Improving Quality of Care in Low- and Middle-Income Countries Workshop Summary (2015) / Chapter Skim
Currently Skimming:
4 Reviewing the Evidence for Different Quality Improvement Methods
4 Reviewing the Evidence for Different Quality Improvement Methods
Pages 35-60
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 35... ...
• Monitoring quality of care in a country depends on accurate clinical and death registries and the ability to link patient data across regis tries. (Klazinga)
|
From page 36... ...
WHAT WORKS? : THE RESULTS OF A SYSTEMATIC REVIEW Alexander Rowe of the CDC presented preliminary results of the Health Care Provider Performance Review, a systematic review of quality improvement methods to change the behavior of health care providers (hereafter, providers)
|
From page 37... ...
As the m database had no studies specifically involving COPE® or SBM-R, Rowe TABLE 4-1 Definitions of the Six Strategies Using Component Variables from the Health Care Provider Performance Review (HCPPR) Database Strategy Definition High-intensity Training >5 days (or ongoing training)
|
From page 38... ...
. In this system, risk of bias is a function of study design, the number of clusters in each arm, data completeness, between-group comparability at baseline, the outcome's reliability, concealment of allocation for studies randomized at the patient level, the likelihood that the intervention could change data collection, and having fewer than six data points before or after an intervention for interrupted time series.
|
From page 39... ...
As about half of the effect sizes were from studies with a public facility–only setting, the adjustment subtracted about 4 percentage points from effect sizes of studies with a public facility–only setting and added 4 percentage points to effect sizes from other settings. For strategies that appeared to have the greatest effectiveness, the analysis checked for confounding by limited variability, the chance that the observed effect came from an idiosyncrasy of study design -- that is, a setting unusually well suited to a strategy.
|
From page 40... ...
Table 4-2 shows the number of comparisons in Rowe's database on the six strategies discussed at the workshop. After removing the strategy groups not mentioned in the database, only 11 percent (n = 7)
|
From page 41... ...
Almost half of the studies were randomized TABLE 4-3 Breakdown of the Risk of Bias in the Strategy Studies Risk of bias Number of Low/ High/ Strategy comparisons Moderate Very high High-intensity training only 9 4 5 Low-intensity training only 36 15 21 Supervision only 7 3 4 Improvement collaborative only 7 0 7 High-intensity training + supervision 4 0 4 Low-intensity training + supervision 5 2 3 Low-intensity training + improvement collaborative 3 0 3 SOURCE: Rowe, 2015. FIGURE 4-1 Breakdown of the strategy studies by region.
|
From page 42... ...
. Figure 4-2 shows the weighted median adjusted median effect sizes (MES)
|
From page 43... ...
The strategies with the highest 4 Afterthe discussion, participants asked if broadening the definition of a strategy changed the effect estimates for other strategies. Rowe said they have observed something similar, though not on the same scale, in the larger database when considering the group-based problem solving and training with supervision strategies.
|
From page 44... ...
SOURCE: Rowe, 2015. FIGURE 4-2 Weighted median adjusted median effect sizes (MES)
|
From page 45... ...
Improvement collaborative + low-intensity training Broadened definition 6 11 (6, 60) Group problem solving + low-intensity training +/− other components SOURCE: Rowe, 2015.
|
From page 46... ...
Similarly, improvement collaboratives showed somewhat larger effect estimates, but such results should be interpreted cautiously because of the high risk of bias and limited variability in the data. Rowe concluded his comments with a request for more studies of rigorous design.
|
From page 47... ...
Then in 2005, I-TECH, a health systems development organization run by the University of Washington and the University of California, San Francisco, with sup port from PEPFAR and the CDC, developed iSanté, an electronic medical records system that supports both individual and population health. iSanté gave clinicians a way to manage longitudinal data and to make data easily accessible to the ministry.
|
From page 48... ...
Agins chose one indicator, the enrollment of eligible patients on antiretroviral therapy (ART) , and shared it as an example of how electronic medical records can be used to improve quality.
|
From page 49... ...
Haiti's national quality advisory board also made use of the electronic medical records system. The real-time data allowed them to set goals, measure their performance, and give feedback to the district and local health offices about their progress.
|
From page 50... ...
Many countries in this region are trying to measure more than the minimum (e.g., infant mortality and vaccination rates) and therefore are investing in improved cancer registries, death registries, and electronic administrative databases; the growing momentum for universal coverage has driven some of these improvements.
|
From page 51... ...
Klazinga recommended that the first priority for low- and middle-income countries be developing clinical registries and administrative databases. Over the past 5 years, the OECD quality experts have been asked to analyze the national quality strategies for a growing list of the nonOECD countries.
|
From page 52... ...
There is an opportunity to use these systems -- all at different stages of development -- to build national quality improvement programs, though Tierney expressed concern that connections between the new electronic systems and quality managers were somewhat haphazard. Mike English built on this point, saying that the electronic medical records system he worked with in Kenya is not as sophisticated as the example Agins shared from Haiti.
|
From page 53... ...
The problem could be avoided in low- and middle-income countries with attention to computerized information systems in primary care. DHIS 2, a health information system used in 47 low- and middle-income countries and 23 international organizations, has the ability to link hospital and primary care data and to control the burden of data collection on health workers (DHIS 2, n.d.)
|
From page 54... ...
54 FIGURE 4-5 The possible outcomes accounted for in a hypothetical cost-effectiveness analysis of a quality improvement program in a Liberian district health center. SOURCE: Broughton, 2015.
|
From page 55... ...
In seven of the studies, researchers found the quality changes to be cost-effective; in the FIGURE 4-6 A quick literature review of cost-effectiveness studies indexed to PubMed found only 10 papers included cost-effectiveness analyses. SOURCE: Broughton, 2015.
|
From page 56... ...
One participant brought up the possibility that the problem of poor quality data is not likely to change and suggested that global health researchers might do well to anticipate this, moving to Bayesian approaches. Classical statistics need a kind of data that quality improvement simply may not be able to provide.
|
From page 57... ...
× × FIGURE 4-7 General structure of economic evaluations. NOTE: QALY = quality-adjusted life year.
|
From page 58... ...
Although the morning speakers explained how quality improvement programs collect considerable data, Broughton made the point that quality improvement research requires information about a comparable control group. Jishnu Das of the World Bank pointed out that publicly available data, such as the Demographic and Health Surveys (DHS)
|
From page 59... ...
Mate echoed Ashish Jha's judgment that the six methods are more similar than different, and that the ultimate success or failure of a strategy has as much to do with its suitability to a particular environment as its technical merits. At the same time, it is imprudent to be too confident about the value of any method, as some of the most promising results come from the weakest study designs.
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.