Acupuncture for insomnia after stroke: a systematic review and meta-analysis

The search was performed without restriction to language or year of publication. We searched Medline, EMBASE, and the Cochrane Central Register of Controlled Trials from database inception through October 2014. For Korean publications, we searched three Korean medical databases (Research Information Service System, National Discovery for Science Leaders, and OASIS). For Chinese articles, we searched the China National Knowledge Infrastructure (CNKI). The keywords used for the search were “stroke OR apoplexy OR cva OR cerebrovascular attack OR cerebrovascular accident OR cerebral infarction OR cerebral hemorrhage” AND “acupuncture OR acupoints OR electroacupuncture OR electro-acupuncture OR auriculotherapy OR auriculoacupuncture” AND “insomnia” in each database language. The search strategy was adjusted for each database (Appendix).

Relevant clinical trials were included if the following criteria were met: 1) they were randomized, controlled trials (RCTs); 2) they included patients diagnosed with insomnia after stroke; 3) stroke patients with insomnia at baseline were enrolled, and 4) they studied insomnia as an outcome measure. Trials were excluded if the study design did not allow evaluation of the effects of acupuncture on insomnia after stroke; that is, studies were excluded if they 1) compared different types of acupuncture, 2) adopted complex treatment without examining the effects of acupuncture alone, or 3) reported insufficient information.

Two reviewers (L.S.H. and L.S.M.) independently extracted data using a standardized data extraction form and reached consensus on all items. Extracted data included authors, year of publication, sample size, interventions, main outcomes, and adverse events.

Instruments of the outcome measurements that were reported in the included studies were the Pittsburgh sleep quality index (PSQI), the efficacy standards of Chinese medicine, the Insomnia Severity Index (ISI), and the Athens insomnia scale (AIS).

The PSQI consists of 19 self-rated questions, which are grouped into seven component scores ranging from 0 to 3 each. The seven component scores are then summed to yield a global PSQI score, which has a range of 0 to 21, with higher scores indicating worse sleep quality. Specifically, a score of “0” indicates no difficulty, whereas a score of “21” indicates severe difficulties in all areas [16]. The ISI is a brief self-report instrument measuring a patient’s perception of insomnia. The seven items are rated on a 0-to-4 scale, and the total score ranges from 0 to 28. A higher score indicates more severe insomnia [17]. The AIS is a self-administered psychometric instrument consisting of eight items. Each item of the AIS can be rated from 0 to 3 for a total score range of 0–24, with a score of “0” indicating no problem at all and a score of “24” indicating very serious problems in all areas [18]. We extracted data on the mean change from baseline measures (the “Mean”, Fig. 3). The standard deviation of changes from baseline was determined using a correlation coefficient from a previously published study [19].

The efficacy standards of Chinese medicine is a measurement tool for the assessment of the states of the patients with complete improvement (recovery of normal sleep duration), partial improvement (increased sleep duration more than three hours), and no improvement after treatment [12]. Response rate was calculated based on proportion of the effective (complete or partial improvement) and not effective (no improvement) patients. We also considered measures of general safety reported for acupuncture as a treatment. We extracted data on the number of participants with improvement as an “Event” (Fig. 3).

The two reviewers independently assessed the methodological quality and the risk of bias of the included studies by means of the risk of bias (ROB) tool in the Cochrane Handbook for Systematic Reviews of Interventions (Version 5.0.2). This instrument consists of 8 domains: random sequence generation; allocation concealment; blinding of patients, personnel, and outcome assessors; incomplete outcome data; selective outcome reporting; and other sources of bias. The tool ranks evidence from research studies as having “high,” “low,” or “unclear” levels of bias; it is also appropriate for evaluating the methodological quality of RCTs. In cases in which the reviewers’ opinions differed, a joint opinion was reached through discussion.

All statistical analyses were performed with Reviewer Manager Software, version 5.3 (Cochrane Collaboration, Oxford, UK). Summary estimates of treatment effects were calculated using a random-effects model. The impact of acupuncture on dichotomous data was expressed as the risk ratio (RR); for continuous outcomes, the mean difference was calculated with a 95 % confidence interval (CI). The statistical heterogeneity in the subgroups was analyzed using the I ² test and was considered to be significant when I ² was greater than 50 %. Even when a low heterogeneity was detected, a random-effects model was applied, because the validity of tests of heterogeneity can be limited with a small number of component studies. Publication bias was detected using a funnel plot.

We identified 165 publications; 13 met the eligibility criteria (Fig. 1). The articles included in the analysis are summarized in Table 1. The 13 articles were published from 2004 to 2012. Two originated in Korea [20, 21] and 11 were from China [8, 22‐31]. The language of publication was English [20, 21] or Chinese [8, 22‐31].

The ROB results are shown in Table 2. With regards to random sequence generation and allocation concealment, five studies had a low ROB [27‐31] and eight studies had an unclear ROB [8, 20‐26]. With regards to blinding of patients and outcome assessors, two studies had a low ROB [20, 21], and 11 studies had an unclear ROB [8, 22‐31]. All RCTs had a low ROB in incomplete outcome data and selective outcome reporting [8, 20‐31]. All RCTs had an unclear ROB in other sources of bias [8, 20‐31].

The majority of the included RCTs stated that the rationale for acupuncture point selection was drawn from Traditional Chinese Medicine theory. Two studies used intradermal acupuncture [20, 21], five used acupuncture alone [22, 24, 28‐30], two used acupuncture and auricular acupuncture [8, 27], two used acupuncture and drugs [23, 26], one used auricular acupuncture [25], and one used electroacupuncture and auricular acupuncture [31]. A total of 33 acupuncture points (24 meridian points and nine auricular acupuncture points) were used for the treatment of insomnia. Acupoints used for insomnia treatment in most trials were Shenmen (HT-7) and Sishencong (EX-HN1) (Fig. 2). The number of acupoints used in each study ranged from two to 11.

We conducted a meta-analysis of the study results based on the insomnia assessment scales used (Fig. 3). In six studies that used the PSQI to assess treatment results, acupuncture appeared to be more effective than drugs for treatment of insomnia after stroke (weighted mean difference, 4.31; 95 % CI, 1.67–6.95; P = 0.001; n = 385, I ²  = 91 %).

In seven studies that used the efficacy standards of Chinese medicine to compare the effects of acupuncture with those of drugs, acupuncture was observed to have a significant difference in reducing insomnia after stroke (RR, 1.25; 95 % CI, 1.12–1.40; P < 0.001; n = 497, I ²  = 54 %).

Studies comparing the effects of intradermal acupuncture with those of sham acupuncture used the ISI or the AIS. In these studies, intradermal acupuncture had a significant difference on insomnia after stroke, as assessed both by the ISI (weighted mean difference, 4.44; 95 % CI, 2.75–6.13; P < 0.001; n = 82, I ²  = 9 %) and by the AIS (weighted mean difference, 3.64; 95 % CI, 2.28–5.00; P < 0.001; n = 82, I ²  = 0 %).

We assessed the publication bias using a funnel plot. However, it was difficult to determine any pattern indicative of publication bias based on the funnel plot’s symmetry owing to the small sample size (fewer than 10 studies, Additional file 1: Figure S1).