Background

Growing expenditure on healthcare and ongoing efforts to improve services give impetus to change in processes and systems [1]. As life expectancy increases, so does chronic disease, which is associated with a greater demand for multidisciplinary care [2, 3]. At the same time, public outlay on healthcare has decreased, inducing potential shortages of healthcare providers [3]. Long-term implications for the quality of care are unclear and should be carefully monitored [3]. According to the Institute of Medicine (IoM), patients do not always receive the most suitable care, at the best time or the best place [2]. Its influential report ‘Crossing the Quality Chasm: A New Health System for the 21st Century’ emphasized the need to redesign healthcare processes and systems in response to this quality gap. It called upon providers to ensure more efficient, safe, timely, effective, patient-centered and equitable care [2, 4].

Although some initiatives were undertaken before 2001, the publication of the IoM report served as a catalyst [2, 5]. Numerous interventions – disease management programs for the chronically ill, quality improvement collaboratives, and change programs – are tested and implemented annually on different scales and within different settings [5]. Nonetheless, progress is slow; evaluations of initiatives are inconsistent and available knowledge fragmented [5]. The effects are not homogeneous and the research designs used to measure them are generally weak [4, 6, 7].

This study seeks to establish, through a review of the literature, what is known about the influence of redesigning healthcare processes on the quality of care delivered in the last ten years. Its specific aims are to report (a) the content of the interventions (their objectives and implementation methods); (b) the characteristics of the redesign investigations (study design and setting); and (c) the outcomes on quality of care (patient safety, effectiveness, efficiency, patient-centeredness, timeliness, and equitability). The objective of this literature review is to summarize the current state of knowledge on redesigning healthcare processes and present an overview of improvement efforts in the field.

The review applies several key concepts. The first is ‘process redesign’, defined as any methodology that focuses on creating new processes or changing existing ones in major ways [8]. That definition is deliberately broad so as to cover as many interventions as possible; recourse to dedicated design concepts – such as ‘lean thinking’, ‘business process re-engineering’ or ‘six sigma’ – might exclude relevant studies. The second is ‘quality of care’, connoting healthcare that is safe, effective, patient-centered, timely, efficient and equitable [2]. The third is ‘healthcare processes’, defined as “the activities that constitute healthcare – including diagnosis, treatment, rehabilitation, prevention, and patient education – usually carried out by professional personnel, but also including other contributions to care, particularly by patients and their families”([9], p. 46).

Methods

Information sources and search strategy

The search strategy was guided by the PRISMA statement [10]. It was designed to access published work and comprised two stages:

  1. 1.

    An extensive search in Pubmed, CINAHL, Business Source Premier and Web of Science, using predefined search terms and free-text words;

  2. 2.

    A search of the reference lists in the included full-text articles.

From March 2014 through April 2014, the databases PubMed, CINAHL, Web of Science and Business Premier Source (EBSCO-host) were searched by one reviewer (JvL). In PubMed, MeSH terms were used; CINAHL Heading terms were used for CINAHL; and Thesaurus terms were used for Business Premier Source. For Web of Science no predefined keywords were available. Additionally, free-text words were used for all databases. An overview of the search terms is given in Appendix 1.

The database search was limited to articles published in English between January 2004 and April 2014. Articles were included if they presented original research on redesign of healthcare processes, quality of care, and if they assessed the same outcome measures before and after an intervention. (See Table 1 for inclusion and exclusion criteria). Three reviewers (JvL, KG & AE) independently screened titles and abstracts for relevance. The reviewers then held a consensus meeting on the inclusion of articles. When that did not yield agreement, the full text was reviewed and discussed to arrive at a decision. Subsequently, reference lists and bibliographies of all included full-text articles from the first stage were searched for additional studies.

Table 1 Inclusion and exclusion criteria
Full size table

Critical appraisal

Studies meeting the criteria were assessed independently for reporting excellence by three reviewers (JvL, AE & KG), prior to inclusion in light of the Standards for Quality Improvement Reporting Excellence (SQUIRE). That checklist provides guidelines for reporting of studies assessing the effectiveness of interventions to improve quality and safety of care. Its 19 items comprise 38 components [11]. Any disagreements between reviewers were resolved through consensus.

Data extraction and analysis

After compliance with the reporting guidelines had been assessed, data were extracted independently by three reviewers (JvL, KG & AE) from the results and discussion/conclusion sections. For that purpose, a form was developed. The form contained variables such as publication year, study objectives, characteristics of the redesign and outcome measures. Any disagreements were resolved through consensus. Meta-analysis could not be performed because the studies used different outcome measures and research designs.

Results

Figure 1 shows the steps leading to inclusion in the review. Initially, after removing duplicates (N = 27), 451 articles were found in the first stage, 11 of which were then included on the basis of their titles and abstracts. Perusal of their reference lists yielded another 24 articles for screening of title and abstract. Based on titles and abstracts, 21 articles were assessed for eligibility. On eight of these, consensus was only reached after reviewing the full text. After assessing the reporting excellence, three articles were excluded. One was removed because it did not describe data collection and timepoints, so it could not be determined whether a before-and-after measurement was performed. Another was removed because it was unclear whether it concerned original research; moreover, the main intervention (presence of a nurse coordinator) did not qualify as process redesign. The third was removed because it was unclear whether the intervention was actually implemented and whether before-and-after measurement was carried out but also because the outcome measures differed at various timepoints. In total, 18 articles were included in the final review.

Fig. 1
figure 1

Search strategy

Full size image

Reporting excellence

Table 2 summarizes the findings according to SQUIRE guidelines. The number of components described range from 11 [12] to 27 [13], with most articles reporting on 20 or more [13–22]. Overall, methods of evaluation and analysis are the least well described. The majority described the research setting (N = 16) [12–27], intervention components and parts (N = 16) [13–16, 18–28], main factors in the choice of intervention (N = 15) [11, 13–18, 20, 22–28], and primary and secondary outcomes (N = 15) [12–14, 16–24, 28, 29]. Thirteen articles presented evidence on the strength of the association between the intervention and changes observed (N = 13) [12, 13, 16–22, 24, 25, 27–29]. Half gave details on the qualitative and quantitative methods applied (N = 9) [13, 17–20, 24, 25, 28, 29] or aligned the unit of analysis with the intervention (N = 9) [13–15, 18–21, 24, 28]. Six described internal and external validity [13, 15, 17–20, 28], whereas two dealt with the validity and reliability of instruments [17, 28]. Whereas none of the articles explicitly stated the study questions, all of them specified the aims of the intervention. Most data concerned changes observed in the care delivery process (N = 12) [13–16, 18, 21–24, 26, 28, 29] or differences in patient outcomes (N = 12) [13, 16–24, 28, 29].

Table 2 Overview of reporting excellence according to the SQUIRE guidelines
Full size table

Types of redesign interventions

Table 3 summarizes the redesign interventions and study methods used. The objective of most studies was the implementation and evaluation of a specific redesign intervention. Improving quality of care was explicitly stated as an objective in seven studies [12, 15, 18, 20, 23, 25, 26]. Half of the redesign interventions implemented the approach known as lean thinking/Toyota production system (N = 9) [12, 14, 15, 21, 24–28]. Two studies described the implementation of the concept of patient-centered medical home [17, 20], and three described more general forms of process redesign (structure redesign vs. process redesign [23], evidence-based redesign [18], nurse practitioner-led practice redesign [29]). Other interventions included a general process improvement project [16], appreciative inquiry [22], a hospitalist-led co-management neurosurgery service [13] and a continuum of care [19].

Table 3 Overview of types of redesign interventions and methods used in included studies
Full size table

Fourteen studies were performed in the USA [12, 13, 15–17, 19–22, 25–29], two in Australia [14, 24], one in South Korea [23] and one in Scotland [18]. Most took place in a hospital setting (N = 12) [13–16, 19, 21–24, 27–29]; others were conducted in primary care (N = 3) [12, 17, 20], a specialized clinic (N = 1) [18] or a laboratory (N = 2) [25, 26]. Length of follow-up ranged from three [18] to 48 [27] months with a median of 12 months, though five studies did not mention its duration [12, 14, 15, 26, 29]. Patients were the most common unit of analysis (N = 14) [13–15, 17, 18, 20–25, 27–29]. However, some studies reported on staff (N = 2) [12, 21] or clinical notes (N = 1) [12] while a few did not define the unit of analysis (N = 3) [16, 19, 26]. Mean sample size was 27,932.87(SD = 61,506.98), ranging from 49 [21] to 228,510 [20]. Thirteen studies used a before-and-after design (N = 12) [12, 14–16, 20–24, 27–29], while five used a controlled before-and-after design [13, 17, 19, 25, 26].

In summary, half of the redesign interventions were characterized as ‘lean thinking’ and took place in a hospital setting. Length of follow-up and sample size diverged widely, and most studies used an uncontrolled before-and-after design to evaluate the effectiveness of the intervention.

Effects of redesign on quality of care

Table 4 summarizes the outcomes of the studies. All reported improvements as a result of process redesign, while three [14, 20, 23] also found declines in quality. Significant improvements were mentioned in 15 studies [13, 14, 16–21, 23–28], mostly gains in effectiveness [16–21, 25, 27] and/or efficiency [14, 17–20, 23, 24, 26, 28]. Outcome measures showed great variance between studies. However, ‘effectiveness’ and ‘efficiency’ were discussed most (11 studies reported on both dimensions [13, 14, 16–22, 25, 29]). Changes in efficiency were demonstrated by 17 studies [12–25, 28, 29]. Efficiency was improved by decreasing hospitalization rates [17, 20], process times (including time to treatment) [14, 23, 24, 28], length of hospital stay [19, 23, 29]; by a shift in the writing of clinical notes [12], savings on (estimated) costs [13, 16, 19, 20, 25, 28], raising provider productivity [21, 22, 26] and reducing process steps and variability [15, 18, 24, 25]. Efficiency also deteriorated: an increase was shown in process time for a sub-category of patients [14, 23], in specialty care visits [20] and in specialty care costs [20].

Table 4 Overview of outcomes of redesign interventions in included studies
Full size table

Changes in effectiveness were demonstrated in 12 studies [13, 14, 16–22, 25, 27, 29]. These reported improvements in disease conditions [17, 20, 29] and adequate treatment usage [16, 22, 29] as well as increases in discharged patients [14, 18] and diagnostic accuracy [25, 27].

Two studies [14, 15] found changes in timeliness as a result of process redesign, which reduced waiting time. Changes in patient-centeredness were demonstrated in three studies [13, 20, 22]: improvements in patient satisfaction or experiences [13, 20, 22]; higher scores on doctor-patient interaction; and better coordination of care [20]. Changes in patient safety were found in 11 studies [12, 14–16, 18, 19, 21, 24, 25, 27, 29]: increased physician identification [12]; improved documentation [12]; a decrease in complications [14, 16, 19, 21, 29]; fewer errors in routing patients to appointments [15]; fewer false-negative diagnoses [25, 27]; and an overall sense of improvement in patient safety [24].

None of the studies measured equity of care. Eight mentioned other outcomes unrelated to the six quality dimensions, such as changes in provider satisfaction [12, 22], staff perceptions of the implemented change [13, 14, 18, 21], changes in team morale [28], or changes in incident rates [18].

Discussion

The need to redesign healthcare processes in order to address deficits in quality of care and create more sustainable care processes is acknowledged worldwide [2, 3, 5]. The effects of process redesign have not been clearly described, however [5, 6]. By synthesizing evidence from 18 studies in the international literature, this systematic review contributes to a better understanding of the influence of process redesign interventions on quality of care. It suggests that they have positive effects on certain aspects of quality. However, the full impact cannot be determined on the basis of the literature. Studies differed in the type of redesign implemented, study setting, methods used for evaluation, and outcome measures. All types of intervention seemed to improve outcomes in one or more respects. Nonetheless, it is not clear which type of redesign has the most potential in a particular setting. Efficiency, effectiveness and patient safety gains were best described in the included studies, while the effects on patient-centeredness, timeliness and equity of care received little attention.

Applying the SQUIRE guidelines demonstrated that overall the reporting was weak. Given the study designs, the results are subject to bias, as changes in the research settings might be responsible for the effects [30, 31]. In addition, changes in process might have been induced by background factors [31]. Longitudinal effects of redesign interventions were hardly evaluated, as follow-up varied from three to 48 months with a median of 12 months. The methodological problems of studies reporting on quality improvement interventions like process redesign are well known [6, 31–34]. Yet the methodology of the studies covered here was no better than in preceding studies. These weaknesses form potential threats to the internal and external validity of the findings. Unless a more uniform and robust evaluation of process redesign interventions is carried out, general conclusions cannot be drawn about their impact on quality of care.

To the best of our knowledge, this is the first systematic review of the effect of process redesign on quality of care, using broad definitions for both study setting and types of redesign. Elkhuizen et al. [6] performed a systematic review of the evidence of business process redesign in hospital settings until 2004. However, that review included studies combining multiple interventions, which made comparison impossible. Those authors concluded that studies were hard to find and lacked a clear and consistent research methodology. In that light, they recommended the development of reporting guidelines.

Specific redesign interventions have been reviewed recently. In one, Mazzocato et al. [35] reviewed the ‘lean-thinking’ literature from a realist perspective, focusing on the mechanisms through which ‘lean thinking’ operated. The authors identified positive effects of lean implementation in all included studies and common contextual factors interacting with components of the lean interventions that triggered the change mechanisms. Here too, the use of unclear study designs or outcome measures is mentioned. The authors suspect publication bias, as only positive effects were being reported.

The impact of quality-improvement collaboratives was reviewed by Schouten et al. [36]. Although the outcomes were positive, the strength of evidence was limited by methodological constraints due to weak study designs, and the authors suspect positively biased findings. Implementation of the concept ‘patient-centered medical home’ was reviewed by Jackson et al. [37], who showed small positive effects on patient experience and care delivery. There too, the strength of evidence was moderate to low. Publications were hard to find, evidence was fragmented, and comparison between studies was hard if not impossible.

The findings of the present review are therefore in line with those of earlier studies on this topic in the sense that a broad perspective on redesign interventions and settings generates similar results.

Limitations

Even though a systematic approach guided this review, the findings might be subject to some bias, which should be kept in mind when interpreting them.

First, publication bias might be present: most of the studies report on positive findings, and there is a general tendency in scientific literature to over-represent positive results [38]. As previous research on this topic also raised concerns about publication bias, this issue is pertinent to this review too. It is unlikely that using predefined redesign concepts would have addressed this problem, as publication bias was a concern in reviews that did use such concepts [35], underlining the need to report all outcomes of redesign in healthcare.

Second, limiting the scope by only including studies that used before-and-after measurement might have led to some selection bias. Nonetheless, limiting the search strategy did ensure a solid basis for comparison of the effects of the redesign interventions.

Third, since the terminology used to describe the interventions varies greatly, we could have missed some relevant studies. We circumvented this problem by searching multiple databases with database-specific headings like MeSH terms and amplifying the strategy by searching with free-text words.

Fourth, the SQUIRE guidelines might not be the only instrument for assessing excellence in reporting. Although they were specifically developed to assess reporting excellence for this type of studies, the checklist does not provide a value judgment on the methodology (or strength of evidence) of the studies [11]. Nonetheless, by covering methodological components, the SQUIRE checklist gives a sense of the methodological strengths of a study.

Finally, using the IoM dimensions of quality of care might have made it difficult to compare findings across studies. Since the IoM does not specify which outcome measures belong to the six dimensions, there is room for interpretation. Even though this might have influenced the presentation of findings in this review, using the IoM dimensions facilitated classification of the outcomes, thereby revealing gaps in the research literature.

Conclusion

Scientific evidence supporting process redesign in healthcare is limited and inconsistent. Outcome measures for the effect of redesign interventions vary across studies to the extent that it is impossible to draw conclusions about the impact on overall quality of care, or even on some of its dimensions. The findings of this systematic review suggest that the evaluation of process redesign interventions should be improved to reveal their full effect. It should meet the basic standards for reporting (SQUIRE guidelines) and apply more robust research designs. The influence of process redesign on patient-centered care, equity of care and timeliness warrants further research, applying outcome measures that capture the full scope of quality of care. Current research tends to ignore the long-term effects of process redesigns. Robust evaluations of their implementation should also identify the mechanisms through which effects were realized. This would help researchers and policymakers determine the value of specific interventions and offer an overview of improvement efforts that is less fragmented.