Department of Occupational and Environmental Medicine, Telemark Hospital, Skien, Norway
Institute of Health and Society, University of Oslo Faculty of Medicine, Oslo, Norway.
BMJ Open. 2023 Mar 30;13(3):e064311. doi: 10.1136/bmjopen-2022-064311.
This study aimed to characterise participants lost to follow-up and identify possible factors associated with non-participation in a prospective population-based study of respiratory health in Norway. We also aimed to analyse the impact of potentially biased risk estimates associated with a high proportion of non-responders.
Prospective 5-year follow-up study.
Randomly selected inhabitants from the general population of Telemark County in south-eastern Norway were invited to fill in a postal questionnaire in 2013. Responders in 2013 were followed-up in 2018.
16 099 participants aged 16-50 years completed the baseline study. 7958 responded at the 5-year follow-up, while 7723 did not.
χ test was performed to compare demographic and respiratory health-related characteristics between those who participated in 2018 and those who were lost to follow-up. Adjusted multivariable logistic regression models were used to assess the relationship between loss to follow-up, background variables, respiratory symptoms, occupational exposure and interactions, and to analyse whether loss to follow-up leads to biased risk estimates.
7723 (49%) participants were lost to follow-up. Loss to follow-up was significantly higher for male participants, those in the youngest age group (16-30 years), those in lowest education level category and among current smokers (all p<0.001). In multivariable logistic regression analysis, loss to follow-up was significantly associated with unemployment (OR 1.34, 95% CI 1.22 to 1.46), reduced work ability (1.48, 1.35 to 1.60), asthma (1.22, 1.10 to 1.35), being woken by chest tightness (1.22, 1.11 to 1.34) and chronic obstructive pulmonary disease (1.81, 1.30 to 2.52). Participants with more respiratory symptoms and exposure to vapour, gas, dust and fumes (VGDF) (1.07 to 1.00-1.15), low-molecular weight (LMW) agents (1.19, 1.00 to 1.41) and irritating agents (1.15, 1.05 to 1.26) were more likely to be lost to follow-up. We found no statistically significant association of wheezing and exposure to LMW agents for all participants at baseline (1.11, 0.90 to 1.36), responders in 2018 (1.12, 0.83 to 1.53) and those lost to follow-up (1.07, 0.81 to 1.42).
The risk factors for loss to 5-year follow-up were comparable to those reported in other population-based studies and included younger age, male gender, current smoking, lower educational level and higher symptom prevalence and morbidity. We found that exposure to VGDF, irritating and LMW agents can be risk factors associated with loss to follow-up. Results suggest that loss to follow-up did not affect estimates of occupational exposure as a risk factor for respiratory symptoms.
本研究旨在描述失访参与者的特征，并确定与挪威前瞻性人群呼吸健康研究中不参与相关的可能因素。我们还旨在分析与高比例不响应者相关的潜在有偏风险估计的影响。
前瞻性 5 年随访研究。
随机选取挪威东南部泰勒马克郡的一般人群作为研究对象，邀请他们在 2013 年填写邮寄问卷。2013 年的应答者在 2018 年进行了随访。
16099 名年龄在 16-50 岁的参与者完成了基线研究。7958 人在 5 年随访时应答，而 7723 人未应答。
使用 χ 检验比较了 2018 年参与和失访的参与者之间的人口统计学和呼吸健康相关特征。使用多变量逻辑回归模型评估了失访、背景变量、呼吸症状、职业暴露和相互作用之间的关系，并分析了失访是否导致风险估计有偏差。
7723 名（49%）参与者失访。男性参与者、最年轻年龄组（16-30 岁）、最低教育水平类别和当前吸烟者的失访率明显更高（均 P<0.001）。在多变量逻辑回归分析中，失访与失业（OR 1.34，95%CI 1.22 至 1.46）、工作能力下降（1.48，1.35 至 1.60）、哮喘（1.22，1.10 至 1.35）、因胸闷而醒来（1.22，1.11 至 1.34）和慢性阻塞性肺疾病（1.81，1.30 至 2.52）显著相关。有更多呼吸症状和接触蒸气、气体、粉尘和烟雾（VGDF）（1.07 至 1.00-1.15）、低分子量（LMW）物质（1.19，1.00 至 1.41）和刺激性物质（1.15，1.05 至 1.26）的参与者更有可能失访。我们没有发现基线时喘息和接触 LMW 物质与所有参与者（1.11，0.90 至 1.36）、2018 年的应答者（1.12，0.83 至 1.53）和失访者（1.07，0.81 至 1.42）之间存在统计学显著关联。
失访 5 年随访的危险因素与其他基于人群的研究报告的危险因素相似，包括年龄较小、男性、当前吸烟、较低的教育水平以及更高的症状患病率和发病率。我们发现，接触 VGDF、刺激性和 LMW 物质可能是与失访相关的危险因素。结果表明，失访未影响职业暴露作为呼吸症状风险因素的估计。
