PDF(Pubmed)
Abstract：
UNASSIGNED: High-income countries offer social assistance (welfare) programs to help alleviate poverty for people with little or no income. These programs have become increasingly conditional and stringent in recent decades based on the premise that transitioning people from government support to paid work will improve their circumstances. However, many people end up with low-paying and precarious jobs that may cause more poverty because they lose benefits such as housing subsidies and health and dental insurance, while incurring job-related expenses. Conditional assistance programs are also expensive to administer and cause stigma. A guaranteed basic income (GBI) has been proposed as a more effective approach for alleviating poverty, and several experiments have been conducted in high-income countries to investigate whether GBI leads to improved outcomes compared to existing social programs.
UNASSIGNED: The aim of this review was to conduct a synthesis of quantitative evidence on GBI interventions in high-income countries, to compare the effectiveness of various types of GBI versus \"usual care\" (including existing social assistance programs) in improving poverty-related outcomes.
UNASSIGNED: Searches of 16 academic databases were conducted in May 2022, using both keywords and database-specific controlled vocabulary, without limits or restrictions on language or date. Sources of gray literature (conference, governmental, and institutional websites) were searched in September 2022. We also searched reference lists of review articles, citations of included articles, and tables of contents of relevant journals in September 2022. Hand searching for recent publications was conducted until December 2022.
UNASSIGNED: We included all quantitative study designs except cross-sectional (at one timepoint), with or without control groups. We included studies in high income countries with any population and with interventions meeting our criteria for GBI: unconditional, with regular payments in cash (not in-kind) that were fixed or predictable in amount. Although two primary outcomes of interest were selected a priori (food insecurity, and poverty level assessed using official, national, or international measures), we did not screen studies on the basis of reported outcomes because it was not possible to define all potentially relevant poverty-related outcomes in advance.
UNASSIGNED: We followed the Campbell Collaboration conduct and reporting guidelines to ensure a rigorous methodology. The risk of bias was assessed across seven domains: confounding, selection, attrition, motivation, implementation, measurement, and analysis/reporting. We conducted meta-analyses where results could be combined; otherwise, we presented the results in tables. We reported effect estimates as standard mean differences (SMDs) if the included studies reported them or provided sufficient data for us to calculate them. To compare the effects of different types of interventions, we developed a GBI typology based on the characteristics of experimental interventions as well as theoretical conceptualizations of GBI. Eligible poverty-related outcomes were classified into categories and sub-categories, to facilitate the synthesis of the individual findings. Because most of the included studies analyzed experiments conducted by other researchers, it was necessary to divide our analysis according to the \"experiment\" stage (i.e., design, recruitment, intervention, data collection) and the \"study\" stage (data analysis and reporting of results).
UNASSIGNED: Our searches yielded 24,476 records from databases and 80 from other sources. After screening by title and abstract, the full texts of 294 potentially eligible articles were retrieved and screened, resulting in 27 included studies on 10 experiments. Eight of the experiments were RCTs, one included both an RCT site and a \"saturation\" site, and one used a repeated cross-sectional design. The duration ranged from one to 5 years. The control groups in all 10 experiments received \"usual care\" (i.e., no GBI intervention). The total number of participants was unknown because some of the studies did not report exact sample sizes. Of the studies that did, the smallest had 138 participants and the largest had 8019. The risk of bias assessments found \"some concerns\" for at least one domain in all 27 studies and \"high risk\" for at least one domain in 25 studies. The risk of bias was assessed as high in 21 studies due to attrition and in 22 studies due to analysis and reporting bias. To compare the interventions, we developed a classification framework of five GBI types, four of which were implemented in the experiments, and one that is used in new experiments now underway. The included studies reported 176 poverty-related outcomes, including one pre-defined primary outcome: food insecurity. The second primary outcome (poverty level assessed using official, national, or international measures) was not reported in any of the included studies. We classified the reported outcomes into seven categories: food insecurity (as a category), economic/material, physical health, psychological/mental health, social, educational, and individual choice/agency. Food insecurity was reported in two studies, both showing improvements (SMD = -0.57, 95% CI: -0.65 to -0.49, and SMD = -0.41, 95% CI: -0.57 to -0.26) which were not pooled because of different study designs. We conducted meta-analyses on four secondary outcomes that were reported in more than one study: subjective financial well-being, self-rated overall physical health, self-rated life satisfaction, and self-rated mental distress. Improvements were reported, except for overall physical health or if the intervention was similar to existing social assistance. The results for the remaining 170 outcomes, each reported in only one study, were summarized in tables by category and subcategory. Adverse effects were reported in some studies, but only for specific subgroups of participants, and not consistently, so these results may have been due to chance.
UNASSIGNED: The results of the included studies were difficult to synthesize because of the heterogeneity in the reported outcomes. This was due in part to poverty being multidimensional, so outcomes covered various aspects of life (economic, social, psychological, educational, agency, mental and physical health). Evidence from future studies would be easier to assess if outcomes were measured using more common, validated instruments. Based on our analysis of the included studies, a supplemental type of GBI (provided along with existing programs) may be effective in alleviating poverty-related outcomes. This approach may also be safer than a wholesale reform of existing social assistance approaches, which could have unintended consequences.
UNASSIGNED: The aim of this review was to conduct a synthesis of quantitative evidence on GBI interventions in high-income countries, to compare the effectiveness of various types of GBI versus \"usual care\" (including existing social assistance programs) in improving poverty-related outcomes.
UNASSIGNED: Searches of 16 academic databases were conducted in May 2022, using both keywords and database-specific controlled vocabulary, without limits or restrictions on language or date. Sources of gray literature (conference, governmental, and institutional websites) were searched in September 2022. We also searched reference lists of review articles, citations of included articles, and tables of contents of relevant journals in September 2022. Hand searching for recent publications was conducted until December 2022.
UNASSIGNED: We included all quantitative study designs except cross-sectional (at one timepoint), with or without control groups. We included studies in high income countries with any population and with interventions meeting our criteria for GBI: unconditional, with regular payments in cash (not in-kind) that were fixed or predictable in amount. Although two primary outcomes of interest were selected a priori (food insecurity, and poverty level assessed using official, national, or international measures), we did not screen studies on the basis of reported outcomes because it was not possible to define all potentially relevant poverty-related outcomes in advance.
UNASSIGNED: We followed the Campbell Collaboration conduct and reporting guidelines to ensure a rigorous methodology. The risk of bias was assessed across seven domains: confounding, selection, attrition, motivation, implementation, measurement, and analysis/reporting. We conducted meta-analyses where results could be combined; otherwise, we presented the results in tables. We reported effect estimates as standard mean differences (SMDs) if the included studies reported them or provided sufficient data for us to calculate them. To compare the effects of different types of interventions, we developed a GBI typology based on the characteristics of experimental interventions as well as theoretical conceptualizations of GBI. Eligible poverty-related outcomes were classified into categories and sub-categories, to facilitate the synthesis of the individual findings. Because most of the included studies analyzed experiments conducted by other researchers, it was necessary to divide our analysis according to the \"experiment\" stage (i.e., design, recruitment, intervention, data collection) and the \"study\" stage (data analysis and reporting of results).
UNASSIGNED: Our searches yielded 24,476 records from databases and 80 from other sources. After screening by title and abstract, the full texts of 294 potentially eligible articles were retrieved and screened, resulting in 27 included studies on 10 experiments. Eight of the experiments were RCTs, one included both an RCT site and a \"saturation\" site, and one used a repeated cross-sectional design. The duration ranged from one to 5 years. The control groups in all 10 experiments received \"usual care\" (i.e., no GBI intervention). The total number of participants was unknown because some of the studies did not report exact sample sizes. Of the studies that did, the smallest had 138 participants and the largest had 8019. The risk of bias assessments found \"some concerns\" for at least one domain in all 27 studies and \"high risk\" for at least one domain in 25 studies. The risk of bias was assessed as high in 21 studies due to attrition and in 22 studies due to analysis and reporting bias. To compare the interventions, we developed a classification framework of five GBI types, four of which were implemented in the experiments, and one that is used in new experiments now underway. The included studies reported 176 poverty-related outcomes, including one pre-defined primary outcome: food insecurity. The second primary outcome (poverty level assessed using official, national, or international measures) was not reported in any of the included studies. We classified the reported outcomes into seven categories: food insecurity (as a category), economic/material, physical health, psychological/mental health, social, educational, and individual choice/agency. Food insecurity was reported in two studies, both showing improvements (SMD = -0.57, 95% CI: -0.65 to -0.49, and SMD = -0.41, 95% CI: -0.57 to -0.26) which were not pooled because of different study designs. We conducted meta-analyses on four secondary outcomes that were reported in more than one study: subjective financial well-being, self-rated overall physical health, self-rated life satisfaction, and self-rated mental distress. Improvements were reported, except for overall physical health or if the intervention was similar to existing social assistance. The results for the remaining 170 outcomes, each reported in only one study, were summarized in tables by category and subcategory. Adverse effects were reported in some studies, but only for specific subgroups of participants, and not consistently, so these results may have been due to chance.
UNASSIGNED: The results of the included studies were difficult to synthesize because of the heterogeneity in the reported outcomes. This was due in part to poverty being multidimensional, so outcomes covered various aspects of life (economic, social, psychological, educational, agency, mental and physical health). Evidence from future studies would be easier to assess if outcomes were measured using more common, validated instruments. Based on our analysis of the included studies, a supplemental type of GBI (provided along with existing programs) may be effective in alleviating poverty-related outcomes. This approach may also be safer than a wholesale reform of existing social assistance approaches, which could have unintended consequences.
摘要：
■高收入国家提供社会援助（福利）计划，以帮助收入很少或没有收入的人减轻贫困。近几十年来，这些计划变得越来越有条件和严格，前提是人们从政府支持过渡到有偿工作将改善他们的状况。然而,许多人最终获得了低薪和不稳定的工作，这可能会导致更多的贫困，因为他们失去了诸如住房补贴，健康和牙科保险之类的福利，同时产生与工作相关的费用。有条件援助计划的管理成本也很高，并造成耻辱。有保障的基本收入（GBI）已被提议作为减轻贫困的更有效方法，在高收入国家进行了一些实验，以调查与现有的社会计划相比，GBI是否可以改善结果。
■本次审查的目的是对高收入国家GBI干预措施的定量证据进行综合，比较各种类型的GBI与“常规护理”（包括现有的社会援助计划）在改善贫困相关结果方面的有效性。
■2022年5月对16个学术数据库进行了搜索，使用关键字和特定于数据库的受控词汇，对语言或日期没有限制。灰色文献的来源(会议，政府,和机构网站)于2022年9月进行了搜索。我们还搜索了评论文章的参考列表，引用的文章，以及2022年9月相关期刊目录。手工搜索最近的出版物一直持续到2022年12月。
■我们包括了除横截面外的所有定量研究设计（在一个时间点），有或没有对照组。我们纳入了高收入国家的研究，这些国家的人口和干预措施都符合我们的GBI标准：无条件，定期支付固定或可预测的现金(非实物)。尽管先验地选择了两个主要的关注结果(粮食不安全，和使用官方评估的贫困水平，国家,或国际措施)，我们没有根据报告的结局筛选研究,因为我们不可能提前确定所有潜在相关的贫困相关结局.
■我们遵循了坎贝尔合作行为和报告准则，以确保严格的方法。在七个领域评估了偏见的风险：混杂，选择,自然减员,动机，实施，测量,和分析/报告。我们进行了荟萃分析，结果可以合并；否则，我们在表格中给出了结果.如果纳入的研究报告了效果估计值或为我们提供了足够的数据来计算它们，我们将其报告为标准平均差异（SMD）。为了比较不同类型干预措施的效果，我们根据实验干预措施的特点以及GBI的理论概念化,开发了GBI类型学.符合条件的与贫困有关的成果被分为类别和子类别，以促进个人发现的综合。因为大多数纳入的研究分析了其他研究人员进行的实验，有必要根据“实验”阶段划分我们的分析(即，设计,招募,干预，数据收集)和“研究”阶段(数据分析和结果报告)。
■我们的搜索产生了来自数据库的24,476条记录和来自其他来源的80条记录。经过标题和摘要筛选，检索和筛选了294篇可能符合条件的文章的全文，在10个实验中纳入了27项研究。八个实验是随机对照试验,一个包括RCT站点和“饱和”站点，一个人使用了重复的横截面设计。持续时间为1至5年。所有10个实验的对照组都接受了“常规护理”(即，无GBI干预)。参与者总数未知，因为一些研究没有报告确切的样本量。在所做的研究中，最小的有138名参与者，最大的有8019名。在所有27项研究中，偏倚风险评估发现至少一个领域存在“一些担忧”，在25项研究中至少一个领域存在“高风险”。在21项研究中，由于减员和22项研究中，由于分析和报告偏倚，偏倚风险被评估为高。为了比较干预措施，我们开发了五种GBI类型的分类框架，其中四个是在实验中实施的，和一个正在进行的新实验中使用的。纳入的研究报告了176项与贫困有关的成果，包括一个预定义的主要结果：粮食不安全。第二个主要结果(使用官方评估的贫困水平，国家,或国际措施)在任何纳入的研究中均未报告。我们将报告的结果分为七类：粮食不安全(作为一个类别)，经济/材料，身体健康,心理/心理健康，社会,教育,和个人选择/机构。在两项研究中报告了粮食不安全，两者均显示改善(SMD=-0.57,95%CI:-0.65至-0.49,和SMD=-0.41,95%CI:-0.57至-0.26),但由于研究设计不同,未合并.我们对一项以上研究中报告的四个次要结局进行了荟萃分析：主观财务幸福感，自我评估的总体身体健康，自我评估的生活满意度，和自我评估的精神困扰。据报道有所改善，除了总体身体健康或干预措施类似于现有的社会援助。其余170项成果的结果，每个人只在一项研究中报告，按类别和子类别在表中进行了总结。一些研究报告了不良反应，但仅限于特定的参与者亚组，并不一致，所以这些结果可能是偶然的。
■纳入研究的结果难以综合，因为报告的结果存在异质性。这部分是由于贫困是多层面的，所以结果涵盖了生活的各个方面(经济，社会,心理,教育,agency,心理和身体健康)。如果使用更常见的方法来衡量结果，那么来自未来研究的证据将更容易评估，经过验证的仪器。根据我们对纳入研究的分析，补充类型的GBI（与现有计划一起提供）可能有效地减轻与贫困相关的结果。这种方法也可能比现有社会援助方法的大规模改革更安全，这可能会产生意想不到的后果。
■本次审查的目的是对高收入国家GBI干预措施的定量证据进行综合，比较各种类型的GBI与“常规护理”（包括现有的社会援助计划）在改善贫困相关结果方面的有效性。
■2022年5月对16个学术数据库进行了搜索，使用关键字和特定于数据库的受控词汇，对语言或日期没有限制。灰色文献的来源(会议，政府,和机构网站)于2022年9月进行了搜索。我们还搜索了评论文章的参考列表，引用的文章，以及2022年9月相关期刊目录。手工搜索最近的出版物一直持续到2022年12月。
■我们包括了除横截面外的所有定量研究设计（在一个时间点），有或没有对照组。我们纳入了高收入国家的研究，这些国家的人口和干预措施都符合我们的GBI标准：无条件，定期支付固定或可预测的现金(非实物)。尽管先验地选择了两个主要的关注结果(粮食不安全，和使用官方评估的贫困水平，国家,或国际措施)，我们没有根据报告的结局筛选研究,因为我们不可能提前确定所有潜在相关的贫困相关结局.
■我们遵循了坎贝尔合作行为和报告准则，以确保严格的方法。在七个领域评估了偏见的风险：混杂，选择,自然减员,动机，实施，测量,和分析/报告。我们进行了荟萃分析，结果可以合并；否则，我们在表格中给出了结果.如果纳入的研究报告了效果估计值或为我们提供了足够的数据来计算它们，我们将其报告为标准平均差异（SMD）。为了比较不同类型干预措施的效果，我们根据实验干预措施的特点以及GBI的理论概念化,开发了GBI类型学.符合条件的与贫困有关的成果被分为类别和子类别，以促进个人发现的综合。因为大多数纳入的研究分析了其他研究人员进行的实验，有必要根据“实验”阶段划分我们的分析(即，设计,招募,干预，数据收集)和“研究”阶段(数据分析和结果报告)。
■我们的搜索产生了来自数据库的24,476条记录和来自其他来源的80条记录。经过标题和摘要筛选，检索和筛选了294篇可能符合条件的文章的全文，在10个实验中纳入了27项研究。八个实验是随机对照试验,一个包括RCT站点和“饱和”站点，一个人使用了重复的横截面设计。持续时间为1至5年。所有10个实验的对照组都接受了“常规护理”(即，无GBI干预)。参与者总数未知，因为一些研究没有报告确切的样本量。在所做的研究中，最小的有138名参与者，最大的有8019名。在所有27项研究中，偏倚风险评估发现至少一个领域存在“一些担忧”，在25项研究中至少一个领域存在“高风险”。在21项研究中，由于减员和22项研究中，由于分析和报告偏倚，偏倚风险被评估为高。为了比较干预措施，我们开发了五种GBI类型的分类框架，其中四个是在实验中实施的，和一个正在进行的新实验中使用的。纳入的研究报告了176项与贫困有关的成果，包括一个预定义的主要结果：粮食不安全。第二个主要结果(使用官方评估的贫困水平，国家,或国际措施)在任何纳入的研究中均未报告。我们将报告的结果分为七类：粮食不安全(作为一个类别)，经济/材料，身体健康,心理/心理健康，社会,教育,和个人选择/机构。在两项研究中报告了粮食不安全，两者均显示改善(SMD=-0.57,95%CI:-0.65至-0.49,和SMD=-0.41,95%CI:-0.57至-0.26),但由于研究设计不同,未合并.我们对一项以上研究中报告的四个次要结局进行了荟萃分析：主观财务幸福感，自我评估的总体身体健康，自我评估的生活满意度，和自我评估的精神困扰。据报道有所改善，除了总体身体健康或干预措施类似于现有的社会援助。其余170项成果的结果，每个人只在一项研究中报告，按类别和子类别在表中进行了总结。一些研究报告了不良反应，但仅限于特定的参与者亚组，并不一致，所以这些结果可能是偶然的。
■纳入研究的结果难以综合，因为报告的结果存在异质性。这部分是由于贫困是多层面的，所以结果涵盖了生活的各个方面(经济，社会,心理,教育,agency,心理和身体健康)。如果使用更常见的方法来衡量结果，那么来自未来研究的证据将更容易评估，经过验证的仪器。根据我们对纳入研究的分析，补充类型的GBI（与现有计划一起提供）可能有效地减轻与贫困相关的结果。这种方法也可能比现有社会援助方法的大规模改革更安全，这可能会产生意想不到的后果。
