Of the 2,969 eligible younger adolescents who participated in Young-HUNT (YH1, 1997), 2,429 (81.8%) also participated in Young-HUNT 2 (YH2, 2001). Their average age in YH1 was 14 years and 53.4% were females. Nearly one in four had felt lonely often or very often while 14% were not content with their life. They reported having on average 4.2 (median 4, inter quartile range (IQR) 2-6) recent complaints, 1.2 (median 1, IQR 0-2) diseases and used 1.1 (median 1, IQR 0-2) types of medicines. One half (52%) had visited a physician within the last year and 92.3% had good or very good health.
Prevalence of CAM visits
One in fourteen (6.9%, 95% CI 5.9–7.9%) had visited a CAM practitioner during the last year at baseline. In YH2, one in eleven (8.7%, 95% CI 7.6-9.8%) had visited a CAM practitioner, an increase of 26% (1.8% points) in 4 years (p<0.001). A total of 13.9% of the adolescents had visited a CAM practitioner in either YH1 or in YH2. Of those who visited a CAM practitioner in YH2, 19.0% also visited in YH1.
Predictors for CAM visits
shows the variables in YH1 that had a statistically significant bivariate prediction with visits to CAM practitioners in YH2 four years later (). Among these, the highest prevalence for CAM visits in YH2 was among those who had visited a CAM practitioner, had self reported poor global health and had limitations due to physical or emotional health in YH1.
To identify the variables predicting visits to a CAM practitioner four years later, a multivariable logistic regression model was used (). It correctly predicted 91.4% of all cases. When controlling for all the other variables in the model, the only variables significantly predicting increased odds that an adolescent would become a CAM visitor (p<0.05), was having visited a CAM practitioner (Adjusted Odds Ratio – adjOR 3.3 (2.2–5.0) and having used one or more of a range of conventional medicines (adjOR 1.6, 1.1–2.3). Being a male predicted reduced odds of visiting a CAM practitioner in the future (adjOR -0.6, 0.4–0.8). Omitting those who had visited a CAM practitioner in YH1 from the multivariable logistic regression analysis did not change the main findings.
To further identify the underlying variables that predict future CAM visits in adolescents, those variables making up the “recent complaints”, “diseases”, and “conventional medicines” were tested in a bivariate analysis. Those with a statistical significant relationship with future CAM visits were then entered into a mulitivariable logistic regression model together with the significant variables from the larger model. In the final model (), the variables predicting increased visits to a CAM practitioner four years later (p<0.05) was visiting a CAM practitioner (adjOR 3.4), having had musculoskeletal pain (adjOR 1.5), had migraine (adjOR 2.3) or other disease that lasted more than three months (adjOR 2.1) or using asthma medicines (adjOR 1.8). Being a male predicted decreased visits (adjOR 0.6).
To look at the differences in prediction and association, the association with CAM visits in YH1 was investigated (). The bivariate analysis showed that in YH1 the highest prevalence for CAM visits was among those who had poor self reported global health, limitations due to physical health, and who had visited a psychologist.
The multivariable logistic regression showed that having good (adjOR 0.7) or fair (adjOR 0.5) self reported global health was associated with reduced odds of CAM visits in YH1 (). Limitations due to physical health (adjOR 2.3), having had one of a range of diseases (adjOR 1.6), and having visited a physician (adjOR 1.9) or a psychologist (adjOR 6.8) increased the odds. The final multivariable model (), showed that having visited a psychologist (adjOR 6.4) or a physician (adjOR 1.9), experiencing limitations due to physical health (adj OR 1.9) or having allergic conjunctivitis (adjOR 1.9), otitis (adjOR 1.6) or eczema (adjOR 1.4) was associated with increased odds of visits to a CAM practitioner.
Prediction vs. association
None of the variables in YH1 that significantly predicted CAM visits in YH2 were associated with CAM visits in YH1 in either the full model () or in the final model ().