The effects of care bundles on patient outcomes: a systematic review and meta-analysis

One of the main purposes of health research is to optimise health and healthcare by identifying effective healthcare interventions. Nevertheless, health research will only improve patient outcomes if the findings of research can be implemented into practice (where this is warranted) [1‐3]. The translation of research findings into practice is a slow process [4]. Thus, a goal of implementation research is to improve patient outcomes by identifying the effective ways of translating research findings into practice [1]. Improving the quality of care and increasing research-informed practice has received much interest over the past decade [5‐8]. Research-informed practice requires healthcare professionals to work and think differently [9] because providing the evidence is necessary but alone is not sufficient [10, 11]. So, more recently, research has looked at how we might change the behaviour of healthcare workers to facilitate the uptake of research-informed practice in healthcare [12‐14].

To improve the quality of care and reduce the variations in care within intensive care units (ICUs), the Institute for Healthcare Improvement introduced the notion of care bundles [15, 16]. Care bundles contain three to five evidence-informed practices, which need to be delivered collectively and consistently with the aim of improving patient outcomes [16]. The Institute for Healthcare Improvement recommends that fidelity with care bundles should be at least 95% and every eligible patient should receive all of the elements included within the care bundle unless medically contraindicated [16]. Care bundles are used within healthcare for many different conditions (e.g. to prevent: ventilator-associated pneumonia, pressure ulcers). Whilst the elements of care within the care bundles formalise care, their success will be influenced by the implementation processes used to support the care bundle use in practice (e.g. shaping of knowledge, monitoring and feedback) [17]. Consequently, the behaviours of healthcare workers need to be targeted as part of the intervention [18].

Interventions aimed at changing health behaviours are often complex and comprise several components which have a synergistic effect [19]. Thus, care bundles are sometimes regarded as ‘complex interventions’ due to the number of components and their interaction within the care bundle; the context within which the care bundle is implemented; the number and variability of outcomes; the extent to which the care bundle can be tailored and the difficulty of performing the care bundle tasks. The Medical Research Council’s framework for developing and evaluating complex interventions recommends grounding complex interventions in theory to increase the likelihood of effectiveness [19]. Capitalising on behaviour change theory is important as the factors which influence the target behaviour; the active components of the intervention and the delivery of the intervention can be identified and selected [13].

Behaviour change techniques are the observable and replicable components of behaviour change interventions, often referred to as the ‘active ingredients’ [20‐22]. Previous studies reporting the use of behaviour change interventions have employed a number of different behaviour change techniques, but they have been defined differently or unclearly which limits the evaluation and replication of these interventions [23]. To address this issue, a taxonomy of 93 behaviour change techniques [22] was developed and can be used to identify intervention components, enabling the standardisation of terms as well as the comparison of behaviour change techniques across studies. Feedback on outcomes of behaviour, prompts/cues and instruction on how to perform a behaviour are examples of behaviour change techniques commonly used to facilitate a behaviour change in healthcare workers [17, 24]. Identifying the specific behaviour change techniques employed during the implementation of care bundles could enable researchers and healthcare workers to understand which components were key when the implementation of a care bundle was successful. Moreover, by using standardised behaviour change language, comparisons with other care bundles in implementation research will be possible. Such standardised language and comparisons will increase our knowledge of the most suitable methods for implementing care bundles and facilitate the prediction and explanation of any subsequent behaviour change [25].

To date, systematic reviews of care bundles have been condition [26‐35] or setting-specific [36‐38]. Very few systematic reviews have explored the common behaviour change techniques employed to facilitate the implementation of care bundles and it is unknown which factors may impact on the success of care bundles. Therefore, the objectives of this review were to evaluate the effects of care bundles as tools for reducing the number of negative patient outcomes, to identify potentially effective approaches to the implementation of care bundles and to explore whether there are plausible factors that modify the effects of care bundles (e.g. healthcare settings, fidelity with the bundle, the number of care bundle elements, different implementation techniques).

To maximise clarity and transparency, we have reported our review in line with the PRISMA statement for the reporting of systematic reviews and the meta-analyses of studies evaluating healthcare interventions [39].

The initial search generated 5796 records, and a total of 37 met the criteria for inclusion in the review (Fig. 1). The reasons for excluding records are stated in Fig. 1.

Existing systematic reviews of care bundles are condition or setting-specific and suggest that care bundles may be effective in preventing and managing a range of conditions such as sepsis [28], central line-associated bloodstream infections [30] and chronic obstructive pulmonary disease [26]. Others focussed on hospital settings [36, 38, 86]. Across all of the existing reviews, the certainty of the evidence was deemed to be low and the high risk of bias in the included studies continues to be reported, limiting the certainty of the conclusions about the effectiveness of care bundles.

It was difficult to assess the effect of fidelity to the care bundles on patient outcomes. Thirteen studies reported levels of fidelity with the care bundle. Levels of adherence varied between the studies suggesting that the full implementation of the elements of care included in the care bundles was rare. This is an important issue as three studies demonstrated fewer occurrences of the negative events (central line-associated bloodstream infections [62], mortality [73] and surgical site infections [60]) when fidelity with the care bundle was high. However, within the analysis, we were generally working with uncertain data, and review findings must be considered in line with the observational nature of subgroup analysis. As noted previously, the quality of the evidence is very low and therefore, we are uncertain whether there was an underlying effect of care bundles that is independent of these study characteristics.

A systematic review of 47 non-randomised studies [29], reporting the strategies used to facilitate the implementation of care bundles employed on ICUs, found the most frequently used strategies were audit and feedback, education and reminders. Unfortunately, the findings were inconclusive as implementation fidelity was rarely reported and the certainty of the evidence was assessed as being low. Thus, it was not possible to determine the most effective strategies used to improve the uptake of the care bundles. These findings are similar to those reported in this review and in a review of 14 studies (five controlled trials, two interrupted time series studies, seven controlled before-after studies) evaluating the effectiveness of chronic obstructive pulmonary disease discharge care bundles [26]. The poor reporting of the implementation fidelity issues may restrict the utility and reproducibility of the systematic review findings [87]. Thus, clear reporting of intervention components and of implementation fidelity are essential to the complete interpretation of data about the effectiveness of behaviour change interventions.

The lack of theory in the development and implementation of the care bundles was evident throughout the systematic review. Eight studies reported using an implementation framework or a psychological theory to guide their implementation [53, 59, 66, 74‐77, 83]. When encouraging healthcare workers to use evidence-based strategies, taking a theory-informed approach is recommended [19, 88]. However, often a pragmatic approach is taken, and this lack of explicit psychological theory during the design and implementation phases of the care bundle may impact on the effectiveness of such interventions [89‐91].

Mechanisms of action are the theoretical constructs through which behaviour change techniques have their effect. Explicitly stating the potential mechanisms of action (e.g. restructuring the environment, training) can facilitate the generalisation of the care bundle findings to other healthcare settings. The most commonly used behaviour change techniques were ‘feedback and monitoring’ and ‘shaping knowledge’. This is in line with previous findings on implementation strategies [37, 92, 93]. However, the frequency of these behaviour change techniques was often not reported, and neither were the levels of engagement with the behaviour change techniques (e.g. attendance at training sessions), or the mechanisms of action. Thus, conclusions regarding the effectiveness of using the behaviour change techniques to facilitate a change in the behaviours of healthcare workers were not possible.

Very low quality evidence from controlled before-after studies (downgraded due to the risk of bias, inconsistencies and potential indirectness of outcomes) suggests that the implementation of care bundles may be an effective strategy to improve patient outcomes when compared with usual care. By contrast, the low-quality evidence from five randomised trials (downgraded due to the risk of bias, inconsistencies and potential indirectness of outcomes) is highly uncertain. Future research should focus on the explicit and transparent reporting of the implementation of the care bundle including issues relating to implementation fidelity such as the frequency with which the behaviour change techniques were used. The higher quality reporting of the research findings will enable stronger conclusions to be drawn about the effectiveness of care bundles.