elderly in United Stated as a vulnerable population

Journal of Gerontology: MEDICAL SCIENCES © The Author 2010. Published by Oxford University Press on behalf of The Gerontological Society of America. Cite journal as: J Gerontol A Biol Sci Med Sci. 2010 July;65(7):734–742 All rights reserved. For permissions, please e-mail: [email protected]. doi:10.1093/gerona/glq037 Advance Access published on May 9, 2010

734

INCREASING numbers of vulnerable older persons chal-lenge primary care resources because of the aging pro- cess and the clustering of chronic illnesses within this age group. Primary care teams are often unaware of the actual health status and functional limitations of these patients (1–3). Primary care physicians (PCPs) increasingly lack the time and tools for proactive care, systematic assessment, long-term monitoring and management of chronic diseases, and disabilities associated with frail health (4). Proactive visiting of older frail patients by nurses may therefore im- prove service levels and prevent adverse health trajectories.

Preventive home visitation to older persons is part of na- tional policies in several countries, including the United Kingdom, Denmark, and Australia. The rationale behind home visitation is that it can trace and counter unmet needs

and thereby prevent and postpone adverse healthy trajecto- ries. This rationale may be threefold: (a) to delay or prevent functional impairment and subsequent nursing home admis- sions by primary prevention (eg, immunization and exer- cise), (b) secondary prevention of untreated symptoms, and (c) delay or prevent further deterioration of chronic dis- eases. Primary care practice rarely integrates home visita- tion by nurses. This can be placed against case management approaches that aim to collaborate more explicitly across disciplines.

Although visitation policies were implemented in several countries, the value of preventive home visitation is still controversial as reported in previous reviews and meta- analyses (5,6). Van Haastregt and coworkers (7) concluded that there was no clear evidence of beneficial effects of

Prevention of Adverse Health Trajectories in a Vulnerable Elderly Population Through Nurse Home Visits:

A Randomized Controlled Trial [ISRCTN05358495]

Hein P. J. van Hout,1 Aaltje P. D. Jansen,1 Harm W. J. van Marwijk,1 Marieke Pronk,1 Dinnus F. Frijters,2 and Giel Nijpels1

1Department of General Practice and 2Department of Nursing Home Medicine, EMGO Institute, VU University Medical Center Amsterdam, The Netherlands.

Address correspondence to Hein P. J. van Hout, PhD, Department of General Practice, EMGO Institute, VU University Medical Center Amsterdam, Van der Boechorststraat 7, 1081 BT Amsterdam, The Netherlands. Email: [email protected]

Objective. Can indicative prevention of home-visiting nurses be effective when targeted at a frail senior population using multidimensional geriatric assessments and personalized care plans?

Methods. We performed an individually randomized controlled trial in 33 blinded primary care practices over 18 months. The 651 participants were aged 75 years or older, lived at home, and were frail but neither terminally ill nor demented. A score in the lowest quartile on at least two of six self-reported functional health domains (COOP–WONCA charts), defined frail health. We compared usual care with proactive home visits by trained community nurses. The nurses (a) assessed the care needs with a multidimensional computerized geriatric instrument, which enabled direct identification of health risks; (b) determined care priorities together with the person; (c) designed and executed individually tailored interventions; and (d) monitored participants by telephone and on average three home visits. Primary outcome measures were functional health and instrumental activities of daily living disability. Secondary outcomes were acute hospital ad- mittance (time until), institutionalization, and mortality.

Results. We found no significant differences between intervention and usual care group on any of the outcome measures. Predefined subgroup analyses revealed a higher risk of hospital admission for persons with poor health in the intervention group.

Conclusions. We could not demonstrate preventive effects of home visits by nurses in vulnerable older persons. Hos- pital admissions increased in the frailest group. The search for effective interventions for vulnerable persons requires further investigation. Future efforts may focus on improved integrated approaches.

Key Words: Primary care—Frailty—Trial—Case management—Nurse—Old age.

Received August 14, 2009; Accepted March 5, 2010

Decision Editor: Luigi Ferrucci, MD, PhD

PREVENTION OF ADVERSE HEALTH TRAJECTORIES 735

preventive home visits among community-dwelling older persons on physical functioning, institutionalization, and mortality. In contrast, Elkan and coworkers (5) reported sig- nificant reductions in mortality and institutional care and stronger effects when targeted to frail older persons. This conclusion was consistent with subgroup findings of a pre- vious Dutch study (8,9). In addition, Stuck and coworkers (6) reported positive effects on mortality, nursing home ad- missions, and functional status based on conditions such as a reasonable number of home visits were offered and care plans were based on multidimensional assessments. A re- cent meta-analyses of Beswick and coworkers (10) could not confirm the importance of visit frequency. In the face of these conflicting results, we hypothesized that preventive home visits by nurses may prevent functional decline, insti- tutionalization, and mortality if they target vulnerable older persons, include multidimensional assessments, and indi- vidualized care plans.

Methods

Design and Setting We performed an individually randomized controlled

trial over a period of 18 months in a frail, elderly (75+) community-dwelling population sample obtained from 33 primary care practices and 55 PCPs. PCPs were blinded to whether their listed patients received preventive home visit- ing by a nurse or usual care. We randomized eligible per- sons living at the same address as one unit. To ensure equal numbers of intervention and usual care patients per prac- tice, we used block randomization with blocks of four cases implicating six possible variations of assignment to Inter- vention (I) and Controls (C) per block (eg, IICC, CCII, ICIC, CICI, ICCI, CIIC). These block variations numbered one to six. Next, we used Pocock’s random number table and assigned up to 10 blocks per practice. An independent statistician kept the assignment lists and assigned individu- als to the intervention or control groups. PCPs were blinded for the group assignment in order to minimize contamina- tion. Data entry personnel were blinded for group assign- ment as well. The region had 1 local hospital, 2 nursing homes (200 beds), and approximately 20 homes for the elderly ( 1,000 beds). The ethical committee of the VU Medical Center approved the study. All participants pro- vided signed informed consent.

Study Population All inclusion and exclusion criteria are summarized in

Table 1. The PCPs provided the names and addresses of all their patients who were aged 75 years or older and lived at home. All persons received a postal health survey including the COOP–WONCA charts in order to identify persons with the frailest health (11,12). COOP–WONCA charts are translated and validated in many languages and are used in

Table 1. Inclusion and Exclusion Criteria

Inclusion Age 75 y and older and listed as primary care practice patient Living at home Frail: self-reported score in the worst quartile of at least two of six COOP–WONCA charts (scoring range: 1, excellent to 5, very bad): overall health ≥4; physical fitness ≥5; changes in health ≥4; daily activities ≥4; mental health ≥3; social activities ≥3 Exclusion Terminally ill as determined by PCPs Persons with dementia symptoms (self-report of memory deterioration and MMSE < 24 or 7-min screen >50%) Living in residential homes Participating in other research projects

Note: MMSE = Mini-Mental State Examination.

primary care settings worldwide to quickly identify func- tional health status. It comprises of six charts on functional health domains. These charts cover (a) overall health, (b) changes in health, (c) physical fitness, (d) daily activities, (e) mental health, and (f) social activities.

For our study, persons scoring in the lowest quartile on at least two of the six charts were considered vulnerable. The quartile cutoffs were based on representative Dutch com- munity–based data supplemented (12,13). Dutch COOP– WONCA scores are comparable with other countries (14,15).

Intervention Protocol Eight trained community nurses executed the preventive

home-visiting program. The visiting program comprised seven key characteristics. First, an assessment of health risks and care needs was performed using the Resident As- sessment Instrument–Home Care version (RAI-HC), a structured multidimensional geriatric instrument developed in the United States (16,17). The assessments were directly entered on laptops, which enabled direct and validated iden- tification of 30 modifiable health risks (Table 3). Assess- ment times ranged between 45 and 75 minutes. The second characteristic is based on the nurses recommending inter- ventions based on the Resident Assessment Instrument manual and a nationally issued nursing guideline. The inter- ventions were created from best practices or evidence-based research and comprised typical geriatric issues, such as home safety, fall prevention, medication adherence, and health promotion. A third feature is exemplified in the de- sign and execution of individually tailored care plans in that they take into consideration the preferences of the individ- ual. The fourth key characteristics was that the nurses left a copy of the care plan at a person’s home to inform other visiting health professionals and to encourage them to add notes to the care plan. A fifth characteristic of the visiting program was that the nurses visited a patient at least four times a year in order to execute and monitor the care plan, evaluate changes in care needs, and adapt the care plan when needed. Another key characteristic was in case of

VAN HOUT ET AL.736

urgent medical matters, the nurses were allowed to consult the PCPs. Lastly, after a year, the nurses reassessed the older person and repeated the protocol.

Training.—The eight community nurses had a long- standing experience in working with frail community- dwelling elderly. Before the study started, we trained the nurses during a 2-day session. The goals of the training session were fivefold. The first training objective taught nurses to apply and interpret the (computerized) geriatric assessment and design and execute care plans based on the RAI-HC outcomes on health risks. Second, the training also taught nurses to use an up to date overview of local social and health care services including their contact per- sons and telephone numbers to easily recommend and re- fer the older persons to appropriate services. A third objective was to train nurses to use a computer for the as- sessments. A fourth objective included teaching nurses to utilize monthly meetings, supervised by two senior staff members to discuss problematic cases that occurred dur- ing the study. The aim was to assure and standardize the quality of care. Lastly, the nurses also received regular educational updates.

Usual Care Usual care varied from no care at all to regular PCP visits

to home care involvement. Table 2 provides an overview of the usual care at baseline in the two groups. During fol- low-up measurements, we asked about their service use over the past 2 months. At 6 and 18 months, they reported to have contacted their PCP on average approximately 1.7 (SD = 2.3) and 3.1 (SD = 1.8) times, respectively. These numbers were comparable with the intervention group. At 6 and 18 months, 35% and 27%, respectively, of the control persons received home care nursing.

Outcomes and Measurements

Primary outcomes

1. Functional status was measured with the COOP– WONCA charts (12) and Short Form 36 item (SF-36) and administered by postal mail (18,19).

2. Disability in activities of daily living (ADL) and instru- mental activities of daily living was measured with the Groningen Activity Restriction Scale (GARS) and ad- ministered by postal mail (20).

Secondary outcomes

1. Hospital admittance data were extracted from the local hospital registry, supplemented with self-report data. We defined emergency visits as visits to the emergency rooms of the hospital or overnight stay at an intensive care unit.

2. Time until placement in nursing homes or homes for dis- abled older persons was surveyed and cross-checked in these homes’ registries or in PCPs’ medical records.

3. Time until death was checked with the PCPs and the hospital database.

Sample Size Calculation Effect estimates were based on previous meta-analyses

and trials (5,21,22). To demonstrate a modest effect (Co- hen’s effect size of .33) on our primary outcome functional health (SF-36 and COOP–WONCA charts), we needed 202 participants per arm based on a two-sided alpha of .05, a 90% probability (23), and an anticipated attrition rate of 30%. Regarding our secondary outcome, we powered the study to be able detect a reduction of 10% in hospital admit- tance, institutionalization or mortality with a two-sided al- pha of .05 and 90% probability. No attrition was anticipated on these measures as we had access to databases with complete and accurate registrations. Therefore, the study required 261 persons per group.

Data Analysis Both “intention-to-treat” (ITT) and per protocol (PP)

analyses were performed. Group differences for the contin- uous outcomes (SF-36 and COOP–WONCA) were ana- lyzed with linear mixed models, a repeated measurement technique (unstructured covariance type) (24). This analysis yields restricted maximum likelihood estimates for the ef- fects of time (T), treatment group (G) and the Time × Treat- ment Group interaction (T × G). Statistically significant Time × Treatment Group interaction indicates that the two groups have different patterns of change in outcome mea- surements over time. A significant time effect only (T) indi- cates changes in time, which are, however, similar for the two groups. Differences in dichotomous outcomes between intervention and usual care patients on hospitalization and emergency visits were tested by logistic regression analyses and expressed in odds ratios (ORs) with 95% confidence intervals (CIs). For time until institutionalization and time until mortality, Cox regression models were used and ex- pressed in hazard ratios with 95% CIs. For the PP analyses, we compared the usual care group with intervention partici- pants who received four or more home visits. To check the robustness of the previous analyses, we also performed sen- sitivity analyses on the self-report measures with two impu- tation techniques: multiple linear interpolation and conventional last observation carried forward.

Although initial analyses revealed baseline imbalance on three variables (previous falls, presence of family caregiver), none of these differed substantially (<10%) and did not affect the effect estimates. We therefore presented unadjusted effect estimates on all outcomes. In addition, we explored potential effect modification by predefined analyses: gender, age

PREVENTION OF ADVERSE HEALTH TRAJECTORIES 737

(75–80 vs >80), number (<2 vs ≥2) and type of chronic disease, disability at baseline (GARS-4 ≥50 vs <50), self- reported Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE < 3.6 vs ≥3.6), and self-rated health (Euroqol-5 Dimensions [EQ5D] <55 vs >55) (29). We finally explored effect modification and effect differ- ences across the nurses. SPSS version 15 was used for all analyses. We set the significance level two tailed at .05. The

Table 2. Baseline Demographic Characteristics (mean [SD] or number of participants [% of group])

Intervention (N = 331)

Control (N = 320)

Age, M (SD) 81.3 (3.9) 81.5 (4.3) Female, n (%) 239 (72.2) 220 (68.8) Marital status, n (%) Married 127 (41.2) 119 (40.3) Never married 18 (5.8) 10 (3.4) Divorced 12 (3.9) 14 (4.7) Widowed 151 (49.0) 152 (51.5) Living alone, n (%) 181 (54.7) 176 (55.0) Education, n (%) None or low 226 (73.1) 213 (71.9) Moderate 55 (17.8) 56 (18.9) High 28 (9.0) 27 (9.1) Profession, n (%) Lower 103 (53.3) 93 (48.7) Middle 65 (33.7) 82 (42.9) Higher 16 (8.4) 25 (13.0) Daily functioning, GARS (18–72), M (SD) 55.6 (10.5) 56.8 (9.9) SF-36 physical score (0–100), M (SD) 31.8 (10.0) 31.9 (9.9) SF-36 mental score (0–100), M (SD) 44.2 (11.4) 45.0 (11.3) Depressive symptoms (0–60) CESD, M (SD) 18.1 (7.5) 17.5 (7.4) Cognitive impairment, IQCODE (≥3.6) n (%) 50 (15.1) 43 (13.4) Number of chronic diseases, M (SD) 2.1 (1.4) 2.0 (1.4) Chronic diseases, n (%) COPD 52 (15.7) 45 (14.1) Heart infarction 131 (39.6) 119 (37.2) Arterial dysfunction 61 (18.4) 60 (18.8) Cancer 56 (16.9) 49 (15.3) Diabetes mellitus 165 (49.8) 156 (48.8) Joint condition 42 (12.7) 43 (13.4) Rheumatism 23 (6.9) 20 (6.3) Hypertension 94 (28.4) 94 (29.4) Fall in the previous 6 mo, n (%)* 79 (23.9)† 108 (33.8)†

Hearing problems, despite aids‡ 132 (39.9) 119 (37.2) Vision problems, despite aids‡ 77 (23.3) 76 (23.8) Presence of family caregiver, n (%) 232 (70.1)† 197 (61.6)†

Used care in past 7 d, n (%) 139 (58.9) 141 (44.1) District nurse 33 (10.0) 27 (8.4) Physiotherapist 46 (13.9) 46 (14.4) Social worker 15 (4.5) 9 (2.8) Day care 62 (18.7) 54 (16.9) Meals on wheels 37 (11.2) 43 (13.4) Hospital admissions previous 2 mo, M (SD) 0.25 (0.57) 0.26 (0.64)

Notes: CESD = Center for Epidemiological Studies-Depression scale; GARS = Groningen Activity Restriction Scale measures both activities of daily living and instrumental activities of daily living. Higher scores indicate worse functioning; IQCODE = Informant Questionnaire on Cognitive Decline in the Elderly; SF-36 = Short Form 36 item version.

* We assigned one fall to persons who fell in the previous 6 months but whose number of falls was unknown (missing).

† Means differed significantly (p ≤ .05). ‡ Severe problems on at least one of three questions on hearing and three on

vision impairment (Strawbridge and coworkers, 1998) (30).

EMGO Institute for Health and Care Research employs a quality assurance policy. All research projects are subject to audits. We assured quality of the data by independent double checks of all forms.

Results

Recruitment The recruitment phase yielded 33 primary care practices

(55 PCPs) with average list sizes of 146 (SD = 53) persons aged 75 years or older. Figure 1 provides an overview of the recruitment, randomization, and follow-up. Approximately 4,823 patients received the health questionnaire by post. Nonresponders were more often woman (41.3% vs 35.8%,

2 = 13.4, p < .001) or older persons (83.5 vs 81.7, t = 13.8, p < .001). Among the 2,949 (61%) responders, 658 vulner- able patients were identified and randomized. The uptake of the intervention was high: only 41 participants in the inter- vention arm and 44 participants in the control arm withdrew because they considered it too much of a burden or had a lack of interest. Follow-up data on mortality, institutional- ization, and hospital admission were complete. Self-report outcomes (eg, COOP–WONCA, SF36, GARS) were com- plete for 72% of all surviving persons.

Compared with the nonvulnerable persons, the persons considered vulnerable by this definition had more frequently two or more chronic diseases (OR = 3.3, 95% CI: 2.8–3.9), more depressive symptoms (Center for Epidemiological Studies-Depression scale ≥ 16; OR = 7.6, 95% CI: 6.4–9.1), cognitive symptoms (IQCODE-SR ≥ 3.6; OR = 4.1, 95% CI: 3.4–5.1), and were less independent in ADL (GARS; OR = 0.89, 95% CI: 0.88–0.9). Moreover, 4.5 years fol- low-up data revealed that the vulnerable persons had a higher risk on hospitalizations and mortality (1.55, 95% CI: 1.30–1.84, and 1.39, 95% CI: 1.17–1.65), respectively. This indicates that we selected a truly vulnerable group.

Baseline Characteristics Table 2 presents the baseline characteristics of the par-

ticipants of the intervention and control groups. The SF-36 physical summary score was 32 (SD = 10) on a scale from 0 to 100. This indicates that we selected a frail portion, as mean values among community-dwelling elderly ranged between 60 and 80 (25). Randomization was successful as there were no relevant differences between intervention and control groups on baseline characteristics.

Performed Interventions In the intervention group, nurses administered the

RAI-HC to 298 of 331 (90%) persons, which demonstrates a good uptake of our intervention. Table 3 lists the health problems identified by the RAI-HC in the intervention group. A mean of 7.7 (SD = 3.4) problems were recorded.

VAN HOUT ET AL.738

“Preventive health” problems were triggered most fre- quently. These concerned mainly issues that are not rou- tinely offered in the Netherlands (for instance, fecal and endoscope screening and mammography of women older than 75 years). Other relevant problems were for instance, pain (n = 194) and medication management issues (n = 181). Of the persons who received the RAI-HC, 231 re- ceived one or more additional home visits. The nurses reg- istered 652 home visits with a mean of 3 (SD = 2.1) home visits per person during the 18 months of the study (includ- ing a second RAI-HC assessment). They registered in total 95 telephone contacts, with on average 0.3 contacts per per- son (SD = 0.7). For 222 (69%) intervention patients, the protocol requirement of four or more visits was met. Typi- cal interventions were referral for balance training to in- crease gait stability and prevent falls; referral to social

services to increase social activity and prevent loneliness, advice; and referral for auditory and visual aids, advice on life style issues to increase physical exercise and healthy nutrition, advice on safety issues, such as medication intake and aids to prevent falls. We compared changes in health risks between baseline and follow-up on 208 persons who completed a second RAI-HC. Of the 30 assessed health risks, 15 decreased significantly, 2 were equally present, and 13 risks had increased.

Functional Health and Disability Outcomes Tables 4 and 5 present the changes from baseline effects at 6

and 18 months across the participants in the intervention and control groups. No statistically significant differences were found between the control and intervention patients on any of

Randomized (N = 651)

Assessed for eligibility (N = 2949) Excluded (N = 2298)

Not frail (2171) Dementia symptoms (54) No informed consent (66) Other reason (7)

Enrolment:

Followed-up: -Hospital data base N = 320 -Self-reported data: T0: N = 250 T1: N = 229 T2: N = 209

Follow-up: Followed-up: -Hospital data base N = 331 -Self-reported data: T0: N = 282 T1: N = 244 T2: N = 215

Lost to follow-up: -Hospital-registered data: none -Self-reported data: N = 111 Reasons: Death: N = 31 Admitted to institution: N = 11 Medical reasons: N = 10 Moved out of region: N = 5 Burden too high: N = 4 No interest: N = 37 Other reason: N = 4 Reason unknown: N = 9

Lost to follow-up: -Hospital-registered data: none -Self-reported data: N = 116 Reasons: Death: N = 27 Admitted to institution: N = 8 Medical reasons: N = 12 Moved out of region: N = 1 Burden too high: N = 6 Lack of interest: N = 33 Other reason: N = 11 Reason unknown: N = 18

Analyses: Analyzed: -Hospital-data base: N = 320 -Self-reported data: T0: N = 250 (29 missing*) T1: N = 229 (8 missing*) T2: N = 209 (0 missing*)

*but completed the next questionnaire

Analyzed: -Hospital-data base: N = 331 -Self-reported data: T0: N = 282 (16 missing*) T1: N = 244 (9 missing*) T2: N = 215 (0 missing*)

*but completed the next questionnaire

Allocation: To control (N = 320)

To intervention (N = 331)

Figure 1. Flow chart.

PREVENTION OF ADVERSE HEALTH TRAJECTORIES 739

the outcomes. All Time × Group interactions were nonsignifi- cant in the linear mixed models. Some modest time trends were found for the group as a whole that included improvement on the SF-36, COOP–WONCA charts, and the GARS.

Mortality, Institutionalization, and Hospital Admission Over 18 Months

Twenty-three (6.9%) control persons compared with 20 (6.4%) intervention individuals died, whereas 39 (11.8%) controls versus 43 (13.4%) intervention persons were institu- tionalized. We found no statistically significant differences between control and intervention groups on time until death and institutionalization (Tables 4 and 5). In the intervention group, 163 (49.2%) persons were admitted at least once com- pared with 141 (55.9%) control persons (OR = 1.2,95% CI: 0.9–1.7, p = .19) and 107 (32.3%) versus 92 (28.8%) had at least one emergency visit (OR = 1.18, 95% CI: 0.85–1.65).

Subgroup Effects Both the ITT and the PP analyses had effect modifica-

tions. It showed that among persons with the poorest self- rated health (EQ5D < 55), intervention persons had a significantly higher risk to be admitted to a hospital com-

pared with the control persons (OR = 1.95, 95% CI: 1.2–3.1, p = .005) and 4.3 (1.2–15.5; p = .028), respectively. In addi- tion, a higher risk on acute hospital visits was found among persons assigned to the intervention group who had two or more chronic diseases (OR = 1.6, 95% CI: 1.04–2.4, p = .03). No differences across the nurses were found.

Vulnerability After 18 months, 61 (28.4%) intervention persons and 50

(23.9%) control persons no longer fulfilled our vulnerabil- ity criterion (OR = 0.79, 95% CI: 0.51–1.23).

Finally, the PP and sensitivity analyses were consistent with the results reported previously (data not shown).

Discussion

Main Results Contrary to our hypothesis, we could not demonstrate

any preventive effects of preventive home visiting by nurses in older vulnerable persons in primary care. Subgroup anal- yses revealed an increased risk of hospital admittance in intervention persons who reported worse health. Interven- tion persons also seemed to have a higher risk of emergency

Table 3. Domains Assessed by Nurses Based on the RAI-HC (16) of Intervention Patients at Baseline: Triggered Health Risks in Order of Frequency

Domains/Triggered Risks Example of Items Assessed n (%)

Preventive health Blood pressure check, flu vaccination, female breast cancer screen, blood check feces 286 (96.6) Pain Signs or complaints 194 (65.5) Medication management Adverse effects, known interactions, presence of supervisor 181 (61.1) Safety environment Lighting, carpets, stairs, bathroom, kitchen, in home climate 166 (56.1) Falls Gait, dementia, parkinsonism 164 (55.4) Health promotion Activity level, endurance, smoking 155 (52.4) Skin and food problems Skin changes, wounds, corn, infections 122 (41.2) Urinal incontinence catheter Frequency, use diapers, use catheter 113 (38.2) Visual functioning Deterioration 107 (36.0) Communication Auditory, visually and verbally 105 (35.4) Social functioning Loneliness, decrease in engagements 77 (26.0) Nutrition Weight change, malnutrition, undernourishment, cancer 47 (15.9) IADL/more formal care Disability 45 (15.2) Depression and anxiety Symptoms and complaints 44 (14.9) Cognition Memory function, orientation, judgment 43 (14.5) Oral/dental health Chewing, swallowing, dry mouth, brushing 41 (13.9) Fecal incontinence Diarrhea, constipation 35 (11.8) ADL revalidation potential Dependence, progression, comprehension 33 (11.1) Reduced service package Decrease services after improvements 29 (9.8) Bedsores Bed mobility, incontinence, history bedsores 26 (8.8) Psychopharmacological medication Behavioral complications, eg, falls, delusions, cognition 25 (8.4) Vulnerable support system Presence and care of family caregiver 17 (5.7) Medication adherence Treatment adherence <80% 16 (5.4) Dehydration Fever, decrease in food intake, liquid intake 15 (5.1) Heart and lungs Breast pain, short of breath, irregular pulse 14 (4.9) Problem behavior Wandering, harassment, resistance, inappropriate 6 (2.0) Palliative care Life expectancy <6 mo, hospice care 4 (1.4) Alcohol abuse Abuse, dependency 3 (1.0) Risk intramural admission ADL dependence, continence, mental stability, mobility 2 (0.7) Physical and mental abuse Fear, neglect, unexplained injuries 1 (0.3)

Note: ADL = activities of daily living; IADL = instrumental activities of daily living.

VAN HOUT ET AL.740

hospital visits when they suffered from multiple chronic conditions.

Strengths and Limitations Our study holds several strengths. First, it concerns one

of largest preventive trials in this field, which strongly con- tributed to the statistical power of the results. Second, we obtained complete outcome data on mortality, institutional- ization, and hospital visits over a long follow-up period. Third, the PCPs were blinded for group assignment, which protected the contrast between the groups. Lastly, a struc- tured intervention was performed in which a validated mul- tidimensional geriatric assessment protocol was used focusing on the patient as well as on the system around the patient. This was expected to increase the odds on positive intervention effects (6).

A limitation to the external validity of our findings may be the nonresponse to the postal health screening. The non- responders were likely to be frailer and perhaps in greater need of nurse support. Limitations to the internal validity of the study were first the limited linkage with primary care physicians, which may be a possible reason why the inter-

Table 4. Primary Outcomes Over 18 mo, Computed by Mixed Models

Functional Health and Disability

Intervention (N = 320)

Usual Care (N = 331)

SF-36 physical component(range 0–100) Baseline, M (SD) 31.8 (10.0) 31.9 (9.9) Time × group

F = 1.36, p = .26 6 mo, M (SD) 31.4 (9.3) 32.1 (9.4) 18 mo, M (SD) 30.7 (9.2) 32.2 (9.3) SF-36 mental component (range 0–100) Baseline, M (SD) 44.2 (11.4) 45.0 (11.3) Time × Group

F = 0.11 p = .90 6 mo, M (SD) 44.5 (10.5) 45.4 (10.6) 18 mo, M (SD) 43.9 (11.2) 45.2 (11.2) ADL and IADL (GARS range 18–72) Baseline, M (SD) 55.5 (9.8) 56.8 (9.8) Time × group

F = 0.00, p = .98 18 mo, M (SD) 51.8 (10.4) 53.0 (10.5)

Note: GARS = Groningen Activity Restriction Scale; SF-36 = Short Form 36 Health Survey.

Table 5. Secondary Outcomes Over 18 months, Usual Care Group Is the Reference Group

Hospitalization, Mortality, Institutionalization

Intervention (N = 320)

Usual Care (N = 331) Risk 95% CI

Hospital admittance ≥1, n (%)

163 (50.9) 141 (42.6) 1.23† 0.91–1.68

Acute hospital visit ≥1, n (%)

128 (40) 101 (30.5) 1.18† 0.85–1.65

Time until Institutionalization, n (%)

23 (6.9) 20 (6.5) 1.04‡ 0.07–16.6

Time until mortality, n (%) 39 (11.8) 43 (13.4) 0.86‡ 0.56–1.34

† OR = odds ratio. ‡ HR = hazard ratio.

vention was negative. Second, we failed to offer assess- ments and home visits to a part (16%) of the intervention group. Third, the frequency of the home visits was modest and might not be sufficient enough to instigate preventive effects. Due to a regional reimbursement policy, only four preventive visits a year were allowed in our study. Yet, a recent meta-analysis could not confirm any effect of visit frequency (10).

Fourth, we remain uncertain about quality of the deliv- ered intervention as we did not audit this systematically. A more elaborate process evaluation alongside the trial would have improved the understanding of our results. Another limitation is it is possible that some PCPs were unblinded to some participant’s group assignment in case a nurse thought it necessary to demand medical action from the PCP. Never- theless, it is not very likely that this would have profoundly influenced the group contrast. Finally, loss of follow-up of the self-report measures was selective for unhealthier and older persons, which may have diluted any preventive effects on self-reported health outcomes.

Explaining Mechanisms Are home visits by nurses ineffective? The absence of a

preventive effect of the home visits might be attributed to several issues. First, the intervention was not integrated in the primary care practices. A better linkage might have en- forced the implementation of the nurses’ care plans. Al- though some studies with “stand alone” nurse interventions have shown beneficial effects (6), a recent review on disease management of complex health care needs suggest the importance of a more integrated or collaborative model (26). Second, the selection of the participants might have been inappropriate. Although we selected a vulnerable population, many functioned relatively well and improved over time. On the other hand, a recent meta-analysis sug- gested that preventive interventions show more effects in nonfrail elderly (27). In addition, the nurses offered a range of interventions, and the outcome measures might not have been sensitive enough to pick up specific effects. On the other hand, a recent meta-analysis advocated that preventive interventions show more effects in nonfrail elderly (10).

Third, the follow-up period of 18 months might have been too short to demonstrate long-term preventive effects. For example, in a study by Hendriksen and coworkers (28) with a follow-up of 3 years, the decrease in hospital admit- tance became obvious only in the second half of their study. Fourth, we underestimated the willingness of many older persons to engage in preventive activities and invested little in motivational skills of the nurses needed to pursue their clients.

A final explanation might be that preventive home visits by auxiliary nurses simply do not add effect in Dutch primary care with its easy access for all persons. In the Netherlands, everyone

PREVENTION OF ADVERSE HEALTH TRAJECTORIES 741

is medically insured and accessibility to medical care including home care services is high. In addition, the PCP or a nurse sees most persons with chronic conditions regularly. Everyone has a medical file kept by one PCP and medical histories of elderly persons are generally well known.

Increased Hospital Admissions in Intervention Subgroups The higher odds of hospital admittance and acute hospital

visits found in certain subgroups of the intervention group may be explained by a nurse-induced increased awareness of the participants concerning the treatability of their health sta- tus after the nurse assessment. The hospitalization rate was comparable with other studies among frail elderly (5,6).

Costs Indicators Hospital admissions are an important cost driver in eco-

nomic evaluations of outpatient studies, and these turned out to be slightly higher for the intervention group. If we add the costs for the nurses and implementation to the inter- vention group as well, the cost effectiveness is very likely to turn out negative.

Future Research To be able to target interventions to selected persons it

might be helpful to evaluate the prognostic value of easily measurable vulnerability indicators in primary care. In addi- tion, the search for effective interventions for vulnerable per- sons needs to be continued. For example, ingredients of the chronic disease management model may be adopted more rigorously, such as coordination and monitoring by nurses, fine-tuning between nurses and PCPs, consultation of old age specialists, motivating and empowering clients for self- management, and preventive activities, combined with better control on the implementation of these processes. Finally, better exploration of subjective care needs may guide interventions.

Conclusions We could not demonstrate any beneficial effects of low-

intensity home visiting by nurses compared with usual care in vulnerable frail older persons in primary care. The search for effective interventions for vulnerable persons requires further investigation. Future efforts may focus on improved integrated approaches.

Funding

This work was partly supported by the Netherlands Organization for Health Research and development (grant no 2200.114).

Acknowledgments

We like to thank all participating PCPs, nurses, patients, and especially Home Care Organization “Omring” who provided the nurses for their co- operation in this project. H.P.J.H., A.P.D.J., H.W.J.M., and M.P. had full access to the data in the study. H.P.J.H. takes responsibility for the integrity of the data and the accuracy of the data analysis.

Authors’ contributions: H.P.J.H. and G.N. designed the study. H.P.J.H. drafted the article and all authors contributed to the final concept.

References 1. Eekhof J, De Bock G, Schaapveld K, Springer M. Effects of screening

for disorders among the elderly:an intervention study in general prac- tice. Fam Pract. 2000;17:329–333.

2. Lagaay AM, van der Meij JC, Hijmans W. Validation of medical his- tory taking as part of a population based survey in subjects aged 85 and over. BMJ. 1992;304:1091–1092.

3. Tremellen J. Assessment of patients aged over 75 years in general practice. BMJ. 1993;305:621–624.

4. Iliffe S, Manthorpe J, Eden A. Sooner or later? Issues in the early di- agnosis of dementia in general practice: a qualitative study. Fam Pract. 2003;20:376–381.

5. Elkan R, Kendrick D, Dewey M, et al. Effectiveness of home based support for older people: systematic review and meta-analysis. BMJ. 2001;323:719–725.

6. Stuck AE, Egger M, Hammer A, Minder CE, Beck JC. Home visits to prevent nursing home admission and functional decline in elderly people, systematic review and meta-regression analysis. JAMA. 2002;287:1022–1028.

7. van Haastregt JC, Diederiks JP, van Rossum E, de Witte LP, Crebolder HF. Effects of preventive home visits to elderly people living in the community: systematic review. BMJ. 2000;320(7237): 754–758.

8. van Rossum E, Frederiks CM, Philipsen H, Portengen K, Wiskerke J, Knipschild P. Effects of preventive home visits to elderly people. BMJ. 1993;307(6895):27–32.

9. Bouman A, van Rossum E, Kempen GI, Knipschild P. Effects of home visits by home nurses to elderly people with health problems: a ran- domised clinical trial in the Netherlands [ISRCTN92017183]. J Am Geriatr Soc. 2008;56(3);397–404.

10. Beswick AD, Rees K, Dieppe P, et al. Complex interventions to im- prove physical function and maintain independent living in elderly people: a systematic review and meta-analysis. Lancet. 2008;371(9614): 725–735.

11. Nelson E, Wasson J, Kirk J, et al. Assessment of function in routine clinical practice: description of the COOP Chart method and prelimi- nary findings. J Chron Dis. 1987;40(suppl 1):55S–66S.

12. Van Weel C, Konig-Zahn C, Touw-Otten F, van Duijn N, Meyboom-de Jong B. Measuring Functional Health With the COOP/WONCA Charts, a Manual. Groningen, The Netherlands: Northern Centre for Health Care Research, University Groningen; 1995.

13. van Hout HP, Nijpels G, van Marwijk HW, et al. Design and pilot results of a single blind randomized controlled trial of systematic demand-led home visits by nurses to frail elderly persons in primary care [ISRCTN05358495]. BMC Geriatr. 2005;5:11.

14. Bentsen BG, Natvig B, Winnem M. Questions you didn’t ask? COOP/ WONCA Charts in clinical work and research. World Organization of Colleges, Academies and Academic Associations of General Practitio- ners/Family Physicists. Fam Pract. 1999;16(2):190–195.

15. Kinnersley P, Peters T, Stott N. Measuring functional health status in primary care using the COOP-WONCA charts: acceptability, range of scores, construct validity, reliability and sensitivity to change. Br J Gen Pract. 1994;44(389):545–549.

16. InterRai Overview Committee. Minimal Data Set, Resident Assessment Interview, Home Care version 2.0. 2005. Boston: Hebrew Senior Life.

17. Landi F, Tua E, Onder G, et al. Minimum data set for home care: a valid instrument to assess frail older people living in the community. Med Care. 2000;38:1184–1190.

18. Kind P. The EuroQol instrument: an index of health related quality of life. In: Spiker B, ed. Quality of Life and Pharmacoeconomics in Clinical Tri- als, 2 ed. Philadelphia, PA: Lippincott-Raven Publishers; 1996:191–201.

19. McHorney CA, Ware JE Jr, Raczek AE. The MOS 36-Item Short- Form Health Survey (SF-36): II. Psychometric and clinical tests of

VAN HOUT ET AL.742

validity in measuring physical and mental health constructs. Med Care. 1993;31(3):247–263.

20. Kempen GI, Miedema I, Ormel J, Molenaar W. The assessment of disability with the Groningen Activity Restriction Scale. Conceptual framework and psychometric properties. Soc Sci Med. 1996;43(11): 1601–1610.

21. Stuck AE, Egger M, Hammer A, Minder CE, Beck JC. Home visits to prevent nursing home admission and functional decline in elderly people. JAMA. 2002;287:1022–1028.

22. Cohen HJ, Feussner JR, Weinberger M, et al. A controlled trial of in- patient and outpatient geriatric evaluation and management. N Engl J Med. 2002;346(12):905–912.

23. Cohen J. Statistical Power Analysis for the Behavioural Sciences. New York: Academic Press; 1977.

24. Fitzmaurice GM, Laird NM, Ware JH. Applied Longitudinal Analysis. Hoboken, NJ: John Wiley & Sons; 2004.

25. van der Zee K, Sanderman R. Measuring the General Health Status with the RAND-36, a Manual. Groningen, The Netherlands: Noordelijk

Centrum voor Gezondheidsvraagstukken, University Groningen; 1990.

26. Bodenheimer T, Berry-Millett R. Care Management of Patients with Complex Health Care Needs. Princeton: Robert Wood Johnson Foun- dation; 2009. Report No.: 19.

27. Beswick AD, Rees K, West RR, et al. Improving uptake and adherence in cardiac rehabilitation: literature review. J Adv Nurs. 2005;49(5): 538–555.

28. Hendriksen C, Lund E, Stromgard E. Consequences of assessment and intervention among elderly people: a three year randomised controlled trial. Br Med J (Clin Res Ed). 1984;289(6457):1522–1524.

29. Jansen AP, van Hout HP, Nijpels G, van Marwijk HW, Gundy C, de Vet HC, Stalman WA. Self-reports on the IQCODE in older adults: a psychometric evaluation. J Geriatr Psychiatry Neurol. 2008;21(2):83–92.

30. Strawbridge WJ, Shema SJ, Balfour JL, Higby HR, Kaplan GA. Antecedents of frailty over three decades in an older cohort. J Gerontol B Psychol Sci Soc Sci. 1998;53(1):S9–16.

Copyright of Journals of Gerontology Series A: Biological Sciences & Medical Sciences is the property of Oxford University Press / UK and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.