Continuous dopaminergic stimulation (CDS)-based treatment in Parkinson's disease patients with motor complications: a systematic review and meta-analysis.

A systematic review of the literature was conducted to identify randomized trials involving continuous dopaminergic stimulation (CDS) in PD patients with motor complications. Difference between n groups was assessed by partitioning heterogeneity and using the χ2 distribution with n-1 degrees of freedom, where n equals the number of groups. We looked for publication bias using funnel plotting, Egger's test and Begg's test. Twenty Randomized Controlled Trials (RCTs) were included. The results showed that CDS could evidently improve the Unified Parkinson's Disease Rating Scale (UPDRS) Part II (p < 0.0001), part III (P < 0.00001) and UPDRS total score (p < 0.00001). There was also a statistical discrepancy in off time reduction (p < 0.00001) and prolongation of on time (p < 0.00001) by the CDS therapy compared with control groups. Meanwhile, the results of this study showed obvious side effects in the CDS therapy compared with the placebo, especially at the expense of increased dyskinesia (23.4% vs 11.7%). The present study showed that CDS was beneficial in the treatment of PD patients with motor complications. But the incidence of the side events is more common than placebo.

Parkinson's disease (PD) is a progressive central nervous system disorder affecting at least six million people worldwide, making it being the second common neurodegenerative disease after Alzheimer's disease1. At the same time, with a growing elderly population and a reduction in other causes of mortality, the prevalence of PD is likely to increase2. The burden of PD is speculated to grow substantially over the next few decades as the size of the elderly population grows. Such projections give impetus to the need for innovative new treatments to prevent, delay onset, or alleviate symptoms of PD3. To our knowledge, the gold standard treatment for PD still is oral L-dopa because of its excellent efficacy and tolerability4. However, long-term use with oral L-dopa administration leads to various motor complications, including the “wearing-off” phenomenon and involuntary movements5. Motor complications occur in up to 50% of patients taking L-dopa more than 5 years and in as many as 100% of patients with young-onset PD, and are a dominating cause of disability. This presents a particular challenge to managing advanced PD patients. The motor complications are supposed to be the result of a combination of progressive denervation of the striatum and long-term abnormal, intermittent drug administration leading to pulsatile stimulation of dopaminergic neurons67. Over time, this pulsatile stimulation results in gene and protein changes in striatal dopaminergic neurons, affecting firing patterns in basal ganglia output neurons and ultimately leads to motor complications8. Moreover, in animal models of PD, intermittent but not continuous administration of L-dopa can lead to the wearing-off phenomenon and dyskinesia9. All of these results provide the rationale for treating patients with continuous dopaminergic stimulation (CDS) approach.

The concept of CDS has received considerable attention as a therapeutic approach to the treatment of Parkinson's disease10. Long-acting dopaminergic agents matched to provide comparable motor improvement to L-dopa treated animal models, but experience an evidently reduced frequency and severity of motor complications11. What is more, clinical studies of L-dopa/carbidopa intraduodenal gel infusion revealed marked reductions in ‘off’ time along with improvement in ‘on’ time when compared with conventional treatment12. Recent numerous evidences suggest wide-ranging symptoms improvement with CDS in patients with severe motor complication, which has beneficial implications for the management of patients with advanced PD in clinical practice13. This theory proposes that CDS can provide more continuous stimulation of brain dopaminergic receptors than intermittent doses of oral L-dopa, which is indicate dyskinesia and motor fluctuations could be delayed even though prevented by realizing continuous stimulation of striatal dopamine receptors using long-acting medicines such as dopamine agonists14.

Nevertheless, the CDS theory has fallen out of favor for a variety of reasons. Some studies reported the incidence of dyskinesia had not been proven to be directly correlated with drug half-life15. In additional, many of these studies are of small sample size and thus cannot fully elucidate the efficacy and safety implications of the therapy. Hence, there is uncertainty whether CDS is more effective than another in the PD patients with motor complications. We therefore conducted the present meta-analysis by pooling the results of these trials in order to provide more useful information about efficacy and safety of CDS in the PD patients with motor complications.

Results

This Meta-analysis is conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses16.

Study inclusion

We identified 652 potentially relevant articles from six databases. After removal of duplicates, 279 records remained. After going through the titles and abstracts, 162 were excluded because they failed to meet the inclusion criteria. By reading the full text of the remaining 117 articles as possibly reporting the CDS for the PD patients with motor complications, 49 studies were excluded because a result of without control groups, 37 were excluded because of the subjects did not have developed motor complications, 11 studies were excluded due to not compared with the placebo. Ultimately, 20 eligible studies satisfied the pre-established inclusion criteria1718192021222324252627282930313233343536 (Figure 1).

Figure 1

PRISMA 2009 Flow Diagram.

Characteristics of studies

Twenty studies with 3234 PD patients with motor complications were retrieved based on the search criteria1718192021222324252627282930313233343536. Among them, only 2 studies belong to L-dopa drug class, remaining 18 studies were dopamine agonist's drug class. The number of participants randomised into the 20 trials included in this meta-analysis ranged from 2428 to 39332 participants. The mean age of the participants in the trials was approximately 64 years. Meanwhile, 64% of the participants were male and they had PD for approximately 8 years. The duration of treatment was varied from 6 weeks28 to 32 weeks22. UPDRS as the outcome measure was observed in 15 studies, the change of the on/off time was observed in 12 studies, adverse effects were reported in 19 studies. Of the L-dopa drug class, L-dopa intraduode infusion and intrajejunal infusion of levodopa/carbidopa intestinal gel were administrated in 1 study, respectively2836. In the dopamine agonists drug class, pramipexole were administrated in 8 studies (8/18, 44.4%), ropinirole were used in 3 studies (3/18, 16.7%), cabegoline and bromocriptine as the intervention therapy in 2 studies. What is more, lisuride infusion27, rotigotine transdermal system30 and subcutaneous apomorphine infusion34 in 1 study, respectively. There were 3 three-arm trials comparing with the placebo. The basic characteristics of each case control study were listed in Table 1.

Table 1

Basic characteristics of included studies

	Basic data: Numbers, Male/Female, MA, DOD (y), MLD
study	Trial group	Control group	Intervention#	Main outcomes
Toyokur 198517	114,59/55,62.5 ± 0.9 Y, NR, 418 ± 20.7 mg	108,50/58,63.2 ± 0.9 Y, NR,465 ± 22.3 mg	Bromocriptine (22.5 mg/d) for 18 w	1.AE
Olanow 199418	189,116/73,62.5 ± 8 Y,11.4 ± 5.5 Y,891 ± 420 mg	189,123/64,63 ± 8 Y,10 ± 5.3 Y,940 ± 496 mg	Pergolide(5 mg/d) for 24 w	1. UPDRSII–III (p1); 2.AE
Steiger 199619	19,NR, 60.8 ± 9.1, 13.6 ± 7.5,NR	18,NR, 63.4 ± 7.2, 11.9 ± 7.2,NR	Cabergoline(5 mg/d) for 3 m	1.on/off time(p1), 2.AE
Hutton 199620	123,83/40,63.4 ± 10 y, 10.6 ± 5.9 y,NR	65,42/23, 62.8 ± 8.9 Y, 10.5 ± 5.0 Y, NR	Cabergoline(1.0 mg/d) for 24 w	1. UPDRSII–III (p1); 2.AE
Rascol 199621	23,14/9,62 ± 7 y,8 ± 2 y,663 ± 302 mg	23,14/9,63 ± 9 y,8 ± 3 y,715 ± 353 mg	Ropinirole (8 mg/d) for 3 m	1.off time (p1); 2.AE
Lieberman 199722	181,119/62, 63.4 y, 9.4 y, NR	179,116/63, 63.3 y, 9.0 y, NR	Pramipexole(4.5 mg/d) for 32 w	1.UPDRSI(P2),2.UPDRSII–IV(P1),3.AE
Guttman 199723	FA:79,48/31.62.89 ± 10.02 y,6 y, NRSA:84,55/29,61.51 ± 9.48 y,7.17 Y, NR	83,53/30.63.72 ± 10.35 y, 7.58 Y, NR	FA: pramipexole(4.5 mg/d) for 10 m;SA: Bromocriptine (30 mg/d)	1.UPDRSI–IV (p1); 2. AE
Wermuth 199824	36,20/16, 63.2 ± 7.9 y, 10.1 ± 5.0 y, NR	33,20/13, 62.1 ± 9.9 y, 9.9 ± 4.1 y, NR	Pramipexole(5 mg/d) for 12 w	1.UPDRS total (p1); 2.AE
Lieberman 199825	95,NR,NR,8.6 ± 4.7 y,759 ± 422 mg	54,NR,NR,9.4 ± 6.3 y,843 ± 517 mg	Ropinirole (24 mg/d) for 6 m	1.off time (p1); 2.AE
Pinter 199926	34,20/14, 59.3 ± 8.3 y, 7.8 ± 4.3 y, NR	34,20/14, 59.3 ± 8.3 y, 7.8 ± 4.3 y, NR	Pramipexole(1.25 mg/d) for 12 w	1.UPDRSI–IV (p1); 2.UPDRS total (p1); 3.AE
Stocchi 200227	20,NR,10 ± 2.1,688.2 ± 133.4 mg	20,NR,8.8 ± 2.7,675 ± 180.6 mg	Lisuride infurision (1 mg/ml) for 4 y	1.0ff time (p1); 2:UPDRS III (p2); 3. AE
Nyholm 200528	12,9/3,64,NR,NR	12,9/3,68,NR,NR	Intraduode infusion Levodopa for 6 w	1.on/off time; 2.UPDRS; 3.AE
Moller 200529	174,108/66, 63.4 y, 7.6 y, 637,7 mg	180,122/58,64.7 y, 7.9 y,648,8 mg	Pramipexole(3.7 mg/d) for 25 m	1.off time (p1); 2. UPDRSII–III, total (p1) 3.AE
Peter 200730	FA:111,63/48.63.2 ± 9.7 y,NR,NRSA:120,71/49,64.3 ± 9.0 y,NR,NR	120,73/47.65 ± 10 y, NR,NR	FA: rotigotine TS(12 mg/d) for 24 w;SA: rotigotine TS(8 mg/d)	1. UPDRSII–III (p1); 2. off time (p1); 3.AE
Poewe 200731	FA:200,112/88.63.2 ± 9.7 y,NR,813 ± 459 mgSA:201,132/69,64.3 ± 9.0 y,NR,795 ± 380 mg	100,71/29.65 ± 10 y,NR,814 ± 398 mg	FA: pramipexole(4.5 mg/d) for 6 m;SA: rotigotine(16 mg/d)	1.on/off time(p1); 2. UPDRSII (p1)3. UPDRSIII(p1); 4.AE
R. Pahwa, 200732	202,118/84,66.3 ± 9.2 y,8.6 ± 4.8 y,824 ± 424.4 mg	191,129/61,66.0 ± 9.7 y,8.6 ± 5.2 y,776 ± 357.3 mg	Ropinirole 24-hour(18.8 mg/d) for 24 w	1.on/0ff time (p1); 2.UPDRS motor (p1)3.UPDRS II(p1); 4.AE
Hauser 200933	FA:24,19/5.64.3 ± 7.8 y,NA,769.8 ± 284.9 mgSA:26,21/5,65.2 ± 9.3 y,NR,775.0 ± 397.1 mg	11,6/5,65.1 ± 7.8 y,NA,638.6 ± 204.1 mg	Pramipexole(4.5 mg/d) for 7 w;	1.on/off time (p1); 2.AE
Pablo 201134	17,NA,59.5 ± 11.7 Y,12.05 ± 4.2 Y,1077.81 ± 446.26 mg	17,NA,66.4 ± 7.0 Y,13.23 ± 4.7 Y,1028.18 ± 388.76 mg	Apomorphine (12–16 hrs/day)subcutaneous Infusion	1.UPDRSIII (p1)
Schapira 201135	164,92/72,61.6 ± 9.7 y, 6.1 ± 4.0 y,NR	178,94/84,60.9 ± 9.7 y, 5.9 ± 3.8 y,NR	Pramipexole(4.5 mg/d) for 18 w	1. UPDRSII–IV; 2.on time (p1); 3.AE
Olanow 201436	37,24/13,63.7 ± 9.5 y,10 ± 4.6 y,NA	34,22/12,65.1 ± 6.8 y,11.8 ± 5.6 y,NA	Intrajejunal infusion of levodopa/carbidopa intestinal gel for 12 w	1. UPDRSII–III (p1); 2.on/off time(p1)3. AE

FA: first arm; SA: second arm; MA: mean age; DLT: Duration of levodopa treatment; MLD: mean levodopa dosage; Y: years; M: months; W: weeks; d: days; NR: no report; #: indicate the maximum dosage in the study; p1: p < 0.05; P2: P > 0.05; TS: transdermal system; UPDRS: Unified Parkinson's Disease Rating Scale; AE: adverse events.

Risk of bias in included studies

We utilized the criteria recommended by Cochrane Handbook for Systematic Reviews to assess the risk of bias in the 20 articles included. Although all included studies claimed randomization, 11 articles described the method of random sequences generation (for example, random number generator, computer generated). But only 5 trials gave information that allowed the assessment of whether an adequate concealment of allocation procedure was used. All the studies reported the blinding of participants except two trials2734. Non-selective reporting was found in 14 trials. Three studies existed certain degree other potential threats to validity. Therefore, all of the included trials were deemed to have a low risk of bias. The methodological quality of each study was summarized in Table 2.

Table 2

The methodological quality of included studies based on the Cochrane handbook

Study	A	B	C	D	E	F	Total
Toyokur 198517	+	−	+	−	?	+	3
Olanow 199418	?	−	+	−	+	−	2
Steiger 199619	?	+	+	−	?	−	3
Hutton 199620	?	−	+	−	+	+	3
Rascol 199621	−	−	+	−	?	?	1
Lieberman 199722	?	−	+	−	+	+	3
Guttman 199723	?	−	+	+	?	+	3
Wermuth 199824	+	−	+	−	+	?	3
Lieberman 199825	+	−	+	−	+	+	4
Pinter 199926	+	−	+	+	?	+	4
Stocchi 200227	?	−	−	+	+	+	3
Nyholm 200528	+	+	+	−	+	+	5
Moller 200529	+	−	+	−	?	+	3
Peter 200730	+	+	+	−	+	+	5
Poewe 200731	?	+	+	−	+	+	4
R. Pahwa, 200732	+	?	+	−	+	+	4
Hauser 200933	+	−	+	−	+	+	4
Pablo 201134	+	−	−	?	+	−	2
Schapira 201135	−	?	+	?	+	+	3
Olanow 201436	+	+	+	+	+	+	6

A: Sequence generation; B: Allocation concealment; C: Blinding of participants, personnel and outcome assessors; D: Incomplete outcome data; E: Selective outcome reporting; F: Other sources of bias; +: Yes; −: No; ?: Unclear.

Meta analysis results

UPDRS

Data on the clinician-rated UPDRS Part II scores were available from 8 trials with 1 trials of L-dopa drug class and 7 trials of dopamine agonist's drug class. For the UPDRS Part II score, there were 2034 participants included in the analysis. We pooled the whole data to process and found significant difference when CDS treatment compared with placebo {p < 0.0001, standard mean difference (SMD) = −3.04, 95% CI: −4.51 to −1.57, Figure 2A}. Nevertheless, there was severe heterogeneity for the analysis of UPDRS Part II between studies (Tau2 = 4.46, Chi2 = 960.54, p < 0.00001, I2 = 99%, Figure 2A). Consequently, we should interpret the pool result prudently. The remaining three studies failed to pool analysis due to the raw data demonstrated in the form of percentage improvement (%) or missing, but all of them reported the significant effects of CDS for ameliorating the UPDRS Part II compared with the placebo group (p < 0.05 or p < 0.01). Compared to placebo treatment, ten independent studies showed significant effects of CDS therapy for improving the UPDRS Part III scores (P < 0.00001, SMD = −2.16, 95% CI −3.10 to −1.23; heterogeneity Tau2 = 2.17, Chi2 = 632.53, p < 0.00001, I2 = 99%, Figure 2B), suggesting that CDS treatment could contribute to improving complications of treatment in patients with PD with motor complications, but the remaining two studies did not provide raw data and thus failed for meta-analysis. There were 3 studies included in the meta-analysis to comparing CDS with placebo according to the UPDRS total score as the outcome index. We pooled the trials and found significant difference between two groups (p < 0.00001, SMD = −0.63, 95%CI: −0.83 to −0.44, Figure 2C) without heterogeneity (Tau2 = 0.00, Chi2 = 2.13, p = 0.34, I2 = 6%, Figure 2C), indicating CDS could apparent ameliorate the symptoms of the PD patients with motor complications. The funnel plot was roughly symmetric for the effect of CDS on UPDRS score. Thus, funnel plots did not suggest an obvious publication bias (Figure 3A). Meanwhile, Egger's weighted regression (p = 0.246 > 0.05, Figure 3B) and Begg's test (p = 0.755 > 0.05, Figure 3B) both suggested low likelihood of publication bias for all analysis.

Figure 2

(A) change from baseline to endpoint in UPDRS Part II of CDS vs Placebo in PD patients with motor complications; (B) change from baseline to endpoint in UPDRS Part III of CDS vs Placebo in PD patients with motor complications; (C) change from baseline to endpoint in UPDRS total score of CDS vs Placebo in PD patients with motor complications.

Figure 3

Bias assessment plot for the effect of CDS on UPDRS score by funnel blot (A),Egger's test (B) and Begg's test (C).

Prolongation of on time and off time reduction

A total of 3 trials evaluated CDS treatment versus placebo in patients of on time increase. Patients had greater response to CDS than placebo as evidenced by obvious significant effects of CDS for prolongation of on time compared with control groups (p < 0.00001, SMD = 2.59, 95%CI: 2.35 to 2.83, Figure 4A). The average on time increased by 2.3 hours for CDS treatment and by 0.7 hours for the placebo group, which was apparent statistical difference between CDS and placebo. Data on off-time reduction were available from 6 trials including 930 participants. Compared to placebo, CDS significant reduced off-time by approximately 2.0 hour a day (p < 0.00001, SMD = −4.34, 95%CI: −5.82 to −2.86, Figure 3C). Nevertheless, there was severe heterogeneity between trials (Tau2 = 3.08, Chi2 = 157.33, p < 0.00001, I2 = 97%, Figure 4B). Some studies failed to pool analysis due to the aforementioned reasons.

Figure 4

(A) change from baseline to endpoint in on time of CDS vs Placebo in PD patients with motor complications; (B) change from baseline to endpoint in off time reduction of CDS vs Placebo in PD patients with motor complications.

Adverse effects

Adverse effects (AE) were reported in 20 studies. CDS in general showed statistically significant higher rates of AE than placebo (Data not showed). There was significant difference of adverse events including incidences of dyskinesia (23.4% vs 11.7%), nausea (14.7% vs 9.3%), dizziness (6.53% vs 4.5%), diarrhea (3.44% vs 1.6%) between the CDS and placebo. However, there was no obvious discrepancy of insomnia (4.7% vs 4.1%).

Discussion

Main findings

Our meta-analysis found that CDS could evidently improve the UPDRS Part II, part III and UPDRS total score compared with placebo in PD patients with motor complications. There was also a statistical significance in off time reduction and prolongation of on time by the CDS therapy. These results displayed that patients receive CDS experienced significant improvement in both “off” and “on” time, as well as in improving the UPDRS subscale score in PD patients with motor complications. However, the results of this study showed obvious side effects in the CDS therapy compared to the placebo, especially at the expense of increased dyskinesia. Hence, it was necessary to detect the optimal therapeutic efficacy to balance the incidence of adverse events. To our best knowledge, it was no question that CDS was beneficial in the treatment of PD patients with motor complications. But it was less clear which class of drug therapy was more effective between L-dopa drug class and dopamine agonist drug class. We failed to draw conclusion which was more effective based on the present meta-analysis due to only two studies focused on the L-dopa drug class. In terms of AE, two kinds of drugs have a nearly similar incidence of AE, both higher rates than placebo.

Interpretation of the results

CDS has been influential in recent years and has lead to increased prevalence of dopamine agonists for PD37. Under normal physiologic condition, the dopamine concentration in the striatum is maintained at constant levels38. Therefore, continuous stimulation of postsynaptic receptors is thought to be essential for normal basal ganglia function39. Meanwhile, motor complications are associated with non-physiologic intermittent or pulsatile stimulation of dopamine receptors40. Based on such theory, CDS has been shown to reduce motor complications in advanced PD patients, and has been hypothesized to prevent their incidence when given as early therapy in PD13. Indeed, some basic researches showed treatment with intermittent injections of a short-acting dopamine agonist induced motor complications while continuous infusion of the same agent did not11. But, whether CDS can reduce the development of dyskinesia has yet to be studied. STRIDE-PD study reported initial therapy with carbidopa-levodopa-entacapone result in an increased frequency of dyskinesia compared to standard carbidopa/levodopa41. Moreover, laboratory and clinical studies have indicated that although dyskinesias are closely linked to the short serum half-life of levodopa, other factors may also play a pivotal role in the pathogenesis7. Consequently, CDS is no doubt effective to manage the PD patients with motor complications, but it may cause serious side effects, especially at the expense of increased dyskinesia.

Limitations

Several limitations of this study should be considered. Firstly, a large proportion of the studies included in this review are of six months or less in duration. Among the 20 studies, a very large proportion of studies treatment duration are less than one years, suggesting that there are insufficient data on the comparative efficacy and tolerability of these drugs beyond six months. In such a chronic long-term disease as PD, it would inevitable be valuable to know whether the drug actions persist for a much longer period. Secondly, The role of potentially useful CDS such as levodopa-carbidopa intestinal gel infusion, continuous duodenal infusion, subcutaneous apomorphine or lisuride infusion, continuous transdermal delivery of rotigotine were very inadequate in this meta-analysis. Although the fact that an extensive search for such studies was performed, most of them were not RCTs. Therefore, it is necessary to carry out more RCTs to assess the efficacy of such CDS in the PD patients with motor complications. Thirdly, patients included in the trials were not always on behalf of patients in clinic. Patients with Mini-Mental State Examination score less than 24, atypical parkinsonian syndromes, any history of deep brain stimulation, psychiatric or clinically significant hypotension were excluded. Thus, the results cannot be generalized to the entire population of PD patients with motor complications. Finally, high heterogeneity occurred several times when comparing CDS with placebo treatment. Inevitably, studies brought together in a systematic review will differ. Any kind of variability among studies in a systematic review or meta-analysis may be termed heterogeneity42. Variability in the participants, interventions, study design and risk of bias all can result in heterogeneity. To solve this problem, in this meta-analysis, random-effects model may be used to incorporate heterogeneity among studies. However, random-effects model is not a substitute for a thorough investigation of heterogeneity. It is intended primarily for heterogeneity that cannot be explained43.

Implications for Practice

This meta-analysis of RCTs mainly focuses on the efficacy and safety of CDS therapy in PD patients with motor complications. According to our meta-analysis, patients receiving CDS therapy exhibit significant ameliorate in their PD symptoms as evidenced by improvements in their UPDRS scores, as well as off time reduction compared with placebo. In terms of the safety assessment of this meta-analysis, CDS therapy in generally showed significant higher rates of adverse events than placebo, especially at the expense of increased dyskinesia. Although acknowledging the limitations of this meta-analysis, our findings present a majority of high-quality trials and provide hopefully evidences that CDS therapy can additionally benefit relieve symptoms of PD patients with motor complications.

Implications for research

A number of implications for research arise from this paper. Firstly, improvement in the methodological quality of RCTs is critical for later trials and more methodologically rigorous studies are needed in this field, and sample size calculation should be conducted before enrollment. Secondly, various kinds of different forms of CDS were administered in the 21 studies included. Thus, it is necessary to identify which form of the CDS displays the most effective anti-Parkinsonian action. Thirdly, a new theory points out standard oral administration of dopamine agonists does not result in constant plasma drug levels, therefore, more continuous drug delivery (CDD) may result in both prolonged reversal of motor deficits and reduced levels of dyskinesia. So, we need carry out more well-designed, randomized, double-blind, placebo-controlled trials and report in detail about continuous duodenal or intravenous L-dopa infusion, subcutaneous apomorphine or lisuride infusion, continuous transdermal delivery of rotigotine in future.

Methods

Inclusion and exclusion criteria

Only Randomized Controlled Trials (RCTs) which evaluate CDS for PD patients with motor complications were included in this review. CDS is a therapeutic concept for the management of PD that proposes continuous, as opposed to intermittent or pulsatile, stimulation of striatal dopamine receptors. However, the form of the CDS is multitudinous. In this meta-analysis, we determine in advance the method to achieve CDS are as follow: (1) levodopa-carbidopa intestinal gel infusion; (2) long half-time dopamine agonists (e.g. ropinorole, bromocriptine and pramipexole); (3) subcutaneous apomorphine or lisuride infusion or continuous transdermal delivery of rotigotine. The former one kind belongs to L-dopa class, and the later two kinds were ascribed to dopamine agonist's class. Moreover, eligible studies had to meet all of the following criteria: (1) published in a peer-reviewed journal; (2) conducted to evaluate the CDS versus placebo, both on a background of L-dopa therapy, with all other aspects of treatment being the same in both arms; (3) had original data being independent from other studies; (4) had adequate data such as absolutely change on/off time or UPDRS score as the main outcome measures.

Literature search

We electronically searched six databases of Cochrane Central Register of Controlled Trials (CENTRAL), PubMed, EMBASE database up to February 2014 for all English language publications. Reference lists from the resulting publications and reviews were used to identify further relevant publications. The following search strategy, using the grouped terms, was used for MEDLINE, and was modified to suit other databases.

Medline (Pubmed) search strategy.

Motor complications

Dyskinesia

Fluctuation

On/off time

Wearing-off.

or/1–5

Parkinson's disease

Parkinson disease

PD

or/7–9

6 and 10

Data extraction

For each study, information was carefully collected from all eligible publications, including first author, year of publication, number of cases and controls, sex ratio, mean age of cases and controls, duration of PD disease, mean L-dopa dosage, intervention of the trial groups, and main outcome measures. For studies including subjects of different arms, data were extracted separately for each arm whenever possible. If more than one paper with the same study was found, the one that contained the most detailed was reviewed. If the data for meta-analysis were missing or only expressed graphically, we tried to contact authors to inquire for further information or calculated by ourselves if available or excluded.

Quality assessment of RCTs

The methodological quality of RCTs was assessed independently using the Cochrane Handbook for Systematic Reviews of Interventions. Two investigators independently evaluated the methodological quality of the included articles. Disagreements were resolved through consensus or discussed with a third author.

Statistical analysis

We considered the main outcome measures as continuous data, and then were given an estimate of the combined overall effect sizes utilizing SMD with a random effects model. We used a random effects rather than a fixed effects model because of this takes into account heterogeneity between multi-studies. We looked for publication bias using funnel plotting, Egger's test and Begg's test. SMD with its 95% CI was used to assess the strength of efficacy of CDS in PD patients with motor complications. If any trials with two or more treatment arms were included, then we predetermine two methods to resolve the matter: (1) if the study was comparing two different drugs within the same drug class versus control, then the data for those drugs were combined to give one comparison of adjuvant therapy versus control; (2) if the study was comparing the same drug but at different dosages versus control, then the arm using generally recommended dose was chosen for inclusion in the analysis. Difference between n groups was assessed by partitioning heterogeneity and using the χ2 distribution with n-1 degrees of freedom (df), where n equals the number of groups. All analysis was performed with Revman version 5.1. Probability value p < 0.05 were considered significant.