M.D. Denkinger1 · A. Lukas1 · F. Herbolsheimer2 · R. Peter2 · T. Nikolaus1
1 Agaplesion Bethesda Clinic, Geriatric Center Ulm University2 Institute of Epidemiology and Medical Biometry, Ulm University
Physical activity and other health-related factors predict health care utilisation in older adults
The ActiFE Ulm study
In Germany, as in all other socio-econom-ically highly developed countries, disease-related cost per capita is exponentially ris-ing with age [25]. This goes along with a linear increase in comorbidity [25]. There-fore, about half of the overall expenditures in German health care have been attribut-ed to those being 65 years and older [17]. However, demographic change does not seem to be major factor responsible for the cost increase in the German health care system and increased health care util-isation does not necessarily reflect higher costs (because costs per day even decrease in the oldest old) [17]. Nevertheless, prev-alence of most chronic diseases rises with age and, therefore, drives the impact on health care utilisation (HCU).
Influences on HCU can be measured using different psychosocial and public health-related constructs [15]. Most stud-ies have utilized the so called “patient-cen-tred” behavioural model of health care utilisation by Andersen and Newman [1]. In Germany, HCU has been mainly ana-lysed in four studies: Hessel et al. [13] used data from a representative German tele-phone survey, Linden et al. [15] analysed the Berlin Ageing Study (BASE), Braune et al. [4] focused on the association of de-pression and HCU using data from the KORA cohort and recently Glaesmer and colleagues [9] emphasized traumatic ex-periences and the role of posttraumatic stress disorders on HCU. All studies have focused on psychosocial aspects of HCU with little emphasis on comorbidity and
physical functioning. A conclusive review of internationally published HCU studies that included people with multiple chron-ic conditions (i.e. mostly elderly) was re-cently published [14]. They found that the greatest body of literature by far result-ed from studies conducted in the Unit-ed States (more than 60%). Because of the highly different health care system sys-tems, results can hardly be compared to European standards.
However, it is noteworthy that HCU was most stably correlated with the pres-ence of multiple chronic conditions and functional aspects and not with mainly psychosocial factors [14]. Therefore, with regard to the behavioural model by An-dersen and Newman [1] the underlying illness level and not predisposing and en-abling factors seem most important.
In this article we concentrated specif-ically on physical conditions including multiple chronic diseases, physical func-tion and physical activity—one of the pri-ority measures of the ActiFE-Ulm study—without neglecting psychosocial factors. This is important, because physical func-tion and activity are potentially modifi-able measures and therefore potential tar-gets for public health programs with re-gard to healthy ageing (primary or sec-ondary prevention).
Methods
The ActiFE Ulm (Activity and Function in the Elderly in Ulm) study is a popula-
tion-based cohort study in people aged 65 years and older, located in Ulm and ad-jacent regions in southern Germany. The study was carried out jointly with the Re-spiratory Survey of the Elderly of the Eu-ropean IMCA II project. A detailed de-scription of the cohort is published else-where [7]. In total, 7,624 inhabitants were randomly contacted by mail and asked to participate. Exclusion criteria were se-vere deficits in cognition, vision or hear-ing that precluded the accomplishment of most assessments or serious German lan-guage difficulties. Between March 2009 and April 2010, 1,506 non-institutional-ized eligible individuals agreed to partic-ipate and underwent the baseline assess-ments while 1,487 could be visited three times within 1 week. All participants pro-vided written informed consent. The study was approved by the ethical com-mittee of Ulm University (No. 312/08).
Assessment of HCU
Participants were asked the total number of times that they had contacted an out-patient physician in the last 12 months. If possible, details were checked with the accompanying person (usually relatives). The number of drugs was estimated by counting every medication that was taken orally (including over the counter medica-tion). Medication assessments were done thoroughly by scanning available packag-es and screening for the most recent pre-scription lists (participants were asked to
290 | Zeitschrift für Gerontologie und Geriatrie 4 · 2012
Beiträge zum Themenschwerpunkt
have their lists and drugs ready before the visits). The total length of stay in a hospital over the last 12 months were also calculat-ed by asking the patient and, if necessary and available the accompanying relatives.
Assessment of comorbidity
A total of 21 chronic conditions (our ex-tended version of the functional comor-bidity assessment) [11] were assessed. Par-ticipants were asked whether a doctor had ever told them that they had the particu-lar disease. Comorbidities were added up without giving them certain weights and a total count was used as confounder [11].
Physical activity measurement
Physical activity was assessed using a uni-axial accelerometer (activPAL®, PAL Tech-nologies Ltd., Glasgow, UK). Several stud-ies confirmed the validity of the acceler-ometer [10, 20].The sensor was attached to the participants’ right thigh and was sealed against water. Participants were instructed to wear the sensor at day and
night over a 1-week period. The signal analysis differentiated between three cat-egories of physical activity: (1) laying or sitting, (2) standing and (3) walking. For the current analysis, average daily walking activity (in minutes) was used. Only com-plete days with activity measurement over 24 h were considered.
Physical performance and (instrumental) activities of daily living
The ActiFE Ulm study included sever-al physical performance measures like a balance test, habitual gait speed and the Five-Chair-Rise test as recommended for the Short Physical Performance Battery (SPPB) [12]. Furthermore the 10-item (instrumental) activities of daily living (IADL) was assessed including questions on dressing, bathing, getting up and down from a chair, walking on the same level as well as going up and down stairs, doing light housework, using transportation, going shopping, organizing medication plans and cutting toenails. Questions were
taken from the LASA study [6] and from a publication by Saliba and colleagues [21] who recommended four of the above named assessments to capture more than 90% of disability in older adults. To avoid collinearity, the ADL measure was used as a proxy for physical functioning.
Other assessments
Further assessments included a question about loneliness and overall pain rated on a numerical scale from 0 (not lonely/no pain) to 10 (very lonely/strongest pain), the professional career/status (9-point scale converted to a 4-point scale), health-related quality of life (the physical compo-nent summary scale of the 12-Item Short Form Health Survey, SF-12), a 6-item so-cial network assessment according to Lub-ben [16], cognition (Mini Mental State Ex-am, MMSE) [8], affective state with a fo-cus on depression (depression scale of the Hospital Anxiety and Depression Scale, D-HADS) [26], body mass index (BMI), age and gender.
291Zeitschrift für Gerontologie und Geriatrie 4 · 2012 |
M.D. Denkinger · A. Lukas · F. Herbolsheimer · R. Peter · T. Nikolaus
Physical activity and other health-related factors predict health care utilisation in older adults. The ActiFE Ulm study
AbstractBackground. Health care utilisation (HCU) can be a useful outcome for estimating costs and patient needs. It can also be used as a surrogate parameter for healthy ageing. The aim of this study was to analyse the associ-ations of formerly described and potentially new parameters influencing health care utili-sation in older adults in Germany.Patients and methods. The ActiFE Ulm (Ac-tivity and Function in the Elderly in Ulm) study is a population-based study in 1,506 community dwelling older adults aged 65–90 years in Ulm and surrounding areas in south-western Germany. Between March 2009 and April 2010 a full geriatric assessment was per-formed including accelerometer-based av-erage daily walking duration, comorbidity, medication, physical and psychological func-tioning, health care utilisation, sociodemo-graphic factors etc. The association between above named measures and health care util-
isation, represented by the number of drugs, the days in hospital and the number of phy-sician contacts over one year was calculated in multiple regression models. Analysis was conducted among subjects with complete in-formation (n=1,059, mean age 76 years, 55% male).Results. The average number of drugs was 4.5 and over 95% of participants visited a physician at least once a year while still more than 65% contacted their physician more than twice a year. Reduced physical activi-ty, BMI, self-rated health and/or comorbidi-ty and male sex were the best predictors of health care utilisation in community dwell-ing older adults when looking at both the number of drugs and the number of physi-cian contacts over 12 months together. With regard to single diseases entities the best predictors of both the number of drugs and the number of physician contacts were asth-
ma, chronic obstructive pulmonary disease (COPD)/chronic bronchitis and chronic neuro-logical diseases (mostly Parkinson’s disease). The number of drugs was most strongly asso-ciated with coronary heart disease, diabetes, and high blood pressure.Conclusion. Reduced walking activity, self-rated health and/or comorbidity and male sex are the best predictors of health care utilisation as measured by the number of drugs and number of physician contacts over 12 months. Walking activity could be regard-ed as the most promising modifiable predic-tor of HCU in older adults.
KeywordsHealth services for the elderly · Aged, 80 and over · Number of drugs · Comorbidity · Prevention
Körperliche Aktivität und andere gesundheitsbezogene Faktoren sind bei Älteren Prädiktoren für die Inanspruchnahme des Gesundheitswesens. Die Studie ActiFE Ulm
ZusammenfassungHintergrund. Die Erfassung einer Inanspruchnahme des Gesundheitswesens kann zur Kostenabschätzung nützlich sein und einen Eindruck vermitteln, welche Er-krankungen Patienten am meisten beein-trächtigen. Somit kann es auch als Surro-gat für gesundes Altern verwendet werden. In dieser Analyse sollten die Zusammenhän-ge bereits bekannter und möglicher neuer Faktoren mit der Inanspruchnahme des Ge-sundheitswesens bei älteren Menschen in Deutschland beschrieben werden.Patienten und Methoden. Die Studie ActiFE Ulm (Activity and Function in the Elderly in Ulm) ist eine populationsbasierte Studie bei 1506 zu Hause lebenden Personen zwischen 65 und 90 Jahren, die in der Stadt Ulm und Umgebung wohnen. Zwischen März 2009 und April 2010 wurde ein komplettes geria-trisches Assessment durchgeführt. Darunter waren u. a. eine akzelerometerbasierte Aktivi-tätsmessung, Assessments der Komorbidität, der Medikation, der physischen und psy-chischen Funktionen, soziodemographische und zahlreiche weitere Faktoren. Die Zusam-
menhänge zwischen den oben genannt-en Parametern und der Inanspruchnahme des Gesundheitswesens, dargestellt durch die Zahl der Medikamente, die stationären Tage im Krankenhaus und die Zahl der Arzt-besuche in einem Jahr, wurden mittels multi-pler Regression (verschiedene Modelle) bere-chnet. Von 1059 Teilnehmern konnten kom-plette Daten ausgewertet werden (Alter im Mittel 76 Jahre, 55% Männer).Ergebnisse. Die mittlere Anzahl an Medika-menten lag bei 4,5, und über 95% der Teil-nehmer hatten im letzten Jahr mindeste-ns einmal einen Arzt aufgesucht, mehr als 65% sogar dreimal und öfter. Eine reduzier-te körperliche Aktivität, der Body-Mass-In-dex, die Selbsteinschätzung der Gesund-heit und die Komorbidität sowie männli-ches Geschlecht waren die besten Prädikto-ren der Inanspruchnahme des Gesundheits-wesens, wenn man die beiden Endpunkte, die Zahl der Medikamente und die Arztbe-suche zusammen betrachtet. Bei der Anal-yse einzelner Krankheitsbilder waren Asth-ma, COPD/chronische Bronchitis und chro-
nische neurologische Erkrankungen führend. Bei alleiniger Betrachtung der Medikamen-tenzahl als Endpunkt zeigten KHK, Diabetes, hoher Blutdruck und andere, vor allem kardi-ologische Krankheitsbilder die stärkste Asso-ziation. Krankenhausaufenthalte waren nur mit der Anzahl der Medikamente signifikant korreliert.Schlussfolgerung. Reduzierte körperliche Aktivität, Selbsteinschätzung der Gesund-heit, Komorbidität sowie das männliche Ge-schlecht sind die besten Prädiktoren für die Inanspruchnahme des Gesundheitswesens. Die körperliche Aktivität kann aus Sicht der Prävention und des gesunden Alterns als in-teressantester modifizierbarer Parameter an-gesehen werden.
SchlüsselwörterGesundheitsversorgung für Ältere · Alte Menschen · Medikamentenzahl · Komorbidität · Prävention
292 | Zeitschrift für Gerontologie und Geriatrie 4 · 2012
Statistical analyses
Analyses were conducted in subjects with complete information (n=1,059). Because numbers were high enough to reach sta-tistical significance and because there was no significant difference to baseline char-acteristics of the missing group, only one imputation was performed: for partici-pants lacking data on falls from the pro-spective falls calendar, the number of falls of the retrospective question (“how many times did you fall in the last 12 months?”) was imputed (n=82). For data description means and standard deviations (SD) were calculated for both the group with miss-ing values in any of the parameters of in-terest and for the population analysed. For the descriptive table t tests and χ2 test were used where appropriate to test for differ-ences between groups.
To assess the strength of the associa-tion of parameters mentioned above with HCU multiple regression analysis was used. Because the outcome variables phy-sician contacts and days in hospital are counts, ill-dispersed (not normally dis-tributed) and have a standard deviation greater than the mean a negative binomi-al regression model was calculated [24]. For the number of days in hospital which exhibits a large number of zero values, a zero-inflated negative binomial regres-sion was performed. To show the strength of association, the p value and the t sta-tistics (useful when p is <0.001 to obtain an estimate of the relative association as compared to other parameters) are given. The t statistic is the estimate divided by the standard error of the respective pre-dictor and was obtained here by calculat-ing the square root of the χ2 statistic. Usu-ally in negative binomial regression only nonparametric χ2 are given and t statistics are not exactly comparable. We still decid-ed to do this simple transformation in or-der to allow comparison of the strength of associations between the models. In all models we were testing the null hypoth-esis that an individual predictor’s regres-sion coefficient is zero, given the rest of the predictors are in the model. The α er-ror was set to 0.05.
Predictors were chosen according to available literature and after calculating bivariate associations using Spearman re-
Tab. 1 Patients’ characteristics (n=1,487)
Parameter Population ana-lysed (n=1059)
Population with missings (n=428)
p value
Sociodemographic characteristics
Age, mean (±SD) 75.84 (±6.55) 75.05 (±6.66) 0.038
Male sex, n (%) 583 (55.1) 255 (59.6) 0.111
Social network (Lubben 6-item scale), n (%) 0.983
<12 suspected supporters (socially isolated)
217 (20.5) 87 (20.7)
≥12 suspected supporters 842 (79.5) 334 (79.3)
Professional prestigeh, n (%) 0.824
Limited (housework, not skilled) 178 (16.9) 67 (15.7)
Acceptable (skilled worker) 300 (28.4) 112 (26.3)
Good (middle level, own small enterprise) 321 (30.4) 130 (30.5)
Maximum (high level, CEO, academic) 256 (24.3) 117 (27.3)
Geriatric domains
Falls per yeara, n (%) 0.847
No falls 685 (64.7) 283 (66.1)
1 fall 222 (21.0) 90 (21.0)
>1 falls 152 (14.3) 55 (12.9)
(Instrumental) Activities of daily livingb, n (%)
0.499
Considerable difficulties 91 (8.6) 44 (11.3)
Little or no difficulties 968 (91.4) 384 (89.7)
Physical health
Body mass index, mean (±SD) 27.86 (±4.23) 26.88 (±3.93) <0.001
Physical activityc, mean (±SD) 104.00 (±40.93) 113.65 (±40.8) <0.001
Self-rated healthd, mean (±SD) 54.44 (±7.39) 55.02 (±6.78) 0.013
Comorbidity (disease counts)e, mean (±SD) 6.1 (±3.42) 4.6 (±3.26) <0.001
Mental health
Mini mental status, mean (±SD) 27.9 (±2.11) 27.5 (±2.88) 0.005
Depression (DHADS)f, n (%) 0.003
Depressed 126 (11.9) 39 (9.1)
Not depressed 933 (88.1) 389 (90.9)
Loneliness NRS Scaleg, n (%) 0.012
Lonely 132 (12.5) 39 (9.1)
Not lonely 927 (87.5) 389 (90.9)
Health care utilisation
Number of drugs, mean (±SD) 4.54 (±3.01) 4.32 (±2.69) 0.300
Hospital length of stay over 12 months 0.864
None 840 (80.1%) 342 (80.9%)
1–14 days 209 (15.3%) 81 (15.8%)
>14 days 49 (4.7%) 14 (3.3%)
Physician contacts (outpatient) over 12 months
0.068
None 39 (3.7%) 37 (8.7%)
1–2 times 269 (25.4%) 146 (34.3%)
>2 times 750(70.9%) 243 (57.0%) aProspective falls calendar over 1 year; b10-item assessment including IADL and ADL tasks; cAccelerometer-based total walking time/day (average over 4–7 days including weekends); d Physical subscale of the SF-12; eExtended functional comorbidity index [11]; fDHADS depression subscore of the HADS; g“How lonely do you feel on a scale from 0 to 10?”; hShorted version of the 9-item question in the ActiFE study. NRS Numerical Rating Scale, SD standard deviation; (D)HADS Hospital Anxiety and Depression Scale.
293Zeitschrift für Gerontologie und Geriatrie 4 · 2012 |
gression coefficients and two by two ta-bles. Age and sex were always forced in-to the models.
All analyses were performed using SAS9.2 and Enterprise Guide 4.1 if appli-cable (SAS Institutes, Cary, NC, USA).
Results
Patients’ characteristics
Patients were classified into one group with missing values in any of the param-eters of interest and into the population analysed. Participant characteristics are demonstrated according to their classi-fication in . Tab. 1. Patients with miss-ing values were significantly younger, less overweight and more often males. They seemed to be slightly better educated, less vulnerable, had higher physical activity levels, fewer comorbidities, lower depres-sion scores, fewer falls, fewer physician contacts and more medications. Howev-
er, they reported more difficulties in the IADL measures. Social support (network), cognition, hospitalisation and self-rated health were comparable.
Overall, in the whole population, the number of community dwelling partic-ipants who visited any doctor in the last year was more than 95% and more than 65% visited their doctors even more than twice a year. Hospitalisations occurred in about 20% while less than 5% had to stay more than 2 weeks.
Multivariate predictors of HCU
Reduced physical activity, lower self-rat-ed health and male sex were consistently and strongly associated with the number of drugs and number of physician visits. BMI was only associated with the num-ber of drugs, and loneliness with num-ber of physician visits. Age and comor-bidity were strongly, but only associated with the number of drugs. Comorbidity,
however, was strongly dependent on self-rated health with regard to physician con-tacts. When omitting self-rated health or comorbidity, each of the parameters be-came strongly associated. With both vari-ables in the model self-rated health pre-vailed. We have, therefore, decided to only control for self-rated health in this mod-el. The number of drugs was associated with both the number of physician con-tacts and was the only variable to signifi-cantly predict the days in hospital.
Single disease predictors
When looking at the individual diseases reported in . Tab. 2, no significant as-sociations were found for the length of stay in a hospital. Having asthma, COPD/chronic bronchitis and chronic neurolog-ical diseases like Parkinson’s disease were associated with both the number of phy-sician contacts and the number of drugs. Coronary heart disease/infarction, heart
Tab. 2 Single diseases associated with three health care utilisation outcomes (n=1,026)
Outcomes Number of drugs Number of physician contacts (1 year total)
Length of stay in a hospital (1 year total)
Diseases t value Estimate p value Rooted χ2 Estimate p value t value Estimate p value
R2, explained variance 0.42 n.a.* n.a.*Association of independent parameters with each outcome in a linear (drugs), a negative binomial (physician contacts) and a zero-inflated negative binomial (hospital stay) regression model. Estimate regression coefficient. t value regression coefficient divided by standard error: a marker for the strength of an association. The models are con-trolled for all variables as shown in Table 2. Please note that for the negative binomial regression the square root of the Wald χ2 statistic is given to allow comparison of the associations between the models. n.a. not applicable, COPD chronic obstructive pulmonary disease *Not available in negative binomial regression.
294 | Zeitschrift für Gerontologie und Geriatrie 4 · 2012
Beiträge zum Themenschwerpunkt
failure, diabetes, high blood pressure, thy-roid problems, osteoporosis, neurological diseases and chronic liver diseases were all significantly associated with the number of drugs, but not with number of physi-cian visits. Concerning the latter, having had a diagnosis of cancer and high cho-lesterol were weakly associated. For re-gression coefficients and detailed p values please refer to . Tab. 3.
Discussion
Reduced physical activity, self-rated health and/or comorbidity and male sex were the best predictors of health care utilisation in community dwelling older adults when looking at both the number of drugs and the number of physician con-tacts over 12 months together. With regard to single diseases entities the best predic-tors of both the number of drugs and the number of physician contacts were asth-ma, COPD/chronic bronchitis and chron-
ic neurological diseases (mostly Parkin-son’s disease). The number of drugs itself was, not surprisingly, most strongly asso-ciated with coronary heart disease, dia-betes, high blood pressure, thyroid prob-lems, osteoporosis and heart failure.
The results correlate to some extent with findings from the Berlin Aging Study (BASE) [15] with regard to the role of co-morbidity that was also clearly the best predictor of the number of drugs taken and which was less significant with re-gard to physician contacts. Important-ly, when using physician contacts as the outcome variable in our analyses self-rat-ed health reduced a highly significant es-timate of comorbidity. This indicates not only a high intercorrelation of the two predictors but also dependency. There-fore, for statistical reasons we had to re-move the weaker variable to reduce multi-collinearity. Self-rated health in our study could represent the correlate to the more thorough assessments on attitudes and be-
liefs in the paper by Linden and colleagues where comorbidity was also only slightly associated with physician contacts. How-ever, Linden and colleagues did not report problems with collinearity. Differences are the significant associations of living alone with both endpoints in BASE while male sex only played a role in the ActiFE pop-ulation. This could be due to the different predictors that were entered in the models such as the ADL measures, BMI and walk-ing activity (ActiFE Ulm) or the number of children in town and more complex psychosocial measures in BASE.
The relationship of comorbidity and self-rated health with physician contacts is of greater interest since Lehnert and colleagues [14] classified comorbidity (= multiple chronic conditions) as the most firmly associated factor with HCU. Self-rated health could for example be one of the most readily modifiable predictors, while comorbidity usually can not. Al-though being highly correlated, self-rated
Tab. 3 Parameters associated with three health care utilisation outcomes (n=1,056)
Outcomes Number of drugs Number of physician contacts (1 year total)
Length of stay in a hospital (1 year total)
Parameters t value Estimate p value Rooted χ2 Estimate p value t value Estimate p value
Age 3.49 0.043 0.001 0.39 0.002 0.695 −0.39 −0.006 0.698
Male sex 4.27 0.636 <0.001 4.14 2.726 <0.001 −0.53 −0.104 0.594
R2, explained variance 0.29 n.a.** n.a.**Association of independent parameters with each outcome in a linear (drugs), a negative binomial (physician contacts) and a zero-inflated negative binomial (hospital stay) regression model. Estimate regression coefficient. t value regression coefficient divided by standard error: a marker for the strength of an association. All parameters that we controlled for are shown. Please note that for the negative binomial regression the square root of the Wald χ2 statistic is given to allow comparison of the associations between the models.aEstimated using a prospective falls calendar; b11-degree numerical rating scale; cphysical subscore of Short Form-12; daccelerometer-based walking activity; eextended functional comorbidity index [11]; f1-degree numerical rating scale; g6-item Lubben social network scale; h4-degree Likert scale. MMSE Mini Mental State Examination, DHADS Depression subscore of the Hospital Anxiety and Depression Scale, n.a. not applicable, *Comorbidity was strongly de-pendant on self-rated health with regard to physician contacts. When adding both confounders, self-rated health prevails. Therefore, comorbidity was omitted from this model.**Not available in negative binomial regression.
295Zeitschrift für Gerontologie und Geriatrie 4 · 2012 |
health includes further aspects of an indi-viduals’ health such as attitudes and be-liefs. One could hypothesize that by try-ing to modify these attitudes HCU could decrease. However, it is not clear whether simply reducing the number of physician contacts (by strengthening an older adults’ self-efficacy, his or her health attitude and level of mastery) could even have a neg-ative impact on mortality for example. Still, the Germans visit their GPs more of-ten than any other population in Europe; hence, is there some room for reduction without a negative impact?
Other modifiable parameters are BMI, loneliness or overall walking activity. In ActiFE Ulm, BMI and loneliness were both positively associated with the amount of HCU. This has not been reported be-fore in all German cohorts mentioned who did not seem to have controlled for these factors. It is rather surprising that in a group of older adults with a mean age of 75 years and balanced age stratification, healthy ageing was associated with a low-er BMI because some studies have shown a U-shaped correlation with a turning point at 75–80 years [23]. On the other hand this finding supports the hypothesis that higher BMI scores are associated with ex-cess mortality until the age group of 75–85 [22]. Interestingly, loneliness was associat-ed with the number of physician visits. It is unclear whether this narrowly statistical-ly significant finding really reflects a need for social communication or is just a sub-group of depression. Still, this surprising result warrants further study.
With regard to walking activity none of the available studies in older adults have demonstrated this association [14]. A similar finding as only been reported in a large Canadian study in patients with type 1 and type 2 diabetes and an average age of 63 years (type 2 diabetes) linking self-reported physical activity with HCU and costs [19]. In addition, a great body of evidence has linked physical activity with many different outcomes including mor-tality [3]. If the association was truly inde-pendent of other confounders, programs to increase activity in older adults might also decrease the amount of HCU in this population.
When looking at single disease en-tities we found a strong relationship of
the common diseases such as high blood pressure, diabetes or cardiovascular dis-ease with multimedication. This is not surprising since the most commonly pre-scribed drugs are those related to vascular risk factors and other studies have dem-onstrated comparable data [19]. Consid-ering both the number of physician vis-its and drug use, only asthma and COPD and chronic neurological diseases became significant. This is not surprising because respiratory diseases have been associated with a largely increased economic burden [2, 18]. This burden is expected to increase in upcoming years with a steadily rising prevalence of COPD in all age groups in-cluding older adults [5]. Considering de-mographic change a concerted public health effort seems mandatory in order to handle rising health care utilization needs and costs of (elderly) patients with respi-ratory diseases.
Other disease that have been report-ed in the literature such as osteoarthritis, rheumatoid arthritis, depression and oth-ers were not or were not as clearly associ-ated with HCU in the ActiFE-Ulm study. The reasons might be two-fold: first we had to solely rely on self-report and could not include disease severity in the analy-ses. Second, other studies have analysed certain pathologies as compared to gen-eral populations while in this study we have compared all 21 disease entities head to head in a population-based cohort with competing comorbidities.
Limitations, recommendations and outlook
In general, the need to rely on information given by the participant without the op-portunity to double check with the gener-al physician (e.g. as compared to the BASE study) can be seen as an important limita-tion of our study. However, the probability of incorrect information should be quite low because of a rather well-function-ing population with mostly intact cogni-tion. Furthermore, our cohort might not be entirely representative of the general German population in the specified age group, because recruitment methods of ActiFE-Ulm [7] might have fostered the selection of a higher functioning popula-tion of older adults. Age stratification with
an overrepresentation of older age groups, on the other hand, could have decreased this selection bias to some extent (because older adults are more vulnerable per se).
To pursue research on these cross-sec-tional findings, future follow-up studies of the ActiFE Ulm population will include measures on HCU as well as another de-tailed assessment of physical activity. We will then have to test whether people who experienced changes in their BMI or who have increased their activity levels have decreased the level of HCU and vice ver-sa. This hypothesis could of course be test-ed in other observational cohort studies in older adults around the world. So again, besides an increased physical activity an increased analytical activity is warranted.
Conclusions for clinical practice
F More than 95% visit any doctor with-in 1 year and more than 65% visited their doctors even more than twice a year. Hospitalisations occurred in about 20% while less than 5% had to stay more than 2 weeks.
F Walking activity (little activity), self-rated health (or comorbidity) and male sex are the best predictors of health care utilisation as measured by the number of drugs and number of physician contacts over 12 months.
F Asthma, COPD/chronic bronchitis and chronic neurological diseases (most-ly Parkinson’s disease) are the three single diseases that are most frequent in health care utilisation (number of drugs and number of physician con-tacts).
F With regard to multimedication only, the most strongly associated diseases are coronary heart disease, diabetes, high blood pressure, thyroid prob-lems, osteoporosis and heart failure.
F Physical activity is the only modifi-able parameter that is significant-ly associated with both measures for health service utilisation, underscor-ing its importance for interventions with regard to healthy ageing
F Loneliness is associated with the number of physician contacts, a find-ing that clearly warrants further study.
296 | Zeitschrift für Gerontologie und Geriatrie 4 · 2012
Beiträge zum Themenschwerpunkt
Corresponding address
Dr. M.D. DenkingerAgaplesion Bethesda ClinicGeriatric Center Ulm UniversityZollernring 26, 89073 [email protected]
Acknowledgement. The authors would like to thank all participants in the study.The ActiFE Ulm study group consists of: B. Böhm, De-partment of Internal Medicine I – Division of Endo-crinology, M. Denkinger, Agaplesion Bethesda Clin-ic, Ulm, Germany, H. Geiger, Department of Dermatol-ogy and Allergology, F. Herbolsheimer, Institute of Ep-idemiology & Medical Biometry, A. Lukas, Agaplesion Bethesda Clinic, Ulm, Germany, J. Kirchheiner, Institute of Pharmacology of Natural Products & Clinical Phar-macology, W. Koenig, Department of Internal Medicine II – Cardiology, G. Nagel, Institute of Epidemiology & Medical Biometry, T. Nikolaus (PI), Agaplesion Bethes-da Clinic, Ulm, Germany, R. Peter (PI), Institute of Epide-miology & Medical Biometry, M. Riepe, Division of Ge-rontopsychiatry, Dept. of Psychiatry and Psychother-apy II, L. Rudolph, Max-Planck Group for Stem Cell Re-search, D. Rothenbacher, Institute of Epidemiology & Medical Biometry, K. Scharffetter-Kochanek, Depart-ment of Dermatology and Allergology, Ch. Schumann, Department of Internal Medicine II – Pneumology, J.M. Steinacker, Department of Sports- and Rehabilita-tion Medicine, C. von Arnim, Department of Neurolo-gy, G. Weinmayr, Institute of Epidemiology & Medical Biometry. (if not differently indicated: All Ulm Universi-ty Ulm, Germany; PI = Principle Investigator)
Funding source. The study was funded by a grant from the Ministry of Science, Research and Arts, state of Baden-Wuerttemberg, Germany, as part of the Geri-atric Competence Center, Ulm University. Michael Den-kinger was supported by a research fellowship pro-gram from the Robert Bosch Foundation, Stuttgart, Germany, which did not have any influence on the content.
Conflict of interest. The corresponding author states that there are no conflicts of interest.
References
1. Andersen R (1973) Societal and individual deter-minants of medical care utilization in the United States. Milbank Mem Fund Q Health Soc 51:95
2. Antonicelli, L, Bucca C, Neri M et al (2004) Asth-ma severity and medical resource utilisation. Eur Respir J 23:723–729
3. Blair SN (2006) Objectively measured physical activ-ity and mortality in older adults. JAMA 296:216–218
4. Braune BT, Berger K (2005) The influence of de-pressive mood on activities of daily living and health care utilization in the elderly–the MEMO study on the KORA platform Augsburg. Gesund-heitswesen 67(Suppl 1):S176–S179
5. Bustacchini S, Chiatti C, Furneri G et al (2011) The economic burden of chronic obstructive pulmo-nary disease in the elderly: results from a system-atic review of the literature. Curr Opin Pulm Med 17(Suppl 1):S35–S41
6. Deeg DJH, Knipscheer CPM, Van Tilburg W (1993) Autonomy and well-being in the aging popula-tion: Concepts and design of the Longitudinal Ag-ing Study Amsterdam. Netherlands Institute of Gerontology, Bunnik, The Netherlands
7. Denkinger MD, Franke S, Rapp K et al (2010) Accel-erometer-based physical activity in a large obser-vational cohort–study protocol and design of the activity and function of the elderly in Ulm (ActiFE Ulm) study. BMC Geriatr 10:50
8. Folstein MF, Folstein SE, McHugh PR (1975) “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psy-chiatr Res 12:189–198
9. Glaesmer H, Braehler E, Riedel-Heller SG et al (2011) The association of traumatic experiences and posttraumatic stress disorder with health care utilization in the elderly – a German population based study. Gen Hosp Psychiatry 33:177–184
10. Grant PM, Ryan CG, Tigbe WW, Granat MH (2006) The validation of a novel activity monitor in the measurement of posture and motion during ev-eryday activities. Br J Sports Med 40:992–997
11. Groll DL, To T, Bombardier C, Wright JG (2005) The development of a comorbidity index with physical function as the outcome. J Clin Epidemiol 58:595–602
12. Guralnik JM, Simonsick EM, Ferrucci L et al (1994) A short physical performance battery assessing low-er extremity function: association with self-report-ed disability and prediction of mortality and nurs-ing home admission. J Gerontol 49:M85–M94
13. Hessel A, Gunzelmann T, Geyer M, Brähler E (2000) Utilization of medical services and medication in-take of patients over 60 in Germany – health re-lated, social structure related, socio-demographic and subjective factors. Z Gerontol Geriatr 33:289–299
14. Lehnert T, Heider D, Leicht H et al (2011) Review: health care utilization and costs of elderly persons with multiple chronic conditions. Med Care Res Rev 68:387–420
15. Linden M, Horgas AL, Gilberg R et al (1997) Pre-dicting health care utilization in the very old. The role of physical health, mental health, attitudinal and social factors. J Aging Health 9:3–27
16. Lubben J, Blozik E, Gillmann G et al (2006) Per-formance of an Abbreviated Version of the Lub-ben Social Network Scale Among Three Europe-an Community-Dwelling Older Adult Populations. Gerontologist 46:503–513
17. Nöthen M (2011) destatis Gesundheit – Magazin Wirtschaft und Statistik. Available at: http://www.destatis.de/jetspeed/portal/cms/Sites/destatis/In-ternet/DE/Navigation/Publikationen/Querschnitts-veroeffentlichungen/WirtschaftStatistik_Gesund-heit.psml. Accessed 25 January 2012
18. Peters D, Chen C, Markson LE et al (2006) Using an asthma control questionnaire and administra-tive data to predict health-care utilization. Chest 129:918–924
19. Plotnikoff RC, Karunamuni ND, Johnson JA et al (2008) Health-related behaviours in adults with di-abetes: associations with health care utilization and costs. Can J Public Health 99:227–231
20. Ryan CG, Grant PM, Tigbe WW, Granat MH (2006) The validity and reliability of a novel activity mon-itor as a measure of walking. Br J Sports Med 40:779–784
21. Saliba D, Orlando M, Wenger NS et al (2000) Iden-tifiying a short functional disability screen for old-er persons. J Gerontol A Biol Sci Med Sci 55:M750–M756
22. Stevens J, Cai J, Pamuk ER et al (1998) The effect of age on the association between body-mass index and mortality. N Engl J Med 338:1–7
23. Thinggaard M, Jacobsen R, Jeune B et al (2010) Is the relationship between BMI and mortali-ty increasingly U-shaped with advancing age? A 10-year follow-up of persons aged 70–95 years. J Gerontol A Biol Sci Med Sci 65A:526–531
24. UCLA: Academic Technology Services, Statistical Consulting Group. Introduction to SAS. Available at: http://www.ats.ucla.edu/stat/sas/notes2/. Ac-cessed 10 January 2012
25. Wernitz MH (2011) Gesundheitsökonomie und das deutsche Gesundheitswesen: ein praxisorien-tiertes Lehrbuch für Studium und Beruf/Martin H, Wernitz; Jörg Pelz. 1. Aufl. Kohlhammer, Stuttgart, Germany
26. Zigmond AS, Snaith RP (1983) The hospital anx-iety and depression scale. Acta Psychiatr Scand 67:361–370
297Zeitschrift für Gerontologie und Geriatrie 4 · 2012 |
