Organizational interventions to implement improvements in patient care: a structured review of reviews

We performed searches in Pubmed (1994–2003) and the Cochrane Library (accessed in January 2004). Older reviews were excluded, because their validity for current decision makers may be limited and we assumed that the reviewed studies would be included in later reviews. The search strategy in Pubmed combined the MeSH terms 'review literature' and 'meta-analysis' with the MeSH term 'healthcare quality'. The search in the Cochrane Library focused on reviews of the Cochrane Effective Practice and Organization of Care (EPOC) Group. We checked references in identified papers in previous overviews of systematic reviews in this field, which were themselves based on exhaustive systematic searches [5‐7]. Authors' personal literature collections were also examined; these were partly based on manual searching in health services research journals over the last 10 years. Only papers written in English were included. Our search was not designed to be comprehensive, but to provide a comprehensive overview of the available research evidence.

A review was included if it reported on its search strategy, if it focused (at least partly) on rigorous evaluations of organizational strategies (defined as planned re-arrangements of one or more aspects of the organization of patient care), and if it was published in 1995 or later. Rigorous evaluations comprised randomized trials, interrupted time-series, controlled before-and-after studies, and prospective comparative observational studies. Papers were included by the first author and the inclusion was checked by the second author. Some reviews also included studies on non-organizational strategies or non-rigorous studies; these sections in the reviews were not used. We did not include reallocation of services from hospital to primary care settings.

A taxonomy of organizational strategies to improve patient care was developed to organize the results (Table 1); this was consistent with other lists of organizational interventions such as that used by EPOC. Two researchers extracted from the reviews information on their focus, methodological quality, and main results. The number of studies reported in the table refers to the number of rigorous evaluations of organizational interventions; this may be lower than the total number of studies in the review as we focused only on rigorous studies of organizational interventions. Two authors independently assessed the methodological quality of the review, as opposed to the included studies, using a previously used 9-item scale (we did not use the summary assessment in the original scale) [7]. A score of 7 or more was taken to indicate good methodological quality. The summaries of the main results regarding effects on professional practice, patient outcomes and costs were based on the text in the original papers, derived from the abstract, results section or discussion section, focusing on quantitative summaries when available.

We expressed effects in terms of average effect size (AES), standardized mean difference (SMD), weighted mean difference (WMD), adjusted odds ratio's (AOR), adjusted relative risk (ARR), median net change (MNC) or percentage studies with improvements (PSI). Except for PSI all figures were extracted from the papers. If a meta-analysis had been performed, we also recorded whether the effect was significant (S) or not (NS). If quantitative summary measures of effectiveness were not used, the range of effects across studies was used. If this was not available, the authors' main qualitative conclusions were reported.

A total of 36 reviews were included [8‐44], of which 21 were of good methodological quality. The reviews with lower scores for methodological quality had not used optimal procedures for data-extraction and data-analysis. The studies were too heterogeneious regarding strategies and context factors to allow statistical pooling; furthermore, information on contextual factors was very limited.

Nine reviews focused on revision of professional roles, of which five were of good methodological quality. All focused on revised roles for non-physicians.

An older review identified 13 (quasi-) randomized trials, which compared nurse practitioners to physicians in primary care [13]. It found that quality of care, resolution of pathological conditions and functional status were not affected, while number of tests ordered and patient satisfaction increased. Similar findings were reported in a more recent review [19]. This latter review also reported that nurse practitioners had longer consultations, while prescriptions, return consultations and referrals did not differ. A review that focused on the effect of specialist nurses in diabetes care found that glycated hemoglobin was not different from usual care over a 12-month period [2]. Outreach nursing in patients with chronic obstructive pulmonary disease did not change patient outcomes, but it increased the use of hospital services [35].

A broad review on quality improvement in hospitals identified 13 studies on improvement strategies, which comprised the component of revision of professional roles [15]. This component significantly contributed to improved professional performance in a meta-regression analysis. A review on adult immunization and cancer screening found 20 trials, which included a component of organizational change – mainly designation of specific prevention responsibilities to nonphysician staff [37]. The meta-regression analysis showed that changing roles was one of the most effective intervention components in increasing use of the clinical and preventive services (compared to educational approaches, feedback and reminding strategies).

A review on enlarged roles of outpatient pharmacists (15 randomized trials, one controlled trial) showed that delivery of pharmacist services influenced the use of services, prescribing patterns, and patient outcomes [11]. Effects on costs were uncertain. Mental health workers replacing primary care providers did not consistently change psychotropic prescribing, consultation rates or mental health referrals [12]. There was some evidence that consultation with primary care providers by mental health workers had a direct effect on prescribing behavior when used as part of complex, multifaceted interventions [12]. A review of advice given by dietitians showed that dietitians did not affect blood cholesterol more than self-help resources [39].

Overall, it seems that revision of professional roles can improve professional performance, while positive effects on patient outcomes remain uncertain. Revision of roles seemed especially effective in preventive care, but the effect in relation to specialized nurses in primary care are still unresolved.

Five papers looked at studies on various interventions to enhance multidisciplinary collaboration, of which one was of good methodological quality [44].

In a review of palliative care teams four of the five randomized trials found that the co-coordinated specialist approach resulted in similar or improved outcomes in terms of patient and family satisfaction, anxiety, pain and symptom control [18]. Those studies that examined costs showed a tendency to reduce hospital days and equal or lower costs.

The involvement of a primary care practitioner in a specialist team was examined in a review, which identified seven randomized trials on programs for chronic or complex conditions [28]. While there were mixed effects for patient outcomes, they improved clinical performance of primary care providers, higher patient knowledge and higher patient satisfaction. Two studies examined costs, showing mixed effects. Only two randomized trials were identified in a review on interventions to promote collaboration between nurses and doctors [44]. These showed reduced hospital stay without change of mortality.

A review on multidisciplinary teams for congestive heart failure patients identified two randomized trials, which showed similar or improved outcomes [31]. Results regarding use of hospital care were inconsistent. The review on multidisciplinary teams for rheumatoid arthritis patients comprised 15 controlled trials (nine of which were randomized) [40]. The six trials of inpatient teams compared with regular outpatient care showed greater improvements in disease activity and in functional status immediately after treatment, which diminished over time. Five of the six trials on outpatient teams showed improvements on various patient outcomes compared with regular outpatient care.

Overall, it seems that multidisciplinary teams can improve patient outcomes. They have primarily been tested in highly prevalent chronic diseases.

Eight reviews focused on integrated services, of which five were of good methodological quality.

A review on stroke considered organized in-patient care, including both dedicated stroke units and mixed assessment/rehabilitation units. It included 19 trials (12 randomized), and showed favorable effects of stroke care [38]. A second review on stroke focused on in-hospital pathways, which were described as 'protocols for well-organized multidisciplinary care' [21]. It identified three randomized trials and seven other studies, which showed no differences regarding objective outcomes, but deterioration of patient reported outcomes.

A review on ambulatory patients with heart failure (11 randomized trials) found that these reduced hospitalization but not all-cause mortality [25]. The programs were cost saving in most studies that reported cost data. A review on secondary prevention programs in coronary heart disease (12 randomized trials) found largely the same results, although only three studies examined costs [26]. There were several studies that showed improved quality of life and functional status in patients from disease management groups.

A review of diabetes care showed improved glycated hemoglobin levels in both disease management (17 studies in a meta-analysis) and case management (11 studies) [29]. The improvement was similar when case management was delivered in addition. Disease management in rheumatoid arthritis had a small non-significant overall effect on functional status [8]. Longer programs or programs with more components were not consistently more effective.

A review on case management programs in primary care (nine randomized trials) focused on comprehensive programs and various conditions, including asthma, congestive heart failure, diabetes, and geriatric conditions [16]. Positive effects were found on patient-centered and clinical outcomes, but not on use of resources. The review on asthma [32] identified only one randomized trial, which showed some improvements in health outcomes.

An extensive review of controlled trials regarding disease management in chronic illness examined the effects of interventions, used within disease management programs [43]. The programs included a wide variety of interventions. While the interventions themselves were not organizational, only applications in the context of organized care for chronic illness were considered. It showed that both provider-directed interventions and patient interventions were associated with effects on provider adherence to guidelines and disease control.

Overall, integrated care systems can improve patient outcomes and save costs. They have been extensively tested in highly prevalent chronic conditions.

A broad range of computerized services was examined in six reviews, of which all but one were of good methodological quality. None of the reviews had a specific disease focus.

A large review on various computerized information services identified 100 randomized trials, mainly in outpatient care settings [9]. Some interventions focused on providers, such as reminders and computer-assisted treatment planning, while others focused on patients, such as computer-assisted interactive education and patient reminders. Most types of interventions showed positive effects, mainly related to specific a process of care, such as diagnostic test use, preventive services, and number of drug prescriptions. Ten of the fourteen studies that reported on patient outcomes found positive effects.

A later review by some of the same authors identified 40 randomized trials of computerized knowledge management in diabetes care [10]. It showed that computerized prompting (9 studies) led to improved overall guideline adherence. Meta-analysis of studies using home glucose records in insulin dose adjustment (16 studies) documented a decrease in glycated hemoglobin and a decrease in blood glucose. Several computerized patient-education programs improved diet and indicators regarding metabolic control.

Computerized physician order entry and clinical decision support systems were found to have effects on medication error rates and prescribing behaviors [21]. A review by Walton et al [12] focused particularly on computerized support for determining drug dose. It identified 23 comparative studies of which 16 were randomized trials. Seven of 11 studies on drug doses used found reductions, but the overall reduction was not significant in a meta-analysis. Six studies measured unwanted effects of drugs and four found significant reductions. Five of six studies on patient outcomes showed benefits. Only two studies considered costs and one study found cost savings, which resulted largely from reduced hospital stay.

A review on computerized record systems in primary care identified 61 studies, of which 39 randomized trials focused on professional performance and 11 randomized trials on patient outcomes [27]. Immunization rates improved in nine studies that focused mainly on reminder systems. Performance of preventive tasks improved. Four studies found improvements in diabetes management. A number of studies showed that computer support improved prescribing and reduced test ordering with implied cost savings. Use of computers increased the number of patients with reduced diastolic blood pressure in three studies, but did not consistently improve outcomes of anticoagulation therapy in two other studies. Five studies showed that consultation length increased.

A review on nursing care record systems identified eight trials, which suggested that documentation was improved but that process or outcomes of care were not influenced [14]. The reviewers concluded that no evidence was found regarding effects on performance attributable to changes in the record systems.

Overall, it seems that professional performance and patient outcomes can be improved by the implementation of computers in clinical practice settings.

Two reviews on quality management were found; both were of moderate methodological quality. A large review reported on 55 studies on the impact of continuous quality improvement, but only three were randomized trials [34]. Notably, these found no positive effects. A second review focused on nursing homes [41]. It identified four controlled trials of heterogeneous interventions (two of these appeared to comprise professional education). It concluded qualitatively that specific components of quality management were particularly effective, such as specific training, assessment procedures, quality assessment cycles and the assistance of a quality consultant. Overall, the effects of quality management on professional performance and patient outcomes remain uncertain.

Seven reviews, all but one of which were of good methodological quality, combined various organizational interventions (such as described in Table 1) into one group for the analysis and interpretation.

A comprehensive review on implementation of preventive services in primary care found four controlled trials on organizational interventions, such as involvement of nurses and a different way of booking appointments [20]. All showed intended positive effects. A review on improving breast cancer screening identified three randomized trials on change in office administrative systems, which all showed increased use of mammography screening [24]. Discharge planning prior to leaving hospital resulted in a small reduction in hospital length of stay for elderly medical patients, mixed effects on re-admission and no effects on patient outcomes [30]. A review on interventions to improve physicians' use of diagnostic tests found that 'enabling interventions' (administrative structure of test ordering) led to change in a majority of the studies if used alone and in most studies when used in combination with predisposing or reinforcing interventions [36].

A review on interventions to implement guidelines in hospitals found 15 trials, which included an organizational component (other than revision of professional roles) [15]. A meta-regression analysis showed that this component did not contribute to effects on process measures.

A review on interventions to improve the management of diabetes mellitus in primary care and outpatient settings identified nine trials [33]. These interventions focused on change in the medical record system, arrangements for follow-up, involvement of a pharmacist, and multidisciplinary collaboration. The authors conclude that regular prompted recall and review of patients improve diabetes management. Higher treatment adherence and patient recovery, and lower costs, were achieved in patients with depression by "collaborative care", a comprehensive package of interventions that included educational and organizational strategies [17].