背景:药物使用障碍(DUD)是全球发病率和死亡率的主要因素。瑞典广泛的国家登记册为研究2001-2015年瑞典DUD发病和复发的时空模式提供了基础。
方法:要识别空间上的DUD模式,对于年龄在15-35岁的瑞典人,采用SaTScan进行时空聚类。我们每年都使用有关瑞典人口统计区(DeSO)居住地点的信息。聚类分析使用泊松概率模型和零假设,即每个DeSO中的预期病例数与DeSO的种群规模成正比。由于SaTScan结果可能不稳定,采取步骤确定稳定的簇,并细化和优化簇大小。使用Gi*统计将每个性别-登记组合的结果与空间聚类的结果进行比较。时空扫描模型还运行了对邻里社会经济状况的调整,以确定与教育有关的DUD患病率,收入,失业和领取社会福利。
结果:DUD患病率随时间增加。在犯罪和医疗登记方面,男性比女性产生更多的集群。女性DUD患病率随着时间的推移而增加,尤其是2012年之后。在两个登记册中,DUD率的相关性高于性别。男性集群从2004年开始出现,而女性犯罪集群在2007年之后出现,女性医疗集群直到2010年才出现。到2013年,所有性别登记组合都存在集群。男性犯罪集团集中在斯德哥尔摩,哥德堡和马尔默是男性和女性医疗集群。与医疗DUD集群相比,邻里SES与犯罪分布的相关性更高。斯德哥尔摩在SES特别高的地区发现了核心集群的持续差距。
结论:确认了瑞典DUD的持续热点以及新的和新兴的热点,尤其是在斯德哥尔摩,哥德堡和马尔默。与性别相比,各登记册的DUD率存在更高的相关性。这些发现有助于监测当前的药物问题,并有助于确定潜在的传播模式和DUD的重要因果途径。
Drug Use Disorder (DUD) is a major contributor to world-wide morbidity and mortality. The extensive national registers in Sweden provide the basis for a study of spatial and temporal patterns of DUD onset and recurrence in Sweden from 2001-2015.
To identify patterns of DUD over space, time and gender for Swedish individuals aged 15-35, space-time clustering using SaTScan was applied. We used yearly information on residential locations in Demographic Statistical Areas (DeSO) for all of Sweden. The clustering analysis used a Poisson probability model and a null hypothesis that the expected number of cases in each DeSO was proportional to the population size of DeSOs. As SaTScan results can be unstable, steps were taken to determine stable clusters and to refine and optimize cluster size. Results for each gender-register combination were compared to the results of spatial clustering using Gi* statistics. The space-time scanning model was also run with an adjustment for neighborhood socioeconomic status to determine DUD prevalence as it relates to education, income, unemployment and receipt of social welfare.
DUD prevalence increased over time. Males yielded more significant clusters than females for both criminal and medical registers. Female DUD prevalence rates increased over time, especially after 2012. Higher correlations in DUD rates existed across the two registers than across gender. Male clusters were present from 2004 onwards while female-criminal clusters appeared after 2007, and female-medical clusters not until 2010. By 2013, clusters existed for all gender-register combinations. Male-criminal clusters were concentrated in Stockholm, Göteborg and Malmö as were male and female-medical clusters. Neighborhood SES was more highly related to the distribution of criminal than medical DUD clusters. A persistent gap in core clusters was identified in Stockholm in an area with notably high SES.
Persistent hotspots of DUD in Sweden were confirmed as well as new and emerging hotspots, especially in Stockholm, Göteborg and Malmö. Higher correlations existed in DUD rates across registers than across gender. The findings are useful for monitoring the current drug problem and for identifying drivers underlying patterns of spread and important causal pathways to DUD.