Psychotropic Medication Use in Parents of Children Diagnosed With Cancer

First, using data from the nationwide Danish Cancer Registry (DCR), with registration of all cancers in adults since 1943 and children since 1975,¹⁰ we examined the risk of a first prescription of psychotropic medication in a cohort of parents of all children with cancer diagnosed between January 1, 1998, and December 31, 2014, compared with a matched cohort of parents of children who were cancer free (Supplemental Fig 2). However, no treatment and prognostic information is available in the DCR. Thus, to identify vulnerable subgroups of parents according to sociodemographic and clinical risk factors for prescription of psychotropic medication, we obtained information on a separate cohort of parents of children diagnosed with cancer between January 1, 2003, and December 31, 2015, according to the Childhood Cancer Registry¹¹ (Supplemental Fig 3). This allowed us to include information on cancer type in 3 main diagnostic groups (hematologic cancer, central nervous system [CNS] tumors, and other solid tumors), relapse (yes versus no), and death (yes versus no). Information on biological parents was obtained from the Central Population Registry.⁷ 

We obtained information on all redeemed prescriptions for psychotropic medication since the start of the Danish National Prescription Registry¹² in 1995, using the codes N06A for antidepressants, N05BA and N05CD for anxiolytics, and N05CF for hypnotics, according to the Anatomical Therapeutic Chemical classification system.

From the Civil Registration System,⁷ we obtained information on characteristics of the parents at the time of cancer diagnosis or index date for parents of children who were cancer free: sex, date of birth, migration, and death. Information on highest attained education (basic [up to 9 years], vocational school [10–12 years], higher education [>12 years], or unknown) was obtained from the Danish education registries,⁷ and annual family income (in quartiles of 5) was obtained from the income statistics registry¹³ both 1 year before cancer diagnosis or index date.

We identified 7821 biological parents (3853 fathers and 3968 mothers) >18 years of age in the Central Population Registry⁷ of all children and adolescents <20 years of age who resided in Denmark and were diagnosed with cancer according to the DCR¹⁰ between January 1, 1998, and December 31, 2014. Because previous psychiatric disorder is a strong predictor of distress, we excluded 1056 parents (14%) with a prescription for psychotropic medication in the years 1 to 3 before the date of diagnosis or index date. Parents who received prescriptions in the year before diagnosis were not excluded because it is possible that their children had symptoms of undiagnosed cancer that were causing distress in the parents. This resulted in a final study sample of 6744 parents of children with cancer. For each child with cancer, we randomly selected 10 children who were cancer free individually matched on year of birth from the Central Population Registry. The comparison cohort comprised parents of these children who, at the index date, were aged >18 years, had resided in Denmark for at least 3 years, and had no prescriptions for psychotropic medication in the 1 to 3 years before the index date. Again, parents who had prescriptions in the year before the index date were not excluded. We identified 65 747 parents (33 416 fathers and 32 331 mothers) and excluded 347 parents because of missing information on either family size, age, education, or income and 9304 (14%) parents who had had a prescription for a psychotropic medication 1 to 3 years before the date of diagnosis or index date, resulting in 65 747 parents of children who were cancer free (33 416 fathers and 32 331 mothers).

To study the impact of socioeconomic and disease characteristics, which is not available in the DCR, we identified 3364 parents of children <15 years of age who resided in Denmark and were diagnosed with cancer according to the Childhood Cancer Registry between January 1, 2003, to June 30, 2015, with no prescription for a psychotropic medication 1 to 3 years before the date of diagnosis. Excluding 74 parents with missing information resulted in a final cohort of 3290 parents of children with cancer.

Parents were followed from the date of diagnosis or the index date until first redeemed prescription, death, emigration, or June 30, 2015 (end of study), whichever came first. To illustrate the development of risk by time since diagnosis, we estimated the cause-specific hazard for a first prescription of psychotropic medication (combined and separately as antidepressants, anxiolytics, and hypnotics) according to the child’s cancer status nonparametrically using a kernel smoother. To investigate the absolute difference in first prescription of psychotropic medication (combined) between parents of children with cancer and parents of children who were cancer free, cumulative incidence proportion of 3 years after cancer diagnosis or index date was estimated.

We used Cox proportional hazard models to estimate whether having a child with cancer was associated with an increased risk for a first prescription for any psychotropic medication or for the 3 types of medication separately. The assumption of proportional hazards was assessed graphically by the observed standardized empirical score process. If proportionality did not hold, the variable was included in the baseline hazard when it was not the main exposure of interest. Otherwise, an interaction with time since diagnosis (0–1, 1–2, 2–3, and ≥3 years) was included in the model, leading to different exposure effects according to time period since diagnosis. Models were adjusted for age at entry (18–33, 34–38, 39–42, 43–47, and ≥48 years) and calendar period at entry (1998–2001, 2002–2005, 2006–2009, and 2010–2013).

Parents were followed from the date of cancer diagnosis in the child until first redeemed prescription, death, emigration, or June 30, 2015, whichever came first. We used Cox proportional hazard models to examine whether sociodemographic factors (education and income obtained 1 year before diagnosis) and disease characteristics (cancer type obtained at diagnosis, relapse, and death as a time-dependent covariate) were associated with an increased risk for a first prescription of any psychotropic medication or for each of the 3 types of medication (antidepressants, anxiolytics, and hypnotics) among parents of children diagnosed with cancer. The assumption of proportional hazards was assessed graphically by the observed standardized empirical score process. If proportionality did not hold, an interaction between time since diagnosis (0–1, 1–2, 2–3, and ≥3 years) was included in the model, leading to different exposure effects according to time period since diagnosis. Models were adjusted for calendar period at entry (1998–2001, 2002–2005, 2006–2009, and 2010–2013), and time since diagnoses was used as the underlying time scale. In addition, individual confounders (education, income, cancer type, and relapse) were included depending on the exposure considered.

For both analyses, risk estimates were presented as hazard ratios (HRs) with 95% confidence intervals (CIs). Because parents within a couple cannot be considered independent, the variance was estimated with adjustment for within-cluster correlation.¹⁴ All analyses were conducted in Stata version 14.¹⁵ The study was approved by the Danish Data protection Agency (2014-41-3405). Research based solely on registry data is exempt from ethical approval as well as from informed consent according to Danish law.¹⁶ 

We followed 72 165 parents until first prescription of any psychotropic medication (n = 15 233), death (n = 471), emigration (n = 1037), cancer in a comparison child (n = 52), or end of follow-up (n = 55 372). Parents were followed up for 0 to 16 years, with a median of 5.6 years. Only small differences were seen between parents of children with cancer (N = 6744) and parents of children who were cancer free (N = 65 421; Table 1). The cumulative incidence of risk of first prescription of psychotropic medication after 3 years for parents of children with cancer was 14% (95% CI, 13.6–15.3), compared with 10% for parents of children who were cancer free (Supplemental Fig 4).

During the first year after diagnosis, parents of children with cancer had an increased risk for a first prescription of any psychotropic medication (HR, 1.83; 95% CI, 1.66–2.01; Table 2). Significantly increased risks were seen in the first year for anxiolytics (HR, 2.99; 95% CI, 2.54–3.52) and hypnotics (HR, 2.64; 95% CI, 2.26–3.09) but not for antidepressants (HR, 1.02; 95% CI, 0.87–1.20; Table 2, Fig 1).

Examining subgroups of vulnerable parents of children with cancer, we followed 3290 parents of children with cancer until first prescription (n = 952), death (n = 9), emigration (n = 27), or end of follow-up (n = 2728). Parents were followed up for 0 to 12.4 years, with a median of 4.4 years. A total of 656 parents (20%) received a first prescription of psychotropic medication during the study period. The only differences across cancer types were seen for age (a larger proportion of younger parents of children with solid tumors than of parents of children with other cancer types; Table 3). Of the 656 parents who received a prescription, 89 parents experienced a first prescription after relapse in their child, and 94 after death of their child (Table 4).

No statistically significant difference was seen for risk of psychotropic medication according to cancer type (Table 4). For the time-dependent covariates of relapse and death, we did observe a difference in effect according to time since diagnosis. Parents who experienced relapse in the child had an increased risk for first prescription of hypnotics in the second year (year 1–2: HR, 4.04; 95% CI, 1.93–8.44) after diagnosis. Parents who experienced the death of a child had an increased risk of a first prescription of anxiolytics in the first year (year 0–1: HR, 4.55; 95% CI, 1.57–13.17) and third year (year 2–3: HR, 4.13; 95% CI, 1.15–14.76) after diagnosis and of hypnotics in the first year (year 0–1: HR, 6.91; 95% CI, 3.50–13.66) and second year (year 1–2: HR, 3.52; 95% CI, 1.63–7.58). Parents with basic education (up to 9 years) had a significantly increased risk of a first prescription of antidepressants (HR, 3.02; 95% CI, 2.23–4.10) and anxiolytics (HR, 2.65; 95% CI, 1.62–4.34) but not hypnotics. Finally, parents with the lowest income had an increased risk of psychotropic medication, especially of hypnotics (HR, 1.81; 95% CI, 1.01–3.24) and antidepressants (HR, 1.46; 95% CI, 1.02–2.09).

To our knowledge, this study is the first registry-based study to document that parents of children with cancer are at increased risk of having psychotropic medication prescribed compared with parents of children who are cancer free. We found an increased risk of use of psychotropic medication, especially for first prescriptions of hypnotics and anxiolytics, potentially indicating psychological reactions of anxiety and sleep disturbances. The primary indication for prescribing antidepressants is moderate-to-severe depression,⁸ whereas the primary indications for anxiolytics and hypnotics include sleep disorders, severe anxiety, and acute stress–related anxiety.⁹ 

Only 9 of 190 previous studies on the risk of psychological distress in parents of children with cancer had a moderate sample (N > 100) and a control group of parents of children who were cancer free, and only 2 of these studies included follow-up of at least 5 years. In the first study, which included 474 parents of childhood cancer survivors and 176 parents of children who were cancer free, the authors found elevated anxiety and depression symptoms 2.5 years after the diagnosis, and up to 12% of these parents had posttraumatic stress symptoms 5 years after diagnosis.¹⁷ In contrast, the authors of the second study, which included 305 parents of children with cancer and 231 parents of children who were cancer free, did not find elevated posttraumatic stress in parents of children with cancer 5 years after diagnosis.¹⁸ Our results are supportive of previous studies suggesting elevated psychological distress, anxiety, and depression in parents of children with cancer, especially in the first 2 years after diagnosis.¹⁹ The mechanisms underlying the results could imply that the parents may experience intrusive disease-related thoughts, ultimately with increased avoidance and worry that disrupts sleep,²⁰ potentially resulting in symptoms of acute or posttraumatic stress symptoms,²¹ depression, anxiety, and sleep disturbances⁶ that are clinically significant enough to warrant intervention. We are not able to determine the mechanisms, but one may speculate if medication management of anxiety and sleep disturbances may have prevented further development of depressive symptoms. The pathogenesis and how symptoms of depression, anxiety, and insomnia develop in co-occurrence in the stress-reaction management phase needs further examination.¹⁷ 

No clear consensus has been reached on who are the most vulnerable parents of children with cancer. Thirty-two studies investigated distress among parents of children with cancer (N >100), and many were limited by their use of self-report measures of psychological symptoms^22,–24 and by potential recall and selection bias because, often, only the most resourceful parents are able to participate.²⁵ The studies suggest that parents of children with cancer are especially vulnerable if the child dies²⁶ and if the family only has basic education and a single mother.²³ Among parents of children with cancer, we found that parents who experienced relapse or death of a child as well as families with basic-level education and low income had the greatest risk of having psychotropic medications prescribed. We had expected differences across cancer types (eg, because of the variation in length of cancer treatment,^1,27 in the care burden,²⁸ and in life-expectancy¹). The lack of variation may suggest that the distress may be less influenced by the care burden than expected; however, further studies should explore this. As expected, we observed an increased risk of hypnotics and anxiolytics in parents of children who relapsed or died. This is in line with a recent study that revealed an increased risk of prescription of antidepressants and anxiolytics in parents 1 year after the death of their child²⁶ as well as several investigations demonstrating the long-term psychological challenges that parents face after the death of their child to cancer.²⁹ Finally, in line with a previous study,²³ our findings suggest that parents with higher education and income may have a lower risk of psychotropic medication because of a higher health literacy, enabling them to establish social support from friends and family, attend counseling, and learn how to cope with their child’s cancer and the changes in their everyday life.

The strengths of this study include identification of all parents of children with cancer in the comprehensive Danish registries and a matched population-based comparison cohort of parents who had not been exposed to cancer in their children. This also gave us a unique opportunity to examine psychotropic medication use established independently of the hypothesis under study. In Denmark, the health care system is free of charge, and the cost of medication is subsidized, making medication affordable to all residents. In addition, mainly general practitioners evaluate the need for prescription of psychotropic medication using validated screening instruments for depression. Together, these conditions minimize selection and reporting biases.

The main limitation is the use of prescriptions for psychotropic medication as an indicator of psychological distress because we may only capture severe psychiatric morbidity and not unrecognized, mild distress, which may be managed through psychotherapy or remain unmanaged. Also, parents of children with cancer may not have the time nor energy during their child’s cancer treatment to pursue long psychotherapeutic treatment, whereas anxiolytics and hypnotics may offer fast symptom relief.⁹ For bereaved parents, this may additionally reflect challenges with confronting the pain of their loss through talk therapy. Thus, we cannot exclude bias related to treatment choices, and we expect to underestimate the number of parents in need of support because we excluded previous use of psychotropic medication. Also, we cannot exclude potential surveillance bias because the parents of children with cancer might receive more attention from pediatric teams perhaps urging them to attend their general practitioner. Lastly, because prescriptions for psychotropic medications have been registered only since 1995, we were unable to take into account lifetime previous use of psychotropic medication, only use within the last 3 years.

Parents of children with cancer are at increased risk of a first prescription of anxiolytics and hypnotics, which may point to increased risk of anxiety symptoms and sleep dysfunction severe enough to seek medical treatment, compared with parents of children who are cancer free. Parents of children with a poor prognosis as well as parents with basic education or low income are especially at risk. Further research is needed to identify parents with unmet need for support and on developing family interventions. Efforts should be made to ensure that medical teams are adequately educated to address stress responses in the whole family.

CI

confidence interval

CNS

central nervous system

DCR

Danish Cancer Registry

HR

hazard ratio