Effects of evidence-based clinical practice guidelines in cardiovascular health care quality improvements: A systematic review

Study searches

We did a systematic search using the following electronic databases for primary studies (randomised controlled trials (RCTs)): Cochrane Central Register of Controlled Trials (CENTRAL). The Cochrane Library, including the Cochrane EPOC (Cochrane Effective Practice and Organization of Care) specialised database; MEDLINE; EMBASE; CINAHL; PsycINFO; LILACS; Health Technology Assessment Databases and Web of Science, Science Citation Index, and Social Sciences Citation Index.

The review authors combined search strategy for indexed terms and developed free text terms. We included searches of grey literature in different sources, such as reports of the world and regional conferences, academic theses and scientific reports not published in indexed journals. We searched for studies published between January 1990 and June 2016, without any language restriction. An advanced search strategy is available as Extended data, Appendix 1 (Ramirez et al., 2019).

Study selection

We analysed the studies found through the search strategy, and two authors independently selected the articles according to the following inclusion criteria: RCT measuring the change in health care when using EB-CPGs on cardiovascular disease. The study should measure the change in any of the three health care dimensions (structure, process, and patient outcomes).

Data extraction

Three authors independently performed data extraction using a modified shorter version of the Cochrane Collaboration EPOC Data Collection Checklist translated into Spanish. Apart from preparing the Spanish version, we eliminated several items not applicable to this review, given that we only included RCTs. We used a standardised digital form for data extraction and analysis. We used a standardised digital form for data extraction and analysis. We used Review Manager software (RevMan 5.3) for the data analysis.

We assessed the risk of bias (quality) according to the Cochrane Handbook (Higgins & Green, 2012). We found very high variability, so the study results were introduced as a narrative in the Results.

Because of the variability between the measurements of the effect of the impact of EB-CPGs on the change of quality in the studies included in this review, it was not possible or appropriate to perform a meta-analysis; therefore, it was not possible to measure the statistical heterogeneity.

Study identification and selection

When the first version of the protocol of this Systematic Review was initiated, consider the inclusion of several clinical topics in a single systematic review. We decided to create a series of three systematic reviews on different clinical topics: Cardiovascular Health, Breast Cancer and Child and Mother Health. The initial searches included all those topics with the same inclusion criteria described. After removing duplicates, the search produced 4279 potential studies. After screening by title and abstract, 4051 were excluded. After the full-text evaluation, 96 studies of all clinical subjects were selected, extended data, Appendix 2 (Ramirez et al., 2019).

Study characteristics

For the analysis, we organised the studies according to the topic or pathology being the core subject of the EB-CPG, and for this report, we only selected RCTs on cardiovascular diseases, because this theme accounts for the higher number of original articles. In total, we selected nine RCTs (Figure 1). Characteristics of the included studies are available as Extended data, Appendix 2 (Ramirez et al., 2019).

Figure 1. Flow chart of the studies (PRISMA, Moher 2009).

We included nine RCTs analysing cardiovascular diseases (Beaulieu et al., 2004; Berner et al., 2003; Ellis et al., 2000; Guadagnoli et al., 2004; Hand et al., 2014; Jäntti et al., 2007; Kiessling & Henriksson, 2002; Tierney et al., 2003; Tsuyuki et al., 2015). All trials were carried out between 2000 and 2015, five in the United States of America, two in Canada, one in Finland and one in Sweden. Eight articles addressed outpatient care and three inpatient care. The EB-CPGs included in the selected trial looked at the following clinical problems: management of stable angina pectoris (over 65 years), unstable angina, dyslipidaemia, acute infarction, heart failure and blood pressure. Besides, the trials included perioperative cardiac evaluations of patients with non-cardiac surgery, cardiopulmonary resuscitation and secondary prevention in patients having coronary artery disease.

A number of the 4,279 studies found in the initial search were not randomised clinical trials, as the authors described in their titles or abstracts. When assessing the full-text articles, we found that many were cluster trials and observational studies with a “before and after” design.

Quality of evidence assessment

As for the quality of the evidence, we observed the presence of a high or unclear risk of bias for allocation concealment (selection bias), blinding of participants and personnel (performance bias), and blinding of outcome assessment (detection bias), which can be explained by the nature of the interventions studied. We found several types of systematic errors: random sequence generation (selection bias), incomplete outcome data (attrition bias) and selective reporting (reporting bias). We found that the interventions measured yielded outcomes assessed with moderate to low evidence certainty according to the GRADE classification.

Risk of bias assessment

Analysing the risk of bias in the nine included RCTs studies found the random sequence generation (selection bias) assessment had a low risk of bias (between 50–75%) for allocation concealment (selection bias) 33% low, 33% unknown and 33% high risk of bias. The blinding of participants and personnel (performance bias) obtained almost 50% high risk of bias. The blinding of outcome assessments (detection bias) obtained a 50% low and 50% uncertain risk of bias. Incomplete outcome data (attrition bias) obtained between a 50 and 75% low risk of bias. For the selective reporting (reporting bias), there was a 100% low risk of bias. Finally, other types of bias had a low risk of between 50 and 75% (Figure 2).

Figure 2. Overall bias risk chart of all included studies.

The studies with the lowest and highest risk of bias (Kiessling & Henriksson, 2002 and Jäntti et al., 2007, respectively) were chosen to be evaluated with the GRADE methodology and to build a summary of findings table. The aim of doing this GRADE table was obtaining the rank of possible grades of the evidence certainty found in the 84 interventions from the nine studies included. The results of these studies produced moderate-to-low evidence certainty, according to the GRADE classification (see Table 1 and Table 2).

Assessment of study outcomes

The authors grouped the outcomes in simple relative and absolute numbers. There was not a global estimate of the measurements of the effects included in the studies because of the considerable variability of measuring units, as well as the clinical heterogeneity found among the studies included. The measurement of the outcomes reported in the studies was dichotomous, continuous or nominal; the majority were dichotomous and were used to measure the mastery of the process, e.g. the number of patients receiving an adequate treatment following a recommendation versus those who did not.

Of the total of 84 interventions included in the nine RCTs evaluated, three (4%) corresponded to the health care structure dimension, 54 interventions (64%) to the dimension health care delivery and 27 interventions (32%) to the dimension of patient outcomes. Regarding the impact of EB-CPG use, we found that in 55 interventions (65%) there was no significant difference between the control and experimental groups.

In four interventions (5%) the outcome favoured the control group (comparison of the measure of compliance of the recommendations (the average adjusted by the patient characteristics, the setting, and the number of measures applied per patient)). The measure of effects (odds ratio (OR)) regarding the conditions acute myocardial infarction and heart failure, separated by care provided by a cardiologists or primary care physicians, were as follows: acute myocardial infarction, cared by a cardiologist OR 0.81 (95% CI: 0.79 to 0.83), and cared by a primary care physician OR 0.73 (95% CI: 0.71 a 0.76); heart failure, cared by a cardiologist OR 0.88 (95% CI: 0.86 to 0.90), and care by a primary care physician OR 0.79 (95% CI: 0.76 to 0.81).

The result favoured the intervention group for 25 interventions (30%). Some of the recommendations were: use of antiplatelet medication during the 24-hour hospital stay, a study of the left ventricle ejection fraction, total cholesterol and LDL measurements, cardiopulmonary resuscitation, and degree of compliance with EB-CPGs recommendations.

Implications for practice

The initial objective of this review was to strengthen the development programs for EB-CPG by evaluating their effects on the quality of health care and to give reliable evidence to sustain the decision-making process related to the construction of EB-CPGs in medium- and low-income countries. Even the research evidence is not strong enough to support the EB-CPGs as a tool to improve the practice for a better quality of care, the results of this review need to be interpreted with caution. Definitely. It seems the standard application used so far must be reviewed and must incorporate new psychosocial strategies oriented toward driving change in clinical practice and the doctor-patient relationship We need reaching the right hands at the right time. It is necessary to emphasise that the standard CPG implementation used so far must be reviewed, and must incorporate new psychosocial strategies oriented toward driving change in clinical practice and the doctor-patient relationships.

For an adequate implementation of a EB-CPG, it is necessary to take into account the possible costs, risks, and benefits that will be necessary to know the expected results precisely. The health systems that build EB-CPGs must maximise their efforts to get health care personnel to follow the recommendations of the EB-CPGs and make an effort to evaluate their impact.

Implications for research

This is the first of series of three systematic reviews on the effects of Evidence-Based Clinical Practice Guidelines (EB-CPGs) on health care quality improvements. The second RD will cover Breast Cancer and the third child and maternal health. We created an interactive https://lnkd.in/dwTxaD8 LIVE GUIDELINES program following the CDC https://lnkd.in/d5r4z6m

Since research in this field is so new, and the results of this study were not conclusive, more research is needed to evaluate the change that EB-CPGs could make to the quality of health care, emphasising the less studied areas, such as the structure of health care services and patient outcomes.

Underlying data

All data underlying the results are available as part of the article and no additional source data are required.

Extended data

Open Science Framework: Effects of Evidence-Based Clinical Practice Guidelines in cardiovascular health care quality improvements- A Systematic Review. https://doi.org/10.17605/OSF.IO/9A5FM (Ramirez et al., 2019).

This project contains the following extended data:

Reporting guidelines

Open Science Framework: PRISMA checklist for “Effects of evidence-based clinical practice guidelines in cardiovascular health care quality improvements: A systematic review”. https://doi.org/10.17605/OSF.IO/9A5FM (Ramirez et al., 2019).