Article Text

Download PDFPDF

Bottom-up implementation of disease-management programmes: results of a multisite comparison
  1. K M M Lemmens1,
  2. A P Nieboer1,
  3. M P M H Rutten-Van Mölken2,
  4. C P van Schayck3,
  5. C Spreeuwenberg3,
  6. J D Asin4,
  7. R Huijsman1
  1. 1Institute of Health Policy and Management, Erasmus University Rotterdam, Rotterdam, The Netherlands
  2. 2Institute for Medical Technology Assessment, Erasmus University Rotterdam, Rotterdam, The Netherlands
  3. 3Research Institute CAPHRI, Maastricht University, Maastricht, The Netherlands
  4. 4ACSION: Advance Care Solutions and Insights for Optimization, Amstelveen, The Netherlands
  1. Correspondence to Dr K M M Lemmens, Institute of Health Policy and Management, Erasmus University Rotterdam, PO Box 1738, 3000 DR Rotterdam, The Netherlands; lemmens{at}


Objectives To evaluate the implementation of three regional disease-management programmes on chronic obstructive pulmonary disease (COPD) based on bottlenecks experienced in professional practice.

Methods The authors performed a multisite comparison of three Dutch regional disease-management programmes combining patient-related, professional-directed and organisational interventions. Process (Assessing Chronic Illness Care survey) and outcome (disease specific quality of life (clinical COPD questionnaire (CCQ); chronic respiratory questionnaire (CRQ)), Medical Research Council dyspnoea and patients' experiences) data were collected for 370 COPD patients and their care providers.

Results Bottlenecks in region A were mostly related to patient involvement, in region B to organisational issues and in region C to both. Selected interventions related to identified bottlenecks were implemented in all programmes, except for patient-related interventions in programme A. Within programmes, significant improvements were found on dyspnoea and patients' experiences with practice nurses. Outcomes on quality of life differed between programmes: programme A did not show any significant improvements; programme B did show any significant improvements on CCQ total (p<0.001), functional (p=0.011) and symptom (p<0.001), CRQ fatigue (p<0.001) and emotional scales (p<0.001); in programme C, CCQ symptom (p<0.001) improved significantly, whereas CCQ mental score (p<0.001) deteriorated significantly. Regression analyses showed that programmes with better implementation of selected interventions resulted in relatively larger improvements in quality of life (CCQ).

Conclusions Bottom-up implementation of COPD disease-management programmes is a feasible approach, which in multiple settings leads to significant improvements in outcomes of care. Programmes with a better fit between implemented interventions and bottlenecks showed more positive changes in outcomes.

  • Chronic disease
  • patient education
  • clinical practice guidelines
  • primary care
  • quality of care

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Disease-management programmes have emerged as a new strategy to enhance quality of care for patients suffering from chronic conditions and to control healthcare costs. Integration of multiple interventions is essential to these programmes1; patient-related, professionally directed and organisational interventions have been found to be key components for success.2 3 Until recently, they were usually implemented through top-down approaches in which the organisation of current practices and processes is basically redesigned with a standard set of interventions, rather than having professionals select the interventions that fit best with the desired changes in their practices.4 However, the emphasis should be on guiding the natural properties and behaviours of sociotechnical systems rather than imposing hierarchical structures and above-down instructions from people who do not actually work at the ‘coal-face.’4 5 Therefore, implementation of change should be linked to professionals' needs and motivation based on problems experienced in practice.6 A bottom-up approach is thought to ensure that programmes are geared to current bottlenecks—factors that cause less-than-optimal chronic care delivery—and based on evidence, as well as on professional commitment. This bottom-up approach contrasts with previous models of innovation by taking the bottlenecks in professional practice as opportunities for improvement. It focuses on locally supported interventions and stimulates professional ownership to flexibly configure their practices. However, evaluating such complex multicomponent interventions is a challenge.7 This would seem to require more systematic use of theories in planning and evaluating the effects of disease-management programmes.6 We developed an evaluation model that allows one relations between the implemented interventions and specific outcomes, such as improvement in quality of life, to be determined.2

To investigate the usefulness of a bottom-up approach that combines the actions of professionals with those of scientific evaluation, we present our research on disease-management programmes in chronic obstructive pulmonary disease (COPD). COPD is a multicomponent disease with a high prevalence that imposes an enormous burden on the patient, medical professionals and society in terms of morbidity, mortality, healthcare resource utilisation and costs.8 Despite this burden and the availability of recommendations for better care, COPD remains a condition that is suboptimally managed.9 It thus presents opportunities for quality improvement.

The aim of this paper is to evaluate the bottom-up implementation of three regional COPD disease-management programmes in The Netherlands. The associations between existing bottlenecks, implementation of selected interventions and outcomes of care are examined.


We performed a multisite comparison of the effects of three different disease-management programmes. Existing bottlenecks were qualitatively evaluated. Quantitative data on processes and outcomes were collected before and after implementation of the programmes. This design has been found to be helpful to better understand relations between problems defined by professionals, implementation of the selected interventions and effects of disease management in multiple representative settings.10

Disease-management programmes

The disease-management programmes studied are part of a Dutch national programme—Partners in Care Solutions (PICASSO) for COPD—that aims to optimise COPD care by supporting regional COPD projects. It offers a bottom-up approach for inducing and sustaining organisational change. This is based on bottleneck analyses with regional stakeholders (such as general practitioners, pulmonologists and patient platforms), consensus on corresponding disease management interventions by local professionals, strategic planning and change management support, and structured evaluation through a theoretically derived framework for understanding how interventions affect outcomes of care.2

Bottleneck analysis

A two-stage bottleneck analysis was performed for each of the three regions, hereafter named A, B and C. First, project leaders invited all regional stakeholders to select potentially missing preconditions (78) for optimal COPD care and rank the 10 most important ones on an online portal (1 point for the least important and 10 points for the most important one). Thirty-one professionals participated in region A, 47 in region B and 30 in region C. Second, the professionals prioritised the obtained regional bottleneck top 10 on the grounds of three aspects: (1) complexity: how complex it is to solve this bottleneck; (2) speed: how long it takes to implement the solution; and (3) impact: what degree of impact on patients' health the solution is expected to have. This was done using a five-point scale, with higher scores indicating less complexity, faster implementation and higher impact. This prioritisation guided the selection by the local professionals of disease-management interventions to be incorporated in a strategic plan of action, ideally a combination of long-term goals and quick wins. Regions varied on complexity and themes, but hardly on speed and impact (table 1). Region C assessed bottlenecks as more complex.

Table 1

Existing bottlenecks

Setting and participants

Table 2 describes the three settings, the bottlenecks and the COPD disease-management interventions designed to tackle these bottlenecks. In each region, general practices cooperated with regional hospital(s) and were supported by their own practice nurses/physician assistants or practice nurses of a regional home care organisation. Regional settings determined inclusion of the stakeholders in the programmes, such as home care organisations. All programmes combined patient-related, professional-directed and organisational interventions, but the focus differed. Patients were systematically educated to increase their understanding of COPD and self-management skills. Professionals were educated on the guidelines and programme. These interventions were supported by organisational interventions, such as structural follow-up of COPD patients and revision of professional roles.

Table 2

Regional selection of disease management interventions in the programmes

Study patients were those registered with the participating general practices, aged 40 or over with a confirmed diagnosis of COPD according to Global Initiative for Chronic Obstructive Lung Disease (GOLD) criteria, which are primarily based on spirometry.11 Patients were recruited between April 2006 and December 2007. Informed consent was obtained. Patients with serious comorbidity, such as congestive heart failure, were excluded by the general practitioner according to protocol.

Data collection and measures

Process data were collected from (electronic) practice registrations after the introduction of the disease-management programme. The process measures were expressed as percentages of patients within practices who actually received a specific intervention—for example, attending follow-up meetings. Furthermore, participating professionals completed the Assessing Chronic Illness Care (ACIC) survey12 at baseline to identify areas for improvement and again at the end of the programme to assess process improvement. Process elements that reflected the programmes' interventions were included: self-management support (patient-related intervention), decision support (professional-directed intervention), as well as delivery system design and clinical information system (both organisational interventions).

The patients completed a set of validated outcome questionnaires twice—before the intervention and 12 months after the intervention. Quality of life was measured by disease-specific questionnaires (CCQ13 and CRQ14). Dyspnoea was measured by the Medical Research Council (MRC) Dyspnoea scale.15 Patients' experiences with quality of care were assessed by means of a questionnaire, the Quality of Care Through the Patients' Eyes COPD.16

Statistical analysis

We used paired-sampled t tests and Wilcoxon signed rank tests (two-sided; α=0.05), where appropriate, to analyse within-programme changes between baseline and 12 months postintervention. We used multiple regression analyses for the between-programme comparisons with health outcomes (quality of life (CCQ) and MRC dyspnoea) as dependent variables. At the first step, we entered dummy variables for programmes B and C in the overall analysis, and programme A was selected as a reference category. A value 1 on the dummy indicates participation in the particular programme, whereas a value 0 indicates participation in one of the other programmes. For example, a value 1 on the Programme B dummy indicates participation in Programme B. At the second step, process changes were added to these programme dummy variables. Finally, age, sex, smoking status and lung function (forced expiratory volume in 1 s percentage predicted) were added to control for patient characteristics. Data processing and analysis were performed using SPSS 15 for Windows. A prior significance level of 0.05 was used for all statistical tests.


Patient inclusion, assignment and follow-up

Seventy-three per cent (189/259) of the eligible patients in region A participated in the disease-management programme; 73% in region B (137/187); 68% in region C (121/178). Table 3 shows baseline characteristics of those participants for whom data collection was fully completed, 150 (79%), 112 (82%) and 108 (89%), respectively. There were no significant differences in baseline characteristics and quality of life measures between these participants and the ‘drop-outs.’ Most of the latter had been found unwilling to complete questionnaires. Significant differences between regions were found for sex, smoking status and COPD severity.

Table 3

Baseline characteristics

From bottlenecks to results

Table 4 summarises the findings for processes (ACIC) and clinical outcomes, quality of life variables and patients' experiences, comparing the baseline measurement with the 12-month measurement.

Table 4

Changes in processes and outcomes of care for chronic obstructive pulmonary disease patients

Most bottlenecks in region A were related to (organising) patient involvement (table 2), and they were also ranked highest. The core of the programme was an Information and Communications Technology-supported diagnostic and treatment protocol, including patient education. Process evaluation showed that all patients had been diagnosed by spirometry and were classified according to the GOLD criteria (n=150). Smoking status had been registered for all. Most of the patients were prescribed airway medication (92%). Inhalation instructions, when applicable, were given to 92% of the patients. However, only 56% received written information material. Ninety-two per cent of patients regularly attended follow-up meetings. All ACIC subscores increased in programme A, but differences for decision support and self-management support did not reach statistical significance. With regard to programme outcomes, a significant decline in lung function was found; dyspnoea improved significantly. Quality-of-life scores did not significantly improve. Overall, patients' experiences with practice nurses improved.

Region B mostly identified and highly ranked organisational problems related to multidisciplinary collaboration (table 2). This programme focused on developing a multidisciplinary diagnostic and treatment protocol, including patient education and smoking cessation referral. All patients (n=112) had been diagnosed by spirometry and were classified by GOLD criteria. Smoking status had been registered for all. Information material was provided to 68%, and 90% received inhalation instruction. Most of the patients regularly attended follow-up meetings (91%). Scores for all ACIC items improved significantly after intervention. As to programme outcomes, lung function did not change, but dyspnoea improved significantly. Statistically significant improvements in CCQ total, functional and symptom, CRQ fatigue and CRQ emotional scales were found. The change in CCQ symptom score was also clinically relevant. Changes in patients' experiences did not reach statistical significance.

Bottlenecks identified for region C related to both patient involvement and organisational problems. Referral for smoking cessation and education by practice nurses were essential parts of the regional diagnostic and treatment protocol. Furthermore, patients were referred to a physiotherapist for reactivation or advice depending on patients' needs. All patients (n=108) had been diagnosed by spirometry and were classified by GOLD criteria. Smoking status had been registered for all. Airway medication was prescribed to 82% of the patients; 91% received inhalation instruction. Eighty-two per cent received information material. Almost all patients regularly attended follow-up meetings (98%). As in programme B, scores on all ACIC items improved significantly after intervention. In this programme, no changes in lung function were found, whereas dyspnoea improved significantly. Statistically significant and clinically relevant improvements were found in CCQ symptom score as well. However, the CCQ mental score deteriorated significantly. Patients' experiences with practice nurses improved significantly.

Regression results for programme outcomes

Table 5 shows the results of the regression analyses for changes in quality of life (CCQ) and dyspnoea. Programme A served as a reference, since it showed the smallest changes in quality of life. The analyses showed a more positive effect of programme B on changes in CCQ total and CCQ functional state as compared with programme A. Both programmes B and C had relatively larger improvements in CCQ symptom change. However, a decrease in CCQ mental state was found in programme C. Changes in MRC dyspnoea were associated with sex and lung function, in so far as being female and having better lung function at baseline are related to improvements in dyspnoea. All programmes resulted in significantly improved MRC dyspnoea; there were no differences in this respect between programmes (table 4). Process measures such as structural follow-up meetings and perceived decision support based on the ACIC instrument correlated significantly with changes in quality of life (data not shown); these correlations disappeared in the regression analyses, however, due to the inclusion of programmes as independent variables.

Table 5

Regression analyses for changes in quality of life and dyspnoea


Principal findings

In this study, bottom-up implementation of COPD disease-management programmes led to statistically significant improvements in various quality-of-life dimensions, dyspnoea and patients' experiences with practice nurses. In each region, interventions were geared to perceived bottlenecks; quantitative analyses suggested that programmes with better implementation of selected interventions were associated with more positive changes in outcomes. Consequently, this study supports the notion that bottlenecks in professional practice can well be used as opportunities for the implementation of disease-management programmes. A recent paper confirms this finding: a bottom-up strategy led by professionals is badly needed to balance the predominantly top-down approaches which frequently result in only modest improvements which are difficult to sustain.5 Although patient baseline characteristics between programmes differed, regression analyses showed that these differences did not explain differences in quality-of-life changes between programmes. Dyspnoea improved in all three programmes, but CCQ symptoms improved only in programmes B and C. This might be explained by better implementation of patient-related interventions focusing on the improvement of self-management. However, analyses also showed deterioration in CCQ mental state in programme C; this might be due to low baseline scores (indicating a better quality of life) which reduced possibilities of detecting improvement. Programme B performed best, which might be explained by well-implemented interventions in relation to the perceived bottleneck.

This study showed that a multifaceted perspective on disease management is indeed effective: all programmes combined patient-oriented, professional-directed and organisational interventions. Previous studies focused on single components or on programmes that did not include all three components. Meta-analyses of previous study results already suggested that multifaceted COPD programmes lead to improvements in quality of life and hospital admissions.17 18 The present study was the first to test the premise of triple intervention programmes within multiple real-life settings. It is not yet possible to specify what aspects of disease-management interventions are most beneficial to COPD patients. Nevertheless, findings from this study show that combining patient-related, professional-directed and organisational interventions is recommendable.

Disease-management interventions selected by the regional professionals mostly concerned rather traditional interventions, such as provision of education instead of motivational interviewing, a directive, patient-centred counselling style for eliciting behaviour change by helping patients to explore and resolve ambivalence.19 Compared with non-directive counselling, it is more focused and goal-directed. The examination and resolution of ambivalence are its central purpose, and the counsellor is intentionally directive in pursuing this goal. This might be explained by the fact that diffusion of innovation models do not provide a framework for assessing the conditions in which such interventions become practically workable in healthcare settings.20 What is more, increasing sense of ownership of professionals does not necessarily bring about more creativity. On the other hand, autonomy to refine innovations and improve their fitness for purpose will accelerate adoption.21

Strengths and limitations

This study combined the best available evidence and appropriate theory with a bottom-up approach suitable for widespread implementation in real-life settings. It fills a gap identified by the MRC for more research into complex interventions—starting with pilot studies targeted at each of the key uncertainties in the design, and moving on to an exploratory evaluation (and then a definitive evaluation).22 Ideally, a design in which an assessment of existing bottlenecks is performed in some sites, while a prespecified programme implemented in other sites without input from professionals would be most suitable. This would enable a test of the implementation approach itself (top-down vs bottom-up); however, this was not feasible in the current study. However, our bottom-up approach did demonstrate that well-implemented interventions geared to perceived bottlenecks are associated with changes in outcomes of care. By measuring processes and multiple outcomes, selected on theoretical grounds, we were able to assess the success of disease-management programmes across a range of domains—based on professionals' and patients' self-reports as well as process data from (electronic) practice registrations—and to provide insights into why programmes will work out well or will have unexpected consequences. For example, outcomes in programme A improved the least, probably because of insufficient implementation of patient-related interventions in relation to the perceived bottleneck with regard to patient involvement. Data on patients who did not participate in the study were not readily available; representativeness therefore cannot be assured. Such data were available for programme C, however, and no significant differences were found in baseline data between those participating and those who are not.

The purpose of this study was to evaluate bottom-up implementation of disease management. Regional bottlenecks are likely to differ. Consequently, regions will differ as to selected interventions. Individual randomisation was not feasible in the current study.7 A quasiexperimental design in multiple settings was therefore applied. Since disease management essentially is applied in real-life settings, this design is an adequate method for understanding the effectiveness of interventions in different clinical settings and in broader patient populations.23 Nevertheless, we fully acknowledge that results have to be interpreted with caution. Also, because internal validity may have been hampered, since professionals were involved in the intervention themselves, which might have influenced the perceived process improvements.

Indications for further research

It is not yet clear what specific characteristics of disease-management interventions (such as intensity, duration or components) are most beneficial to COPD patients. Previous research in other chronic diseases found that effects appeared to be somewhat stronger for delivery system design and self-management support.24 Adams et al25 deduced from a review that self-management interventions were more powerful. Future research should therefore focus on multifaceted interventions that include more robust self-management interventions.

From our study, we learnt that disease management should be geared to perceived bottlenecks in a region. Knowledge that underpins the adoption, dissemination and implementation of complex interventions within healthcare settings is either not objective or lacking.21 It is recommended to connect the natural properties of the sociotechnical system that comprises healthcare, and guide the existing practices which come naturally to professionals, so that these can be developed and used to promote continuous quality improvement. Engaging professionals by creating ownership of a problem and involving them in the selection of (complex) solutions (ie, innovation) should be further explored as a factor in the associations between innovation, dissemination, diffusion and adoption.


We would like to thank the staff of practices and organisations who participated in the study for assistance with data acquisition and insights.



  • Funding This work was supported by an unrestricted grant from PICASSO for COPD, an initiative of Pfizer BV and Boehringer Ingelheim BV in cooperation with research institute Caphri (Care and Public Health Research Institute) of Maastricht University. One disease-management programme was also supported by Menzis care and income insurer.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.