Background: There is increasing attention on machine learning (ML)-based clinical decision support systems (CDSS), but their added value and pitfalls are very rarely evaluated in clinical practice.... Show moreBackground: There is increasing attention on machine learning (ML)-based clinical decision support systems (CDSS), but their added value and pitfalls are very rarely evaluated in clinical practice. We implemented a CDSS to aid general practitioners (GPs) in treating patients with urinary tract infections (UTIs), which are a significant health burden worldwide. Objective: This study aims to prospectively assess the impact of this CDSS on treatment success and change in antibiotic prescription behavior of the physician. In doing so, we hope to identify drivers and obstacles that positively impact the quality of health care practice with ML. Methods: The CDSS was developed by Pacmed, Nivel, and Leiden University Medical Center (LUMC). The CDSS presents the expected outcomes of treatments, using interpretable decision trees as ML classifiers. Treatment success was defined as a subsequent period of 28 days during which no new antibiotic treatment for UTI was needed. In this prospective observational study, 36 primary care practices used the software for 4 months. Furthermore, 29 control practices were identified using propensity score-matching. All analyses were performed using electronic health records from the Nivel Primary Care Database. Patients for whom the software was used were identified in the Nivel database by sequential matching using CDSS use data. We compared the proportion of successful treatments before and during the study within the treatment arm. The same analysis was performed for the control practices and the patient subgroup the software was definitely used for. All analyses, including that of physicians' prescription behavior, were statistically tested using 2-sided z tests with an alpha level of .05. Results: In the treatment practices, 4998 observations were included before and 3422 observations (of 2423 unique patients) were included during the implementation period. In the control practices, 5044 observations were included before and 3360 observations were included during the implementation period. The proportion of successful treatments increased significantly from 75% to 80% in treatment practices (z=5.47, P<.001). No significant difference was detected in control practices (76% before and 76% during the pilot, z=0.02; P=.98). Of the 2423 patients, we identified 734 (30.29%) in the CDSS use database in the Nivel database. For these patients, the proportion of successful treatments during the study was 83%-a statistically significant difference, with 75% of successful treatments before the study in the treatment practices (z=4.95; P<.001). Conclusions: The introduction of the CDSS as an intervention in the 36 treatment practices was associated with a statistically significant improvement in treatment success. We excluded temporal effects and validated the results with the subgroup analysis in patients for whom we were certain that the software was used. This study shows important strengths and points of attention for the development and implementation of an ML-based CDSS in clinical practice. Trial Registration: ClinicalTrials.gov NCT04408976; https://clinicaltrials.gov/ct2/show/NCT04408976 Show less
Background: There is increasing attention on machine learning (ML)-based clinical decision support systems (CDSS), but their added value and pitfalls are very rarely evaluated in clinical practice.... Show moreBackground: There is increasing attention on machine learning (ML)-based clinical decision support systems (CDSS), but their added value and pitfalls are very rarely evaluated in clinical practice. We implemented a CDSS to aid general practitioners (GPs) in treating patients with urinary tract infections (UTIs), which are a significant health burden worldwide.Objective: This study aims to prospectively assess the impact of this CDSS on treatment success and change in antibiotic prescription behavior of the physician. In doing so, we hope to identify drivers and obstacles that positively impact the quality of health care practice with ML.Methods: The CDSS was developed by Pacmed, Nivel, and Leiden University Medical Center (LUMC). The CDSS presents the expected outcomes of treatments, using interpretable decision trees as ML classifiers. Treatment success was defined as a subsequent period of 28 days during which no new antibiotic treatment for UTI was needed. In this prospective observational study, 36 primary care practices used the software for 4 months. Furthermore, 29 control practices were identified using propensity score-matching. All analyses were performed using electronic health records from the Nivel Primary Care Database. Patients for whom the software was used were identified in the Nivel database by sequential matching using CDSS use data. We compared the proportion of successful treatments before and during the study within the treatment arm. The same analysis was performed for the control practices and the patient subgroup the software was definitely used for. All analyses, including that of physicians’ prescription behavior, were statistically tested using 2-sided z tests with an α level of .05.Results: In the treatment practices, 4998 observations were included before and 3422 observations (of 2423 unique patients) were included during the implementation period. In the control practices, 5044 observations were included before and 3360 observations were included during the implementation period. The proportion of successful treatments increased significantly from 75% to 80% in treatment practices (z=5.47, P<.001). No significant difference was detected in control practices (76% before and 76% during the pilot, z=0.02; P=.98). Of the 2423 patients, we identified 734 (30.29%) in the CDSS use database in the Nivel database. For these patients, the proportion of successful treatments during the study was 83%—a statistically significant difference, with 75% of successful treatments before the study in the treatment practices (z=4.95; P<.001).Conclusions: The introduction of the CDSS as an intervention in the 36 treatment practices was associated with a statistically significant improvement in treatment success. We excluded temporal effects and validated the results with the subgroup analysis in patients for whom we were certain that the software was used. This study shows important strengths and points of attention for the development and implementation of an ML-based CDSS in clinical practice. Show less
Background: There is increasing attention on machine learning (ML)-based clinical decision support systems (CDSS), but their added value and pitfalls are very rarely evaluated in clinical practice.... Show moreBackground: There is increasing attention on machine learning (ML)-based clinical decision support systems (CDSS), but their added value and pitfalls are very rarely evaluated in clinical practice. We implemented a CDSS to aid general practitioners (GPs) in treating patients with urinary tract infections (UTIs), which are a significant health burden worldwide.Objective: This study aims to prospectively assess the impact of this CDSS on treatment success and change in antibiotic prescription behavior of the physician. In doing so, we hope to identify drivers and obstacles that positively impact the quality of health care practice with ML.Methods: The CDSS was developed by Pacmed, Nivel, and Leiden University Medical Center (LUMC). The CDSS presents the expected outcomes of treatments, using interpretable decision trees as ML classifiers. Treatment success was defined as a subsequent period of 28 days during which no new antibiotic treatment for UTI was needed. In this prospective observational study, 36 primary care practices used the software for 4 months. Furthermore, 29 control practices were identified using propensity score-matching. All analyses were performed using electronic health records from the Nivel Primary Care Database. Patients for whom the software was used were identified in the Nivel database by sequential matching using CDSS use data. We compared the proportion of successful treatments before and during the study within the treatment arm. The same analysis was performed for the control practices and the patient subgroup the software was definitely used for. All analyses, including that of physicians’ prescription behavior, were statistically tested using 2-sided z tests with an α level of .05.Results: In the treatment practices, 4998 observations were included before and 3422 observations (of 2423 unique patients) were included during the implementation period. In the control practices, 5044 observations were included before and 3360 observations were included during the implementation period. The proportion of successful treatments increased significantly from 75% to 80% in treatment practices (z=5.47, P<.001). No significant difference was detected in control practices (76% before and 76% during the pilot, z=0.02; P=.98). Of the 2423 patients, we identified 734 (30.29%) in the CDSS use database in the Nivel database. For these patients, the proportion of successful treatments during the study was 83%—a statistically significant difference, with 75% of successful treatments before the study in the treatment practices (z=4.95; P<.001).Conclusions: The introduction of the CDSS as an intervention in the 36 treatment practices was associated with a statistically significant improvement in treatment success. We excluded temporal effects and validated the results with the subgroup analysis in patients for whom we were certain that the software was used. This study shows important strengths and points of attention for the development and implementation of an ML-based CDSS in clinical practice. Show less
BackgroundStructured primary diabetes care within a collectively supported setting is associated with better monitoring of biomedical and lifestyle-related target indicators among people with type... Show moreBackgroundStructured primary diabetes care within a collectively supported setting is associated with better monitoring of biomedical and lifestyle-related target indicators among people with type 2 diabetes and with better HbA1c levels. Whether socioeconomic status affects delivery of care in terms of monitoring and its association with HbA1c levels within this approach, is unclear. This study aims to understand whether, within a structured care approach, 1) socioeconomic categories differ concerning diabetes monitoring as recommended; 2) socioeconomic status modifies the association between monitoring as recommended and HbA1c.MethodsObservational real-life cohort study with primary care registry data from general practitioners within diverse socioeconomic areas, who are supported with implementation of structured diabetes care. People with type 2 diabetes mellitus were offered quarterly diabetes consultations. 'Monitoring as recommended' by professional guidelines implied minimally one annual registration of HbA1c, systolic blood pressure, LDL, BMI, smoking behaviour and physical activity. Regarding socioeconomic status, deprived, advantageous urban and advantageous suburban categories were compared to the intermediate category concerning 1) recommended monitoring; 2) association between recommended monitoring and HbA1c.ResultsAim 1 (n=13,601 people): Compared to the intermediate socioeconomic category, no significant differences in odds of being monitored as recommended were found in the deprived (OR 0.45 (95%CI 0.19-1.08)), advantageous-urban (OR 1.27 (95%CI 0.46-3.54)) and advantageous- suburban (OR 2.32 (95%CI 0.88-6.08)) categories. Aim 2 (n=11,164 people): People with recommended monitoring had significantly lower HbA1c levels than incompletely-monitored people (-2.4 (95%CI -2.9;-1.8)mmol/mol). SES modified monitoring-related HbA1c differences, which were significantly higher in the deprived (-3.3 (95%CI -4.3;-2.4)mmol/mol) than the intermediate category (-1.3 (95%CI -2.2;-0.4)mmol/mol). Conclusions Within a structured diabetes care setting, socioeconomic status is not associated with recommended monitoring. Socioeconomic differences in the association between recommended monitoring and HbA1c levels advocate further exploration of practice and patient-related factors contributing to appropriate monitoring and for care adjustment to population needs. Show less
Background Dutch standard diabetes care is generally protocol-driven. However, considering that general practices wish to tailor diabetes care to individual patients and encourage self-management,... Show moreBackground Dutch standard diabetes care is generally protocol-driven. However, considering that general practices wish to tailor diabetes care to individual patients and encourage self-management, particularly in light of current COVID-19 related constraints, protocols and other barriers may hinder implementation. The impact of dispensing with protocol and implementation of self-management interventions on patient monitoring and experiences are not known. This study aims to evaluate tailoring of care by understanding experiences of well-organised practices 1) when dispensing with protocol; 2) determining the key conditions for successful implementation of self-management interventions; and furthermore exploring patients' experiences regarding dispensing with protocol and self-management interventions. Methods in this mixed-methods prospective study, practices (n = 49) were invited to participate if they met protocol-related quality targets, and their adult patients with well-controlled type 2 diabetes were invited if they had received protocol-based diabetes care for a minimum of 1 year. For practices, study participation consisted of the opportunity to deliver protocol-free diabetes care, with selection and implementation of self-management interventions. For patients, study participation provided exposure to protocol-free diabetes care and self-management interventions. Qualitative outcomes (practices: 5 focus groups, 2 individual interviews) included experiences of dispensing with protocol and the implementation process of self-management interventions, operationalised as implementation fidelity. Quantitative outcomes (patients: routine registry data, surveys) consisted of diabetes monitoring completeness, satisfaction, wellbeing and health status at baseline and follow-up (24 months). Results Qualitative: In participating practices ( = 4), dispensing with protocol encouraged reflection on tailored care and selection of various self-management interventions nA focus on patient preferences, team collaboration and intervention feasibility was associated with high implementation fidelity Quantitative: In patients ( = 126), likelihood of complete monitoring decreased significantly after two years (OR 0.2 (95% CI 0.1-0.5), < 0.001) npSatisfaction decreased slightly (- 1.6 (95% CI -2.6;-0.6), = 0.001) pNon-significant declines were found in wellbeing (- 1.3 (95% CI -5.4; 2.9), p = 0.55) and health status (- 3.0 (95% CI -7.1; 1.2), p = 0.16). Conclusions To tailor diabetes care to individual patients within well-organised practices, we recommend dispensing with protocol while maintaining one structural annual monitoring consultation, combined with the well-supported implementation of feasible self-management interventions. Interventions should be selected and delivered with the involvement of patients and should involve population preferences and solid team collaborations. Show less
Dijk, W.J. van; Saadah, N.H.; Numans, M.E.; Aardoom, J.J.; Bonten, T.N.; Brandjes, M.; ... ; Kiefte-de Jong, J. 2021
Background Structured primary diabetes care within a collectively supported setting is associated with better monitoring of biomedical and lifestyle-related target indicators amongst people with... Show moreBackground Structured primary diabetes care within a collectively supported setting is associated with better monitoring of biomedical and lifestyle-related target indicators amongst people with type 2 diabetes and with better HbA1c levels. Whether socioeconomic status affects the delivery of care in terms of monitoring and its association with HbA1c levels within this approach, is unclear. This study aims to understand whether, within a structured care approach, (1) socioeconomic categories differ concerning diabetes monitoring as recommended; (2) socioeconomic status modifies the association between monitoring as recommended and HbA1c.Methods Observational real-life cohort study with primary care registry data from general practitioners within diverse socioeconomic areas, who are supported with the implementation of structured diabetes care. People with type 2 diabetes mellitus were offered quarterly diabetes consultations. "Monitoring as recommended" by professional guidelines implied minimally one annual registration of HbA1c, systolic blood pressure, LDL, BMI, smoking behaviour and physical activity. Regarding socioeconomic status, deprived, advantageous urban and advantageous suburban categories were compared to the intermediate category concerning (a) recommended monitoring; (b) association between recommended monitoring and HbA1c.Results Aim 1 (n = 13 601 people): Compared to the intermediate socioeconomic category, no significant differences in odds of being monitored as recommended were found in the deprived (OR 0.45 (95% CI 0.19-1.08)), advantageous urban (OR 1.27 (95% CI 0.46-3.54)) and advantageous suburban (OR 2.32 (95% CI 0.88-6.08)) categories. Aim 2 (n = 11 164 people): People with recommended monitoring had significantly lower HbA1c levels than incompletely monitored people (-2.4 (95% CI -2.9; -1.8) mmol/mol). SES modified monitoring-related HbA1c differences, which were significantly higher in the deprived (-3.3 (95% CI -4.3; -2.4) mmol/mol) than the intermediate category (-1.3 (95% CI -2.2; -0.4) mmol/mol).Conclusions Within a structured diabetes care setting, socioeconomic status is not associated with recommended monitoring. Socioeconomic differences in the association between recommended monitoring and HbA1c levels advocate further exploration of practice and patient-related factors contributing to appropriate monitoring and for care adjustment to population needs. Show less
Brakema, E.A.; Kleij, R.M.J.J. van der; Poot, C.C.; Chavannes, N.H.; Tsiligianni, I.; Walusimbi, S.; ... ; FRESH AIR collaborators 2021
Effectiveness of health interventions can be substantially impaired by implementation failure. Context-driven implementation strategies are critical for successful implementation. However, there is... Show moreEffectiveness of health interventions can be substantially impaired by implementation failure. Context-driven implementation strategies are critical for successful implementation. However, there is no practical, evidence-based guidance on how to map the context in order to design context-driven strategies. Therefore, this practice paper describes the development and validation of a systematic context-mapping tool. The tool was cocreated with local end-users through a multistage approach. As proof of concept, the tool was used to map beliefs and behaviour related to chronic respiratory disease within the FRESH AIR project in Uganda, Kyrgyzstan, Vietnam and Greece. Feasibility and acceptability were evaluated using the modified Conceptual Framework for Implementation Fidelity. Effectiveness was assessed by the degree to which context-driven adjustments were made to implementation strategies of FRESH AIR health interventions. The resulting Setting-Exploration-Treasure-Trail-to-Inform-implementatioN-strateGies (SETTING-tool) consisted of six steps: (1) Coset study priorities with local stakeholders, (2) Combine a qualitative rapid assessment with a quantitative survey (a mixed-method design), (3) Use context-sensitive materials, (4) Collect data involving community researchers, (5) Analyse pragmatically and/or in-depth to ensure timely communication of findings and (6) Continuously disseminate findings to relevant stakeholders. Use of the tool proved highly feasible, acceptable and effective in each setting. To conclude, the SETTING-tool is validated to systematically map local contexts for (lung) health interventions in diverse low-resource settings. It can support policy-makers, non-governmental organisations and health workers in the design of context-driven implementation strategies. This can reduce the risk of implementation failure and the waste of resource potential. Ultimately, this could improve health outcomes. Show less
Bonten, T.N.; Verkleij, S.M.; Kleij, R.M.J.J. van der; Busch, K.; Hout, W.B. van den; Chavannes, N.H.; Numans, M.E. 2021
Introduction Lifestyle interventions are shown to be effective in improving cardiovascular disease (CVD) risk factors. It has been suggested that general practitioners can play an essential role in... Show moreIntroduction Lifestyle interventions are shown to be effective in improving cardiovascular disease (CVD) risk factors. It has been suggested that general practitioners can play an essential role in CVD prevention. However, studies into lifestyle interventions for primary care patients at high cardiovascular risk are scarce and structural implementation of lifestyle interventions can be challenging. Therefore, this study aims to (1) evaluate (cost-)effectiveness of implementation of an integrated group-based lifestyle programme in primary care practices; (2) identify effective intervention elements and (3) identify implementation determinants of an integrated group-based lifestyle intervention for patients with high cardiovascular risk. Methods and analysis The Healthy Heart study is a non-randomised cluster stepped-wedge trial. Primary care practices will first offer standard care during a control period of 2-6 months, after which practices will switch (step) to the intervention, offering participants a choice between a group-based lifestyle programme or standard care. Participants enrolled during the control period (standard care) will be compared with participants enrolled during the intervention period (combined standard care and group-based lifestyle intervention). We aim to include 1600 primary care patients with high cardiovascular risk from 55 primary care practices in the area of The Hague, the Netherlands. A mixed-methods process evaluation will be used to simultaneously assess effectiveness and implementation outcomes. The primary outcome measure will be achievement of individual lifestyle goals after 6 months. Secondary outcomes include lifestyle change of five lifestyle components (smoking, alcohol consumption, diet, weight and physical activity) and improvement of quality of life and self-efficacy. Outcomes are assessed using validated questionnaires at baseline and 3, 6, 12 and 24 months of follow-up. Routine care data will be used to compare blood pressure and cholesterol levels. Cost-effectiveness of the lifestyle intervention will be evaluated. Implementation outcomes will be assessed using the RE-AIM model, to assesses five dimensions of implementation at different levels of organisation: reach, efficacy, adoption, implementation and maintenance. Determinants of adoption and implementation will be assessed using focus groups consisting of professionals and patients. Ethics and dissemination This study is approved by the Ethics Committee of the Leiden University Medical Center (P17.079). Results will be shared with the primary care group, healthcare providers and patients, and will be disseminated through journal publications and conference presentations. Show less
In the iZi study in The Hague, use and acceptance of commercially available technology by home-dwelling older citizens was studied, by comparing self-efficacy and perceived physical and mental... Show moreIn the iZi study in The Hague, use and acceptance of commercially available technology by home-dwelling older citizens was studied, by comparing self-efficacy and perceived physical and mental Quality of Life (QoL)-related parameters on an intervention location of 279 households and a control location of 301 households. Technology adoption was clinically significantly associated with increased perceived physical QoL, as compared with control group, depending on the number of technology interventions that were used. A higher number of adopted technologies was associated with a stronger effect on perceived QoL. We tried to establish a way to measure clinical significance by using mixed methods, combining quantitative and qualitative evaluation and feeding results and feedback of participants directly back into our intervention. In general, this research is promising, since it shows that successful and effective adoption of technology by older people is feasible with commercially available products amongst home-dwelling older citizens. We think this way of working provides a better integration of scientific methods and clinical usability but demands a lot of communication and patience of researchers, citizens, and policymakers. A change in policy on how to target people for this kind of intervention might be warranted. Show less
In the iZi study in The Hague, use and acceptance of commercially available technology by home-dwelling older citizens was studied, by comparing self-efficacy and perceived physical and mental... Show moreIn the iZi study in The Hague, use and acceptance of commercially available technology by home-dwelling older citizens was studied, by comparing self-efficacy and perceived physical and mental Quality of Life (QoL)-related parameters on an intervention location of 279 households and a control location of 301 households. Technology adoption was clinically significantly associated with increased perceived physical QoL, as compared with control group, depending on the number of technology interventions that were used. A higher number of adopted technologies was associated with a stronger effect on perceived QoL. We tried to establish a way to measure clinical significance by using mixed methods, combining quantitative and qualitative evaluation and feeding results and feedback of participants directly back into our intervention. In general, this research is promising, since it shows that successful and effective adoption of technology by older people is feasible with commercially available products amongst home-dwelling older citizens. We think this way of working provides a better integration of scientific methods and clinical usability but demands a lot of communication and patience of researchers, citizens, and policymakers. A change in policy on how to target people for this kind of intervention might be warranted. Show less
Objective Whether care group participation by general practitioners improves delivery of diabetes care is unknown. Using 'monitoring of biomedical and lifestyle target indicators as recommended by... Show moreObjective Whether care group participation by general practitioners improves delivery of diabetes care is unknown. Using 'monitoring of biomedical and lifestyle target indicators as recommended by professional guidelines' as an operationalisation for quality of care, we explored whether (1) in new practices monitoring as recommended improved a year after initial care group participation (aim 1); (2) new practices and experienced practices differed regarding monitoring (aim 2).Design Observational, real-life cohort study.Setting Primary care registry data from Eerstelijns Zorggroep Haaglanden (ELZHA) care group.Participants Aim 1: From six new practices (n=538 people with diabetes) that joined care group ELZHA in January 2014, two practices (n=211 people) were excluded because of missing baseline data; four practices (n=182 people) were included. Aim 2: From all six new practices (n=538 people), 295 individuals were included. From 145 experienced practices (n=21 465 people), 13 744 individuals were included.Exposure Care group participation includes support by staff nurses on protocolised diabetes care implementation and availability of a system providing individual monitoring information. 'Monitoring as recommended' represented minimally one annual registration of each biomedical (HbA1c, systolic blood pressure, low-density lipoprotein) and lifestyle-related target indicator (body mass index, smoking behaviour, physical exercise).Primary outcome measures Aim 1: In new practices, odds of people being monitored as recommended in 2014 were compared with baseline (2013). Aim 2: Odds of monitoring as recommended in new and experienced practices in 2014 were compared.Results Aim 1: After 1-year care group participation, odds of being monitored as recommended increased threefold (OR 3.00, 95% CI 1.84 to 4.88, p<0.001). Aim 2: Compared with new practices, no significant differences in the odds of monitoring as recommended were found in experienced practices (OR 1.21, 95% CI 0.18 to 8.37, p=0.844).Conclusions We observed a sharp increase concerning biomedical and lifestyle monitoring as recommended after 1-year care group participation, and subsequently no significant difference between new and experienced practices-indicating that providing diabetes care within a collective approach rapidly improves registration of care. Show less
Objective Management of type 2 diabetes mellitus (T2DM) requires frequent patient monitoring. Within a collective care group setting, doubts on the clinical effects of registration are a barrier... Show more Objective Management of type 2 diabetes mellitus (T2DM) requires frequent patient monitoring. Within a collective care group setting, doubts on the clinical effects of registration are a barrier for full adoption of T2DM registration in general practice. We explored whether full monitoring of physiological, biomedical and lifestyle-related target indicators within a care group approach is associated with lower HbA1c levels. Design Observational, real-life cohort study Setting Primary care data registry from the EerstelijnsZorggroepHaaglanden care group. Exposure The care group provides general practitioners collectively with organisational support to facilitate structured T2DM primary care. Patients are offered quarterly medical and lifestyle-related consultation. Main outcome measure Full monitoring of each target indicator in patients with T2DM, which includes minimally one measure of HbA1c level, systolic blood pressure, LDL, BMI, smoking behaviour and physical exercise between January and December 2014; otherwise, patients were defined as ´incompletely monitored´. HbA1c levels of 8,137 fully-monitored and 3,958 incompletely-monitored patients were compared, adjusted for the confounders diabetes duration, age and gender. Since recommended HbA1c values depend on age, medication use and diabetes duration, analyses were stratified into three HbA1c profile groups. Linear multilevel analyses enabled adjustment for general practice. Results Compared to incompletely-monitored patients, fully-monitored patients had significantly lower HbA1c levels [95%CI] in the first (-2.03 [-2.53;-1.52]mmol/mol) (-0.19% [-0.23%;-0.14%]), second (-3.36 [-5.28;-1.43]mmol/mol) (-0.31% [-0.48%;-0.13%]) and third HbA1c profile group (-1.89 [-3.76;-0.01]mmol/mol) (-0.17% [-0.34%;0.00%]). Conclusions/interpretation This study shows that in a care group setting, fully-monitored patients had significantly lower HbA1c levels compared with incompletely-monitored patients. Since this difference might have considerable clinical impact in terms of T2DM-related risks, this might help general practices in care group settings to overcome barriers on adequate registration and thus improve structured T2DM primary care. From population health management perspective, we recommend a systematic approach to adjust the structured care protocol for incompletely-monitored subgroups. Show less
Ommen, B. van; Wopereis, S.; Empelen, P. van; Keulen, H.M. van; Otten, W.; Kasteleyn, M.; ... ; Pijl, H. 2018
Background: Worldwide nearly 3 million people die from chronic obstructive pulmonary disease (COPD) every year. Integrated disease management (IDM) improves quality of life for COPD patients and... Show moreBackground: Worldwide nearly 3 million people die from chronic obstructive pulmonary disease (COPD) every year. Integrated disease management (IDM) improves quality of life for COPD patients and can reduce hospitalization. Self-management of COPD through eHealth is an effective method to improve IDM and clinical outcomes.Objectives: The objective of this implementation study was to investigate the effect of 3 chronic obstructive pulmonary disease eHealth programs applied in primary care on health status. The e-Vita COPD study compares different levels of integration of Web-based self-management platforms in IDM in 3 primary care settings. Patient health status is examined using the Clinical COPD Questionnaire (CCQ).Methods: The parallel cohort design includes 3 levels of integration in IDM (groups 1, 2, 3) and randomization of 2 levels of personal assistance for patients (group A, high assistance, group B, low assistance). Interrupted time series (ITS) design was used to collect CCQ data at multiple time points before and after intervention, and multilevel linear regression modeling was used to analyze CCQ data.Results: Of the 702 invited patients, 215 (30.6%) registered to a platform. Of these, 82 participated in group 1 (high integration IDM), 36 in group 1A (high assistance), and 46 in group 1B (low assistance); 96 participated in group 2 (medium integration IDM), 44 in group 2A (high assistance) and 52 in group 2B (low assistance); also, 37 participated in group 3 (no integration IDM). In the total group, no significant difference was found in change in CCQ trend (P=.334) before (-0.47% per month) and after the intervention (-0.084% per month). Also, no significant difference was found in CCQ changes before versus after the intervention between the groups with high versus low personal assistance. In all subgroups, there was no significant change in the CCQ trend before and after the intervention (group 1A, P=.237; 1B, P=.991; 2A, P=.120; 2B, P=.166; 3, P=.945).Conclusions: The e-Vita eHealth-supported COPD programs had no beneficial impact on the health status of COPD patients. Also, no differences were found between the patient groups receiving different levels of personal assistance. Show less
