Effect of THNPs on Heroin OD Mortality | Methodology

Chapter 2. Methods

2.1 Introduction

This chapter contains a precise description of the process used to search and appraise the evidence for this systematic literature review. The search was carried out during August and September 2014. An in-depth description of how evidence and research available related to the PICO question were generated, will be discussed. As mentioned in the previous chapter, the aim of this dissertation is to determine the effect of THNPs to heroin users, on heroin OD mortality.

The analysing process of the PICO question, whether THNPs reduces the incidence of fatal heroin OD deaths, will be amplified through a description of the actual search strategy, and the methods used to appraise the retrieved studies. Through specific criteria, a systematic approach was applied in obtaining and selecting the relevant literature. Identified keywords used for the search syntaxes will be presented, together with the databases used for the retrieval of evidence, and the number of hits produced from each database. Search syntax is the relationship of a sequence of words used in a search strategy. A discussion of the inclusion and exclusion criteria set for this systematic review (SR) will follow, together with the methods of appraisal chosen to evaluate the studies.

2.2 Keywords and databases used

According to Timmins and McCabe (2005) the cornerstone of an effective search strategy is the use of proper keywords. A number of keywords were formulated from the PICO question elements, and from the preliminary search of the literature. These were mainly identified from titles and keywords used in articles found from the preliminary search. Table 2.1 below, shows the keywords used in relation to each PICO element.

Table 2.1 Keywords developed to describe the PICO framework

The main electronic systematic search with the search syntaxes explained further on in this chapter was carried out using different databases accessed from the Hybrid Discovery (HyDi) available from the online library at the University of Malta. Relevant databases were found on HyDi by choosing the category ‘Health Sciences’ and sub-category ‘Nursing’ since the other sub-category is ‘Midwifery’ which is unrelated. Six databases were used to conduct a search for the literature. These were Academic Search Complete, Cumulative Index to Nursing and Allied Health – CINAHL Plus with Full Text, Cochrane Central Register of Controlled Trials, and Cochrane Database of SRs; via EBSCO host search engine together with Medline via ProQuest search engine, and Science Direct. In addition to these databases Google Scholar was also utilised, but the search syntaxes combinations could not be used since the search engine does not have an advanced search strategy, and the search resulted in large thousands of articles. Therefore simple keywords alone were used to search for any other relevant studies, or to obtain full text versions which are not available in other databases. Strengths and limitations of each database are described below in table 2.2.

Table 2.2 Databases strengths & Limitations

2.3 Inclusion and Exclusion Criteria

For readers to assess reliability and validity of a SR they have to be provided with a set of inclusion and exclusion criteria (Cronin, Ryan, Coughlan, 2007). Furthermore, Torgerson (2003) suggests that a set of inclusion and exclusion criteria are important so that throughout the search, only the articles that will answer the PICO question are extracted while others are eliminated.

During the preliminary search it was noted that literature about the subject lacks randomized controlled trials (RCTs’), meta-analysis, and SRs. As shown in figure 2.1, these study designs are found at the top of the hierarchy of evidence as described by Aslam et.al (2012). Hence, all types of study designs were part of the inclusion criteria. Sibbald & Roland (1998) said that clinical trials are known to be a gold standard to perform good evidence based practice and although they were in the inclusion criteria, the results were negative since up-to date no RCTs have taken place. Although this is a major limitation for this literature review, one must keep in mind that currently, the first and largest RCT is being carried out as described in chapter 1.

Source type selected in all databases was ‘Academic Journals’ since books, newspaper articles, and magazines are not of strong evidence for evidence based healthcare because sometimes they do not have any type of study design, and may be narrative. Each search was limited to English language literature only. Studies that administered intranasal (IN) naloxone were excluded since this method is an off-label method (Doe-Simkins, Walley, Epstein, & Moyer, 2009), and in an RCT by Kelly et al. (2005) it was found that IM naloxone was clinically more effective than IN naloxone (6 minutes vs 8 minutes). Table 2.3 and table 2.4 below describe further in detail the inclusion and exclusion criteria respectively.

Figure 2.1 The Hierarchy of Evidence Pyramid (adapted from Aslam et al. 2012, p.3)

Table 2.3 Inclusion Criteria

Table 2.4 Exclusion Criteria

2.4 Search Strategy

The keywords mentioned in table 2.1 were developed so that each PICO element has its own search syntax as identified by Bettany-Saltikov, (2012). The use of search syntaxes has assisted in narrowing the search to the most relevant studies to the PICO question. Table 2.5 below, shows the search syntax used for each PICO item.

Table 2.5 Search syntaxes

Search syntaxes were combined as described here:

• Population (syntax 1) AND Intervention (syntax 2)
• Population (syntax 1) AND Outcome (syntax 3)
• Intervention (syntax 2) AND Outcome (syntax 3)

Boolean logic operator ‘OR’ was used to combine each keyword of each PICO element syntax, while the operator ‘AND’ was used to combine the PICO elements together as described above.

In all search fields of the databases used except Google Scholar, ‘In All Text’ was selected for all searches. In Google Scholar the keywords searched independently were ‘take-home naloxone’ and ‘naloxone distribution’. These keywords were searched in titles only, since when searched for anywhere in the article the results were over 50,000 with most of them irrelevant to the dissertation.

As shown in table 2.6 below, the defined search resulted in 2,260 articles. After the titles were evaluated whether they fit with the inclusion and exclusion criteria mentioned previously, and removing duplicate articles the search results were reduced to 160 articles by reading through the titles. The abstracts of these 160 articles were read and consequently 25 articles were noted to be directly related to the PICO question addressed in this dissertation, as they all met the inclusion criteria. One of the 25 articles which is a study by Lenton & Hargreaves (2000) was not available online for free. The author, Professor Lenton was contacted via e-mail. Fortunately there was an immediate response, and the article was provided for free. However, after reading these 25 articles, 15 were eliminated because they had one or more of the exclusion criteria fit. The remaining 10 articles were identified and had fit all of the inclusion and exclusion criteria and were chosen for the critical appraisal of this systematic literature review. Furthermore the search was enhanced by looking up at all the reference lists of the final articles chosen for critical appraisal, but no new studies were found. Identified studies for critical appraisal are the following: Strang et al. (2008), Piper et al. (2008), Tobin, Sherman, Beilenson, Welsh, & Latkin (2009), Enteen et al. (2010), Wagner et al. (2010), Bennett, Bell, Tomedi, Hulsey, & Kral (2011), Bennett & Holloway (2012), Lankenau et al. (2013), Banjo (2014), and Clark, Wilder, & Winstanley (2014). In addition to the electronic search, a hand search was done within the University of Malta library which did not produce any results. Table 2.6 below presents the number of hits obtained from each of the databases above and the different search syntaxes combination.

Table 2.6 Number of articles found from each search

2.5 Refining of the research question

After finding the relevant literature to the original PICO question, and considering properly the population being studied, the research question was duly refined since this systematic literature review will be based on the impact of THNP effect on heroin OD deaths amongst heroin injection drug users. Initial PICO question was:

‘Do naloxone take-home programs provided to drug users in the community reduce the incidence of fatal opioid OD?’

The question was refined to:

Do naloxone take-home programs reduce the incidence of fatal heroin OD amongst heroin injection users?

The refining of the PICO question was essential since some of the literature concerned prescription opioid users or general drug users, but the reviewer’s main interest is heroin injection users.

2.6 Methods of Appraisal

After identifying the relevant studies and refining the PICO question, the next step for this dissertation was to critically appraise the literature. This will be covered in chapter 3. Critical appraisal is the evaluation of strength and quality of research studies in order to deliver evidence-based clinical decisions and after determining the worthiness of different studies through the use of different critical appraisal tools, multiple research findings can be synthesized together to develop evidence-based-practice (LoBiondo-Woods & Haber, 2013). Furthermore, Burls (2009) argues that studies being biased can result in false outcomes and conclusions, because they are not composed of an acceptable quality. A bias is an orderly deviation of the results occurring from the way that a study is led, investigated or reported. In light of this, critical appraisal tools were developed to help researchers in identifying valid and reliable studies. Specific critical appraisal tools in the form of checklists are used to analyse different study designs. Commonly used appraisal tools are:

• Critical Appraisal Skills Programme (CASP)
• Scottish Intercollegiate Guidelines Network (SIGN)
• Centre for Evidence Based Mental Health (CEBMH)
• Critically Appraise Topic (CAT)

After studying and considering all of the above methods of appraisal, together with other appraisal tools, it was decided that the CASP appraisal tools are to be used for critically appraising the identified studies relevant to the PICO question. The reason for choosing the CASP appraisal tools is because it is clear, concise, and easy to understand. CASP offers different checklists appropriate for different research designs where one can critically evaluate research through a set of questions and helps in determining the quality of the literature being appraised. At this phase, the studies were also rated and arranged according to the hierarchy of evidence adopted from Aslam (2012) as shown in figure 2.1.

2.7 Conclusion

In conclusion, the study’s main objective is to determine whether THNPs reduces the mortality rate of heroin OD. Several keywords were identified from the PICO question elements, and from preliminary search results. These keywords were systematically searched in different databases using different search syntaxes combinations. A set of inclusion and exclusion criteria were then set up in order to get to the studies relevant to the PICO question. PICO question was then refined and then it was decided to use the CASP appraisal tools for evaluating the relevant studies which will be discussed in the next chapter.

Words 1763