WHICH QI INTERVENTIONS HAVE BEEN STUDIED IN CAP?
2. WHICH QI INTERVENTIONS HAVE BEEN STUDIED IN CAP?
There is little evidence to suggest that any of the specialist society guide- lines developed over the past decade have improved the management of CAP at a national level, although a number of studies have demonstrated that adher- ence with certain aspects of guidelines is associated with improved outcomes (Battleman et al., 2002; Gordon et al., 1996; Menendez et al., 2002). In the United Kingdom, for example, it is unlikely that the British Thoracic Society (BTS) guidelines will have a nationwide impact on the quality of CAP care until they are widely and appropriately implemented in primary and secondary care. Researchers have used a range of interventions, incorporating various aspects of specialist society guidelines, at a local or regional level, however, to try and improve the quality of CAP healthcare. The majority of these studies are from North America and many have been published in the health manage- ment or QI literature (e.g., Florida Medical Quality Assurance, Inc., 1998; Fortune et al., 1996; Gottlieb et al., 1996; Halley, 2000; McGarvey and Harper, 1993; Phillips and Crain, 1998; Reddy et al., 2001; Rollins et al., 1994; Ross et al., 1997; Sperry and Birdsall, 1994), which is generally inaccessible to
Pneumonia Guidelines in Practice
39 practising European clinicians. Details of the studies published in the English
language medical literature are summarised in Tables 1 and 2. The objective of most of these studies has been to optimise healthcare process while maintaining or improving clinical or economic outcomes. Only
a few studies have measured the impact of QI interventions on patient-centred outcomes and none have assessed cost-effectiveness. All appear to have used
a multidisciplinary development process and complex (multifaceted) interven- tions. While interventions and implementation strategies have varied, common threads are seen in the use of: (1) practice guidelines/protocols, care pathways, or decision support and (2) presentations to healthcare professionals. Research nurses have sometimes been used to support interventions, which is unlikely to
be sustainable in the long term. Audit and feedback, reminders and academic detailing (Solomon et al., 2001) have also been used in some studies. Unfortunately, these studies provide little data on the clinical effectiveness and costs of the single components of multifaceted interventions and there is little information therefore about which single interventions and implementa- tion strategies provide the best value for money. Additionally, the health policy implications of many of these studies are limited because of the common use of uncontrolled observational study designs. For example, the positive results seen in many studies could simply be due to regression to the mean or prevail- ing temporal trends (American College of Physicians–American Society of Internal Medicine, 1999), such as the known decreasing trend in the length of hospital stay for CAP patients in the United States (Metersky et al., 2000). Because of the predilection of authors and journals to publish positive data, there is also likely to be publication bias.
Particularly in North America, care pathways are commonly used interven- tions to try and optimise the management of CAP. Care pathways were first developed as an industry tool to identify and manage the rate limiting steps of industrial processes. Over the past two decades they have increasingly been used in healthcare, initially to improve the efficiency of resource use in man- aged care systems and more recently to improve the quality of healthcare (Pearson et al., 1995). The National Pathway Association (2001) of the United Kingdom defines a care pathway as:
An integrated care pathway determines locally agreed, multidisciplinary practice based on guidelines and evidence where available, for a specific patient/user group. It forms all or part of the clinical record, documents the care given and facilitates the evalua- tion of outcomes for continuous quality improvement.
The main distinguishing feature from guidelines is that care pathways inform the healthcare professional about an optimal timing and sequence of key healthcare events rather than purely instructing on, for example, the indications for ordering a specific test.
Table 1. A summary of controlled CAP QI studies published in the English language medical literature Study year (country)
Intervention and
Study design
(total sample size)
Marrie et al., 2000 (Canada)
1. Lower admission of low-risk patients in To evaluate the safety and
1. Educational plan to
Cluster randomised controlled
intervention hospitals (31% vs 49% effectiveness of a CAP
reinforce adherence
trial of 19 hospitals
2. Critical pathway using the (1743 patients)
P ⫽ 0.01)
critical pathway
PSI, levofloxacin, and
2. Mean bed-days per patient managed lower in
IV-oral switch and discharge
pathway arm (4.4 vs 6.1, P ⫽ 0.04)
criteria
3. No difference in patient quality of life
3. Study nurse prompts of switch/discharge criteria
Dean et al., 2001(USA)
1. 30-day mortality in hospitalised patients To assess the impact of a
1. ATS-based decision
Controlled before–after
decreased from 13.4% to 11% (pathway) vs pneumonia guideline on
support
(28661 patients)
13.2% to 14.2% (non-pathway) mortality
2. Presentations
3. Letters and reminders
(OR ⫽ 0.69, 0.49–0.97)
4. Academic detailing
2. Overall use of guideline antibiotics increased
5. Outpatient and admission
from 28% to 56%
order sheets 6. Feedback of outcomes
Chu et al., 2003 (USA)
1. Patients receiving antibiotics within 4 hr To demonstrate that an
1. Feedback of quality
Controlled before–after with
increased from 57% to 69% (P ⬍ 0.001) in the external QI project can lead
indicators
control hospitals subsequently
intervention group, but decreased in controls to an improvement in care
2. The use of either:
being exposed to the
(a) Clinical pathways or
intervention
2. Sputum and blood culture procurement
Gavin Bar
not accounted for by
within 4 hr also significantly improved temporal trends
(b) Standing orders
(2087 patients)
3. Following exposure to the intervention, care at control hospitals also significantly improved
lo w
Note: ATS—American Thoracic Society.
Pneumonia Guidelines in Pr
Table 2. A summary of uncontrolled CAP QI Studies published in the English language medical literature Study year (country)
Intervention and
Study design
(Total sample size)
1. Patients managed as outpatients increased from To safely increase the % of
Atlas et al., 1998 (USA)
1. Identification of low-risk
Uncontrolled before–after
patients (using PSI and
(313) with a PORT “control”
42% to 57% (P ⫽ 0.01)
patients managed as
2. Eight intervention patients were subsequently outpatients
nurse support)
cohort for patient satisfaction
2. Enhanced outpatient
hospitalised within 4 weeks compared to no
services
controls
3. Oral clarithromycin
3. Patients in the intervention cohort were less
4. Protocol presented to
satisfied with care (71% vs 90%, P ⫽ 0.04)
physicians
actice
Rhew et al., 1998 (USA)
1. No differences in quality of life (SF-36), To examine the effect of
1. Physician and nurse
Uncontrolled before–after
patient satisfaction or readmission to hospital physician decision support for
champions
2. Underpowered to detect clinically important IV-oral switch and discharge
2. Educational sessions
3. Provision of “real-time”
differences
decision support about the risk of IV-oral switch and discharge
Benenson et al., 1999 (USA)
1. Time to antibiotics decreased from 315 to 175 To determine if a clinical
1. Clinical pathway
Uncontrolled before–after with
and 171 minutes in the first and second post- pathway resulted in a shorter
(implementation strategy
two post-intervention cohorts
intervention periods, respectively (P ⬍ 0.0001) antibiotic delivery time and
not detailed)
2. Length of stay decreased from 9.7 to 8.9 (first) length of hospital stay and
and 6.4 (second) days (P ⬍ 0.0001) improved mortality
1. IV antibiotic duration decreased (6–2 mean days (Northern Ireland)
Al-Eidan et al., 2000
1. Protocol for inpatient
Uncontrolled before–after
LRTI care presented to
(P ⬍ 0.001)
To examine the impact of a
2. Treatment failure decreased, 31% to 8% (P ⬍ 0.001) community-acquired LRTI
medical staff
3. Mean length of stay decreased, 9 to 4.5 mean days treatment protocol
2. Distribution of a written
summary to medical staff
(P ⬍ 0.001)
3. Posting of the protocol
4. Total costs decreased from £2024 to £1020 per patient
algorithm on wards
(P ⬍ 0.001). The only UK-based study of
4. Pharmacy involvement
a LRTI protocol
Table 2. Continued Study year (country)
Intervention and
Study design
(Total sample size)
Suchyta et al., 2001; Dean et al.,
1. Hospital admission decreased from 14% to 6% 2000 (USA)
1. A practice guideline for
Uncontrolled before–after
CAP in an outpatient
(P ⫽ 0.02)
To study the use of a practice
2. Use of recommended antibiotics increased guideline in urban urgent care
setting
from 45% to 72% (P ⬍ 0.001) clinics
2. Use of 1993 ATS
guidelines
3. Decreases in mean antibiotic costs ($186 to
3. Presented to physicians
$141, P ⫽ 0.009) and direct costs per patient
and nurses
($678 to $319, P ⫽ 0.008)
4. Distributions of literature
4. Decreased length of stay (154–89 hr, P ⫽ 0.04)
and guideline forms
Meehan et al., 2001 (USA)
1. Antibiotics within 8 hr of admission To study the effects of
A statewide pathway with
Uncontrolled before–after
increased from 83% to 89% (P ⫽ 0.001) a statewide quality
locally customised
2. O 2 assessment within 24 hr of admission improvement project
implementation focusing on
rapid antibiotic delivery,
increased from 94% to 95% (P ⬍ 0.001)
blood culture collection, and
3. Mean length of stay decreased from 6.6 to 5.5 days
O 2 assessment
(P ⬍ 0.001)
Dobbin et al., 2001
1. No difference in length of stay, time to (Australia)
An antibiotic protocol
Retrospective cohort study
defervescence, and in-hospital mortality To compare outcomes
aimed at limiting use of 3 rd
comparing two hospitals
2. IV antibiotic cost per patient lower ($18 vs between hospitals using
generation cephalosporins
post-intervention (158)
$95) in penicillin-based protocol using hospital penicillin-based and
in CAP (implementation
strategy not detailed)
cephalosporin-based CAP antibiotic protocols
Gavin Bar
Lawrence et al., 2002 (USA)
1. Time to antibiotics decreased from 413–291 To improve time to
1. Protocol presented to
Uncontrolled before–after
physicians and nurses
minutes (P ⬍ 0.01)
antibiotics and sputum
2. Sputum procurement increased from 11% to procurement
2. Continuing education,
audit and feedback
25% (P ⫽ 0.06)
lo
Notes: ATS—American Thoracic Society; LRTI—Lower respiratory tract infection; PORT—Pneumonia Outcomes Research Team; PSI—Pneumonia Severity Index.
Pneumonia Guidelines in Practice
43 Often embedded in care pathways, or provided to healthcare professionals
in other formats, is decision support. There are now a number of evidence- based decision support “rules” to guide important clinical management deci- sions in CAP. These include severity prediction rules, for example, PSI (Fine et al., 1997), to guide the decision about an appropriate site of care (i.e., inpa- tient vs outpatient), criteria for admission to intensive care (Ewig et al., 1998), and markers of clinical stability for IV to oral switch and discharge from hospital (Halm et al., 2002). As demonstrated by their moderate sensitivity and specificity, however, none of these rules are perfect and they should all be used in combination with clinical judgement (Angus et al., 2002). Further details are provided in Table 3.