WHAT IS THE EVIDENCE THAT QI INITIATIVES IMPROVE PROCESS OF CARE IN CAP?

4. WHAT IS THE EVIDENCE THAT QI INITIATIVES IMPROVE PROCESS OF CARE IN CAP?

Process of care measures may be more pragmatic endpoints than “tradi- tional” outcomes (e.g., mortality) in CAP research or QI. Process measures, however, should be linked to clinical, patient-centred, or economic outcomes. This section will review the evidence that QI interventions improve process of care in CAP.

4.1. Aspects of process that are likely to impact on clinical and patient-centred outcomes

In a cleverly designed controlled before–after study, Chu et al. (2003) showed that feedback from an external QI agent resulted in a higher proportion of patients receiving antibiotics within 4 hr of admission at intervention com- pared to control hospitals (69% vs 57%). This change appears to have been mediated through the use of either clinical pathways or standing orders. When control hospitals were subsequently exposed to the intervention, delivery of antibiotics at these sites also improved (66% vs 53%). This evidence is sup- ported by three uncontrolled studies. Benenson et al. (1999) used a before– after study with the use of two post-implementation cohorts to evaluate the impact of a care pathway that emphasised the importance of the early diagno- sis and delivery of antibiotic therapy. Door to antibiotic time decreased from 315 to 175 min and 171 min at 1 and 3 years post-implementation, respec- tively. Although uncontrolled, this is an interesting finding as there has been concern about the sustainability of QI interventions. Meehan et al. (2001) found that a locally customised statewide care pathway improved the delivery of antibiotics within 8 hr of admission to hospital from 83% to 89%. Lawrence et al. (2002) used an educational intervention with monthly audit and feedback to reduce door to antibiotic time from 413 to 291 min. In some of these studies (Chu et al., 2003; Lawrence et al., 2002), sputum and blood culture procure- ment have also increased. As discussed above, however, it is debatable if these processes of care have an impact on outcomes.

Pneumonia Guidelines in Practice

49 In another uncontrolled study, Suchyta et al. (2001) used a practice guide-

line incorporating the American Thoracic Society’s (ATS’s) antibiotic guidance to increase the use of recommended antibiotics from 45% to 72%. Using a wide range of interventions and in the largest published controlled CAP QI study to date; Dean et al. showed an improvement in overall use of guideline antibi- otics in study hospitals from 28% to 56% (Dean et al., 2001). In the only UK study, Al-Eidan et al. (2000) described a reduction in the diversity of antibiotics used in CAP (from 12 to 3) following implementation of a practice guideline.

4.2. Aspects of process that are likely to impact on economic outcomes

In an uncontrolled before–after study, Atlas et al. (1998) found that the use of the PSI as admission decision support, increased low-risk patients managed as outpatients from 42% to 57%. The intervention included a study nurse during weekday working hours, enhanced outpatient services, and the use of oral clari- thromycin. These findings were subsequently confirmed in the only randomised CAP QI trial. A care pathway, which incorporated the PSI with levofloxacin, reduced the proportion of low-risk patients admitted to hospital from 49% to 31% without impacting on patients’ quality of life or clinical outcomes (Marrie et al., 2000). Using an adapted version of the ATS admission guidance, Suchyta et al. (2001) also safely reduced hospital admission from 14% to 6%.

An early switch from IV to oral therapy and subsequent discharge from hospital has been shown to be a safe and cost-effective strategy for CAP in randomised controlled trials (Castro-Guardiola et al., 2001; Paladino et al., 2002). Only a few studies have evaluated this strategy, however, as part of a QI intervention. Al-Eidan et al. (2000) demonstrated a reduction in IV antibiotic administration from 6 to 2 mean days. In contrast, Rhew et al. (1998) were unable to demonstrate a significant change in guideline adherence following the implementation of physician decision support for this aspect of care. This study was underpowered, however, to detect clinically important changes. In another small study (not shown in the tables), Weingarten et al. (1996) found similar results.