Teamwork and Adherence to Recommendations Explain the Effect of a Care Pathway on Reduced 30-day Readmission for Patients with a COPD Exacerbation

ABSTRACT

This study aimed to increase our understanding of processes that underlie the effect of care pathway implementation on reduced 30-day readmission rate. Adherence to evidence-based recommendations, teamwork and burnout have previously been identified as potential mechanisms in this association. We conducted a secondary data analysis of 257 patients admitted with chronic obstructive pulmonary disease exacerbation and 284 team members caring for these patients in 19 Belgian, Italian and Portuguese hospitals. Clinical measures included 30-day readmission and adherence to a specific set of five care activities. Teamwork measures included team climate for innovation, level of organized care and burnout (emotional exhaustion, level of competence and mental detachment). Care pathway implementation was significantly associated with better adherence and reduced 30-day readmission. Better adherence and higher level of competence were also related to reduced 30-day readmission. Only better adherence fully mediated the association between care pathway implementation and reduced 30-day readmission. Better team climate for innovation and level of organized care, although both improved after care pathway implementation, did not show any explanatory mechanisms in the association between care pathway implementation and reduced 30-day readmission. Implementation of a care pathway had an impact on clinical and team indicators. To reduce 30-day readmission rates, in the development and implementation of a care pathway, hospitals should measure adherence to evidence-based recommendations during the whole process, as this can give information regarding the success of implementation

KEYWORDS:

• adherence to recommendations
• quality of care
• burnout
• level of organized care
• team climate for innovation

Introduction

Chronic Obstructive Pulmonary Disease (COPD) is the fourth leading cause of death worldwide (1). The exacerbations related with this disease are a leading cause of unplanned medical admissions (2, 3). The management of COPD is well described in the Global Initiative for Obstructive Lung Disease (GOLD) recommendation (4). Recent European research however revealed much variation in the adherence to these recommendations, and an overall low performance (3,5). An importance-performance analysis visualized these care gaps, which are highest for care activities related to non-pharmacological management (6).

Care pathways (CPs) are considered as a useful method to increase the adherence to clinical recommendations (7). CPs present a patient-related, professional-directed and organizational intervention, according to Cochrane Effective Practice and Organization of Care (EPOC) (8,9). The aim of CPs is to improve outcomes by providing a mechanism to coordinate care, reduce fragmentation and ultimately reduce costs (10). In general, CPs tend to increase the documentation of care and to decrease in-hospital complications (11). Findings for other patient outcomes are however inconclusive. But overall results are promising as triple intervention programmes, based on EPOC, could lead to improvements in quality of life and reduction in hospitalizations (12). The impact of CP for COPD can lead to significantly shorter length of stay (13-15), increased interval before readmission (13), decline in mean dyspnea score at discharge (14), decline in complications (15), decline in mean anxiety score (13,14), increased pulmonary function test at discharge (14), improved albumin/nutrition profile was obtained (13), increased referral to rehabilitation program (15) and increase arterial blood gases values at discharge (13,14). However, it is hard to compare the adherence to evidence-based recommendations and patient outcomes for different CPs because there is large variation in implemented care activities and lack of consistency in the measurement of these recommendations and patient outcomes.

The European Quality of Care Pathway (EQCP) study was set up to evaluate the impact of CPs on the adherence to clinical recommendations, patient outcomes and teamwork. This European study was the first international cluster randomized controlled trial that evaluated the impact of CPs for patients admitted with a COPD exacerbation (16-18). A CP for patients admitted with a COPD exacerbation led to a reduction in 30-day readmission rate. No impact was found on the total adherence to clinical recommendations, although some care activities were significantly better performed in the intervention group compared to the control group. In general, the diagnostic and pharmacological management were well performed (17). Next to clinical impact, a CP led to better interprofessional teamwork, higher level of organized care and lower risk of burnout (18,19).

Low adherence to recommendations may contribute to suboptimal patient outcomes (20). On the other hand, interprofessional teams were linked with improved quality of treatment and better clinical outcomes (21-23). This suggests that adherence to recommendations and teamwork could both be improved after implementation of a CP and these variables may be potential mediators in the association between CP and reduced 30-day readmission. Burnout is negatively correlated with quality of healthcare and safety for patients (24,25).

Until now, CP research had focused exclusively on evaluating the impact of CP implementation on clinical and team outcomes independently. The aim of this study is to increase understanding of the mechanisms through which CPs affect patient outcomes. We hypothesized that the reduction in 30-day readmission, after implementation of a CP, is due to improved adherence to a set of evidence-based recommendations and better teamwork.

Methods

Study design and participants

This study is a secondary data analysis exploring the impact of CP implementation on adherence to recommendations, patient outcomes and teamwork based on data of a cRCT (16,26). Teaching status, size of hospitals and annual volume of patients were used as stratification criteria for the cRCT and hospitals were randomized in a control and intervention group. In the intervention group, the usual care was evaluated before the implementation of a CP. Then, the intervention, a CP, was developed and implemented. The intervention consists of three active components: (1) evaluation on the organisation and quality of care before implementation of a CP; (2) providing set of evidence-based key interventions to each team and (3) training on how to develop and implement a CP based on the 7-phase protocol (27). After implementation of the intervention, the care in the intervention group was evaluated. At this moment, the care in the control group was also measured (16,17). The intervention was implemented at team level. Nineteen teams were involved in the care for patients admitted with a COPD exacerbation in Belgian, Italian and Portuguese hospitals. Data collection for Belgian data took place between October 2010 and January 2012 and between January 2013 and May 2014 for Italian and Portuguese data. The study protocol and prior analysis of team data are described more in detail elsewhere (16,18,28). This study was registered as a cRCT at ClinicalTrials.gov (identifier NCT00962468).

Clinical measures

In the EQCP study, 24 care activities and five patient outcomes were measured (30-day and 6-month mortality and COPD-specific readmission rate and length of stay) (17). For this study the clinical measures were limited to adherence to a specific set of evidence-based recommendations and 30-day readmission rate, which refers to COPD-specific readmission (17,29,30). We included 30-day readmission rate, as this was the only patient outcome on which our CP had a significant impact (17). The clinical measures were evaluated at individual patient level. The set of evidence-based recommendations was defined as 1) arterial blood gas measurement 1 or 2 days prior to discharge in patients hypoxemic during COPD exacerbation, 2) referral to pulmonary rehabilitation during the past year, 3) nutritional assessment (BMI), 4) nutritional management of patients with overweight and 5) nutritional management of patients with underweight. For each of the 24 care activities measured in the EQCP study, the impact on 30-day readmission rate was measured. In the set of evidence-based recommendations we only included the care activities which had a significant impact on 30-day readmission rate. Adherence is a proportion of the number of individual care activities adhered to and the number of relevant care activities.

Team measures

Teamwork measures included team climate for innovation (32), the level of organized care (18,28,33) and burnout (34). These were measured at the individual healthcare worker level and were also hypothesized mediating variables. Mediation analysis assesses whether a given variable accounts for an association, i.e. how or why does an effect occur (31)? Team climate for innovation was measured by the Team Climate Inventory (TCI) (score between 1 and 5) (32). The level of organized care was measured by the Care Process Self Evaluation Tool (CPSET) (score between 1 and 100) (18,28,33). Burnout was measured at the individual healthcare worker level by the Burnout Inventory (34) (score between 1 and 10). Burnout was measured by three dimensions: emotional exhaustion (the lower, the better), mental detachment or depersonalisation (the lower, the better) and level of competence or personal accomplishment (the higher, the better) (34). The data collection took place during one specific week between October 2010 and January 2012 for Belgian data and between January 2013 and May 2014 for Italian and Portuguese data (28).

Statistical analysis

Teamwork was measured at team^{Figure 1} member level and analysed at team level. Team scores were calculated by averaging team member scores. Only fully completed questionnaires for TCI, burnout and CPSET were included in the statistical analyses. Team scores were calculated by averaging team member scores. COPD severity at admission, cardiac failure, diabetes, BMI and^{Table 1} intervention-control group were used as covariates for the impact of a CP on 30-day readmission (17).

To understand mechanisms underlying the association between CP implementation and 30-day readmission, two-level mediation analysis was applied (31,35). Mediation analysis assesses whether a given variable accounts for an association, i.e. how or why does an effect occur (31). Mediation analysis thus allows examination of process. Variables that reflect this process are called mediators. Adherence and teamwork were hypothesized mediators in the relationship between CP implementation and 30-day readmission rate (36). E.g. Full mediation of adherence means that the development and implementation of a CP leads to a significant higher adherence and that adherence, on its turn, leads to a significantly reduced 30-day readmission rate. To test our hypotheses of mediation, five sequential models were tested (36). The first four models used simple and multiple multilevel regression analysis, after which multilevel parallel mediation analysis was used in the fifth model. Our hypotheses are graphically depicted in Figure 1.

Figure 1. Overview mediation analysis.

In model 1, the association between the intervention, i.e. CP implementation (main explanatory variable) and 30-day readmission (main outcome variable) was evaluated (18,19).

Next, we tested the association between CP implementation and the proposed mediators: adherence (model 2a), TCI (model 2b), level of organized care (model 2c), emotional exhaustion (model 2d), level of competence (model 2e) and mental detachment (model 2f).

Subsequently, the impact of adherence (model 3a), TCI (model 3b), level of organized care (model 3c), emotional exhaustion (model 3d), level of competence (model 3e) and mental detachment (model 3f) on 30-day readmission were examined.

In model 4, we examined whether CP implementation, adherence and level of competence together explained 30-day readmission rate (combining model 1, model 3a and model 3e). Note that a mediating effect of the four hypothesized teamwork mediators was no longer pursued, since these did not relate to 30-day readmission rate (models 3b, 3c, 3d and 3f).

In model 5, we simultaneously examined the joint effect of CP implementation, adherence and level of competence on 30-day readmission rate (model 4) and the effect of CP implementation on adherence and level of competence (model 2a and model 2e). Mediation for adherence was calculated by multiplying the estimate of the effect of CP implementation on adherence with the estimate of the effect of adherence on 30-day readmission rate. Mediation for level of competence was calculated by multiplying the estimate of the effect of CP implementation on the level of competence with the estimate of the effect of level of competence on 30-day readmission rate.

Statistical significance was defined as a 2-sided P-value of 0.05. Model fit evaluation was performed according to standard procedures (37). Multilevel regression and parallel mediation analyses were conducted in Mplus 7.31.

Results

Descriptive findings

Table 1 describes patient and hospital characteristics. In total 257 patients admitted in 19 hospitals were included. In the intervention group, patients had a lower GOLD severity compared to the control group (P = 0.031). Adherence was significantly higher in the intervention group (P = 0.012), and the 30-day readmission rate was significantly lower in the intervention group compared to the control group (P = 0.042). In total 284 team members caring for these patients completed the team questionnaires Intervention teams had a higher proportion of female team members (P = 0.026). No significant differences were found for the other team member characteristics (Table 2).

Table 1. Patient and hospital characteristics.

	Intervention group (n=145)	Control group (n=112)	P-value^a
Individual patient level
Age (years)	69.5 (10.8)	70.2 (9.7)	0.337
Sex			0.773
Men	97 (66.9%)	73 (65.2%)
Women	48 (33.1%)	39 (34.8%)
COPD severity at admission			0.031
GOLD I: mild	4 (2.8%)	7 (6.3%)
GOLD II: moderate	50 (34.5%)	22 (19.6%)
GOLD III: severe	60 (41.4%)	49 (43.8%)
GOLD IV: very severe	31 (21.4%)	34 (30.4%)
Smoking status			0.994
no smoker	9 (6.2%)	7 (6.3%)
ex-smoker	74 (51.0%)	57 (50.9%)
active smoker	62 (42.8%)	48 (42.9%)
Hospitalisation in the year before index admission^b			0.058
0	94 (66.7%)	61 (55.0%)
≥ 1	47 (33.3%)	50 (45.0%)
Unknown	4	1
Charlson comorbidity index			0.119
<=2	59 (40.7%)	35 (31.3%)
>2	86 (59.3%)	77 (68.8%)
Cardiac failure^c	45 (31.0%)	43 (38.4%)	0.218
Diabetes^d	18 (12.4%)	18 (16.1%)	0.402
BMI			0.059
<20	39 (26.9%)	19 (17.0%)
>=20	106 (73.1%)	93 (83.0%)
30-day readmission rate^e	15 (10.3%)	22 (19.6%)	0.042
Adherence	47.3 ± 32.6	24.8 ± 29.2	0.012
ABG measurement^f	21 (42.0%)	15 (39.5%)	0.825
Referral to pulmonary rehabilitation during last year	72 (42.9%)	50 (34.0%)	0.463
Nutritional assessment (BMI)	108 (64.3%)	28 (19.0%)	0.007
Nutritional assessment – underweight	10 (18.2%)	2 (5.7%)	0.233
Nutritional assessment - overweight	10 (37.0%)	2 (9.5%)	0.140
Hospital level
Teaching hospitals	5/10 (50.0%)	4/9 (44.4%)	0.809
>600 beds	2/10 (20.0%)	3/9 (33.3%)	0.509
>300 patients admitted with COPD exacerbation yearly	2/10 (20.0%)	1/9 (11.1%)	0.592
Dedicated caring wards for pneumology	1/10 (10.0%)	3/9 (33.3%)	0.213
Possibility for non-invasive ventilation	3/10 (30.0%)	4/9 (44.4%)	0.514

Notes: Data are mean (SD) or n/N (%). Abbreviations: BMI, body mass index; GOLD, Global Initiative for Obstructive Lung Disease.

^a Chi-square test, Mann-Whitney U-test and t-test for categorical, ordinal and continuous variables, respectively.

^b Index admission: first admission during the study, at point of inclusion.

^c Includes heart failure, arrhythmia, valvular disease, acute myocardial infarction, ischemic heart disease.

^d Includes diabetes uncomplicated and complicated.

^e Adjusted for COPD severity at admission, cardiac failure, diabetes, BMI and intervention-control group.

^f Arterial blood gas measurement 1 or 2 days prior to discharge in patients hypoxemic during COPD exacerbation.

Table 2. Team characteristics.

	Intervention group (n = 152)	Control group (n = 132)	P-value^a
Individual team level
Age of participating team members	20-29 = 32.9%	20-29 = 28.0%	0.434
(% within a given category)	30-39 = 24.3%	30-39 = 29.5%
	40-49 = 19.7%	40-49 = 20.5%
	50-59 = 23.0%	50-59 = 20.5%
	60-69 = 0.0%	60-69 = 1.5%
Sex (% within a given category): male team member	24 (15.8%)	35 (26.5%)	0.026
female team member	128 (84.2%)	97 (73.5%)
Average number of years of being team member (SD)	9.4 (8.4)	9.1 (9.6)	0.785
Mean years of experience with COPD patients (SD)	9.5 (8.4)	11.4 (10.5)	0.091
Team processes
Team climate for innovation	3.6 (0.6)	3.2 (0.7)	0.007
Team vision	3.8 ± 0.6	3.4 ± 0.7	0.004
Participative safety	3.7 ± 0.7	3.3 ± 0.8	0.019
Task orientation	3.4 ± 0.7	3.0 ± 0.8	0.027
Support for innovation	3.4 ± 0.7	2.9 ± 0.8	0.001
Team outcomes
Level of organized care	67.9 ± 14.1	60.1 ± 15.0	0.005
Patient focused organization	73.0 ± 15.3	65.8 ± 16.7	0.008
Coordination of care	69.7 ± 14.0	62.8 ± 15.9	0.013
Communication with patient and family	64.4 ± 19.3	56.3 ± 20.6	0.042
Collaboration with primary care	66.1 ± 17.3	62.0 ± 18.7	0.158
Follow up of care	64.4 ± 15.6	55.6 ± 18.3	0.005
Level of burnout
Emotional exhaustion	3.5 ± 1.7	4.1 ± 1.9	0.026
Level of competence	7.9 ± 1.4	7.5 ± 1.3	0.026
Mental detachment	3.3 ± 1.4	4.0 ± 1.6	0.027

Notes: ^aChi-square test, Mann-Whitney U-test and t-test for categorical, ordinal and continuous variables, respectively.

Regression and mediation analysis

All models used in the mediation and moderated mediation analysis indicated a good model fit (the Root Mean Square Error for Approximation (RMSEA) ≤ 0.05, comparative fit index (CFI) ≥0.95 and Tucker-Lewis Index (TLI) ≥0.95) (37).

The regression coefficients of the four models used in the simple and multiple multilevel regression analysis are shown in Table 3.

Table 3. Simple and multiple multilevel regression analysis.

	Care Pathway				30-day readmission
	Estimate	Lower 2·5% CI	Upper 2·5% CI	P-value	Estimate	Lower 2·5% CI	Upper 2·5% CI	P-value
Model 2a: effect of CP on adherence (proposed mediator)
Adherence	14.449	3.240	25.658	0.012	—	—	—	—
Model 2b: effect of CP on TCI (proposed mediator)
TCI	0.365	0.108	0.621	0.005	—	—	—	—
Model 2c: effect of CP on level of organized care (proposed mediator)
Level of organized care	6.655	1.785	11.525	0.007	—	—	—	—
Model 2d: effect of CP on emotional exhaustion (proposed mediator)
Emotional exhaustion	−0.588	−1.319	0.142	0.115	—	—	—	—
Model 2e: effect of CP on level of competence (proposed mediator)
Level of competence	0.244	−0.205	0.692	0.287	—	—	—	—
Model 2f: effect of CP on mental detachment (proposed mediator)
Mental detachment	−0.675	−1.345	−0.005	0.048	—	—	—	—
Model 3a: effect of adherence (proposed mediator) on 30-day readmission
Adherence	—	—	—	—	−0.037	−0.053	−0.020	<0.001
Model 3b: effect of TCI (proposed mediator) on 30-day readmission
TCI	—	—	—	—	−0.827	−1.976	0.321	0.158
Model 3c: effect of level of organized care (proposed mediator) on 30-day readmission
Level of organized care	—	—	—	—	−0.024	−0.120	0.072	0.631
Model 3d: effect of emotional exhaustion (proposed mediator) on 30-day readmission
Emotional exhaustion	—	—	—	—	0.129	−0.366	0.623	0.609
Model 3e: effect of level of competence (proposed mediator) on 30-day readmission
Level of competence	—	—	—	—	−1.640	−2.899	−0.380	0.011
Model 3f: effect of mental detachment (proposed mediator) on 30-day readmission
Mental detachment	—	—	—	—	−0.027	−0.628	0.574	0.930
Model 4: effect of CP, adherence and level of competence (proposed mediators) on 30-day readmission
Care pathway	—	—	—	—	0.591	−0.865	2.047	0.426
Adherence	—	—	—	—	−0.039	−0.061	−0.016	0.001
Level of competence	—	—	—	—	−2.128	−0.213	−4.044	0.029

Notes: 30-day readmission was adjusted for COPD severity at admission, cardiac failure, diabetes and BMI.

Abbreviations: CI, Confidence Interval; BMI, Body Mass Index; GOLD, Global Initiative for Obstructive Lung Disease; TCI, Team Climate for Innovation.

Model 1 shows that the implementation of a CP leads to a significant reduction in 30-day readmission rate (β = −0.898, 95%CI (−1.765; −0.030), P = 0.042).

Model 2 shows that the implementation of a CP leads to a significantly better adherence (β = 14.449, P = 0.012), TCI (β = 0.365, P = 0.005), level of organized care (β = 6.655, P = 0.007) and lower mental detachment (β = -0.675, P = 0.048).

Of these six proposed mediators, only better adherence (β = −0.037, P < 0.001) and higher level of competence (β = −1.640, P = 0.011) led to a significant reduction in 30-day readmission (Model 3). As such, only these proposed mediators will be included in the next steps.

The effect of CP implementation on 30-day readmission was strongly reduced and became insignificant when adding adherence and level of competence to the model (Model 4) (β = 0.591, P = 0.426) (Table 3).

The final parallel mediation model (Model 5) shows that better adherence fully mediates the association between CP implementation (β = 18.152, P = 0.001) and reduced 30-day readmission (β = −0.038, P = 0.001). Full mediation is concluded from: 1) the product of the path from CP implementation to adherence and the path from adherence to 30-day readmission (β = −0.693, P = 0.003), indicating mediation; 2) the direct effect of CP implementation on 30-day readmission is no longer significant (β = 0.526, P = 0.464), indicating full mediation. If this effect was still significant, this would have indicated partial mediation. Higher level of competence led to a significant reduction in 30-day readmission (β = −2.044, P = 0.022) but did not show any explanatory mechanism in the association between CP implementation and 30-day readmission (Table 4).

Table 4. Parallel mediation analyses: estimating the mediating effect of adherence and teamwork in the association between CP implementation and 30-day readmission.

	30-day readmission				Level of competence				Adherence
		Lower	Upper			Lower	Upper			Lower	Upper
	Estimate	2·5% CI	2·5% CI	P-value	Estimate	2·5% CI	2·5% CI	P-value	Estimate	2·5% CI	2·5% CI	P-value
Model 5: mediation model assessing the mediating effect of adherence and level of competence in the association between CP and 30-day readmission
Care pathway	0.526	−0.880	1.931	0.464	0.163	−0.319	0.646	0.508	18.152	7.474	28.830	0.001
Adherence	−0.038	−0.060	−0.016	0.001	—	—	—	—	—	—	—	—
Level of competence	−2.044	−3.789	−0.298	0.022	—	—	—	—	—	—	—	—

Notes: 30-day readmission was adjusted for COPD severity at admission, cardiac failure, diabetes and BMI.

Abbreviations: CI, Confidence Interval; BMI, Body Mass Index; GOLD, Global Initiative for Obstructive Lung Disease; TCI, Team Climate for Innovation.

Discussion

This is the first study examining whether adherence and teamwork explain why CP implementation has an effect on a patient outcomes. In our study, CP implementation was significantly associated with better team climate for innovation, better level of organized care, better adherence and reduced 30-day readmission. Better adherence and higher level of competence reduced 30-day readmission rate. Only better adherence fully mediated the association between CP implementation and reduced 30-day readmission. Team climate for innovation, level of organized care and level of competence did not show any explanatory mechanisms in the association between CP implementation and 30-day readmission.

This study confirms that burnout plays an important role in quality of care and may have an indirect impact on patient outcomes. A recent meta-analysis showed a negative relationship between burnout and quality of healthcare, patient safety and reduced patient satisfaction (25). Burnout was significantly associated with urinary tract infection and surgical site infection (38). Mental detachment can have an impact on nurse-reported frequency of patient falls, nosocomial infections and medications errors but also on patient and family verbal abuse and patient and family complaints. The latest was also predicted by emotional exhaustion (39). Of the three dimension of burnout, emotional exhaustion had the strongest relationship with quality of care (25). On the other hand, the perception of patient-centered care was associated with improved patient outcomes (25,40). A CP led to more collaborative care, and organizations should pay attention to burnout as it may have an impact on the care patients receive (25). However, in this study, collaboration with primary care did not significantly improved.

In general, the adherence to clinical recommendations for patients with a COPD exacerbation is suboptimal (6,41). Previous results of the EQCP study showed a significant impact of CP implementation on some but not all evidence-based care activities (17). If we performed the same mediation analysis with the complete evidence-based recommendations, described in the GOLD recommendations (4), no significant impact was found on 30-day readmission. The idea behind GOLD recommendations is that, ideally, all these care activities should be performed to improve patients' outcomes (4). This may indicate that the effect of our CP is fragmented as no coherent impact was found. A possible explanation of this fragmented effect can be due to the fact that hospitals, in the intervention group, received the evidence-based care activities but, due to the high number of care activities, they selected only a few on which they could work on and improve. This supports the idea of using a priority bundle for quality improvement initiatives. In a priority bundle, teams or hospitals select a few care activities on which they will focus on for improving the care patients receive. This set of care activities should mainly be indicators with a high room for improvement but also a high level of evidence. Each hospital should perform an analysis which combines both the performance rates and the importance rate of the different care activities (6). On the other hand, some of the care activities were already “well” performed and there is less room for improvement for these care activities (17).

To improve patient outcomes, our results suggest that the focus for organizations and teams could be on both improving the healthcare worker wellbeing and improving the adherence to clinical recommendations. This is in line with the expansion from triple aim to quadruple aim in optimizing health system performance (42). Triple aim contains patient experience, improving population health and reducing costs. Burnout imperils triple aim and the goal of improving work life of health care providers should be added to triple aim, which lead to quadruple aim (42). The idea of quadruple aim seems to be also applicable for CP research.

This study had several limitations. First, as this is an international study, the standardization of the intervention was difficult because of differences in culture and in the translation of the evidence-based key interventions from English to Dutch, Italian and Portuguese. Second, as the team data was collected based on self-reported measurements, social desirability bias is possible. Third, the study coordinator distributed the questionnaires to the team members and could select in that way the participated team members. Last, no distinction could be made between missing data or care activities that were not performed. When something was not mentioned in the patient record, this was seen as not performed and this may lead to under performance of that care activity. The idea behind this is that when an action is not mentioned in the patient record, the other healthcare workers are not aware of this which can lead to duplication of actions (43).

This study generates new insight in how CPs might work, by linking both clinical and team measurements. Future research is needed to generalize these results for different pathologies. To compare the impact of a CP for different pathologies and between organizations, the context in which the CP was developed and implemented should also be measured (44).

Conclusion

In conclusion, this is the first study evaluating the impact of clinical processes and teamwork on the association of CP implementation and patient outcomes. Better adherence to recommendations fully mediated the association between CP and 30-day readmission. This finding suggests that organizations could measure adherence to recommendations during the development and implementation of a CP, as it can give information regarding the success of implementation and it can have an impact on patient outcomes.

Declaration of interest section

We acknowledge Pfizer SA who supported this research by providing an unrestricted educational grant. We also acknowledge the support of the Clinical Research Fund of University Hospitals Leuven, Belgium. The funder had no role in the design, data collection, analysis, interpretation of data, writing of the report, or decision to submit the report for publication.

Additional information

Funding

The European Pathway Association obtained an unrestricted education grant from Pfizer SA, and this study was partially funded by Clinical Research Fund of University Hospitals Leuven. The autonomy of E-P-A and all involved academic institutions with regard to scientific independence and intellectual property on the methodology was guaranteed.