Reliable Implementation of Clinical Pathways: What Will It Take—That Is the Question
Article Outline
Several studies of inpatient and emergency department (ED) implementation of care pathways for the management of acute asthma in children have demonstrated positive effects on length-of-stay, costs, misuse of therapies such as intravenous corticosteroids and oral albuterol, hospitalization and readmission rates, overuse of laboratory tests, and underuse of therapies such as oral corticosteroids, controller medications, and structured education.1, 2, 3, 4, 5
See related article, p 315
Cunningham et al have provided further rationale for the use of evidence-based, clinical pathways for management of an acute asthma exacerbation.6 In particular, their 30% reduction in prescribing errors within the group of patients treated with the integrated care pathway (ICP) demonstrates that standardization of practice through a pathway can be a key safety strategy in complex medical systems. Although the authors note that when stratified by severity, the decrease in errors came only from the mild stratum, the study was not powered to detect differences in less frequent, more severe errors. Furthermore, the authors document improvements in the use of evidence-based therapies. The ICP increased prescribing of oral corticosteroids for asthma exacerbations at discharge from the ED by 29%, with a number needed to treat of 5.5. Patients discharged after hospital admission for asthma received a 48-hour action plan twice as often when cared for using the ICP, with a corresponding number needed to treat of 2.
Results for the primary outcome, length-of-stay (LOS), estimated a decrease, but did not reach statistical significance. We believe there are several important points to draw from these negative findings, but do not believe this study should be interpreted as supporting the null hypothesis. Earlier studies of care pathways and effect on LOS have found, as the authors note, decreases between 25% and 50%. However, in these studies, the baseline LOS was 3 to 4 days, with the care pathways producing a LOS between 1.5 and 2.5 days.1, 2, 3 In the Cunningham study, the baseline LOS in the control group was 41 hours, and we agree with the author’s inference that this striking difference in baseline LOS suggests enough difference between the inpatient system they studied in Scotland and the US systems studied by previous authors that assuming a similar relative reduction in LOS (the authors chose a conservative estimate of 20%) may have led to an under-powered study. Qualitatively, in each of these studies, one key aspect of the care pathway is a mechanism by which patients can be discharged by nurses, on the basis of predetermined criteria, without a physician returning to the bedside to directly verify that the criteria have been met. In the Cunningham “per protocol” analysis (ie, in those patients with a completed and documented discharge checklist) the reduction in LOS was 15% (P = .07), double the estimate from the intention-to-treat analysis. Our interpretation, then, is that the Cunningham study, although under-powered to statistically confirm an effect, provides further evidence that care pathways reduce LOS.
Most important, however, in the attempt to address methodologic critiques of past studies of care pathways, the authors’ choice of methodology causes a practical bias toward the null hypothesis and is an example of misapplication of “traditional” methods of design and analysis to the study of complex systems. Specifically, the authors defend the choice of a cluster-randomized design, in which the ICP intervention was applied in an on-again, off-again manner to the ED and inpatient system, to avoid the past critique of designs in which patients are randomized to different wards (eg, intervention versus control ward). The critique in those cases is that any treatment effect of the intervention may actually result from, in the authors’ words, “different skill mixes” of the different wards. A cluster-randomized study of different centers might be the gold standard to address that concern. In this case, by having one ward turn on and off the ICP for 28 weeks, the authors’ systematically removed one of the prime mechanisms by which an ICP works—allowing the delivery system to become better and better at what it does. The authors’ analysis actually demonstrates this. To address the concern of contamination (ie, that ICP-related behaviors were adopted with time during control weeks) the authors performed a treatment-by-time analysis and showed no interaction. Although this may show that the LOS in the control group did not decrease during the course of the study (ie, no contamination effect), it also showed that LOS in the intervention group did not decrease with time either. Much like a transplant-surgery team should be more efficient and effective after their 100th operation together compared with after their first, an inpatient care delivery system should improve with time also. We believe that Cunningham’s methodology, in which the delivery system could not effectively adapt to the ICP because it was applied intermittently, prohibited or slowed such adaptation, thereby biasing results toward no effect on LOS. In this light, then, we feel the important effects of the asthma ICP that were demonstrated by Cunningham—prescribing safety and less under-use of oral steroids and education plans—may be underestimates.
Finally, we propose that investigators interested in clinical pathways turn to harder questions. Our view is that the evidence base is clear—clinical pathways are effective, and to improve the quality, safety, and equity of the health care delivery system standardization and clinical pathways are an essential starting point. In negative trials of guideline or pathway implementation, one common feature is incomplete or uneven implementation.7, 8 Future trials of clinical pathways should test how to best implement and maintain pathways and explore how to best adapt them to the match the needs of individual patients. In this way, we believe the care a patient receives will vary on the basis of his or her individual needs, rather than the habits and opinions of his or her provider.
References
- . Impact of a pediatric asthma clinical pathway on hospital cost and length of stay. Pediatr Pulmonol. 2001;32:211–216
- Improved outcomes for hospitalized asthmatic children using a clinical pathway. Ann Allergy Asthma Immunol. 2000;84:509–516
- . Effectiveness of a clinical pathway for inpatient asthma management. Pediatrics. 2000;106:1006–1012
- . Effect of a clinical pathway on the hospitalization rates of children with asthma: a prospective study. Arch Dis Child. 2007;92:60–66
- . Structured discharge procedure for children admitted to hospital with acute asthma: a randomized controlled trial of nursing practice. Acta Paediatr. 2005;94:226–233
- . Effect of an integrated care pathway on acute asthma/wheeze in children attending hospital: cluster randomized trial. J Pediatr. 2007;152:315–320
- Effect of computerized evidence based guidelines on management of asthma and angina in adults in primary care: cluster randomized controlled trial. BMJ. 2002;325:941–948
- A randomized controlled trial of an asthma clinical pathway for children in general practice. Acta Paediatrica. 2005;94:226–233
PII: S0022-3476(07)01163-8
doi:10.1016/j.jpeds.2007.12.017
© 2008 Mosby, Inc. All rights reserved.
Refers to article:
- Effect of an Integrated Care Pathway on Acute Asthma/Wheeze in Children Attending Hospital: Cluster Randomized Trial , 19 November 2007
