A cross-sectional study was conducted of workers aged 40 or over from 7 private companies in various industries such as electrical appliances, pharmaceuticals, the wholesale industry, and the food industry in Japan. The data were obtained from the results of annual medical examinations in each company in 2016, questionnaire surveys conducted with workers, and health insurance claims. According to Article 44 (2) of the Occupational Safety and Health Regulations, blood tests are a legal requirement for those aged 40 or over in medical examinations. In addition, the prevalence of diabetes increases with age, and prevalence under 40 years is approximately one-third that of those in their 40s.31 For these reasons, we considered that including workers under 40 would affect the analysis, and therefore we limited our focus to those aged 40 or over in this study. In conducting this study, we explained the purpose to the management and workers via e-mail, intranet homepage, or the Safety and Health Committee, and the consent of the workers was obtained. The research protocol was approved by the Ethics Committee of Medical Research, University of Occupational and Environmental Health, Kitakyushu, Japan.
Classification of participants
On the basis of the diabetes diagnosis criteria of the Japan Diabetes Society,32 the participants with fasting blood glucose of less than 110 mg/dL and HbA1c [National Glycohemoglobin Standardization Program (NGSP)] (hereinafter HbA1c) of less than 6.0% were defined as the normal group. From the health insurance claims, if participants had taken anti-diabetic drugs from 3 months before the questionnaire through to the response month, they were classified as being in a diabetic treatment group regardless of their blood tests. Participants with insulin treatment were also included in the diabetic treatment group. In addition, information on the diagnosis name and prescription content was also collected from the health insurance claims. The reason for confirming the prescription history for 3 months was that the prescription of anti-diabetic drugs is mostly for 90 days or less in Japan.33 Regarding insulin treatment, some previous studies reported that it was a factor in the occurrence of presenteeism,28,34 but no significant difference was found in the insulin treatment group in our previous studies.15 Thus we considered that insulin had almost no effect and decided to include it in the diabetic treatment group in this study. We excluded participants with data for only one test and those with casual blood glucose instead of fasting blood glucose because both fasting blood glucose and HbA1c were used. In addition, type 1 diabetes and other specified diabetes were excluded. We did not consider typical symptoms of diabetes such as dry mouth or polyuria.
Treatment control targets are the same in the United States16 and Japan32 where the target for prevention of complications is HbA1c <7%, and if it is difficult to strengthen the treatment due to side effects such as hypoglycemia, the target is HbA1c <8%. Therefore, we divided participants into three groups according to treatment control targets: good control group—HbA1c <7%; intermediate control group—7% ≤ HbA1c <8%; and poor control group—8% ≤ HbA1c. In addition, the number of anti-diabetic drugs was considered and classified as monotherapy group or combination therapy group if the participants were taking two or more drugs. Currently, some compounding drugs for diabetes are used, and participants taking these compounding drugs were included in the combination therapy group as two types of drugs were essentially prescribed.
We defined productivity loss due to presenteeism as “presenteeism loss” and evaluated the loss using the Quantity and Quality (QQ) method.35 The evaluation using this method was performed through the following steps. First, we asked whether participants had any health problems or conditions during their work in the past month. If the answer was “no,” the presenteeism loss was set to zero. If the answer was “yes,” we asked the participants to identify their health problems from a list of 14 conditions and to select the one condition that most affected their work. If the conditions did not affect their work, the presenteeism loss was also set to 0. The 14 conditions were as follows: (1) troubled by allergies (e.g. hay fever); (2) skin diseases/itchiness (e.g. eczema, atopic dermatitis); (3) disorders caused by infections (e.g. cold, influenza, gastroenteritis); (4) gastrointestinal disorders (e.g. recurrent diarrhea, constipation); (5) pain in arm and leg joints or lack of mobility (e.g. arthritis); (6) back pain; (7) painful neck or stiff shoulder; (8) headaches (e.g. migraine, chronic headache); (9) tooth trouble (e.g. toothache); (10) mental health problems (e.g. depression, anxiety); (11) insomnia, insufficient sleep; (12) a sense of weariness or fatigue; (13) eye problems (e.g. loss of vision, eyestrain, dry eye, glaucoma); and (14) other.
Second, we asked participants to describe the quantity and quality of the work when they had the identified problem compared with those when they had no problems. The answers were scored from 0 (unable to work at all) to 10 (normal). Finally, presenteeism loss was calculated using the following equation:
Presenteeism loss = 100 – Quantity (range: 0–10) × Quality (range: 0–10)
In our previous study,15 the top 10% with a presenteeism loss score of 51 or higher, was defined as high presenteeism loss, and the top 20%, with a score of 36 or higher, was defined as moderate presenteeism loss, but similar results were obtained with these two indicators. Therefore, only 36 or higher was defined as presenteeism loss in this study.
Participant characteristics were summarized using means and standard deviations (SDs) for continuous variables and percentages for categorical variables. The presenteeism loss was calculated in each treatment control group compared with that of the normal group, and was also calculated for each therapy type compared with that of the normal group. Furthermore, the presenteeism loss in each treatment control group of combination therapy was also calculated. We performed logistic regression analysis with each treatment control group, number of anti-diabetic drugs, and each treatment control group of combination therapy as the independent variable, and presenteeism loss using the QQ method, that is, 36 or higher as the dependent variable. For all analyses, the normal group was the reference category, and age, sex, company, and occupation were adjusted. Adjusted odds ratios (aORs) and corresponding 95% confidence intervals (CIs) were calculated. In all analysis, p-values <0.05 were considered statistically significant. All analyses were performed using STATA Version 16 (StataCorp LLC, College Station, TX).