Effectiveness of case management interventions in reducing common and potentially preventable complications associated with Parkinson’s disease:A systematic review and meta-analysis

Objective We sought to systematically examine the effectiveness of case management interventions on common and potentially preventable complications associated with Parkinson’s disease, both in persons with Parkinson’s disease and in persons with other chronic health conditions. We specically focused on falls, depression, anxiety, hallucinations, urinary tract infections and swallowing impairments. There is no systematic insight in the effect of case management on common complications associated with Parkinson’s disease. This is an important knowledge gap given that people living with Parkinson's have identied care coordination as one of their highest priorities. Furthermore, it remains unclear whether the putative benecial effects of case management would vary by key patient characteristics, such as their age, gender or disease characteristics. Such insights would contribute to a shift from “one size ts all” healthcare resource allocation to personalized medicine.


Introduction
Parkinson's disease (PD) is a progressive, neurodegenerative disorder affecting more than 6 million people worldwide [1,2]. It is characterized by a combination of motor and non-motor symptoms. Given its complex nature, the treatment of PD requires continuous monitoring and ongoing interdisciplinary collaboration between health care providers of various disciplines, who ideally deliver proactive care [3]. However, current health care systems are typically designed to treat chronic diseases such as PD using a "one size ts all" approach instead of tailoring care to each patient's individual needs [4,5]. Consequently, persons with PD often become responsible for coordinating their own care. As a result, patients receive care that is fragmented and ineffective in achieving the desired health outcomes [3,5]. This not only increases health care costs, but also causes an unnecessary burden on patients and their carers which, in turn, negatively affects their quality of life.
To address these challenges, case management has been introduced as an approach to improve care coordination [6,7]. Case management has been de ned in different ways [6,8], however, the common basis for each de nition is that case management is a collaborative process involving one case manager or a small team that plans, coordinates and reviews the delivery of health care services to meet a patient's individual needs [9]. According to this integrated care approach, a case manager takes over the responsibility to manage the non-acute services. The interpretation of case management can vary substantially, but common core elements include the development and review of individualized care plans, organization of multidisciplinary case meetings, screening and monitoring of risk factors and symptoms, use of evidence-based guidelines, information support for involved physicians, empowerment of patients through providing education, and enhancing self-management skills [9][10][11].
There is emerging evidence for the promising effect of case management interventions on reducing hospital (re-)admissions and length of stay in patients with chronic illnesses other than PD, such as asthma [12,13], diabetes [14,15], chronic heart failure [15,16] or chronic obstructive pulmonary diseases [17,18]. However, the effect of case management on common complications associated with PD remains unknown. This is an important knowledge gap given that people living with PD have identi ed care coordination as one of their highest priorities. Furthermore, it remains unclear whether the putative bene cial effects of case management on common complications would vary by key patient characteristics, such as demographics (i.e., age or gender), disease severity or duration, or key characteristics of the intervention (e.g., the number of individual patient contact). Such insights would facilitate a wider deployment of case management to susceptible subgroups of patients, thereby contributing to a broader shift from "one size ts all"-based healthcare resource allocation to "personalized" medicine.
To close these gaps in knowledge, we conducted a systematic review and meta-analysis to examine: 1) the extent and quality of evidence for the effectiveness of case management interventions on common and potentially preventable complications associated with PD; and 2) to what extent putative effects of case management vary across patient subgroups.

Methods
This systematic review is guided according to the PRISMA checklist (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) [19].

Search focus
We focused our search on the following common and potentially preventable complications associated with PD [20,21] : (1) mood disorders, including depressive symptoms and anxiety; (2) fractures or injuries caused by falls; (3) swallowing impairment; (4) urinary tract infections; and (5) neuro-psychiatric disorders, including hallucinations. Although cognitive decline is also a well-known common complication associated with PD, we did not expect any direct causal effect of case management interventions on the level of cognitive impairment and thus excluded this complication from this systematic review.
Our initial search focused exclusively on data on case management interventions in people with PD. That search yielded very few results and, therefore, we expanded our search strategy by including data on case management interventions in people with other chronic health conditions with one or more of the same ve common and potentially preventable complications, in whom we hypothesized case management interventions would have similar effects as in people with PD. Speci cally, we broadened the search by including data from people diagnosed with Alzheimer Disease, asthma, cancer, chronic obstructive pulmonary disease (COPD), chronic heart failure, dementia, diabetes, hypertension, multiple sclerosis or rheumatoid arthritis.

Search strategy
An initial and limited search for empirical literature was undertaken by one reviewer [ADvH] in PubMed to identify important Medical Subject Headings (MeSH) and key words describing relevant articles. Next, a systematic search for research-based literature was performed by two independent reviewers [ADvH, JMJD] using the identi ed MeSH terms and keywords, by which we exclusively focused on published articles referenced in PubMed and Embase online in July 2019. The following MeSH terms were used to identify studies on case management: "Case Management", "Disease Management", "Patient Care Management", "Patient Care Planning" and "Patient-Centered Care" which were combined with relevant key words. The detailed search strategy for PubMed can be found in Supplemental Data I. Finally, the reference lists of included studies were screened to identify studies missed by the search. A veri cation search was performed in February 2021.

Selection criteria
Abstracts and titles of all obtained studies were independently, systematically examined for the selection criteria by two reviewers [ADvH, JMJD] and disagreements were resolved during consensus meetings with a third reviewer [SKLD]. Search limits were applied to include only articles in English and those articles published in a peer-review journal. We included in this systematic review studies that (1) used an observational (prospective and retrospective) or interventional study design; (2) included results on the association of a case management intervention with at least one of the ve common and potentially preventable complications; (3) included populations who were diagnosed with one of the selected chronic diseases [aged ≥ 18 years]; and (4) described the case management intervention clearly and contained at least three core elements; (5) de ned a clear control group, usually receiving usual care; and (6) reported at the least measures of the distribution of age and gender for the intervention and control group, as potential confounders (Table 1). We excluded studies involving participants living in residential nursing homes, as those participants receive 24-hour care and were thus not comparable to other populations receiving case management. Also excluded were studies that described only study protocols or conference abstracts.

Data extraction and analysis
The quality of the articles was evaluated by two authors (JMJD and ADvH) using the Cochrane risk of bias tool for randomized controlled trials (RCTs) [22] and the ROBINS-I tool for nonrandomized studies [23]. Two review authors [ADvH, JMJD] independently screened and evaluated the studies. The following data were extracted from each study: author, publication year, trial design, country, aim of the study, study design, distribution of participant characteristics (sex, age and diagnosis of disease), characteristics of case management intervention (table 1), characteristics of care received by the control group, follow-up duration, study outcomes, and primary outcome results (mean, con dence intervals, standard deviation, standard error, interquartile range). Where possible, standard deviations (SD) were calculated from standard errors (SE). In case no SD or SE was reported, authors were contacted and asked to provide additional data. For ve articles we did not receive additional data [24][25][26][27][28], so these articles could not be included in the meta-analysis and/or random-effects regression analysis. We calculated standardized effect sizes for the main results of each study. Heterogeneity was assessed using the I 2 statistics. We used funnel plot visualization and Egger's test for funnel plot asymmetry to assess whether there was evidence for small study effects.
We conducted meta-analyses using restricted maximum likelihood (REML) and DerSimonian-Laird estimator in a random-effects model, because of signi cant heterogeneity between estimates across studies. In addition, prede ned random-effects regression analyses were performed to identify effect modi ers of the effects of case management interventions, if su cient studies were available within a complication category (e.g. depressive symptoms, anxiety). In fact, two studies were excluded for this analysis on depression, as the necessary data for regression analysis was not available [29,30]. We included the following study population characteristics as potential effect modi ers: mean age, percentage of female participants, disease group (neurodegenerative vs internal medicine diseases). We also included the number of case management components (out of nine components, which outlined in Table 1) as a potential effect modi er. Due to restrictions in the number of potential effect modi ers we were not able to include individual components in the random-effects regression analyses for feelings of anxiety. Instead, we included the number of included case management components as potential effect modi er. Multicollinearity of covariates -the modifying effect of covariates on each other -was assessed using variance in ating factors (VIF).We considered associations with p<0.05 to be statistically signi cant across meta-analyses and meta-regression analyses. Analyses were conducted in R [31], using packages metafor [32] and ggplot2 for visualization [33].

Study selection
The combined PubMed and Embase searches yielded 4765 unique records. After abstract review, 57 fulltext articles were assessed for eligibility, of which 23 ful lled our selection criteria. The other 34 were excluded for the following reasons: 25 studies did not provide relevant data on the topic under study; four studies were not identi ed as case management interventions; two studies dealt with a different health population; two studies did not include original data; and one study presented the same cohort. Two further studies [29,34] were identi ed through cross-reference checking and two study [27,35] was added through veri cation search, bringing the total to 27 included studies. Figure 1 provides an overview of the search and study selection process.

Risk of bias and quality assessment
Of the four non-randomized intervention studies, three were rated as carrying a high risk of bias and one as a moderate risk. Of the RCTs, only eight studies [24,29,34,35,37,38,40,49] were judged to be of a good methodological quality and at low risk of carrying bias. The remaining 15 studies were of low or moderate quality with a high or unknown risk of bias among several domains. However, most RCTs were rated as having a high or unknown risk on the domains of blinding of participants and blinding of outcome assessment, which is less applicable to this kind of intervention. Details on the quality assessment are presented in Supplemental Data II.

Components of case management intervention
Across the 27 included studies, there was substantial heterogeneity across the content and duration of the various case management interventions. Table 3 displays the different strategies used in each study. Common components of case management among the 27 studies (i.e., applied in at least … studies) were (1) regular telephone contacts combined with in-person visits; (2) monitoring of signs, symptoms and risk factors; (3) ensuring therapy adherence; and (4) providing educational support on disease management and treatment, or trainings on self-management skills. However, the content and structure of these components varied highly among these studies. For instance, the purpose of in-person home visits ranged from developing and discussing a therapeutic plan with the patient in one study [41], to monitoring changes in signs and symptoms as well as reviewing patient's safety in their own home environment in another study [37]. The location of in-person visits also varied from the patient's home to a clinical setting. Most studies were conducted through a combination of in-person and telephone contacts, with only seven studies [25,27,38,42,45,46,52] reporting an intervention that was exclusively conducted through telephone contact.
Providing assistance for social and nancial support, organization of multidisciplinary case meetings, medication review and the development of individualized care plans were more commonly reported case management interventions. In only four studies, case management interventions also included the support of informal carers [25,26,35,36]. And of these, only one study included a regular assessment of the carer's physical health and provision of education on carer's coping skills, in addition to educating carers on disease management [36].
Furthermore, several studies incorporated the use of technological support systems which not only supported the implementation of case management strategies, but also offered new possibilities. For instance, a web-based service facilitated communication between the patient and the care team, to schedule patient contacts and to keep track of progress and current disease treatment [36]. In a different study, a web-based collaborative intervention facilitated peer-to-peer support for patients with cancer through a chat room connecting all enrolled participants to each other [30].
Of the 27 studies, 23 [25-28, 30, 34-45, 48-53] reported that case management interventions were delivered by a nurse case manager or a team consisting of a nurse case manager and other health care specialists. In one study [29] the case manager was a depression clinical specialist without further clari cation of the background of the specialist and in another one [24] two research coordinators ful lled the role as care managers. In the other two studies the background of the case manager was not speci ed any further [46,47]. Length of follow-up ranged from one and a half month to 24 months, with 18 studies reporting a 12-month or even longer follow-up period ( Table 2).

Overview of outcome measurements
Results of the narrative data synthesis are summarized and presented in Table 2. Nearly all included studies evaluated the effectiveness of case management interventions on depressive symptoms, whereas anxiety was addressed in only eight studies [25,28,41,43,[46][47][48]51]. None of the included studies reported on falls, urinary tract infections, swallowing impairment or hallucinations.

Effect of case management on anxiety
Eight studies (n = 1239 participants) reported outcomes on symptoms of anxiety [25,28,41,43,[46][47][48]51]. The most commonly used scale was the Hospital Anxiety and Depression Scale (HADS) [25,28,41,43,46,48], followed by 7-item Generalized Anxiety Disorder [47], and State Trait Anxiety Index (STAI) [51]. Six studies reported su cient data and were included in a random effects meta-analysis, the results of which revealed a signi cant effect of case management interventions in decreasing anxiety
Twenty studies reported su cient data and were included in the meta-analysis (Figure 3). Two studies [40,50] reported two different depression outcomes (CES-D/SCL-20 PHQ-9), but for analytical reasons only one (PHQ-9) was included in the meta-analysis. A random-effects meta-analyses revealed a signi cant effect of case management intervention on depression (SMD= -0.48; CI: -0.71, -0.25), but heterogenity was high (I 2 = 92.3%). A funnel plot and Egger's test of funnel plot asymmetry (p = 0.57) showed no evidence for publication bias.

Discussion
This systematic review and meta-analysis shows that case management is more effective than usual care at reducing depressive symptoms and anxiety, two common and potentially preventable complications associated with PD. This effect persisted for less complex case management interventions (whereby complexity was based on the number of included elements), across countries, for different chronic diseases and given the nearly equal representation of female and male participants, also across both genders. I to our hypotheses, found no evidence for effect modi cation by case management intervention, study population characteristics, or duration of follow-up. Of note, we did not identify any studies reporting on falls, hallucinations, swallowing impairment and urinary tract infections, so the effect of case management on those complications remains unclear. Also, only one of the included interventions was conducted in people with PD [37], which is a crucial remaining gap that should be addressed by future studies. The randomized study reported that a PD nurse-led care management intervention among veterans led to better adherence to quality-of-care indicators and the screening instrument showed signi cant improvement among the intervention group compared to usual care [37].
In this trial, the PD nursed used four strategies to specify PD problems of each veteran and to develop an action plan: (1) a telephone-administered assessment to identify 28 problem areas; (2) evidence-based care protocols and if not available, use of expert consensus on care management; (3) patient portal for communication purposes; and (4) documentation templates to provide care that is patient-centered and coordinated. However, this study is limited by geographical and time factors. Our study extends the ndings of previous studies indicating a bene cial effect of case management on patients' clinical health outcomes and functioning in everyday life [7,54,55]. While previous research on case management has focused on its effects on reducing hospital (re-)admissions, length of stay and costs, little research has been done regarding its potential to reduce complications. Our systematic review addresses this gap in knowledge. The ndings of this systematic review favor case management interventions over usual care and suggest that even less complex case management interventions have a bene cial effect on nonmotor features that can also be encountered in PD. Case management reduces feelings of anxiety and depressive symptoms, which are both complications that can have an immense impact on the quality of life in patients with PD [56, 57]. Several indirect working mechanisms may have contributed to the bene cial effect on depressive symptoms and anxiety (Figure 4), which can both have an immense impact on the quality of life and even mortality in patients with PD [56-58]. First, patients across studies received personalized management of their chronic health condition, including individually tailored health information and problem-solving strategies provided by a case manager. Second, the availability of one main contact person for newly arising issues and the ensuing establishment of a personal relationship between patients and case managers may have further helped to reduce feelings of anxiety and depressive symptoms. Notably, previous research on improving PD care revealed that having a single point of access was rated as the top priority by people with PD [4]. Third, having little information about the rate of disease progression and treatment options is known to enhance feelings of anxiety in patients.
Patient education eventually allows for more shared decision-making and thereby treatment that is better tailored to a patient's individual needs and their coping behavior which, in turn, might alleviate anxiety. Several methodological considerations need to be considered. First, heterogeneity across studies was high and effect estimates varied substantially between studies, which might have affected the results of our meta-regression, leading to inability to identify any covariates of our effect estimates. Furthermore, since only two studies [39,48] followed patients for longer than two years, the long-term effectiveness of case management interventions remains largely unclear. Second, only a few of the included studies were of good methodological quality. In particular, limited blinding of outcome assessment, insu cient details on speci cation of e cacy of speci c case management elements and the lack of participant selection limited our ability to more accurately assess case management impact. In addition, as a systematic search strategy cannot screen full-text articles, the risk remains that relevant articles are not captured by the used search strategy. Eggers et al. [59], for instance, conducted a RCT with a nurse case-manager led intervention with patients with PD and reported on the effectiveness of CM on reducing hospitalization caused by falls. As these ndings have not been reported in the abstract of the study, this study was not captured through our search strategy. Our data showed that bene cial effects of case management interventions on anxiety were somewhat more distinct in studies with a higher percentage of men and in studies with a relatively old population, although both observations were not statistically signi cant. No similar trends were observed for the effects of case management interventions on depressive symptoms. This systematic review identi ed the need for further research into the most effective case management interventions and components, on the optimal intensity and frequency of the individual case management strategies, and on their interaction with patient characteristics. Once this knowledge becomes available, case management interventions can be implemented that are better tailored to individual needs and as such presumably more effective. Moreover, this systematic review revealed that case management implementation is more common among certain chronic diseases than in others; the commonest ones were diabetes and heart failure, while only one included study concerned patients with Forest plot on the effect of case management interventions on feelings of anxiety (random-effects model) Figure 3 Forest plot on the effect of case management interventions on depressive symptoms (random-effects model)