cific characteristics are important to takeinto account.

(_ Epidemliol Community Health 1 997;51:676-685)

The role of chronic diseases as determinants ofmobility limitations is intuitively important,but not well defined. A higher number ofchronic diseases is consistently associated witha higher prevalence of mobility limitations,1 3

and longitudinally with a higher incidence ofmobility loss.4 However, these associations donot provide information on the influence of

different specific chronic diseases. Recently,more studies have been focused on associationsbetween specific chronic diseases andmobility.` In elderly people, the specificchronic diseases that are most consistentlyassociated with either a higher prevalence orhigher incidence of mobility limitations includearthritis,2 6-9 cardiac diseases,2 4 7 9 cerebrovas-cular disorders,2 4 6-8 chronic obstructive pul-monary disease, 79 diabetes mellitus2 4 8 9and, toa lesser extent, cancer2 8 9and atherosclerosis.-

It may, however, be doubted whether cur-rently used measurements of chronic diseasesare sufficiently detailed to obtain accurateinformation on the associations between spe-cific chronic diseases and mobility. The meas-urement of chronic diseases in population sur-veys is mostly confined to self reportspertaining to the presence or absence of a spe-cific disease. The accuracy of patients' selfreports regarding the presence of chronicdiseases is generally considered adequate,although this was shown to be different forspecific chronic conditions.'-3 It has beenstated that the clinical severity of the diseaseshould be taken into account in studies on theassociations between specific chronic diseasesand mobility.4 6Methods that are used by physicians in clini-

cal practice can be of value in the assessment ofclinical severity of chronic diseases. Physiciansimplicitly judge the clinical severity of a diseasein their patients according to the presence ofdisease specific signs and symptoms. In acquir-ing the necessary knowledge, medical historyquestions are of primary importance. This isparticularly the case for most chronic diseases.Surprisingly, the use of medical history ques-tions, conditional on disease presence, as amethod of improving the explanation of mobil-ity limitations in survey studies has never beensystematically evaluated. Information aboutthe additional influence of these parameters ofclinical severity is important for epidemiologi-cal research and health planning by providing amore detailed measure of chronic diseases thatcan be used to obtain a more precise picture ofthe burden of disease for both individuals andpopulations, and its association with outcomemeasures, such as mobility limitations. Whenthe presence of certain disease specific symp-toms, reflecting clinical severity, would seemto be strongly associated with mobility

Institute for Researchin ExtramuralMedicineD M W KriegsmanJ Th M van EijkB W J H PenninxA J P Boeke

Department ofGeneral Practice,Nursing HomeMedicine and SocialMedicineD M W KriegsmanJ Th M van EijkA J P Boeke

Department ofPsychiatry andDepartment ofSociology and SocialGerontologyD J H Deeg

Vrije Universiteit,Amsterdam, theNetherlands

Correspondence to:Dr D M W Kriegsman,EMGO Institute, Van derBoechorststraat 7, 1081 BTAmsterdam, the Netherlands.

Accepted for publication22 April 1997

676

on March 23, 2020 by guest. P

rotected by copyright.http://jech.bm

j.com/

J Epidem

iol Com

munity H

ealth: first published as 10.1136/jech.51.6.676 on 1 Decem

ber 1997. Dow

nloaded from

677Disease characteristics and mobility

limitations, it may be possible to increase theaccuracy ofpredictions for the future regardingthe prevalence of mobility limitations, disabil-ity, and the use of health care facilities. In addi-tion, intervention strategies directed at particu-larly those parameters of clinical severity ofspecific chronic diseases with the strongestassociations with health outcomes may bedeveloped to diminish the negative conse-quences of chronic diseases in the future.

In this study, we investigate whether theexplanation of mobility limitations can beimproved by adding questions directed atassessment of clinical severity, conditional onthe self reported presence of a specific chronicdisease. The influence of these parameters ofclinical disease severity on the associationsbetween chronic diseases and mobility limita-tions is examined for several chronic diseasesthat often afflict the elderly, and that have beenrepeatedly shown to influence physical func-tioning and thereby the ability of older peopleto live independently in the community. Sevenspecific chronic diseases with a high prevalencein the elderly population were selected: chronicnon-specific lung disease (asthma, chronicbronchitis or pulmonary emphysema), cardiacdisease (including myocardial infarction), pe-ripheral atherosclerosis, cerebrovascular acci-dent, diabetes mellitus, arthritis (rheumatoidarthritis or osteoarthritis) and malignant neo-plasms.The following questions will be considered:(1) Does the self reported presence of

specific chronic diseases explain self reportedmobility limitations in community dwellingelderly people?

(2) Does the presence of signs and symp-toms, reflecting clinical disease severity, add tothe explanation of mobility limitations inpatients with a specific chronic disease?

MethodsThis study uses a cross sectional design. Thedata were collected as part of the first data col-lection cycle of the Longitudinal Aging StudyAmsterdam (LASA),'4 a 10 year longitudinalstudy on predictors and consequences ofchanges in physical, cognitive, emotional, andsocial functioning in older persons.

HYPOTHESESFor the first research question, we expected theassociation with mobility limitations to be mostpronounced for those chronic diseases ofwhichthe clinical picture is characterised by symp-toms of the locomotor system (stroke andarthritis), or by decreased endurance capacity(chronic non-specific lung disease, cardiac dis-ease). The associations of the other specificchronic diseases (peripheral atherosclerosis,diabetes mellitus, and cancer) with mobilitylimitations are expected to be less pronounced.The inclusion of disease specific parameters

of clinical severity, as measured by medical his-tory questions, is hypothesised to add to theexplanation of mobility limitations especially inthose chronic diseases that have the strongestassociation with mobility limitations. Thesymptoms that are expected to add most, are

those that most directly affect physical func-tioning (for example, symptoms of the lowerbody joints in patients with arthritis, and loco-motor sequelae in stroke victims), or endur-ance capacity (for example, shortness of breathin patients with chronic non-specific lungdisease, congestive heart failure in patients withcardiac disease). On the other hand, diseasespecific symptoms that are not directly associ-ated with the locomotor system (for example,aphasia in stroke victims), or with endurancecapacity (for example, having ever had wheez-ing breath, or a history of a recent exacerbationin patients with chronic non-specific lungdisease, or a history of myocardial infarction inpatients with cardiac disease) are not expectedto add significantly to the explanation ofmobility limitations.

STUDY POPULATION

A sample of people aged 55 to 85 years, strati-fied by age and sex according to expected attri-tion resulting from mortality at mid-term ofLASA (after five years) in each age group, was

drawn from the population registries of 11municipalities in three culturally distinct geo-

graphical areas in the Netherlands. The cohortwas recruited in 1991 for the NESTOR-LSNstudy "Living arrangements and social net-works of older adults" (response rate 62.3%).'5Although there was a decline in response withincreasing age (p<0.05), the stratification hasensured representativity on age, sex, and levelof urbanisation of the sample.'5 After 11months, the participants in NESTOR-LSNwere approached for the first LASA cycle.From the initial sample of 3805 persons whoparticipated in NESTOR-LSN, a total of 3107participated in the main interview of LASA(81.7%). Data were collected in the periodfrom September 1992 through October 1993.Details of the procedures and results of thefield work are described elsewhere.'6 Of thepersons who did not participate, 260 (6.8% ofinitial sample) proved to be ineligible (de-ceased or not able because of severe physical or

cognitive disturbances, or both). Of the other438 non-participants, 394 (10.4% of initialsample) refused and 44 (1.2% of initial sample)could not be contacted. The corrected re-

sponse percentage, excluding those who were

ineligible after all, therefore was 87.6%. Older

KEY POINTS* Self reports of chronic diseases are insuf-

ficient measures to explain mobility limi-tations in elderly people.

* Symptoms reflecting clinical disease se-verity add to the explanation of mobilitylimitations.

* Use of medical history questions is anadequate and cheap method to obtaininformation about the clinical severity ofchronic diseases in survey studies.

* The accuracy of future predictions re-garding prevalence of disability may beimproved by using information frommedical history questions.

on March 23, 2020 by guest. P

rotected by copyright.http://jech.bm

j.com/

J Epidem

iol Com

munity H

ealth: first published as 10.1136/jech.51.6.676 on 1 Decem

ber 1997. Dow

nloaded from

Kriegsmzan, Deeg, vani Eijk, Pennilx, Boeke

age was significantly associated with refusal toparticipate (p<0.001) and ineligibility becauseof severe physical or cognitive disturbances(p<0.0001). Excluded for this study were par-ticipants who, at the time of the main interview,were institutionalised (living in a residential ornursing home or in a psychiatric hospital;n=126), and those who were living independ-ently but were unable to complete the fullinterview (n=151), leaving a total number of2830 (91.1% of all participants). As a result,the study population is a comparatively healthyselection of the original sample, although thestratified sampling frame guaranteed that suffi-cient numbers of subjects in the highest agegroup, as well as subjects with physicalproblems, were included.

MEASUREMENTSDeterminantsThe presence of chronic diseases was assessedby asking the participants explicitly whetherthey had any of the following seven chronicdiseases: chronic non-specific lung disease(asthma, chronic bronchitis or pulmonaryemphysema), cardiac disease (including myo-cardial infarction), peripheral atherosclerosis,cerebrovascular accident, diabetes mellitus,arthritis (rheumatoid arthritis or osteoarthri-tis), and malignant neoplasms. If a disease wasreported present, additional questions wereasked concerning medical treatment (as apotential confounder) and disease specificsymptoms and signs reflecting clinical severityof the disease. Pertinent medical history ques-tions (questions that physicians use to get animpression of disease severity) were selected inclose cooperation with general practitionerswith expert knowledge on the specific chronicdiseases. As far as possible, the items werederived from existing questionnaires such asthe Rose Questionnaire for cardiac disease.'7For malignancies, a radiation oncologist wasconsulted. Table 1 presents the variables usedin the analyses, as well as the absolute numbersof subjects in the "yes" category.For cardiac diseases, the answers to the dis-

ease specific questions were used to distinguishseparate diagnoses. The symptoms of diabetesmellitus pertained to the presence of diabeticcomplications: macrovascular complications(angina pectoris or intermittent claudication,or both), retinopathy, and peripheral neuropa-thy. For stroke, the presence of specific seque-lae was included, namely locomotor disabili-ties, visual handicap, and expressive (such asdifficulties in finding words) or perceptive(understanding written text) aphasia. Forarthritis, both the presence of symptoms andthe extent of joint involvement, as well as a his-tory of joint surgery were taken into account.Because the numbers of respondents with aspecific primary malignancy were too small toinclude them separately, malignancies werecategorised according to overall five yearsurvival rates after diagnosis.'8 When thepatient reported to have had two or moreprimary tumours, the category was based onthe tumour with the worst prognosis.

The presence of other chronic diseases wasassessed by asking respondents whether theyhad, at the moment of the interview, any otherchronic disorder of at least three months' dura-tion in addition to those that were explicitlyasked. The category "other chronic diseases"includes, among others, neurological diseases,osteoporosis, neck and back problems, liverand kidney diseases, endocrinological diseases,varicose veins and venous insufficiency, hyper-tension, gastrointestinal diseases, allergies,consequences of accidents or surgery, chroniclocomotor problems not covered elsewhere,dizziness, anaemia, disorders of the eyes orears, and mental disorders.

Outcome measureThe presence of mobility limitations wasassessed using three self report items pertain-ing to mobility activities in daily life: ability towalk up and down a 15 step staircase withoutstopping, use of private or public transporta-tion, and cut own toenails. Answers on eachitem were originally coded as 0 "can beperformed without difficulty", 1 "can beperformed, but with difficulty", 2 "can only beperformed with help" or 3 "cannot be per-formed at all". Based on the results of pilotstudies,'9 these items were shown to constitutethe best scale out of a set of nine items. Theselection of the three items was based on itera-tive reliability analysis, in which the items withthe highest item rest correlation were retainedwithout an important influence on the reliabil-ity coefficient. To determine whether the threemobility items, which cover a range of mobilityrelated activities necessary to maintain au-tonomy in daily life, are equally important forthe underlying dimension "mobility limita-tions" in subjects with specific chronic dis-eases, additional analyses were performed.Factor analyses were conducted for the threemobility items for each chronic disease sepa-rately, including only those subjects whoreported to have only one of the specificchronic diseases. In this way, the relativeimportance of the separate items as part of theunderlying dimension "mobility limitations"could be compared. The factor structureproved to be comparable between the specificchronic diseases, and was not different fromthat in subjects without any disease. Thus, theseparate mobility items can be considered torepresent a general underlying dimension forsubjects with different diseases, and summingthe three item scores results in a global index ofmobility limitations, which is equally valid foreach of the specific chronic diseases studied.Although the reliability of this three itemmobility scale in the total study population wasadequate (Cronbach's ca 0.72), and all itemsloaded on one factor (all factor loadings>0.75), use of a total score in linear regressionanalysis was precluded by an extremely skewedfrequency distribution with >60% having atotal score of 0 (range 0-9), and a non-randomdistribution of the residuals after multiplelinear regression with the score on the mobilityscale as the dependent variable. Therefore, themobility scale was dichotomised into codes 0

678

on March 23, 2020 by guest. P

rotected by copyright.http://jech.bm

j.com/

J Epidem

iol Com

munity H

ealth: first published as 10.1136/jech.51.6.676 on 1 Decem

ber 1997. Dow

nloaded from

679Disease characteristics and mobility

Table 1 Variables used in the analyses

Variables Criteria No Yes

316Chronic non-specific lungdiseaseMedical treatmentDaily coughDaily phlegm productionExacerbation past yearShortness of breathWheezing breath everWheezing breath during restDisturbed night rest

Cardiac diseaseMedical treatmentMyocardial infarctionAngina pectoris

Other ischaemic heart disease

Congestive heart failure

Valvular diseaseCardiac pacemakerOther cardiac disease

AtherosclerosisMedical treatmentIntermittent claudication

Arterial surgeryOther peripheral arterydisease

Diabetes mellitusMedical treatmentMacrovascular complications

Diabetic retinopathyDiabetic neuropathy

StrokeMedical treatment¢ 2 strokesLocomotor sequelaeVisual sequelaeExpressive aphasiaPerceptive aphasia

ArthritisMedical treatmentJoint pain past 3 monthsJoint stiffness past 3 monthsJoint swelling past monthComplaints lower body jointsComplaints upper body jointsJoint surgery lower bodyJoint surgery upper body

MalignanciesMedical treatment_ 2 primary tumoursPrognostic category:Excellent5-years survival >50%

5-year survival S 50%Haematogenous metastasesSurgical treatmentChemotherapyRadiation therapy

Other variables:Other chronic disease

Use of prescribed drugs and/or regular contact with physician for chronic non-specific lung diseaseYesYesYesDuring light exertion or at restYesYesOften or always

Use of prescribed drugs and/or regular contact with physician for cardiac diseaseEver had myocardial infarctionPain, or heavy, uncomfortable feeling on the chest during exertion, disappearing within 10 minutesafter stopping or taking sublingual nitroglycerine, or avoidance of exertion because of chest-painHistory of coronary artery surgery (including dotter procedure) without myocardial infarction andwithout symptoms of angina pectorisSleeping with >1 pillow because of shortness of breath and/or oedema of ankles, feet or legs whenwaking up in the morningHistory of valvular surgeryHistory of pacemaker implantationNot fitting in any of the categories described above (for example: relatively benign arrhythmia,valvular lesions without a history of surgery)

Use of prescribed drugs and/or regular contact with physician for atherosclerosisPain or cramps in the calf (calves) when walking, which disappears within 10 minutes afterstoppingArterial bypass or dotter procedure of the arteries of the abdomen or lower limb(s)History of other arterial surgery

Use of prescribed drugs and/or regular contact with physician for diabetesAngina pectoris and/or intermittent claudication (for definition: see cardiac disease andatherosclerosis) without positive self report of cardiac disease and atherosclerosisHistory of laser coagulation therapy of the retina (e)Pain in legs and/or feet during rest

Use of prescribed drugs and/or regular contact with physician for strokeHistory of _ 2 strokesLimited in use of arms and/or legs because of strokeVisual problems because of strokeProblems with speech because of strokeProblems understanding written text because of stroke

Use of prescribed drugs and/or regular contact with physician for rheumatoid arthritis/osteoarthritisPain in one or more joints during most days of past 3 monthsStiffness in one or more joints when getting up in the morning during most days of past 3 monthsSwelling of one or more joints during most days of the past monthComplaints of one or more of the joints in toes, feet/ankles, knees or hipsComplaints of one or more of the joints in fingers, hands/wrists, elbows, shoulders or neckHistory of surgery on one or more of the lower body joints because of arthritisHistory of surgery on one or more of the upper body joints because of arthritis

Use of prescribed drugs and/or regular contact with physician for cancerHistory of ¢ 2 primary malignant tumours

Non-melanoma skin cancerMalignant melanoma, sarcomas, lymphomas, cancer of the breast, genitourinary tract (except theovaries), brain, larynx, colon or rectumLeukaemia, cancer of the lungs, oesophagus, stomach, pancreas, liver, gall bladder or ovariesMetastases in skeleton, liver, brain or lungsYesYesYes

Comorbidity (the presence of another chronic disease in addition to the index disease):Chronic non-specific lungdisease + comorbidityCardiac disease + comorbidityAtherosclerosis + comorbidityDiabetes mellitus +comorbidityStroke + comorbidityArthritis + comorbidityMalignancies + comorbidity

All variables (except "prognostic category" in malignancies) are coded O=no and 1 =yes. When the index disease is coded 0, all disease specific variables are coded 0.

when all three activities could be performedwithout difficulty, and 1 when the respondentreported to have at least difficulty in perform-ing one or more activities.

Potential confoundersBecause of the associations of age and sex withboth the presence of specific chronic diseases

and with mobility limitations, both were

adjusted for in the analyses.Medical treatment for the index disease was

considered a potential confounder because thismay influence the association between a

specific chronic disease and mobility in two

directions: people who are medically treatedmay either have a lower odds for mobility

22915114918415024915130

542481259240

33

138

161880

262

205

18384

52147

19434

1453

1381003868293122

98141860947723379261018439

24816725

37185

2611

2076271

1127

262

407225169

119684197

-, . 11

on March 23, 2020 by guest. P

rotected by copyright.http://jech.bm

j.com/

J Epidem

iol Com

munity H

ealth: first published as 10.1136/jech.51.6.676 on 1 Decem

ber 1997. Dow

nloaded from

680Kriegsnman, Deeg, van Eijk, Penninx, Boeke

limitations because of this treatment, or have ahigher odds when treatment has been startedbecause of relatively severe disease. For all spe-cific index diseases, continuous medical treat-ment was considered present when the patientreported to use prescribed drugs, or to haveregular contact with a physician for the indexdisease, or both.

Finally, to answer the second research ques-tion, the presence of any other chronic diseasein addition to the index disease (comorbidity)was adjusted for in the analyses as a potentialconfounder. Comorbidity was operationalisedas a dichotomous variable and defined presentwhen either one or more of the other explicitlyasked diseases was present or the patient indi-cated to have any other chronic disease, orboth.

ANALYSESTo answer the first research question, logisticregression analysis was performed using thetotal study sample. The analysis was adjustedfor age and sex. To compare the strengths ofthe associations between specific chronic dis-eases and mobility limitations, all specificchronic diseases, including the category "otherchronic disease", were entered in the modelsimultaneously.To answer the second research question,

logistic regression analyses were performed foreach specific chronic disease separately. Sub-jects included in each model were thosewithout any chronic disease (reference group)and those with at least the specific diseasestudied (the index disease). For example, in thelogistic regression model examining the addi-tional influence of parameters of clinical sever-ity on the explanation of mobility limitations inpatients with cardiac disease, subjects includedwere those who reported cardiac disease to bepresent and those who reported no chronicdisease at all. Independent variables wereincluded in the models in four steps.

In the first step, the potential confounderssex, age, and the presence of any other chronicdisease in addition to the index disease(comorbidity) were entered. In the second step,the index disease was entered. In the third step,medical treatment for the index disease wasentered because of its role as a potentialconfounder. Finally, in the fourth step the dis-ease specific characteristics reflecting the clini-cal severity of the index disease were consid-ered for stepwise entry. Variables included inthis final step were entered into the modelwhen a significant contribution to the explana-tion of mobility limitations was present(pi,<0.05). These parameters of clinical diseaseseverity have an additional influence on mobil-ity, conditional on self reported disease pres-ence, when the final model is a significantimprovement compared with the previous onethat is adjusted for age, sex, comorbidity, andmedical care for the index disease. This isjudged by the change in -2 log likelihood andthe significance of its improvement from thethird to the fourth step.The disease specific analyses may be sensi-

tive to the problem of colinearity because of

our decision to use the subjects without anychronic disease as a reference group for all dis-ease specific models. As a result, colinearitybetween the independent variables may havebeen introduced because the variables con-

cerning medical treatment, clinical diseaseseverity, and comorbidity could be scored onlyfor those subjects who reported the indexdisease (in the reference group, these variablesare coded 0). To determine the extent to whichcolinearity was present, additional logisticregression analyses were performed for each ofthe specific chronic diseases separately, includ-ing only those subjects who reported the indexdisease. In these analyses, adjusted for thepotential confounders age, sex, comorbidity,and medical treatment for the index disease,the disease specific characteristics reflectingclinical disease severity were considered forstepwise entry as described before (fourth stepof the disease specific analyses). For each spe-cific chronic disease, (the index diseases:chronic non-specific lung disease, cardiacdisease, peripheral atherosclerosis, diabetesmellitus, stroke, arthritis, and malignancies)the strengths of the associations between thedisease specific characteristics and mobilitylimitations among the subjects with the indexdisease were comparable to those resultingfrom the analyses described in the previousparagraph (in which the reference group ofsubjects without any chronic disease was

included), as judged from the odds ratios and95% confidence intervals. Thus, colinearitywas shown not to be an important problem inthe disease specific analyses.

It should be noted that subjects whoreported more than one chronic disease areincluded in more than one regression model:for instance, a person with diabetes mellitusand chronic non-specific lung disease appearsboth in the model with diabetes as the indexdisease, and in that with chronic non-specificlung disease as the index disease.

ResultsTable 2 shows the background characteristicsof the study population. Table 3 shows theresults pertaining to the first research question.Adjusted for age and sex, the presence of any ofthe specific chronic diseases is significantlyassociated with a higher odds for mobility limi-tations, with odds ratios (ORs) varying from1.38 for malignancies to 3.37 for arthritis (seetable 3). As was expected, stroke and arthritisshow the largest ORs (3.29 and 3.37, respec-tively), followed by chronic non-specific lungdisease and cardiac disease (2.34 and 2.11,respectively).

In table 4, the results are shown pertaining tothe second research question, concerning theadditional explanation of mobility limitationsby parameters of clinical disease severity. Theresults of the disease specific logistic regressionanalyses are presented for each step in theanalysis for each disease separately.

In subjects with chronic non-specific lungdisease, after adjustment for the potential con-founders age, sex, and comorbidity, thepresence of chronic non-specific lung disease

on March 23, 2020 by guest. P

rotected by copyright.http://jech.bm

j.com/

J Epidem

iol Com

munity H

ealth: first published as 10.1136/jech.51.6.676 on 1 Decem

ber 1997. Dow

nloaded from

681Disease characteristics and mobility

Table 2 Basic characteristics of the study population(n=2830)

Determinants Number %

SexMale 1379 48.7Female 1451 51.3

Age (y)55-64 904 31.965-74 909 32.275 1017 35.9

Chronic diseasesChronic non-specific lung disease 316 11.2Cardiac disease 542 19.2Peripheral atherosclerosis 262 9.3Diabetes mellitus 205 7.2Stroke 138 4.9Arthritis 981 34.7Malignancies 248 8.8Other chronic diseases 1127 39.80 Chronic diseases 724 25.61 Chronic disease 960 33.9¢2 Chronic diseases 1146 40.5

Mobility limitationsWalking up and down 15 stepsWithout difficulty 2140 76.0With difficulty 412 14.6Only with help/not at all 265 9.4

Using own/public transportationWithout difficulty 2470 87.4With difficulty 150 5.3Only with help/not at all 206 7.3

Cutfing toenailsWithout difficulty 2018 71.6With difficulty 374 13.2Only with help/not at all 428 15.2

Difficulty with _s 1 activityNo 1733 61.8Yes 1072 38.2

Table 3 Results of multiple logistic regression analyses(odds ratios and 95% confidence intervals) of age, sex, andspecific chronic diseases on mobility limitations in the totalstudy sample (ntO,=2805)

Number 95%who Odds Confidencereplied yes ratios intervals

Age (per year) 1.10 1.09,1.11Sex (female versus male) 1.76 1.45, 2.13Chronic diseases (yes versus no)

Chronic non-specificlung disease 316 2.34 1.76, 3.10Cardiac disease 542 2.11 1.67, 2.68Peripheral atherosclerosis 262 1.56 1.14, 1.97Diabetes mellitus 205 1.89 1.33, 2.68Stroke 138 3.29 2.10, 5.17Arthritis 981 3.37 2.78, 4.07Malignancies 248 1.38 1.02, 1.88Other chronic disease 1127 1.92 1.60, 2.31

-2 log likelihood 2824.95% predicted correctly 75.0

itself does not add significantly to the explana-tion of mobility limitations, compared withsubjects without any chronic disease (step 2:OR 1.49; 95% CI 0.70, 3.16). However, inpatients with chronic non-specific lung dis-ease, medical treatment for the disease is an

explanatory factor with regard to the presence

of mobility limitations (step 3: OR 4.44; 95%CI 2.40, 8.21). In addition to medicaltreatment for chronic non-specific lung dis-ease, a positive self report of shortness ofbreath during light exertion or at rest (step 4:OR 3.24; 95% CI 1.79, 5.86) and regular dis-turbance of night rest resulting from chronicnon-specific lung disease (step 4: OR 3.65;95% CI 1.12, 11.83) are associated with thepresence of mobility limitations. Thus, fromthe model resulting after the final step, it seemsthat the presence of mobility limitations in

chronic non-specific lung disease patients isexplained by shortness ofbreath and disturbednight rest as parameters of clinical severity ofthe disease, as well as by medical treatment forthe disease.A self report of cardiac disease is an explana-

tory factor for mobility limitations in thesepatients after adjustment for age, sex, andcomorbidity (step 2: OR 2.56; 95% CI 1.58,4.15). However, the presence of cardiac diseasedoes not explain mobility limitations whenmedical treatment is adjusted for (step 3: OR1.03; 95% CI 0.47, 2.29). Medical treatmentfor cardiac disease is an explanatory factor formobility limitations in cardiac patients (step 3:OR 2.69; 95% CI 1.37, 5.31). Conditional onthe self reported presence of cardiac disease,symptoms of angina pectoris (step 4: OR 2.03;95% CI 1.33, 3.11) or congestive heart failure(step 4: OR 1.98; 95% CI 1.19, 3.30) add tothe explanation of mobility limitations inpatients with cardiac disease. In the finalmodel, adjusted for angina pectoris andcongestive heart failure, medical treatment forcardiac disease (either use of medication orregular physician contacts), still adds to theexplanation of mobility limitations, irrespectiveof the other variables in the model.Adjusted for the potential confounders age,

sex, and comorbidity, peripheral atherosclero-sis does not add significantly to the explanationof mobility limitations (step 2: OR 0.66; 95%CI 0.22,2.01). Also, in patients with peripheralatherosclerosis, medical treatment and diseasespecific characteristics do not add to the expla-nation of mobility limitations.Compared with subjects without any chronic

disease and adjusted for age, sex, and comor-bidity, the self reported presence of diabetes isan explanatory factor for mobility limitations(step 2: OR 2.53; 95% CI 1.14, 5.62). Neithermedical treatment nor disease specific charac-teristics of diabetes (diabetic complications)have any additional influence on the explana-tion of mobility limitations. However, the netexplanatory power of diabetes mellitus onmobility limitations loses its significance, whenmedical treatment is included in the regressionmodel (step 3: OR 2.21; 95% CI 0.43, 11.33).In this case, colinearity between a self report ofdiabetes and medical treatment is probably thecause, because 95% of the subjects reportingdiabetes also reports to be medically treated forthe disease.

Adjusted for age, sex, and comorbidity, ahistory of stroke explains the presence ofmobility limitations (step 2: OR 4.73; 95% CI1.59, 14.05). This is particularly so for patientswho report to be medically treated for theirstroke (step 3: OR 2.96; 95% CI 1.13, 7.78).Certain disease specific characteristics, apartfrom medical treatment, significantly explainthe association between stroke and mobilitylimitations. This applies to whether more thanone stroke occurred (step 4: OR 4.03; 95% CI1.25, 13.04), and the presence of visual seque-lae (step 4: OR 5.45; 95% CI 1.21, 24.59). Thepresence of locomotor sequelae did not addsignificantly to the explanation of mobilitylimitations in stroke patients. Although this

on March 23, 2020 by guest. P

rotected by copyright.http://jech.bm

j.com/

J Epidem

iol Com

munity H

ealth: first published as 10.1136/jech.51.6.676 on 1 Decem

ber 1997. Dow

nloaded from

Kriegsman, Deeg, van Eijk, Penninx, Boeke

Table 4 Results of multiple logistic regression analyses of mobility limitations (odds ratios and 95% confidence intervals) on age, sex, comiorbidity anddisease specific characteristics in subjects with a specific chronic disease, compared with subjects without any chronic disease

Step 1 - enter Step 2 - eniter Step 3 - enter Step 4 - stepwise

nY, OR 95% CI OR 95% CI OR 95% CI OR 95% CI

Chronic non-specific lung disease n,,,=1026Age (per year)Sex (female versus male)Comorbidity (yes versus no)Chronic non-specific lung diseaseMedical treatmentShortness of breathDisturbed night rest

-2 log likelihood (p valueimprovement)

'Y0 predicted correctlyCardiac disease ni,,,=1243Age (per year)Sex (female versus male)Comorbidity (yes versus no)Cardiac disease

Medical treatmentAngina pectorisCongestive heart failure

-2 log likelihood (p valueimprovement)

% predicted correctlyAtherosclerosis n,,,,=980Age (per year)Sex (female versus male)Comorbidity (yes versus no)Atherosclerosis

Medical treatment-2 log likelihood (p valueimprovement)

% predicted correctlyDiabetes niellitus n,,,,=922Age (per year)Sex (female versus male)Comorbidity (yes versus no)Diabetes mellitusMedical treatment

-2 log likelihood (p valueimprovement)

% predicted correctlyStroke n,,,,=852Age (per year)Sex (female versus male)Comorbidity (yes versus no)StrokeMedical treatmentB 1 strokeVisual sequelae

-2 log likelihood (p valueimprovement)

% predicted correctlyArthritis n,,,,=1681Age (per year)Sex (female versus male)Comorbidity (yes versus no)Arthritis

Medical treatmentJoint stiffnessComplaints lower bodySurgery lower body

-2 log likelihood (p valueimprovement)

'/0 predicted correctlyMalignancies n,,,,=959Age (per year)Sex (female versus male)Comorbidity (yes versus no)Malignancies

Medical treatmentHaematogenous metastases

-2 log likelihood (p valueimprovement)

'7, predicted correctly

26231622915030

1.12 1.10, 1.15 1.12 1.10, 1.151.72 1.21, 2.45 1.75 1.22, 2.49

12.21 8.47, 17.59 8.47 3.90, 18.431.49 0.70, 3.16

818.0782.4

817.0582.3

(NS)

1.11 1.09, 1.13 1.11 1.09, 1.131.78 1.32, 2.41 1.97 1.44, 2.708.12 6.01, 10.97 3.81 2.37, 6.15

2.56 1.58, 4.15407542481240138

1091.0080.3

1077.3080.1

(<0.001)

1.13 1.11, 1.16 1.13 1.11, 1.162.09 1.44, 3.03 2.08 1.44, 3.0211.18 7.58, 16.48 16.66 5.33, 52.06

0.66 0.22, 2.01225262183

741.1783.4

740.5983.4

(NS)

1.14 1.11, 1.17 1.14 1.11, 1.172.36 1.60, 3.47 2.37 1.61, 3.509.44 6.16, 14.47 3.99 1.71, 9.30

2.53 1.14, 5.62169205194

694.1583.4

689.3183.6

(<0.05)

1.14 1.11, 1.17 1.14 1.11, 1.171.91 1.26, 2.90 2.02 1.32, 3.08

15.77 9.21,26.99 3.56 1.11, 11.404.73 1.59, 14.05

1191381003829

684

981418477792184

609.2085.5

1.12 1.10, 1.131.80 1.41, 2.305.89 4.63, 7.49

1660.6375.9

1.13 1.10, 1.151.53 1.05, 2.217.69 5.21, 11.35197

24816711

759.0782.1

602.1585.5

(<0.01)

1.12 1.10, 1.141.52 1.18, 1.962.33 1.71, 3.18

4.62 3.26, 6.55

1584.6176.9

1.13 1.10, 1.151.76 1.23, 2.52

10.68 4.76, 24.010.42 0.16, 1.094.44 2.40,8.21

793.73 (<0.0001)83.0

1.11 1.09, 1.132.00 1.46,2.743.92 2.43, 6.341.03 0.47, 2.292.69 1.37,5.31

1068.74 (<0.01)80.2

1.13 1.11, 1.162.14 1.47, 3.1117.09 5.43, 53.800.46 0.14, 1.581.61 0.84, 3.09

738.53 (NS)83.8

1.14 1.11, 1.172.38 1.61, 3.503.98 1.70, 9.282.21 0.43, 11.331.16 0.25, 5.29

689.27 (NS)83.6

1.14 1.11, 1.172.00 1.31, 3.064.46 1.36, 14.641.86 0.47, 7.292.96 1.13, 7.78

597.36 (<0.05)85.5

1.12 1.10, 1.141.46 1.13,1.882.28 1.67,3.113.73 2.58, 5.391.78 1.32, 2.40

(<0.0001) 1569.95 (<0.001)77.5

1.13 1.10, 1.151.51 1.04, 2.196.00 2.60, 13.811.31 0.58, 2.94

758.6682.1

(NS)

1.13 1.10, 1.151.49 1.03, 2.175.89 2.55, 13.601.52 0.60, 3.850.82 0.44, 1.53

758.28 (NS)81.8

1.13 1.10, 1.161.72 1.19, 2.479.84 4.22, 22.950.31 0.11,0.863.03 1.58, 5.833.24 1.79, 5.863.65 1.12, 11.83

771.63 (<0.0001)84.5

1.11 1.09, 1.131.86 1.35, 2.563.20 1.95, 5.260.84 0.37, 1.912.38 1.18, 4.802.03 1.33, 3.111.98 1.19, 3.30

1047.64 (<0.001)80.6

1.13 1.11, 1.162.14 1.47,3.11

17.09 5.43, 53.800.46 0.14, 1.581.61 0.84, 3.09

738.53 (NS)83.8

1.14 1.11, 1.172.38 1.61, 3.503.98 1.70, 9.282.21 0.43, 11.331.16 0.25, 5.29

689.2783.6

(NS)

1.14 1.11, 1.182.07 1.35, 3.193.11 0.88,10.931.53 0.37, 6.372.58 0.97, 6.844.03 1.25, 13.045.45 1.21, 24.59

585.1085.7

1.121.382.24

1.511.401.872.281.54

1525.7978.8

1.131.555.901.510.754.99

754.1082.1

(<0.01)

1.11, 1.141.06, 1.791.63, 3.08

0.93, 2.451.03, 1.921.37, 2.541.55, 3.341.02, 2.31

(<0.000 1)

1.10, 1.151.06, 2.252.55, 13.620.59, 3.830.40, 1.410.98, 25.54

(<0.05)

might have been caused by colinearity the lack and the presence of disease specific character-of an association between the occurrence of istics reflecting clinical severity (step 4: ORmore than one stroke and locomotor sequelae 1.51; 95% CI 0.93, 2.45). In addition to therenders this explanation unlikely. presence of medical treatment (step 4: ORThe association between arthritis and mobil- 1.40; 95% CI 1.03, 1.92), disease specific vari-

ity limitations (step 2: OR 4.62; 95% CI 3.26, ables that are associated with mobility limita-6.55) is largely explained by medical treatment tions in arthritis patients are joint stiffness (step

682 on M

arch 23, 2020 by guest. Protected by copyright.

http://jech.bmj.com

/J E

pidemiol C

omm

unity Health: first published as 10.1136/jech.51.6.676 on 1 D

ecember 1997. D

ownloaded from

Disease characteristics and mobility

4: OR 1.87; 95% CI 1.37, 2.54), complaints ofthe joints in the lower body (step 4: OR 2.28;95% CI 1.55, 3.34), and a history of joint sur-gery of the joints in the lower body, such astotal hip replacement (step 4: OR 1.54; 95% CI1.02, 2.31).Adjusted for age, sex, and comorbidity, a

history of cancer is not of influence on thepresence of mobility limitations (step 2: OR1.31; 95% CI 0.58, 2.94). Moreover, neithermedical treatment nor disease specific charac-teristics add to the explanation of mobilitylimitations in subjects with a history of cancer,compared with those without any chronicdisease. The only disease specific characteristicthat was of borderline significance in explain-ing mobility limitations among cancer patients,was the presence ofhaematogenous metastases(step 4: OR 4.99; 95% CI 0.98, 25.54), but thenumber of patients with metastases is small(n= 11). We also performed the analyses withexclusion of those subjects with a history ofnon-melanoma skin cancer, but these resultsdid not differ from those including all cancer

patients.

DiscussionIn the total study sample, each self reportedchronic disease is significantly associated withthe presence of mobility limitations. As ex-

pected, the strongest associations are found forchronic diseases of which the clinical pictureincludes disability (stroke and arthritis), or thatare generally associated with decreased endur-ance capacity (chronic non-specific lung dis-ease and cardiac disease). This finding isconsistent with results from previousstudies.2 69

This study, however, shows that the selfreported presence of specific chronic diseases istoo crude a measure to be sufficiently accuratein explaining the presence of mobility limita-tions. After adjustment for the potentialconfounders age, sex, and comorbidity, theassociations between the specific chronic dis-eases and mobility limitations seem to belargely explained by disease specific character-istics (including medical treatment for theindex disease) reflecting clinical severity of thedisease.Symptoms explaning mobility limitations in

case of a positive self report of chronicnon-specific lung disease (shortness of breathduring light exertion, and regular disturbanceofnight rest) are associated with a continuouslydecreased endurance tolerance. This is not thecase for the other symptoms of chronicnon-specific lung disease (for example, cough-ing and wheezing), which are, at the most, onlylimiting during an episode. In addition, comor-bidity and medical treatment for chronic non-

specific lung disease were shown to beexplanatory factors for mobility limitations inpatients with chronic non-specific lung disease.Thus, the association of chronic non-specificlung disease with mobility limitations can beexplained by parameters of clinical diseaseseverity (shortness of breath and disturbednight rest, as well as medical treatment) and the

presence of other chronic diseases in additionto the chronic non-specific lung disease.For cardiac disease, symptoms of angina

pectoris and congestive heart failure, bothassociated with decreased endurance capacity,added significantly to the explanation ofmobility limitations, whereas the other charac-teristics studied (such as a history of myocar-dial infarction or cardiac surgery) did not. Theassociation between the self reported presenceof cardiac disease and mobility limitationsseems to be explained completely by medicaltreatment and specific cardiac diagnoses in-dicative of the clinical severity (angina pectorisand congestive heart failure), as well as by thepresence of comorbidity.For atherosclerosis, disease specific charac-

teristics did not add to the explanation ofmobility limitations. For peripheral atheroscle-rosis, it has been shown that the accuracy ofpatients' self reports, compared with generalpractitioners' information, is low.'0 Therefore,it might be expected that symptoms would addto the explanation ofmobility limitations in thischronic disease, but this was not the case. Rel-evant bias can be suspected because ofconfusion between the Dutch words "slagad-eren" (arteries) and "spataderen" (varicoseveins, venous insufficiency). Venous insuffi-ciency may also be associated with mobilitylimitations, but symptoms are different fromthose of peripheral atherosclerosis. The ex-planatory power of comorbidity on mobilitylimitations in patients reporting peripheralatherosclerosis is high, compared with that inpatients with other index diseases.

In patients with diabetes mellitus, diseasespecific characteristics reflecting clinical sever-ity (that is, the presence of microvascular andmacrovascular complications) did not add sig-nificantly to the explanation of mobility limita-tions. The possibility that these disease specificcharacteristics, particularly macrovascularcomplications, could be shown to influence theexplanation of mobility limitations in diabetesmellitus, was diminished by the decision not toconsider the additional presence of a selfreportof cardiac disease or atherosclerosis as manifes-tations of macrovascular complications, but toinclude these as comorbidity. The prevalenceof cardiac disease and peripheral atherosclero-sis among patients with diabetes mellitus washigh, compared with the prevalence in the totalsample (28.8% versus 19.2% for cardiacdisease, 19.0% versus 9.3% for atherosclero-sis), and compared with the prevalence ofmacrovascular complications in diabetic pa-tients (16.6%). In contrast with the diseasespecific complications, the presence of comor-bidity adds to the explanation of mobility limi-tations in patients with diabetes mellitus. Thefact that virtually all diabetic patients aremedically treated for the disease, which intro-duces the problem ofcolinearity in the analysis,probably explains why the association of diabe-tes mellitus with mobility limitations is not sig-nificant anymore, once medical treatment isentered as a potential confounder.

In contrast with expectations, the presenceof locomotor sequelae in stroke patients did

683

on March 23, 2020 by guest. P

rotected by copyright.http://jech.bm

j.com/

J Epidem

iol Com

munity H

ealth: first published as 10.1136/jech.51.6.676 on 1 Decem

ber 1997. Dow

nloaded from

Kriegsmnan, Deeg, van Eijk, Penninx, Boeke

not add to the explanation of mobilitylimitations. Although the prevalence of thissymptom is high (49.3%), a history of morethan one stroke and the presence of visualsequelae, which were both not associated withthe presence of locomotor sequelae, seem to bemore important. These aspects of clinicalseverity largely explain the association betweenstroke and mobility limitations. In contrastwith the results in the other index diseases,comorbidity offers no additional explanation.

In patients with arthritis, affliction of thelower body joints has the largest explanatorypower on mobility limitations, as would beexpected considering the content of the mobil-ity items included. Also, a history of lower bodyjoint surgery (an intervention with the purposeof restoring function and alleviating symp-toms) adds to the explanation of mobility limi-tations. The conclusion that can be drawn fromthis cross sectional analysis is that, althoughjoint surgery undoubtedly has a beneficialinfluence on functioning and symptoms, mo-bility is not restored fully, compared with peo-ple of the same age and sex, but withoutchronic diseases. The association betweenarthritis and mobility limitations is thusexplained by certain parameters of diseaseseverity (joint stiffness, complaints of lowerbody joints, and a history of surgery on thelower body joints), as well as by continuousmedical treatment and the presence of comor-bidity.The cross sectional design of this study is a

possible explanation for the negative results wefound for patients with a history of cancer: inall prognostic categories, only survivors areincluded (the mean duration of time sincediagnosis was approximately eight years). Inthe absence of haematogenous metastases andadditional chronic diseases, the initial severityof cancer (as reflected by the prognosticcategory) does not add to the explanation ofmobility limitations.

Medical treatment was adjusted for, becauseit was expected that the associations betweenthe specific index diseases and mobility couldbe modified by treatment in both directions.However, when medical treatment was ofinfluence (in chronic non-specific lung disease,cardiac disease, and arthritis), it was consist-ently associated with a higher odds for mobilitylimitations. A possible explanation is that onlypatients with comparativley severe chronicnon-specific lung disease, cardiac disease orarthritis will receive continuous medical treat-ment (medication or regular check ups by aphysician). Although their symptoms may bealleviated to a great extent, they may still expe-rience mobility limitations more often thanpeople with less severe disease. For thosediseases for which continuous medical treat-ment is the rule, such as diabetes mellitus,stroke or cancer, inclusion of medical treat-ment in the model will not add significantly tothe explanation of mobility limitations becauseof its low variance among patients.As described in the methods section, our

study population is a comparatively healthyselection of the original sample, which may

have caused an underestimation of the ob-served associations.

Summarising, this study provides evidencethat a more accurate explanation of theassociations between specific chronic diseasesand mobility limitations can be obtained byincluding questions on medical treatment anddisease specific signs and symptoms reflectingclinical disease severity, conditional on apositive self report of the index disease. Thismethod of acquiring additional informationreflecting the clinical severity of chronicdiseases provides a cost saving alternative forother methods that are frequently used, such asmedical record extraction,20 22 laboratorytests,23 or clinical examination.21 For large scalepopulation surveys, the organisational andfinancial demands associated with the use ofthese methods will often exceed the availableresources. Moreover, there may be some doubtas to whether these measures, which are oftendeveloped in specific clinical settings, areequally applicable in the general population.Our results also provide evidence supportingthe usefulness of distinguishing between spe-cific chronic diseases when studying the impactof disease on aspects of physical functioning,such as the presence of mobility limitations,instead of using measures in which theinformation about the specific chronic diseasesis pooled into one overall score, as is the casefor many other measures of diseaseseverity.22 25-28 Use of pooled information ex-cludes the possibility to investigate the impactof a specific chronic disease on the outcomethat is being studied, be it mobility, mortality orthe use of health care facilities.

In view of our results, future predictionsregarding the prevalence of mobility limitationsmay have to be adjusted, but additional investi-gations are needed to determine to whatextent. Currently, predictions are mostly com-puted using trends in the prevalences of selfreported presence of diseases and mobilitylimitations. The most important gain of thisstudy, compared with previous ones, is that it isshown that the presence of disease specificcharacteristics reflecting clinical severity inpatients with a specific chronic disease, ratherthan only the positive self report of a disease,explains the presence of mobility limitations.The use of this type of additional information,reflecting the clinical severity of a disease, mayincrease the accuracy of future predictionsregarding the development of the prevalence ofmobility limitations or disability. Moreover,differences between studies on the associationsbetween chronic diseases and physical func-tioning may be explained by differences in theprevalences of disease specific signs and symp-toms in diagnosed and treated patients. Forinstance, the conflicting evidence with regardto future developments of disability free lifeexpectancy as far as "compression" or "decom-pression" of morbidity is concerned29 34 maypartly result from a different distribution ofclinical disease severity (in other words, differ-ent prevalences of signs and symptoms) acrosscountries. Differences in this distribution may,for example, reflect differences in the extent to

684

on March 23, 2020 by guest. P

rotected by copyright.http://jech.bm

j.com/

J Epidem

iol Com

munity H

ealth: first published as 10.1136/jech.51.6.676 on 1 Decem

ber 1997. Dow

nloaded from

685Disease characteristics and mobility

which diseases are diagnosed in early stages (alarge proportion of diseases diagnosed byscreening may result in relatively more peoplewith less severe disease). From a longitudinalperspective, using information on both the

presence of chronic diseases and on disease

specific characteristics reflecting clinical dis-

ease severity, makes it possible for future

predictions to take into account the effects of

changes in the distribution of clinical severityof specific chronic diseases. The occurrence of

these type of changes is likely to occur. For

example, increases in the early detection of

chronic diseases and improvements in medical

treatment for these early stages, includingtreatment of risk factors for other diseases

(such as treatment of hypertension to preventcardiovascular and cerebrovascular disease)may lead to a shift of the distribution of clinical

severity towards less severe disease, and to a

decrease in the associations of specific chronicdiseases with disability when the changes in

clinical severity are not taken into account.

This study is based on data collected in the context of the Lon-

gitudinal Aging Study Amsterdam (LASA), conducted at the

Department of Psychiatry and the Department of Sociology andSocial Gerontology of the Vrije Universiteit, Amsterdam, theNetherlands.

Funding: LASA is largely funded by the Ministry of Health,Welfare and Sports of the Dutch Government.Conflicts of interest: none.

1 Cornoni-Huntley JC, Foley DJ, Guralnik JM. Co-morbidityanalysis: a strategy for understanding mortality, disabilityand use of health care facility of older people. Int _7Epidenil ol 1991 ;20 (suppl l):S8-17.

2 Verbrugge LM, Lepkowski JM, Imanaka Y Comorbidityand its impact on disability. Milbank Q 1989;67:450-84.

3 Guralnik JM, LaCroix AZ, Everett DF. Comorbidity ofchronic conditions and disability among older personsUnited States, 1984. Amn Med Assoc 1990;263:209-10.

4 Guralnik JM, LaCroix AZ, Abbott RD, et al. Maintainingmobility in late life. I. Demographic characteristics andchronic conditions. Am ]f Epidemiol 1993;137:845-57.

5 Verbrugge LM, Lepkowski JM, Konkol LL. Levels ofdisability among US adults with arthritis. Gerontol 1991;46:S71-83.

6 Boult C, Kane RL, Louis TA, Boult L, McCaffrey D.Chronic conditions that lead to functional limitation in theelderly. ]f Gerontol 1994;49:M28-36.

7 Guccione AA, Felson DT, Anderson JJ, et al. The effects ofspecific medical conditions on the functional limitations ofelders in the Framingham study. Am _7 Publ Health1994;84:351-8.

8 Bos GAM van den. The burden of chronic diseases in termsof disability, use of health care and healthy life expectan-cies. Eur] Publ Health 1995;5:29-34.

9 Verbrugge LM. New thinking and science on disability inmid- and late life. Eur7Publ Health 1995;5:20-8.

10 Kriegsman DMW, Penninx BWJH, Eijk JThM van, BoekeAJP, Deeg DJH. Self-reports and general practitionerinformation on the presence of chronic diseases in commu-nity dwelling elderly: a study on the accuracy of patients'self-reports and on determinants of inaccuracy. _7 Clin Epi-demniol 1996;49: 1407-17.

11 Harlow SD, Linet MS. Agreement between questionnairedata and medical records. The evidence for accuracy ofrecall. Am ]f Epidemniol 1989; 129:233-48.

12 Bos GAM van den. Care for the chronically ill. (In Dutch:Zorgen van en voor chronisch zieken). [PhD thesis].Amsterdam: Bohn, Scheltema and Holkema, 1989.

13 Kehoe R, Wu S-Y, Leske MC, Chylack LT. Comparing self-reported and physician-reported medical history. AmnEpidemiol 1994;139:813-8.

14 Deeg DJH, Knipscheer CPM, Tilburg W van, eds.Autononmy anid well-being in the aginlg populationz. Conicepts anzddesign of the Longitudinal Aging Study Amsterdamii. Bunnik:Netherlands Institute for Gerontology, 1993.

15 Broese van Groenou MI, Tilburg TG van, Leeuw ED de,Liefbroer AC. Data collection. In: Knipscheer CPM, Jong-Gierveld J de, Tilburg TG van, Dykstra PA, eds. Livingarrangenmenits and social tnetworks of older adults. First results.Amsterdam: VU University Press, 1995:185-97.

16 Smit JH, Vries MZ de. Procedures and results of the fieldwork. In: Deeg DJH, Westendorp-de Serriere M, eds.Autonomy, and well-being in the aging populationi. I. Reportfronm the Lonigitudinal Aginzg Study Anmsterdanm 1992-1993.Amsterdam: VU University Press, 1994:7-13.

17 Rose GA. The diagnosis of ischaemic heart pain and inter-

mittent claudication in field surveys. Bzull World Health Org1962;27:645-58.

18 Perez CA, Brady LW, eds. Prilciples anid practice of radiationoncology. Philadelphia: Lippincott, 1992.

19 Deeg DJH, Kriegsman DMW, Zonneveld RJ van. Preva-lence of four chronic conditions and their association withhealth limitations in older persons in The Netherlands,1956-1993. (In Dutch: Prevalentie van vier chronischeziekten en hun samenhang met gezondheidsbeperkingenbij ouderen in Nederland, 1956-1993). Tijdschr SocGezondheidsz 1994;72:434-41.

20 Asenjo MA, Bare L, Bayas JM, et al. Relationship betweenseverity, costs and claims of hospitalized patients using theSeverity of Illness Index. Eur]7 Epidemtiol 1994;10:625-32.

21 Hartz AJ, Guse C, Sigmann P, Krakauer H, Goldman RS,Hagen TC. Severity of illness measures derived from theUniform Clinical Data Set (UCDSS). Med Care 1994;32:881-901.

22 Parkerson GR, Broadhead WE, Tse CJ. The Duke Severityof Illness Checklist (DUSOI) for measurement of severityand comorbidity. Clin Epidemiol 1993;46:379-93.

23 Damiano AM, Bergner M, Draper EA, Knaus WA, WagnerDP. Reliability of a measure of severity of illness: acute

physiology of chronic health evaluation - II. Clin Epidemiol1992;45:93-101.

24 Kalra L, Dale P, Crome P. Evaluation of a clinical score forprognostic classification of elderly stroke patients. Age Age-ing 1994;23:492-98.

25 Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new

method of classifying prognostic comorbidity in longitudi-nal studies: development and validation.] Chron Dis 1987;40:373-83.

26 Charlson M, Szatrowski TP, Peterson J, Gold J. Validation ofa combined comorbidity index.

_Clin Epidemiol 1994;47:

1245-51.27 Parmelee PA, Thuras PD, Katz IR, Lawton MP. Validation

of the Cumulative Illness Rating Scale in a geriatricresidential population. Arn Geriatr Soc 1995;43: 130-7.

28 Waite K, Oddone E, Weinberger M, Samsa G, Foy M,Henderson W. Lack of association between patients' meas-

ured burden of disease and risk for hospital readmission.]Clin Epideniol 1994;47:1229-36.

29 Verbrugge LM. Recent, present and future health of Ameri-can adults. Annu Rev Public Health 1989;10:333-61.

30 Olshansky SJ, Ault AB. The fourth stage of the epidemio-logic transition: the age of delayed degenerative diseases.Milbank Q 1986;64:355-91.

31 Manton KG, Corder LS, Stallard E. Estimates of change inchronic disability and institutional incidence and preva-lence rates in the US elderly population from the 1982,1984, and 1989 National Long Term Care Survey. _7Gerontol 1993;48:S 153-66.

32 Roos NP, Havens B, Black C. Living longer but doingworse: assessing health status in elderly persons at two

points in time in Manitoba, Canada, 1971 and 1983. SocSciMed 1993;36:273-82.

33 Corder LS. Improving survey measurement: health statustransitions in a longitudinal list sample design. In: MathersC, McCallum J, Robine J-M, eds. Advances in healthexpectancies: proceedinzgs of the 7'" meeting of the InternationalNetwork on Health Expectancy (REVES), Canberra, February1994. Canberra: Australian Institute of Health andWelfare, 1994:408-23.

34 Verbrugge LM. The experience and measure of disability.In: Mathers C, McCallum J, Robine J-M, eds. Advances inhealth expectancies: proceedings of the 7' nmeeting of theInternational Network on Health Expectancy (REVES), Can-berra, February 1994. Canberra: Australian Institute ofHealth and Welfare, 1994:19-33.

on March 23, 2020 by guest. P

rotected by copyright.http://jech.bm

j.com/

J Epidem

iol Com

munity H

ealth: first published as 10.1136/jech.51.6.676 on 1 Decem

ber 1997. Dow

nloaded from