Psychiatric comorbidity among terminally ill patients in general practice in the Netherlands: a comparison between patients with cancer and heart failure

Characteristics of the patients

Due to the change of computer system during the study period data could not be extracted in two of the six GPRNU practices. In the four remaining GP practices with 29 GPs and 44 030 patients, 162 terminal cancer cases and 116 terminal heart failure cases were identified in the study period. After excluding cases with inaccurate coding of diagnoses and incomplete data (n = 78), a total of 191 patients were eligible for analysis (111 with cancer and 80 with heart failure). There was no significant difference in age and sex between those included or excluded from the study. For those included in the study, there were more female patients in the terminal heart failure group (60.0%) as compared to the terminal cancer group (46.8%) but the difference was not statistically significant (OR = 1.702, 95% CI = 0.951 to 3.047). The average age at death for patients with terminal cancer was 70.8 years (standard deviation [SD] = 12.8), and for patients with heart failure was 85.75 years (SD = 9.24) (Table 1). The most common types of cancer were: gastrointestinal (rectal, colon and pancreas, stomach) (43.2%) followed by pulmonary (23.4%), haematological (9.9%), breast (9.0%), genital-urinary (5.4%), and others/unknown (9.1%).

Somatic comorbidity

The prevalence of somatic comorbidity in the complete study group was 43.5%. Somatic comorbidities were significantly more common in patients with terminal heart failure (65.0%) compared to patients with terminal cancer (27.9%) (OR = 4.793, 95% CI = 2.581 to 8.899). Diabetes mellitus (20.9%) was the most common comorbidity followed by chronic obstructive pulmonary disease (COPD) (13.1%) and ischaemic heart disease (IHD; 11.5%). Chronic disease comorbidity was also more common among patients with terminal heart failure than patients with cancer (diabetes mellitus, 32.5% versus 12.6%, OR = 3.336, 95% CI = 1.608 to 6.922; COPD, 26.2% versus 3.6%, OR = 9.521, 95% CI = 3.121 to 29.050 and IHD, 21.2% versus 4.5%, OR = 5.721, 95% CI = 2.012 to 16.262).

Psychiatric comorbidity and psychotropic prescription

The overall prevalence of psychiatric diagnoses in the study group was 13.6% (Table 1). The most prevalent psychiatric comorbidities were other organic psychosis (6.3%) and depression (4.7%). There was no statistical significance in the prevalence of psychiatric diagnoses between patients with cancer and heart failure.

Based on the prescription data, half of the terminally ill patients were prescribed at least one type of psychotropic drugs (Table 2). Benzodiazepines were the most common followed by antipsychotics and antidepressants.

There was no significant difference in the prevalence of psychotropic drugs prescribed to the patients with cancer or heart failure, except for methylphenidate that was only used by patients with cancer. Some patients in the terminal phase used more than one drug; 9.4% used two psychotropic drugs and 5.2% used three types. There was no significant difference between patients with cancer and heart failure in the frequency of multiple psychotropic drug prescription.

Association of psychotropic drugs prescription

Logistic regression analysis showed that the presence of a psychiatric diagnosis was the only factor significantly associated with psychotropic drug prescription in the terminal phase of life, after adjustment for sex, age, presence of physical comorbidities and type of terminal illness (Table 3).

Summary

In patients with terminal cancer and heart failure the study found no difference in the prevalence of psychiatric diagnoses or in the prescription of psychotropic drugs. Half of the patients were prescribed at least one type of psychotropic drug, mostly benzodiazepines and antipsychotics. Although psychotropic drug prescription was associated with psychiatric comorbidity, only 13.6% of the patients actually had a psychiatric diagnosis, mostly psychosis and depression.

Strength and limitations

The study used the patient data from the GPRNU database, which is a representative sample of the Dutch primary care population.29 The patient sample studied is most probably an adequate representation of the patients with cancer and heart failure in the terminal phase of life.

Patient data came from GPs from four different primary care centres only, which may have hampered the representativeness of the results. The fact that they work closely together may have affected the registration routine of these GPs for psychiatric disorders.

In general, routine registration data do have limitations when used for research. In the review of the records it was noticed that GPs often identified different psychiatric complaints within each cancer episode, but did not use separate ICPC codes for each of these symptoms. In addition, the ICPC coding structure does not allow all complaints to be registered specifically.

In addition, the focus of palliative care in general practice is primarily on maintaining the best quality of life and not on accurate diagnosis and registration. GPs are probably more concerned about relieving the pain and anguish at the end of life, including the use of psychotropic prescriptions, than to diagnose difficult -to-distinguish psychiatric symptoms and classify them.

In this study, patients with cancer and heart failure were compared, but patients in these two groups differed in many respects. Patients with heart failure were on average 10–15 years older than patients with cancer when they died. Also, unexpectedly, there was no difference in sex in the heart failure group. Age and sex may have biased the presence of psychiatric symptoms. Furthermore, the cancer group was heterogeneous, taking different cancer types with different disease progression together. As distress levels of patients in different age and disease groups differ a priori, this may have affected the comparison between the two groups.

Comparison with existing literature

This study found that a psychiatric diagnosis among terminally ill patients in a primary care setting, either patients with cancer or heart failure, was relatively infrequent. The prevalence of depression was 5.4% for patients with cancer. This is in line with an earlier report of Reeve et al, who found a prevalence of 4.1% in patients with cancer.30 However, in a previous meta-analysis the study found a prevalence of 10.8% depression in patients with cancer.31Differences in patient selection between studies may play a role in this discrepancy. Particularly, more severely depressed patients may not stay in the GP setting.

In this study the prevalence of depression in patients with heart failure was 3.8%. An earlier meta-analysis showed that the prevalence of depression in patients with heart failure was 21%.26 The overall prevalence of psychiatric comorbidity in patients with terminal cancer in the study was 13.6%, which is strikingly low compared to other studies which reported prevalence up to 50% in patients with cancer.15,–,17 The main reason for this under-reporting of psychiatric comorbidity by GPs is probably the difficult diagnostic process in terminally ill patients. The various somatic complaints in patients with cancer, resulting from both disease progression and treatment side effects, overlap with and may even mask psychiatric symptoms.18,19,20,32 Although in the past, there were many studies on diagnostic validity or case-finding tools in palliative and cancer patients, their reliability and validity were never demonstrated.32,33 Therefore GPs who suspect psychiatric symptoms in terminally ill patients, lack evidence-based criteria to adequately diagnose psychiatric disease. In addition, depression may be underdiagnosed because a depressed mood is considered as a physiological phenomenon in the terminal phase of life.

The psychotropic drug prescription rate of 50.3% seems relatively high and in sharp contrast with the low prevalence of psychiatric diagnoses. This contrast is also visible for separate drug subclasses. The antidepressant rate was 14.1%, while only 4.7% of patients were diagnosed with a depression. The same dissociation was demonstrated for benzodiazepines and antipsychotics. This may be explained by the fact that psychotropic drugs are also prescribed for many non-psychiatric complaints.34 For instance, antidepressants are used for fatigue, insomnia, and pain, while antipsychotics are used in case of vomiting, agitation, and confusion in terminal phase patients.35 The finding is in line with the generalist approach adopted by GPs; they recognise the needs of patients and treat the symptoms accordingly. In the terminal stage of disease with complex psychiatric comorbidity an individually-tailored model prevails and a disease oriented model according to diagnostic criteria is abandoned.36,–,38 This also suggests that psychotropic drug prescription is not the optimal proxy indicator of psychiatric disease in patients with terminal stage cancer.

Regarding the main objective of this study, it was demonstrated that psychiatric symptoms are frequent both in patients with terminal cancer and heart failure. In this study neither the prevalence of psychiatric comorbidity nor the use of psychotropic drugs differed between the two patient groups, despite the difference in mean age. In both groups it was found that there was a similar discrepancy between the frequency of psychiatric diagnoses and that of psychotropic drug prescription. Patients from different age classes and distinctive illnesses used the same amount of psychotropic drugs in the terminal stage before dying. A previous study showed that shortly before dying the chances for prescription of a new psychotropic drug increase.39 It can now be added that the increased incidence of psychiatric comorbidity and the use of psychotropic drugs are related to the terminal disease phase and are not disease specific.

Implications for practice and research

Psychological symptoms and psychiatric comorbidity are common in terminally ill patients, as indicated by the high use of psychotropic drugs. The study demonstrated that this is primarily related to the terminal disease stage and not to the type of illness. This warrants more attention for psychiatric morbidity, not only in patients with cancer, but in the terminal phase of all chronic diseases. The findings in this study should be replicated in a prospective cohort study to examine the experience of psychological suffering in patients at terminal stage. In addition, intervention studies may address the cost-effectiveness of psychotropic drug use in the terminal stage. Alternative therapeutic intervention with rapid responses like psychostimulants or less adverse effects like mindfulness-based psychotherapy need to be explored.

The rate of formally registered psychiatric diagnoses in the terminally ill patients reported by the GP was relatively low. In contrast, the frequency of psychotropic drugs prescription was much higher. The study observed no differences in psychiatric diagnoses and prescriptions between patients with cancer and heart failure, regardless of the age differences. These findings suggest that psychiatric disorders in terminal patients in the general GP practice are primarily related to the terminal disease phase, and not disease specific.