Evaluation of a primary care nurse case management intervention for chronically ill community dwelling older people


  • Cheryl Schraeder PhD, RN, FAAN,

  • Cynthia W Fraser MA,

  • Ida Clark MSN, RN, APN/FNP,

  • Barbara Long RN,

  • Paul Shelton EdD,

  • Valerie Waldschmidt BS,

  • Christine L Kucera BFAID,

  • William K Lanker MD

Cheryl Schraeder
Director of Policy & Practice Initiatives Institute for Healthcare Innovation UIC College of Nursing Urbana IL
Telephone: +1 217 586 4164
E-mail: cheryls@uic.edu


Aim.  The purpose of this study was to test the effectiveness of a collaborative primary care nurse case management intervention emphasising collaboration between physicians, nurses and patients, risk identification, comprehensive assessment, collaborative planning, health monitoring, patient education and transitional care on healthcare utilisation and cost for community dwelling chronically ill older persons.

Background.  Primary care teams comprised of nurses and primary care physicians have been suggested as a model for providing quality care to the chronically ill, but this type of intervention has not been systematically evaluated.

Design.  A non-randomised, 36 month comparison of two geographically distinct primary care populations was conducted.

Methods.  Six hundred and seventy-seven persons aged 65 and older were determined to be at high-risk for mortality, functional decline, or increased health service use. The treatment group (= 400) received the intervention and the comparison group (= 277) received usual care. Health plan claims files provided data on number of hospitalisations and bed days, emergency department (ED) visits, physician visits and total cost of care.

Results.  After adjustment for baseline variables, there were no significant differences between the treatment and comparison group in the percentage of patients hospitalised or ED visits. However, among those hospitalised in the treatment group, the likelihood of being re-hospitalised was significantly reduced by 34% (= 0·032). After adjusting for the cost of the intervention, although not statistically significant, the reduced hospital use resulted in cost savings of $106 per patient per month in the treatment group.

Conclusions.  The results indicate that a collaborative primary care nurse case management intervention has the potential to be an effective alternative to current primary care delivery system practice.

Relevance to clinical practice.  The study suggests that a chronic care intervention emphasising collaboration between physicians, nurses and patients, may be more effective when implemented in integrated provider networks.


By the year 2030, it is estimated that over 71 million adults in the United States (US) will be 65 years of age or older (He et al. 2005), and account for approximately 20% of the nation’s population. It is also projected that 80% will have at least one chronic condition, and 50% will have two or more (Centers for Disease Control 2007). Chronic illness is the leading cause of mortality, disability, diminished quality of life and health care resource use for this age group (CDC & Merck Foundation 2007). Because 98% of older people live in the community (National Center for Health Statistics 2006), healthcare systems should have an increased interest in community-based programmes to address the needs of older people with chronic conditions.

Chronic care is complex and challenging. Older adults with chronic conditions are at increased risk for poor outcomes and preventable hospital readmissions. Exacerbations and remissions of symptoms punctuate the course of chronic disease. They repeatedly experience the fragmentation of medical and community-based services, finding the current healthcare system poorly suited to the special needs associated with chronicity (Leveille et al. 1998, Lesser & Ginsburg 2000).

The management and care of individuals with chronic conditions comprise the single largest cost to the U.S. health care system for a disproportionate number of people. Despite a tremendous investment of resources, the health care industry has been reluctant to develop and implement new systems to improve the delivery of clinical care (McGlynn et al. 2003), especially for individuals with chronic conditions (Reuben et al. 1999). Although the medical community concentrates on treating specific chronic illnesses, an emerging perspective is challenging the current health care status to a different way of providing care. Chronic disease, rather than a specific category, is viewed as a condition that requires a different framework of care, linking the pre and postacute continuum of care with the community. Within this framework, persons with chronic illnesses are viewed as active participants in their health care and have specific responsibilities for self-management (Wagner et al. 1996, 1999).

The Institute of Medicine (2001) emphasised that the integration of care delivery, evidence-based medicine, and chronic care management could play a significant role in improving care quality and reducing medical costs for individuals with chronic illnesses. A number of alternative models of chronic care management have emerged in the ambulatory setting. The primary goals of these models are to control health care costs, improve clinical outcomes and enhance patient and provider satisfaction. Care co-ordination typically includes a combination of patient screening and assessment, careplan development, and implementation of recommendations through actions such as health education, referrals/facilitation of community resources, and treatment adherence monitoring. These activities are generally provided as a complement to ongoing usual care rather than modifying the basic structure of primary care (Reuben 2002).

These models are generally grouped under three different, but similar, care co-ordination strategies: collaborative, multidisciplinary care teams consisting of primary care physicians (PCP), nurses, social workers and other non-physician personnel; case management, performed primarily by nurses; and disease management (Wolff & Boult 2005, Case Management Society of America 2007). Multidisciplinary teams have improved select outcomes for patients with stroke, heart failure, mental illness and terminal conditions through reduced hospitalisations, improved mortality rates and symptom management (McDonald et al. 2007). Disease management programmes have shown promise in improving depression symptoms, glycemic control for patients with diabetes, and mortality and hospital readmission rates for patients with heart failure (McDonald et al. 2007). However, their relevance to older adults with more than one chronic condition is questioned (Boyd et al. 2005).

There are current examples of community-based nurse case management programmes for chronically ill older persons that have improved patient outcomes and demonstrated cost savings. ‘The Frail Elderly Community-Based Case Management Project’ showed positive outcomes for enrollees, who averaged 12 comorbidities each, including significant reductions in ED visits, hospital admissions, total bed days and costs along with ‘an improved perception on quality of life’ (Duke 2005). Additionally, nursing case management studies have demonstrated cost-effectiveness during the post-hospitalisation period (Naylor et al. 2004) when older persons are at an increased risk for poor outcomes and preventable readmissions (Thornton et al. 2002).

Although studies that have reported cost results have not substantiated their ability to significantly impact overall costs of care (Congressional Budget Office 2004, Ofman et al. 2004, Krause 2005, Wise et al. 2006), positive health outcomes from case management studies are not in short supply. A randomised trial of a community-based nurse case management intervention for older women with breast cancer documents benefits for treatment group members ranging from emotional and educational support to managing co-existing chronic conditions (e.g. hypertension and diabetes) (Jennings-Sanders & Anderson 2003). Medicare + Choice members enrolled in a community-based nursing case management programme demonstrated improvements in obtaining preventive health measures (e.g. pneumonia vaccine, mammograms, pap smears, colonoscopies and annual lipid profiles), and self-management of chronic diseases (e.g. regular weight monitoring, articulation of necessary survival skills, annual foot and eye exams) (Schifalacqua et al. 2000). Other studies of nursing case management programmes reveal benefits to patients including lowered HbA1c and cholesterol levels among diabetes patients (Taylor et al. 2003), decreased functional decline (Gitlin et al. 2006), and improved access to appropriate treatment (Goodwin et al. 2003) and resources that would otherwise go unused (Egan et al. 2002).


We developed a model of collaborative primary care nurse case management that uses registered nurses as case managers operating within a multi-specialty physician group practice and a vertically integrated health care system (Schraeder & Shelton 1997). The model reduced the risk of hospitalisation among caregivers of individuals with Alzheimer’s disease (Shelton et al. 2001), and reduced all-cause mortality within another predominately rural, older adult population, but not hospitalisation (Schraeder et al. 2001). The purpose of this study is to report the effects of a similar intervention implemented in a primary care practice in central Illinois. Our hypotheses were that the intervention would reduce service utilisation, lower total costs of care and prove revenue neutral.


Study sites

The treatment site consisted of eight counties in east central Illinois, a primarily rural geographic area. The managed care plan contracted with one multi-specialty, physician group practice and nine smaller, independent physician groups for primary care, and two urban and five rural hospitals to provide care to enrolled patients. The comparison site (usual care) consisted of 13 counties in west central Illinois, with a more diverse population base. The managed care plan contracted with 49 small, PCP practices, and three urban and six rural hospitals to provide care to enrolled patients. All treatment and comparison group patients had the same covered health benefits regardless of where they lived. The study was exempt from ethics committee approval because all study data was being collected as part of usual care for both the intervention and usual care groups by the health plan.

Study sample

All individuals who joined the health plan during the first year of operation were mailed a 50-item health questionnaire (= 4053). After telephone contact with non-responders, a total of 3562 (treatment site = 2102; comparison site =1460) questionnaires were returned, for an overall response rate of 88%.

The study sample consisted of those individuals who: (i) voluntarily completed the health questionnaire and were identified at high-risk for mortality, functional decline or health service use based on the answers to 24 selected screening questions, a process which has been described elsewhere (Schraeder et al. 1997, Shelton et al. 2000); (ii) were aged 65 and older; and (iii) did not reside in a nursing home when they joined the managed care plan. This resulted in a study sample of 677 patients; 400 in the treatment group and 277 in the comparison group. This ‘high risk’ patient group accounted for 19% of the overall assessed patient population who joined the health plan during the first year of operation.


The intervention is based on the premise that individuals with chronic illness face long-term challenges of disease symptom management, wavering illness trajectories and multiple treatment regimens. While the aetiologies of specific chronic diseases are different, their illness components, disease course and clinical and self-management strategies have many similarities (Corbin & Strauss 1991). The intervention emphasises that effective chronic care should be organised, coordinated and delivered by multi-disciplinary primary care teams (Bodenheimer et al. 2002). This approach requires team members to have new skills, including the ability to collaborate, share patient care responsibilities and participate in focused efforts to improve care (O’Connor et al. 1998, Bodenheimer et al. 1999, Larson 2001).

The 36-month intervention emphasised collaboration between physicians, nurses and patients, risk identification, comprehensive assessment, collaborative planning, health monitoring, patient education and transitional care among chronically ill older persons living in the community. The specific details of the intervention have been published previously (Schraeder et al. 1997, 1999, Shelton et al. 1998), and are summarised below.

The intervention included the addition of a registered nurse to the primary care practice of intervention group physicians. The team’s goal was to enhance existing primary care by providing patient assessments conducted in the home or office, flexible home or office visits, detailed care planning, and coordination and procurement of supportive services. The intervention also included routine telephone monitoring to identify changes in condition and adherence to treatment regimes, proactive post illness follow-up, and disease education. Team communication occurred via telephone, voice mail, written notes and summaries, careplan letters to patients and families, informal office conversations and formal meetings in which patient treatment plans were discussed. The longitudinal nature of the intervention was an important element in the study design as it was felt that intervention benefits would accrue over time.

The intervention was implemented in the following steps:

  • 1An education series was developed for PCP, nurses and administrators that consisted of the following topics: the goals and responsibilities of the collaborative care teams, clinical decision making with older adult patients, and measuring and reporting major study outcomes. This series was co-presented to providers at their practice location by a physician leader and nurse administrator of the participating group practice.
  • 2A nurse case manager (NCM) and case assistant (CA) were assigned and located in each of eight primary care clinics and worked with approximately five PCP and their ‘high risk’ health plan patients.
  • 3The NCM completed an initial home visit and assessment with all identified high risk patients when they joined the managed care plan, and then developed a plan of care with their PCP. Each patient was contacted or visited at least monthly by the NCM to review and update the plan of care, monitor their health status, provide education on managing their health, and arrange any necessary health-related services. The team completed a comprehensive review of each patient at least every 12 months or more frequently if there was a change in health status.
  • 4If a patient was hospitalised, the NCM contacted them at least weekly for the first month following hospital discharge to address changes in their plan of care, and monitor additional health care needs.
  • 5The NCM provided ongoing patient education on available community resources, medication management, disease processes and exacerbation of symptoms that signified a change in health condition. Patient educational materials and counselling were provided for specific health problems and conditions.
  • 6The NCM made regular calls to patients to proactively monitor their health and health management techniques. Calls were also made after visits to their primary and specialty physicians, and following emergency department (ED) visits. In addition, the NCM received e-mail alerts if a patient had a physician office visit, ED visit or hospital admission, which served as a trigger to contact the patient, review the situation and the plan of care.
  • 7Monthly reports were distributed to the collaborative care teams (PCP/NCM). The reports included individual patients and their characteristics (i.e. health status, medications, health conditions, utilisation) and overall characteristics of the patient panel of each PCP and clinical site (Schraeder et al. 2000).

Cost of the intervention

The per patient per month (PPPM) cost of the intervention was $54 during the study period. The average cost of the intervention per patient was $1491. This cost included all salary and benefits for the NCM (5·3 FTEs) and CA (3·7 FTEs) plus administrative and overhead expenses.

Data collection and outcome measures

Self-report data were collected by mailed health questionnaires, or when necessary, by trained telephone interviewers if individuals were unable to complete the health questionnaire. Data included demographics, health and functional status and prior health service utilisation (Schraeder et al. 1997). Demographics included age, gender, race, education and living arrangement. Health status assessment included the number of prescription medications taken on a daily basis, and the presence of chronic conditions (cancer, chronic obstructed pulmonary disease, depression, diabetes, heart disease, incontinence, myocardial infarction and stroke). Functional status included limitations in activities of daily living (ADL) and instrumental activities of daily living (IADL), and the use of durable medical equipment. Health service utilisation in the prior 6 months was comprised of the number of primary care and specialist physician visits, number of hospital admissions, number of ED visits and any visits from a home health nurse. Enrolment, provider, utilisation and cost data were obtained from the managed care plan’s administrative claims system. Enrolment data included the dates of enrolment and disenrolment, and reason for disenrolment. Provider information included the name, office location and type (e.g. adult medicine, family practice, specialist) of the PCP. Utilisation data included the number of hospitalisations and bed days, the number of PCP, physician specialist and mid-level provider visits (nurse practitioner and physician assistant), ED visits, and billed charges. Total billed charges were then converted into the Medicare allowable costs for each service provided. The cost of the intervention was added to that total amount for each intervention patient, based on the number of months they were enrolled in the study, for a total cost of care amount.

All utilisation and cost data were collected for each patient from their date of enrolment to the end of the study period. There were significant differences in the number of enrolled months and disenrolment rates between the two groups. Treatment group patients were enrolled approximately 3 months longer in the study than comparison group patients [27·6 months (SD 7·9) vs. 24·8 (SD 8·6), < 0·001]. Voluntary disenrolment was 8% for the treatment group and 22% for the comparison group (< 0·001).

Primary study outcomes were health service resource use and cost of care (total allowed Medicare costs plus the cost of the intervention) during the study period. Health resource outcomes included hospitalisation rates, especially multiple hospitalisations, because older persons with chronic disease are especially vulnerable to rehospitalisation (Ottenbacher et al. 2000, Naylor et al. 2004); and ED visit rates without a subsequent hospitalisation, because comorbid older persons are more likely to be hospitalised when they use emergency services (Aminzadeh & Dalziel 2002, McCusker & Verdon 2006). Total cost of care was calculated on a PPPM basis. Secondary study outcomes included the total number of hospitalisations, hospital bed days and ED visits.

Statistical analysis

Study evaluation occurred at enrolment (baseline) and annually at 12, 24 and 36 months. Comparisons of baseline measures between the treatment and comparison group were made using the Student t-test or Wilcoxon rank sum test for continuous variables and the chi-squared test for categorical variables. A two-part model was used to test intervention effects (Diehr et al. 1999). The first part of the model used logistic regression to estimate the intervention effect on the likelihood of any hospital or ED use. The second part of the model used linear regression to estimate intervention effects on the amount of service use (hospitalisations, hospital bed days and ED visits) among service users only, and PPPM cost of care. Significant baseline variables (< 0·10) were used to adjust for population differences, as well as the total number of months patients were enrolled in the study. In addition, a propensity score, determined by logistic regression to estimate the probability of assignment to the treatment group, was calculated for each study participant using all the baseline variables in Table 1 (Rubin 1997, Sinn & McAlister 2002), and included in the regression models. The use of propensity scores has been suggested as an appropriate analytical strategy when randomisation is not possible or feasible, and also when there are significant baseline differences between experimental and comparison groups (Linden et al. 2005).

Table 1.   Baseline characteristics of the study population
CharacteristicsTreatment group (= 400)Comparison group (= 277)p-value
  1. *Heart disease = congestive heart failure and/or angina.

  2. ADL = activities of daily living include limitations in bathing, dressing, toileting, transfers in/out of chairs or bed, eating and walking.

  3. IADL = instrumental activities of daily living include limitations in shopping, using the telephone, transportation, taking medications and managing personal finances.

  4. §Possible cognitive impairment answered ‘A Lot’ to the question: During the past month, how much difficulty have you had remembering things?

  5. DME, any use of the following: quad cane, walker, wheelchair, hospital bed, ramp or oxygen equipment. COPD, chronic obstructed pulmonary disease; MI, myocardial infarction; ED, emergency department.

Age, mean ± SD, years75·4 ± 7·176·4 ± 7·90.067
Gender, female (%)53·460·30.084
Minority race (%)7·323·1<0.001
<High school education (%)26·334·70.021
Lives alone (%)26·833·90.049
Health conditions (%)
 Heart disease*42·341·90.937
 Incontinence, urine14·313·70.911
2 or more ADL limitations (%)23·123·30.965
2 or more IADL limitations (%)16·817·30.836
Any falls in past month (%)10·013·40.179
Possible cognitive impairment (%)§5·85·81·000
Taking ≥5 prescription medications daily (%)45·343·30.638
Health status fair/poor (%)44·446·90.610
Using any DME (%)21·322·40.776
Complete physical in past 12 months (%)67·253·4<0.001
Health care utilisation in previous 6 months (%)
 No primary care physician visit (%)13·515·20.576
 Any specialist physician visit (%)53·361·00.049
 Any hospital admission (%)24·026·40.527
 Any ED visit (%)29·128·20.863
 Any home health visits (%)12·314·40.420

As the cost of care data was highly skewed, it was transformed prior to analysis. To make the data distribution more normal, the dollar value was transformed to its logarithm form and the regression analysis performed to calculate p-values and compare adjusted means on the log scale (Schwartz & Ash 1997, Diehr et al. 1999). To derive mean costs on the original scale, the adjusted log means were then transformed back into the dollar scale using a smearing estimate (Duan 1983). All statistical analyses were conducted by the intention-to-treat principle using spss statistical software, version 14.0 (SPSS Inc, Chicago, IL, USA). A p-value of ≤ 0·05 (2-sided) was considered statistically significant.


Patient characteristics

Table 1 displays the baseline characteristics of the patients enrolled in the treatment (= 400) and comparison group (= 277). There were significant baseline demographic differences between the two groups, reflecting differences in the populations residing in the two distinct geographic regions. The comparison group had a higher percentage of patients who were of non-white race, had less education and lived alone compared to the treatment group. Baseline health conditions and self-report service utilisation data were very similar between the two groups, except that the comparison group had a higher proportion of patients with cancer, specialist physician visits, and a lower proportion who had a yearly physical in the previous year before enrolment.

Health resource use and cost of care: unadjusted outcomes

The unadjusted study outcomes are shown in Table 2. During the study period, a little over half the patients in both the treatment and comparison groups were hospitalised at least once. More patients in the comparison group had two or more hospitalisations compared with treatment group patients (= 0·006). For those patients who were hospitalised at least once during the study period, treatment group patients had significantly fewer hospitalisations (= 0·001) and hospital bed days (= 0·002) per person than the usual care group. ED visit rates without any hospitalisation were similar for both groups, as were ED visits per person for those who had at least one ED visit. The treatment group had significant mean PPPM costs of care that were $485 less than the comparison group (< 0·001) during the study period.

Table 2.   Unadjusted results of study outcomes
OutcomeTreatment group (= 400)Comparison group (= 277)p-value
  1. SD, standard deviation; PPPM, per patient per month; ED, emergency department.

Any hospital admission %51·053·80·352
2 or more hospital admissions %19·228·80·006
Mean hospitalisations for service users only ± SD1·76 ± 1·272·30 ± 1·830·001
Mean hospital bed days for service users only ± SD8·19 ± 10·1513·89 ± 16·540·002
Any ED visit without hospitalisation %16·812·10·086
Mean ED visits for service users only ± SD1·48 ± 0·871·79 ± 1·200·135
Mean cost of care PPPM, ± SD$1193 ± $1953$708 ± $1028<0·001

Health resource use and cost of care: adjusted outcomes

Table 3 shows the results of the multivariate regression analyses. There were no significant differences between the treatment and comparison groups after adjusting for baseline differences in the likelihood of any hospital admission, or any ED visit without hospitalisation during the study period. There were significant differences between the two groups in the proportion of patients who had multiple hospitalisations. The intervention reduced the likelihood of multiple hospitalisations by 34% (odds ratio = 0·66, 95% confidence interval = 0·45–0·97; p = 0·032) compared with the comparison group. There were significant differences in the total number of hospitalisations for those patients who were hospitalised. Treatment group patients were hospitalised an average of 0·5 fewer times compared to the comparison group (= 0·002) for an average of 5·3 fewer hospital days (= 0·001). The treatment group had an adjusted mean cost of care difference of $106 PPPM less than the comparison group during the study period, but this difference was not statistically significant.

Table 3.   Adjusted results of study outcomes
OutcomeOdds ratio95% CIp-valueB95% CIp-value
  1. CI, confidence interval; PPPM, per patient per month; ED, emergency department. Adjusted for the following baseline covariates: age, gender, minority race, less than high school education, lives alone, cancer, not having had an annual physical, any specialist MD visits, number of months enrolled in the study and propensity score.

Any hospital admission0·930·67–1·300·683   
2 or more hospital admissions0·660·45–0·970·032   
Mean hospitalisations for service users only   −0·54−0·89–0·200·002
Mean hospital bed days for service users only   −5·25−8·23–2·270·001
Any ED visit without hospitalisation1·390·87–2·250·173   
Mean ED Visits for service users only   −0·32−0·81–0·180·205
Cost of care PPPM   −$106−$138–$750·253


This study evaluated the impact of a collaborative primary care nurse case management intervention on the health resource use of a sample of chronically ill community dwelling older persons enrolled in a managed care plan in central Illinois. After adjusting for baseline differences, the most significant findings were reductions in repeat hospitalisation rates, and reductions in total number of hospitalisations and bed days in the treatment group. While the percentage of persons hospitalised was similar for each study group, the treatment group was significantly less likely to be re-hospitalised, thus accounting for the observed differences. Although not statistically significant, the adjusted PPPM cost of care difference between the treatment and comparison group was $106, which included the cost of the intervention. This cost of care difference can be considered budget neutral, and although not statistically significant, was more than enough to offset the cost of the intervention.

Our results are similar to other research findings that tested comparable nurse-physician collaborative care management interventions. Duke (2005) reported significant reductions in hospital admissions and length of stay in a sample of chronically ill older adults. Schifalacqua et al. (2000) found a significant decrease in hospital days per thousand, 30 day readmission rates and length of stay with high-risk members of a Medicare managed care plan. Fisher & McCabe (2005) achieved a 15% reduction in hospital days per thousand, and an 11% decline in hospital admissions in a sample of Medicaid patients. Doughty et al. (2001) also reported decreased hospital readmission rates and bed days for patients with congestive heart failure.

There were many aspects of the intervention that may have contributed to these outcomes. Collaboration between physicians and NCM improved the health management of the patients because the partnership allowed for fuller assessment of a patient’s overall health, development of more comprehensive careplans with specific interventions, and follow-up via telephone, office visit or in the home. Patient education was focused on problem areas identified by the patient and provided in increments that promoted understanding and behavioural change. Active intervention around key transition periods provided an opportunity to adjust patient careplans to meet current needs and helped reduce exacerbations in health conditions.

Collaboration was developed between the participating physicians and their NCM by clearly identifying roles, scope of practice and accepted communication venues. Positive professional relationships were strengthened as the NCM demonstrated clinical competence, availability for consultation, and accountability for following through with planned arrangements. Timely and efficient communication about the patient’s current health status was an essential component of this collaborative relationship and was facilitated by NCM participation in patient clinic visits with the physician, brief updates via email, impromptu conversations between visits and scheduled team conferences.

As the collaborative relationship between the physicians and the NCM progressed, communications regarding patient issues were increasingly initiated by both parties. The NCM conducted frequent monitoring and ongoing assessment of the patient. This allowed for proactive identification of problems or changes in patient’s health status that might affect the trajectory of care. Follow-up recommendations were then given to the physician, prompting timely and effective interventions. The physician would also contact the NCM following a patient visit to discuss concerns, request specific follow-up action, or recommend specialised education for the patient. The collaborative partnership gave the NCM opportunities to share in-depth patient information, assist in solving and avoiding potential exacerbations, and provide information about available community services with which a physician or his staff may not be familiar. As a result of the collaboration, patients received more complete medical care than they would have had otherwise.

Upon program enrolment, patients participated in a face-to-face, comprehensive health assessment with their NCM that gave them the opportunity to share their individual strengths, weaknesses and health care goals. Information gained from this initial assessment was integrated with the PCP’s medical assessment, and a mutually agreed upon comprehensive healthcare assessment was formed. This thorough view of a patient’s individual needs provided the foundation for an extensive plan of care and promoted a strong alliance between PCP, NCM and patient.

The initial interaction between the patient and their NCM fostered the growth of a personal relationship that often afforded NCM a more complete view of the older person’s complex needs and home situation than the physician would have. Time constraints during physician office visits, the patient’s comfort level with the physician, and the tendency for physician visits to focus on a patient’s acute rather than comprehensive needs, contribute to this phenomenon. For example, one female patient had confided symptoms of depression to her NCM; however, during a routine visit with her physician she was unable to tell him about her concerns. Since physicians and NCM communicated regularly, situations like this could be easily addressed, and worked to further strengthen the collaborative care team.

The collaborative care planning process involved patient and family, the NCM and the PCP in discussion of healthcare problems, evaluation of health care needs and creation of a careplan. Changes in status (hospitalisation, health or new diagnoses), family and support systems, availability of community services or patient preference triggered careplan revisions. NCM continuously monitored patients health status and living situation which allowed for timely updates to the physician and more expedient adjustments in care planning. The NCM would also discuss questions the patient had that they were uncomfortable asking their physician so all insights were shared by the team. Typically NCM also provided a brief synthesis of new issues, lab results, and clinical findings and then offered suggestions for care changes to the physician. This was appreciated by the physicians as it enabled efficient and well-informed careplan revisions with minimal demand on their time.

The NCM provided patients with specific information regarding health education and promotion in small, relevant segments. Changes in health status, frequent disease exacerbations, inability to perform ADL or IADL, symptoms of depression, or a new diagnosis of a chronic condition were events that prompted the need for educational interventions. Instructional materials were customised to accommodate each patient’s level of understanding and ability. The patient was central to planning how to use the new information and in deciding how to monitor their personal progress. During follow-up discussions, the NCM assessed the patient’s comprehension as well as their ability to use the information to improve their healthcare management.

During transitional periods of care surrounding, a hospitalisation or nursing home placement, the NCM visited the patient and family during admission and remained involved throughout the transition. Once discharged, the NCM conducted a home visit with the patient and family where medications were scrutinised to ensure accurate follow-up and disease symptoms were assessed to verify stability of medical conditions. Afterward, the NCM maintained ongoing contact with the patient and family to address issues that transpired. The NCM and physician communicated when necessary during this period until the issues were resolved and the patient’s health was stable. This process of ongoing monitoring and assessment during care transitions decreased the likelihood of disease exacerbations and may have prevented repeat hospitalisation and avoidable physician office visits.

Overall, patient information was consolidated, which gave a broad picture of a patient’s health status over time. As a result, prioritising treatment and educational needs for patients became less complicated and collaborative decision-making processes and communication between team members became more effective. Proactive efforts to identify and manage disease exacerbations and monitor transitions in care were key components to the success of the intervention. This approach to chronic care management including collaborative care planning, ongoing monitoring and support for the patient and family to assure adherence to care recommendations was well received by the patients, their families, and the physicians and may have contributed to reduced hospitalisations and bed days. An added benefit was that patients and families felt more in control of the healthcare process.


This study had several limitations. The study was a treatment-comparison group design which may not have been able to control for system differences that might have influenced study outcomes. A propensity score was calculated for each study participant to statistically account for the non-randomised design. Also, we repeated the same analytical strategy with the patients who were not screened ‘high risk’ (treatment site = 1701; comparison site = 1183), and therefore not part of the treatment or comparison study populations. We found no statistically significant differences in their utilisation or cost patterns (data not shown, available from the authors) during the study period, lending support to our conclusions that the observed treatment effects were due to the intervention and not to project design.

Another limitation was that the analyses were limited to only those variables collected from patient self-report. The screening questionnaire included those variables that have been shown to be significant predictors for increased utilisation and cost among older persons. Access to actual patient utilisation prior to program enrolment, had it been available, might have influenced patient classification into risk categories.

Finally, the results of this study may not be generalisable to other patient populations residing in different geographical areas. The patients enrolled in this study were from both urban and rural areas of central Illinois. Patients residing in predominately urban areas and in different health care markets may have access to different types of medical services and facilities and respond differently to changes in primary care approaches.


Currently, healthcare systems and providers are experiencing a great deal of instability and are reluctant to implement new systems to improve the delivery of clinical care. This study suggests that collaborative primary care nurse case management interventions can positively impact service use and cost of care for selected older persons with chronic conditions. The results indicate that perhaps this type of collaborative care intervention may be an effective alternative to current primary care delivery system practice. The data also suggest that new models of chronic disease care, which emphasise collaboration between nurses and physicians working in teams, may be more effective when implemented in multi-specialty physician group practices and vertically integrated provider networks than traditional primary care practices (Rundall et al. 2002). These health care systems can create and maintain the clinical, data and care management resources necessary to improve disease management in their older patient populations.


Financial support for this study was provided by Grant No. 97111-G from the John A. Hartford Foundation, New York, New York.


Study design: CS, CWF, PS, IC; data collection and analysis: collection CLK, CS, CWF, VW; analysis PS and manuscript preparation: CS, CWF, IC, VW, PS, CLK, BL, WKL.