Types of studies

We included randomised placebo‐controlled trials (RCTs) as well as controlled before‐and‐after studies. Before‐and‐after studies were included, because it has been shown that there is no placebo effect of statins on lipid parameters and that a placebo control is therefore not essential (Tsang 2002). We included data from cross‐over trials if the authors reported data for the initial treatment period versus parallel treatment groups followed by an adequate washout period before crossing over to the other active treatments and if data were reported in a similar manner during all treatment periods.

Types of participants

Participants may be of any age, with and without cardiovascular disease. They could have normal lipid parameters or any type of hyperlipidaemia or dyslipidaemia. We included participants with various comorbid conditions, including type 2 diabetes mellitus, hypertension, metabolic syndrome, chronic renal failure or cardiovascular disease.

Types of interventions

Pitavastatin had to be administered at a constant daily dose defined as a single dose per day compared to placebo or alone defined as a single pitavastatin dose per day for a period of three to 12 weeks. We chose this administration time window to allow at least three weeks for a steady‐state effect of pitavastatin to occur and to keep it short enough to minimise participants dropping out. We included studies where pitavastatin was administered at any time during the day. Trials required a washout baseline dietary stabilisation period of at least three weeks, where all previous lipid‐altering medication was withdrawn. This baseline phase ensured participants follow a standard lipid‐regulating diet and helped to stabilise baseline lipid values prior to treatment. In trials where participants were not receiving lipid‐altering medications or dietary supplements before receiving the test drug, we did not require washout baseline dietary stabilisation periods.

Types of outcome measures

Pitavastatin 1 mg/day, 2 mg/day and 4 mg/day were the primary doses as these are the recommended and predominantly prescribed doses.

Lipid parameters: for the RCTs, we presented the mean percentage change from baseline for different doses of pitavastatin minus the mean percentage change from baseline with placebo for LDL cholesterol, total cholesterol, HDL cholesterol and triglycerides. For the before‐and‐after studies, we presented the mean percentage change from baseline of different doses of pitavastatin. RCT data and before‐and‐after study data were combined because it was shown previously (Adams 2014; Adams 2015) that the two study designs yielded similar results (Tsang 2002).

Electronic searches

The Cochrane Hypertension Information Specialist searched the following databases without language, publication year or publication status restrictions:

• The Cochrane Central Register of Controlled Trials (CENTRAL) via the Cochrane Register of Studies (CRS‐Web) (searched 4 March 2019);

• MEDLINE Ovid (from 1946 onwards), MEDLINE Ovid Epub Ahead of Print, and MEDLINE Ovid In‐Process & Other Non‐Indexed Citations (searched 4 March 2019);

• Embase Ovid (from 1974 onwards) (searched 4 March 2019);

• ClinicalTrials.gov (www.clinicaltrials.gov) (searched 4 March 2019);

• World Health Organization International Clinical Trials Registry Platform (www.who.it.trialsearch) (searched 4 March 2019).

• Epistemonikos (https://www.epistemonikos.org) searched 4 March 2019).

The Information Specialist modelled subject strategies for databases on the search strategy designed for MEDLINE. We present the search strategies for major databases in Appendix 1.

Systematic searches in the following databases were conducted for before‐and‐after studies in MEDLINE, Embase, CENTRAL, CRS‐Web, ClinicalTrials.gov, World Health Organization International Clinical Trials Registry Platform, and Epistemonikos. No filters were used to retrieve RCTs and before‐and‐after studies because we wanted to retrieve all available documents on the subject.

Searching other resources

• The Cochrane Hypertension Information Specialist searched the Hypertension Specialised Register segment (which includes searches of MEDLINE, Embase, and Epistemonikos for systematic reviews) to retrieve existing reviews relevant to this systematic review, so that we could scan their reference lists for additional trials. The Specialised Register also includes searches for controlled trials in the Allied and Complementary Medicine Database (AMED), CAB Abstracts & Global Health, CINAHL, and Web of Science.

• We checked the bibliographies of included studies and any relevant systematic reviews identified for further references to relevant trials.

• Where necessary, we contacted authors of key papers and abstracts to request additional information about their trials.

We included grey literature by searching other resources.

1. ProQuest Dissertations and Theses (search.proquest.com/pqdtft/).

2. OpenTrials (opentrials.net).

3. US Food and Drug Administration (www.fda.gov/).

4. European Patent Office (worldwide.espacenet.com).

These resources were searched using the following keywords: pitavastatin, alipza, itavastatin, nisvastatin, livalo, livazo, NK‐104.

Selection of studies

Initial selection of RCTs and before‐and after‐studies involved retrieving and reading the titles and abstracts of each paper found from the electronic search databases or bibliographic citations (see Figure 1 for PRISMA flow diagram (Moher 2009)). Two review authors (SA and NA) analysed the full‐text papers independently, to decide on the studies to be included. We resolved disagreements by recourse to the third review author (JMW). Two review authors (SA and NA) independently extracted the appropriate data from each of the included studies. 

Figure 1

---

---

Study flow diagram for pitavastatin.

Data extraction and management

We extracted the mean percentage change directly from the data, or we calculated it from the baseline and endpoint values using the calculation found in Appendix 2. We added the calculated data to the Data and analyses section of the review. If the calculated data differed from the given data by more than 10%, the data were not included in the review. We extracted standard deviations (SDs) and standard errors (SEs) from the report or calculated them when possible using the following calculations (Appendix 3). We entered the data from RCT and controlled before‐and‐after studies into Review Manager 5 (RevMan 2020) as continuous and generic inverse variance data, respectively.

Assessment of risk of bias in included studies

We assessed all RCTs and before‐and‐after studies using the Cochrane 'Risk of bias' tool under the categories of allocation (selection bias), blinding (performance bias and detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias), and other potential sources of bias. We produced 'Risk of bias' tables' as outlined in the Cochrane Handbook for Systematic Reviews of Interventions, Chapter 8 (Higgins 2011). Controlled before‐after‐studies were considered 'high risk' and 'low grading' compared to RCTs. With respect to before‐and‐after studies, having only one study group is considered 'high risk' for random sequence generation and allocation concealment but other features in this study design might be at lower risk, and there may be unidentified differences between the intervention and control groups that may affect changes in the outcome measure. We appreciate that blinding of participants and personnel and blinding of outcome assessment are inappropriate for before‐and‐after studies and that this is a limitation. However, because the lipid parameter measurements are unlikely to be influenced by lack of blinding and were measured in a remote laboratory, they were considered unlikely to be affected by the study design. We were able to use the Cochrane 'Risk of bias' tool for the controlled before‐and‐after studies because there was a lack of difference in the mean differences between the two type of studies (Tsang 2002).

Measures of treatment effect

We analysed the treatment effects as mean difference (MD) for each dose in the RCTs and generic inverse variance for each dose in the before‐and‐after controlled studies separately. In the event that the mean effects from the two study designs were not statistically different, we re‐analysed all efficacy study data using the generic inverse variance to determine the overall weighted treatment effects and their 95% confidence intervals (CIs) for blood total cholesterol, LDL cholesterol, HDL cholesterol, and triglycerides.

Unit of analysis issues

The unit of analysis is the mean value for the people completing the study. We expected follow‐up to be reasonably high for these short‐term studies. The data, however, represented treatment efficacy and not real world effectiveness of pitavastatin on these lipid parameters.

Dealing with missing data

We expected follow‐up to be reasonably high for these short‐term trials. The data, however, represented treatment efficacy and not real‐world effectiveness of pitavastatin on these lipid parameters. When data were missing, we requested them from the authors. The most common type of value that was not reported was the SD of the change.

In the case of a missing SD for the change in lipid parameters, we imputed the SD using the following hierarchy (listed from high to low preference).

1. SD calculated either from the t statistics corresponding to the exact P value reported or from the 95% CI of the mean difference between treatment groups.

2. Average weighted standard deviation of the change from other trials in the review (Furukawa 2006).

Because it is common for the SD to be miscalculated and in order not to overweight studies where it was inaccurately calculated and lower than expected, when SD values were less than 40% of the average weighted SDs, we used the imputed value by the method Furukawa 2006.

Assessment of heterogeneity

The Chi² test to identify heterogeneity was not appropriate because it has low power when there are few studies but has excessive power to detect clinically unimportant heterogeneity when there are many studies (Higgins 2002). The I² is a better statistic. I² calculates between‐study variance/(between‐study variance + within‐study variance). This measures the proportion of total variation in the estimate of the treatment effect that is due to heterogeneity between studies. This statistic is also independent of the number of studies in the analysis (Higgins 2002).

The I² was assessed as moderate heterogeneity when ranging from 30% to 50% and high heterogeneity when greater than 50%.

Assessment of reporting biases

We assessed publication bias using funnel plots, as outlined in the Cochrane Handbook for Systematic Reviews of Interventions, Chapter 13 (Page 2019) when there were ten studies or more examining the same outcome (dose).

Data synthesis

We entered all RCTs into Review Manager 5 (RevMan 2020) as mean difference fixed‐effect model data to determine the weighted treatment effect and 95% CIs for blood total cholesterol, LDL cholesterol, HDL cholesterol, and triglycerides. We entered all controlled before‐and‐after studies as generic inverse variance fixed‐effect model data to determine the weighted treatment effect. If the effects in the RCTs were not statistically significantly different from the before‐and‐after studies, we entered all trials for each dose as generic inverse variance to determine the best overall weighted treatment effect for each dose.

We recorded data of each study and dose in GraphPad Prism 4, to yield a weighted least squares analysis based on the inverse of the square of the SE for each lipid parameter, to generate weighted log dose response curves. We entered the number of participants in placebo‐controlled trials, who prematurely withdrew due to at least one adverse effect in Review Manager 5 (RevMan 2020) as dichotomous data for each dose and all combined doses of pitavastatin and reported these as RR versus placebo.

The relative potency of pitavastatin with respect to fluvastatin, atorvastatin, rosuvastatin and cerivastatin, was determined as the ratio of the milligram (mg) amount of pitavastatin to the mg amount of fluvastatin or atorvastatin or rosuvastatin or cerivastatin needed to produce the same specified effect. These values were calculated from the log dose response curves of pitavastatin, fluvastatin, atorvastatin, rosuvastatin and cerivastatin for LDL cholesterol, total cholesterol and triglycerides. The relative potencies were estimated from these dose ratios. The relative potency results are mentioned in the Effects of interventions subsection of the Results section.

Subgroup analysis and investigation of heterogeneity

The main subgroup analyses were the different doses of pitavastatin. We assessed heterogeneity using the I² (Higgins 2002). If an I² value was ≥ 50%, we attempted to identify possible causes for this by carrying out a number of planned subgroup analyses, provided there were sufficient numbers of trials (see below).

We analysed subgroups based on the following factors.

1. Placebo‐controlled trials versus before‐and‐after trials (described above)

2. Men versus women

3. Morning administration time versus evening administration time as defined by the cut‐offs 6:00 am to noon and 6:00 pm to midnight

4. Kowa funded versus non‐Kowa funded trials

5. Twice‐daily versus once‐daily administration

Sensitivity analysis

We conducted sensitivity analyses to assess the effect of different comorbidities, such as familial hyperlipidaemia, on the treatment effect. We compared the treatment effects as generic inverse variance between trials whose subjects were reported to have type IIa or familial hypercholesterolaemia versus trials whose subjects were not reported to have genetic hypercholesterolaemia. Trials were not included in the comparison if the subjects had both familial and non‐familial hypercholesterolaemia. We conducted sensitivity analyses to assess the effect of different methods of dosing, such as twice daily versus single dose, on the treatment effect.

RCTs and before‐and‐after studies were analysed separately in the Data and analyses section.

Summary of findings and assessment of the certainty of the evidence

We used Grading of Recommendations, Assessment, Development and Evaluation (GRADE) approach to assess the certainty of the supporting evidence behind each estimate of treatment effect (Schünemann 2019a; Schünemann 2019b). We presented key findings of the review, including a summary of the amount of data, the magnitude of the effect size and the overall certainty of the evidence, in the summary of findings Table 1, summary of findings Table 2 and summary of findings Table 3. We did not summarise the findings on HDL cholesterol in a 'Summary of findings' table because pitavastatin doses ranging from 1 mg/day to 8 mg/day had no dose‐related effect on HDL cholesterol. There were not enough data to determine risk of withdrawal due to adverse effects due to pitavastatin, therefore, there was no 'Summary of findings' table for WDAEs.