Comparative effectiveness and acceptability of non-pharmacological interventions for insomnia in older people: a systematic review and network meta-analysis

다국어 초록 (Multilingual Abstract)

Objectives

The prevalence of insomnia, which has been reported to be 20-55%, increases with the physiological aging process, multiple aging-associated pathological conditions, and increased prescription medications in the elderly. Elderly insomnia has been reported to be associated with various negative outcomes such as cognitive impairment, dementia, depression and cardiovascular events. Therefore, elderly insomnia has been regarded as causing huge socioeconomic burden, not simply a health problem. The treatment of elderly insomnia is mainly focused on cognitive behavioral therapy (CBT) for insomnia, and in consideration of the therapeutic benefits and risks, pharmacotherapies, in particular, the use of benzodiazepines is very limited. Despite the importance of non-pharmacological treatments for treating elderly insomnia, the comparative advantage between the available non-pharmacological treatments has not been determined. The purpose of this study was to determine the comparative advantage of the effectiveness and acceptability of non-pharmacological treatments available in elderly insomnia based on the evidence to date, using the methodology of conventional pair-wise meta-analysis and network meta-analysis.

Methods

Comprehensive searches in 13 English, Korean and Chinese databases were performed to search randomized controlled trials (RCTs) evaluating the effectiveness of non-pharmacological treatments for elderly insomnia, up to August 5, 2019. The non-pharmacological treatments analyzed in this review were based on the treatment options in evidence-based recommendations made by international sleep disorder experts in 2009, the most recent clinical guidelines for insomnia in the elderly. The methodological quality of the studies included was assessed using the Cochrane’s risk of bias assessment tool. Review Manager version 5.3 for Windows and Stata version 16.0 were used to perform conventional pair-wise meta-analysis and network meta-analysis, respectively. Priorities of interventions for each outcome were estimated using surface under the cumulative ranking probabilities (SUCRA).

Results

Twenty-eight RCTs involving 2,391 participants were included in this review. The following treatments were compared in terms of effectiveness as assessed by the Pittsburgh sleep quality index (PSQI) total score: (1) acupuncture, (2) acupuncture combined with benzodiazepines, (3) acupuncture combined with relaxation, (4) behavioral treatment (BT), (5) benzodiazepines, (6) benzodiazepines combined with CBT, (7) benzodiazepines combined with exercise, (8) CBT, (9) sleep education, (10) melatonin, (11) qigong, (12) relaxation and (13) wait-list. The results showed that compared to wait-list, acupuncture (mean difference (MD) -4.37, 95% confidence interval (CI) -8.53 to -0.12), acupuncture combined with benzodiazepines (MD -5.20, 95% CI -9.82 to -0.57), BT (MD -10.44, 95% CI -17.31 to -3.58), benzodiazepines (MD -4.28, 95% CI -8.45 to -0.11), benzodiazepines combined with CBT (MD -7.18, 95% CI -12.17 to -2.19), and CBT (MD -4.92, 95% CI -8.63 to -1.22) showed significant superiority. Sleep education (MD -4.31, 95% CI -9.78 to 1.15) or relaxation (MD -2.44, 95% CI -5.98 to 1.09) did not show significant or near-significant differences compared with the wait-list. The sensitivity analysis removing the outlier did not significantly affect the results. On the other hand, no significant comparative superiority or inferiority was found between the treatments in terms of acceptability as all-cause drop-out. In terms of safety as the incidence of adverse event, nearly half of the interventions were excluded from network meta-analysis because they did not form a network. As results, acupuncture (odds ratio (OR) 0.00, 95% CI 0.00 to 0.08), acupuncture combined with relaxation (OR 0.00, 95% CI 0.00 to 0.01), benzodiazepines (OR 0.00, 95% CI 0.00 to 0.25), melatonin (OR 0.00, 95% CI 0.00 to 0.30), and relaxation (OR 0.00, 95% CI 0.00 to 0.08) seemed significantly safer than selective serotonin reuptake inhibitor (SSRI). Moreover, there was insufficient evidence in pair-wise meta-analysis that acupuncture is safer than benzodiazepines (OR 0.28, 95% CI 0.09 to 0.89) and acupuncture combined with relaxation is safer than SSRI (OR 0.00, 95% CI 0.00 to 0.01).

Conclusion

In this systematic review, the methodology of conventional pair-wise meta-analysis and network meta-analysis was used to determine the comparative advantage of non-pharmacological treatments for elderly insomnia in terms of the effectiveness and acceptability, based on current evidences. In terms of effectiveness in PSQI total score, compared to wait-list, acupuncture, acupuncture combined with benzodiazepines, BT, benzodiazepines, benzodiazepines combined with CBT, and CBT showed superior benefits. Combined treatments including benzodiazepines combined with CBT and acupuncture combined with benzodiazepines were generally superior to monotherapies. In terms of acceptability, there was not enough data to draw conclusions. In terms of safety, there was insufficient evidence that acupuncture is overall safe. The results of this review suggest that more head-to-head trials of non-pharmacological interventions for insomnia in the older people should be conducted and national supports are also needed because the condition in this population causes serious medical and social burdens. Moreover, in treating insomnia in older people, clinicians need to establish the treatment plan by balancing the patient’s benefits and risks in context of shared-decision making.

국문 초록 (Abstract)

연구 목적

불면증은 노인에서 생리적 노화과정이나 동반된 병리적 상태, 복용 약물의 증가와 함께 그 유병률이 증가하여, 연구에 따라 유병률이 20-55%에 이르는 것으로 보고되고 있다. 노인에서 불면증과 인지장애, 치매, 우울증, 심혈관계 사건 등 각종 부정적인 결과와의 관련성이 보고되고 있으며, 따라서 노인 불면증은 단순히 건강 관련 문제가 아닌 사회경제적 부담을 야기하는 것으로 간주된다. 한편, 노인 불면증의 치료는 불면증에 대한 인지행동치료 (cognitive behavioral therapy for insomnia)를 위주로 하고 있으며, 치료 상의 이득과 위해를 감안하여 약물치료, 특히 benzodiazepines계 약물은 매우 제한적으로 사용된다. 따라서 노인 불면증 치료에 있어 비약물요법의 중요성이 높음에도 불구하고, 사용할 수 있는 비약물요법 간의 비교 우위는 결정되지 않은 상황이다. 본 연구의 목적은 일반 메타분석과 네트워크 메타분석의 방법론을 사용하여, 현재까지의 근거를 토대로 노인 불면증에서 사용 가능한 비약물요법의 효과성과 수용가능성에 관한 비교 우위를 결정하는 것이었다.

연구 방법

2019년 8월 5일까지 13개의 영어, 한국, 중국 데이터베이스 검색을 통해, 노인 불면증에 대한 비약물요법의 효과성을 평가하는 무작위 대조군 임상시험을 수집했다. 본 연구에서 비약물요법은 노인 불면증에 대한 가장 최근의 임상진료지침인 2009년에 다국적 전문가에 의해 제작된 근거-기반 권고문에 수록된 요법을 기준으로 하였다. 포함된 연구들의 방법론적 질은 코크란 연합의 비뚤림 위험 평가 도구를 사용하여 평가되었으며, Review Manager version 5.3 for Windows와 Stata version 16.0를 사용하여 각각 일반 메타분석과 네트워크 메타분석을 실시했다. 각 지표에 대한 우선순위는 SUCRA를 사용하여 추정되었다.

연구 결과

검색 및 선별 결과, 2,391명의 연구대상자를 포함하는 28편의 무작위 대조군 임상시험이 본 연구에 포함되었다. 피츠버그 수면의 질 지수 총점으로 평가한 효과성 측면에서 다음의 치료들이 비교되었다: (1) 침치료, (2) 침치료와 benzodiazepines의 병용, (3) 침치료와 이완요법의 병용, (4) 행동요법, (5) benzodiazepines, (6) benzodiazepines과 인지행동치료의 병용, (7) benzodiazepines과 운동의 병용, (8) 인지행동치료, (9) 수면교육, (10) 멜라토닌, (11) 기공요법, (12) 이완요법, (13) 대기군. 분석 결과, 대기군과 비교하여, 침치료 (mean difference (MD) -4.37, 95% confidence interval (CI) -8.53 to -0.12), 침치료와 benzodiazepines의 병용 (MD -5.20, 95% CI -9.82 to -0.57), 행동요법 (MD -10.44, 95% CI -17.31 to -3.58), benzodiazepines (MD -4.28, 95% CI -8.45 to -0.11), benzodiazepines과 인지행동치료의 병용 (MD -7.18, 95% CI -12.17 to -2.19), 인지행동치료 (MD -4.92, 95% CI -8.63 to -1.22)는 유의한 우수성을 보였다. 한편, 수면교육 (MD -4.31, 95% CI -9.78 to 1.15)과 이완요법 (MD -2.44, 95% CI -5.98 to 1.09)은 대기군과 비교하여 유의한 차이 또는 유의성에 가까운 차이를 보이지 않았다. 이상치 데이터(outlier)를 제외한 민감도 분석은 본 결과에 유의한 영향을 미치지 않았다. 한편, 모든-원인 탈락으로 평가한 수용가능성에서는 치료들 간의 유의한 차이가 발견되지 않았다. 이상반응 발생으로 평가한 안전성에서는 네트워크를 이루지 못한 절반에 가까운 중재가 네트워크 메타분석에서 제외되었다. 그 결과, 선택적 세로토닌 재흡수 억제제와 비교하여, 침치료 (odds ratio (OR) 0.00, 95% CI 0.00 to 0.08), 침치료와 이완요법의 병용 (OR 0.00, 95% CI 0.00 to 0.01), benzodiazepines (OR 0.00, 95% CI 0.00 to 0.25), 멜라토닌 (OR 0.00, 95% CI 0.00 to 0.30), 이완요법 (OR 0.00, 95% CI 0.00 to 0.08)은 유의하게 우수한 안전성을 보였다. 또한, 일반 메타분석 결과에서 침치료는 benzodiazepines에 비해 (OR 0.28, 95% CI 0.09 to 0.89), 침치료와 이완요법의 병용은 선택적 세로토닌 재흡수 억제제 (OR 0.00, 95% CI 0.00 to 0.01)에 비해 유의하게 더 안전하다는 것이 확인되었다.

결론

본 체계적 문헌고찰에서는 일반 메타분석과 네트워크 메타분석 방법론을 사용하여, 현재까지의 근거를 토대로 노인 불면증에서 사용 가능한 비약물요법의 효과성과 수용가능성에 관한 비교 우위를 결정하고자 하였다. 그 결과, 효과성 측면에서 대기군과 비교하여 침치료, 침치료와 benzodiazepines의 병용, 행동요법, benzodiazepines, benzodiazepines과 인지행동치료의 병용, 인지행동치료는 유의한 우월성을 보였다. 특히 benzodiazepines과 인지행동치료의 병용 혹은 침치료와 benzodiazepines의 병용은 개별 요법에 비해 유의한 우월성을 보였다. 한편, 수용가능성에서는 결론을 내릴 만한 충분한 데이터가 존재하지 않았다. 안전성 측면에서는 침치료가 전반적으로 안전하다는 제한된 수준의 근거가 존재했다. 하지만 포함된 연구의 수가 적고, 특히 수용가능성과 안전성 측면에서의 데이터가 부족하므로, 확고한 결론을 도출하기 위해서는 노인 불면증에 비약물요법을 사용한 잘 설계된 대규모 무작위 대조군 임상시험이 더 수행될 필요성이 있다.

목차 (Table of Contents)

• LIST of TABLES iv
• LIST of FIGURES iv
• ABSTRACT vi
• ABBREVIATIONS ix
• I. INTRODUCTION 1
• II. METHODS 4
• 1. Search strategy 4
• 2. Eligible criteria 4
• 1) Types of studies 4
• 2) Types of participants 5
• 3) Types of interventions and comparators 5
• 4) Types of outcome measures 5
• 5) Timing of outcome assessment 6
• 6) Hierarchy of outcome measures 6
• 3. Study selection and data extraction 7
• 4. Risk of bias assessment 7
• 5. Data analysis 8
• 1) Pair-wise meta-analysis (conventional meta-analysis) 8
• 2) Network meta-analysis 8
• 2-1) Data coding 9
• 2-2) Network model establishment 10
• 2-3) Consistency and inconsistency test 10
• 2-4) Effect size 10
• 3) Additional analysis 11
• 4) Publication bias 11
• III. RESULTS 12
• 1. Characteristics of included studies 12
• 2. Risk of bias included studies 19
• 3. Comparative effectiveness 21
• 1) PSQI total score (Primary outcome) 21
• 1-1) Network model establishment 21
• 1-2) Consistency and inconsistency test 22
• 1-3) Effect size 23
• 1-4) Pair-wise meta-analysis 28
• 1-5) Sensitivity analysis 30
• 2) Polysomnography data (Secondary outcome) 33
• 2-1) Network model establishment 33
• 2-2) Pair-wise meta-analysis 34
• 4. Comparative acceptability 36
• 1) Drop-outs for any reasons (Primary outcome) 36
• 1-1) Network model establishment 36
• 1-2) Consistency and inconsistency test 37
• 1-3) Effect size 38
• 1-4) Pair-wise meta-analysis 43
• 2) Drop-outs for AEs (Secondary outcome) 44
• 2-1) Network model establishment 44
• 2-2) Consistency and inconsistency test 45
• 2-3) Effect size 46
• 2-4) Pair-wise meta-analysis 51
• 5. Safety 52
• 1) Incidence of AEs (Secondary outcome) 52
• 1-1) Network model establishment 52
• 1-2) Consistency and inconsistency test 53
• 1-3) Effect size 54
• 1-4) Pair-wise meta-analysis 58
• 6. Publication bias 59
• IV. DISCUSSION 61
• 1. Summary of evidence 61
• 2. Clinical interpretation 64
• 3. Role of network meta-analysis in complementary and integrative medicine field 66
• 4. Strengths and limitations 68
• V. CONCLUSION 70
• REFERENCES 71
• APPENDICES 77
• APPENDIX 1. PRISMA NMA checklist 77
• APPENDIX 2. Search strategy used in each database 80
• APPENDIX 3. Excluded studies and reason after full-text review 87
• APPENDIX 4. Results of individual studies for each outcome 97
• KOREAN ABSTRACT 104