In this large prospective cohort study, we found that parental risk constellations were differently associated with receiving medications for sleep problems in individuals depending on whether they also received medications for anxiety/depression during the study period. Offspring in the four risky constellations were not more likely than offspring in the overall low-risk group to receive sleep medications when they did not also receive anxiety/depression medication. Two risk constellations were associated with increased risk of receiving sleep medication when the offspring also received anxiety/depression prescription drugs: the first is families with frequent drinking in both parents and mental health symptoms in fathers (LP5), and the second is binge drinking, mental health symptoms and low education in both parents (LP2). Offspring from families with low education in both parents were more likely to receive only prescription drugs for anxiety/depression and not for sleep problems.
This was the first study to use a cluster approach to study the association between parental drinking, mental health and education, and subsequent offspring prescription drug use for sleep problems, which enabled us to disentangle the role of parental risk constellations in predicting prescription drug use for sleep problems with and without preceding, co-occurring or following anxiety/depression prescription drug use. We found that parental risk constellations were not associated with the likelihood of being dispensed of sleep medication when this prescription was not preceded or followed by medication for anxiety/depression. This is in contrast to findings in previous studies that showed an effect of parental drinking, parental socioeconomic status, parental mental health on other offspring sleep outcomes19–21, 27. For instance, two US studies showed an association between parental drinking on preadolescents sleep problems, measured by actigraphs, and that the effect was stronger in families with low socioeconomic status19,20. One of these studies controlled for early indicators of anxiety and depression at baseline20, the other did not19, and none of the studies addressed whether sleep problems at follow-up occurred in conjunction with anxiety/depression. Our findings contrast those of a recent study from the US, which examined the association between parental depression and sleep problems in preadolescent offspring27. Offspring whose parents had depression were at increased risk of sleep problems compared to offspring of non-depressed parents27. Further, parent-reported lower socioeconomic status was also associated with increased risk of offspring sleep problems27. The study controlled for offspring anxiety and depression symptoms, and to avoid confounds, sleep items were removed from the anxiety and depression measures27. Our findings also contrast those of a Canadian study which showed that adolescents low subjective socioeconomic position, were associated with lower self-reported sleep quality21, also after controlling for adolescent anxiety and depression.
When considering our findings that two parental risk constellations were associated with increased risk of being dispensed sleep medication, when also being dispensed anxiety/depression prescription drugs, this may be interpreted as supporting the abovementioned studies, which showed an association between parental drinking, mental health, and SES, on offspring sleep problems19–21, 27. However, most adolescents with depression, also have sleep problems, and those who do, tend to have more severe depression34,43. A plausible explanation, therefore, is that what the parental risk constellations predict are the more severe cases of anxiety/depression. The finding that only one parental risk constellation was associated with receiving anxiety/depression drugs only, adds support to the interpretation that parental risk factors are primarily associated with the most severe cases of anxiety/depression, which typically occur in conjunction with sleep problems. This sheds light on our previous work, where we found an association between some parental risk constellations and subsequent treatment for anxiety/depression in offspring18. Seen in the light of the finding from the current report, the associations observed in our previous work were likely largely driven by more severe cases, with both sleep problems and anxiety/depression.
Our study extends and adds nuance to the research on parental risk factors as predictors of offspring sleep problems. Taken together, our findings suggest that parental risk constellations of drinking patterns, mental health, and education, are not associated with offspring receiving prescription drug treatment for sleep problems, in cases where they do not also receive medications for anxiety and/or depression. The findings underscore the importance that studies primarily examining sleep problems, control for anxiety and depression to tease out the associations pertaining to sleep. As our results show, there may otherwise be a risk that associations that stem from anxiety/depression be misattributed to sleep. This problem does not extend to anxiety/depression research, where sleep symptoms are included among the symptoms of anxiety/depression.
Methodological considerations
Major strengths of the study include the large sample size, the prospective design, practically no attrition, the use of person-centered approaches that identified parental risk profiles based on information obtained from both mothers and fathers. The study has several limitations, that should be taken into consideration when interpreting the results. Our sample consisted of two-parent families where mothers, fathers and adolescents participated in HUNT. While this limits generalizability to other family constellations, an advantage with the approach is that it steers clear of the single data source limitations and biases that previous studies are fraught with, e.g. information about parental characteristics is provided by the adolescents only, or by the adolescent and just one parent44. We used self-report data at baseline, this may have resulted in underreporting, inaccurate recall and selective reporting 45. Our registry-based outcomes are conservative as they only include individuals that have self-selected into seeking help for, been diagnosed with a disorder, and received prescription drugs for treating these disorders. We used NorPD to capture the outcomes of interest instead of healthcare utilization databases. The rationale is that in consultations with general practitioners (GPs), patients often raise several health issues, and while some GPs records all the issues in the primary healthcare database, many record only the primary reason for the consultations, thus if sleep problems, or anxiety/depression were not the primary reason, these problems are not captured in the database. Thus, for looking at anything but primary diagnoses/main reason for contact, the primary healthcare database is less reliable than the NorPD, which captures all prescriptions dispensed. For our purpose, NorPD is therefore more reliable, because all who receive prescription drugs are issued prescriptions after consultation and diagnosis by a medical doctor. Consequently, this approach does not capture the population prevalence of sleep problems and/or anxiety/depression, which is likely substantially higher, only cases that are both diagnosed and receive treatment with prescription drugs.
Sleep problems, as captured in NorPD, may include cases that receive prescription medication for sleep problems due to transient situational factors, such as stress, bereavement, relationship problems. It seems plausible that parental factors are of less importance for these. However, given the nature of the of the data, we have no way of identifying whether the prescription drug use was due to transient situational factors or not. Further, while we know that patients were dispensed the prescription medications, we do not know whether they used them. It should also be noted that some of the prescription drugs used to treat anxiety/depression are also sleep inducing; thus, some patients that experienced both depression/anxiety and sleep problems, who received these prescription drugs, may have had both of their problems solved with one medication. Finally, we use the long follow-up time as a rationale for separating between sleep problems with and without preceding, co-occurring or following anxiety/depression. However, nine years is not lifetime, and we can only generalize to the years for which we have data. It is possible that study participants received prescription drugs that contradicts this assumption after follow-up ended in 2016, and for the oldest cohort, during the years preceding registry follow-up start.